pmc.ncbi.nlm.nih.gov

Oral evening primrose oil and borage oil for eczema

  • ️Mon Mar 04 2024

Abstract

Background

Eczema is a chronic inflammatory skin condition, which usually develops in early childhood. Many children outgrow this disorder as they reach secondary school age, and although It may improve with age, there is no cure. Constant itch makes life uncomfortable for those with this condition, no matter what age they are, so it may have a significant effect on a person's quality of life. Its prevalence seems to be increasing as populations move from rural locations to cities. Some people, who do not see an adequate improvement or fear side‐effects of conventional medical products, try complementary alternatives to conventional treatment. This is a review of evening primrose oil (EPO) and borage oil (BO) taken orally (by mouth); these have been thought to be beneficial because of their gamma‐linolenic acid content.

Objectives

To assess the effects of oral evening primrose oil or borage oil for treating the symptoms of atopic eczema.

Search methods

We searched the following databases up to August 2012: Cochrane Skin Group Specialised Register, CENTRAL in The Cochrane Library, MEDLINE (from 1946), EMBASE (from 1974), AMED (from 1985), and LILACS (from 1982). We also searched online trials registers and checked the bibliographies of included studies for further references to relevant trials. We corresponded with trial investigators and pharmaceutical companies to try to identify unpublished and ongoing trials. We performed a separate search for adverse effects of evening primrose oil and borage oil in November 2011.

Selection criteria

All randomised controlled, parallel, or cross‐over trials investigating oral intake of evening primrose oil or borage oil for eczema.

Data collection and analysis

Two review authors independently applied eligibility criteria, assessed risk of bias, and extracted data. We pooled dichotomous outcomes using risk ratios (RR), and continuous outcomes using the mean difference (MD). Where possible, we pooled study results using random‐effects meta‐analysis and tested statistical heterogeneity using both the Chi² test and the I² statistic test. We presented results using forest plots with 95% confidence intervals (CI).

Main results

A total of 27 studies (1596 participants) met the inclusion criteria: 19 studies assessed evening primrose oil, and 8 studies assessed borage oil. For EPO, a meta‐analysis of results from 7 studies showed that EPO failed to significantly increase improvement in global eczema symptoms as reported by participants on a visual analogue scale of 0 to 100 (MD ‐2.22, 95% CI ‐10.48 to 6.04, 176 participants, 7 trials) and a visual analogue scale of 0 to 100 for medical doctors (MD ‐3.26, 95% CI ‐6.96 to 0.45, 289 participants, 8 trials) compared to the placebo group.

Treatment with BO also failed to significantly improve global eczema symptoms compared to placebo treatment as reported by both participants and medical doctors, although we could not conduct a meta‐analysis as studies reported results in different ways. With regard to the risk of bias, the majority of studies were of low risk of bias; we judged 67% of the included studies as having low risk of bias for random sequence generation; 44%, for allocation concealment; 59%, for blinding; and 37%, for other biases.

Authors' conclusions

Implications for practice  

Oral borage oil and evening primrose oil lack effect on eczema; improvement was similar to respective placebos used in trials. Oral BO and EPO are not effective treatments for eczema.

In these studies, along with the placebos, EPO and BO have the same, fairly common, mild, transient adverse effects, which are mainly gastrointestinal.

The short‐term studies included here do not examine possible adverse effects of long‐term use of EPO or BO. A case report warned that if EPO is taken for a prolonged period of time (more than one year), there is a potential risk of inflammation, thrombosis, and immunosuppression; another study found that EPO may increase bleeding for people on Coumadin® (warfarin) medication.

Implications for research

Noting that the confidence intervals between active and placebo treatment are narrow, to exclude the possibility of any clinically useful difference, we concluded that further studies on EPO or BO for eczema would be hard to justify.

This review does not provide information about long‐term use of these products.

Keywords: Adult; Child; Humans; Administration, Oral; Dermatitis, Atopic; Dermatitis, Atopic/drug therapy; Dermatologic Agents; Dermatologic Agents/therapeutic use; Eczema; Eczema/drug therapy; gamma-Linolenic Acid; gamma-Linolenic Acid/administration & dosage; Linoleic Acids; Linoleic Acids/administration & dosage; Oenothera biennis; Plant Oils; Plant Oils/administration & dosage; Randomized Controlled Trials as Topic

Plain language summary

Oral evening primrose oil and borage oil for eczema

Eczema is an itchy and red skin condition, which may affect 20% of people world wide at some time in their life. Though it may improve with age, there is no cure. Many children outgrow this disorder as they reach secondary school age. Constant itch makes life uncomfortable for those with this condition, no matter what age they are.

Conventional medical treatments make life better for people; however, some people who do not see an adequate improvement in their eczema or fear side‐effects of conventional medical products, turn to complementary alternatives to conventional medical treatment. This review is of two such products: evening primrose oil (EPO) and borage oil (BO) taken orally (by mouth), which have been thought to have benefits for eczema.

We included 27 studies, with 1596 adults and children from 12 countries. Of these, 19 studies compared EPO with a placebo (dummy) treatment, and 8 used BO compared with placebo. We looked for evidence of overall improvement in eczema and in quality of life. All 27 studies evaluated overall improvement of eczema, but only 2 studies of EPO measured improvement in quality of life. There was no statistically significant advantage demonstrated for either EPO or BO compared to placebo. In summary, we did not find evidence that eczema improved by taking these products any more than it did by taking placebo.

There was some evidence of mild and temporary side‐effects for participants with either product or placebo, which were mainly mild, and included temporary headache and upset stomach or diarrhoea. With EPO there is an anticoagulant (blood‐thinning) effect when taking these products. There is a warning with the blood thinner warfarin (Coumadin®) that taking EPO can increase bleeding. One report warns that if EPO is taken for a prolonged period of time (more than one year), there is a potential risk of inflammation, thrombosis, and immunosuppression due to slow accumulation of EPO in the tissues. Another reports a single case in which EPO was thought to have produced harms. We found no clinical evidence of such harm in these short‐term trials.

This systematic review found no evidence that either BO or EPO are effective in treatment of eczema. Both of these products and the placebos used in the studies had similar mild, temporary side‐effects, which were mainly gastrointestinal.

Background

Description of the condition

Eczema is a chronic inflammatory skin condition, which usually develops in early childhood, often in the first three months of life. There is often a fluctuating course, but most cases clear up or improve by adulthood, the majority within the first five years of life.

Eczema can appear on all areas of the skin. In infants it often affects the face, forearms, knees, and ankles, and in older children mainly the flexures of the elbows and knees. Adults who suffer from eczema are often affected on their face and flexural areas of the neck, elbows, knees, and ankles. There is a frequent association of eczema with other atopic disorders, such as asthma and hay fever (Williams 2000), and environmental factors play an important role (Wang 2007).

Definition

After international discussions regarding the word 'atopic', a task force has suggested the term 'atopic eczema/dermatitis syndrome' (AEDS) to summarise all forms of eczema that have previously been called 'prurigo Besnier' or 'atopic dermatitis' (Johansson 2001). We used the term 'eczema' throughout this review.

The diagnosis of eczema is made clinically. In 1980, Hanifin and Rajka published a list of all the possible signs and symptoms of eczema, divided into major and minor features (Hanifin 1980). They include pruritus or itch, a chronic relapsing course, typical morphology and distribution of lesions, and familial or personal history of other atopic disorders.

Sampson defined "three major criteria for atopic dermatitis: family history of atopic disease; typical facial, or extensor, eczematous or lichenified dermatitis; and evidence of pruritus (itching)" (Sampson 1992).

The most recent form of diagnosis was developed by a UK working group (Williams 1994). It is based on a range of sensitive and specific symptoms and signs and depends on a person having an itchy skin condition, plus three or more of the following:

  • history of involvement of skin creases;

  • personal history of asthma or allergic rhinitis, or history of atopic disease in a first‐degree relative of children;

  • history of general dry skin in past year;

  • visible flexural eczema or eczema on cheeks, forehead, or outer limbs; or

  • onset before two years of age.

A variety of criteria have been proposed that allow a diagnosis to be made (Bos 2010; Tofte 2001). In this review, we use a clinician's diagnosis of eczema (Nankervis 2010).

Causes

The causes of eczema remain unclear, but it is likely to be of a multi‐factorial nature. Genetic (Friedmann 2002), environmental (McNally 2001), and socioeconomic (McNally 2001) factors may all be important. Twin studies have shown a concordance rate of 72% in monozygotic twins and 23% in dizygotic twins (Schultz‐Larsen 1993). Eczema is possibly "inherited more often from the mother than from the father and tends to run true to type within each family" (Weller 2009). It is more prevalent in higher social classes in industrialised societies (McNally 1998). Low birth order and small family size also appear to be risk factors for atopic disorders. "A cross‐sectional parental postal survey in the UK showed statistically significant associations between eczema symptoms and dampness in the home, the use of a radiator to heat the child's bedroom and the use of synthetic pillows. Frequent vacuuming in the home was associated with a decreased prevalence of eczema" (McNally 2001). The perception that eczema is caused by house dust mites and food allergies often dominates the lives of people it affects (Barnetson 2002).

"A model for the immune pathogenesis proposes a predominant Th2 cytokine milieu in the initiating stages of acute atopic dermatitis lesions, and a mixed Th1 and Th2 pattern in chronic lesions" (Pastar 2005).

A defective skin barrier function may be of major importance (Gfesser 1997), although other studies have contradicted the finding of a disturbed skin barrier (Matsumoto 2000). Molecules, called structural lipids, which are located between stratum corneum cells, are also thought to "play a considerable role in the water‐holding potential of the stratum corneum" (Imokawa 1991). "One of the major structural lipids is ceramide. Natural ceramides are too expensive at present to be made commercially available; however, several pseudo‐ceramides have been synthesised and shown to be clinically effective in preventing and improving dry skin" (Draelos 2008). Epidermal barrier dysfunction has been suggested as a primary event in the development of this disease (Cork 2009).

Breastfeeding had been believed to help prevent or delay the development of atopic conditions such as eczema (van Odijk 2003). One systematic review found "no strong evidence of a protective effect of exclusive breastfeeding for at least 3 months against eczema, even in those with a positive family history of atopy" (Batchelor 2010). A recent systematic review found no conclusive evidence that breastfeeding lessens established eczema in infants (Kramer 2012).

A deficiency in essential fatty acids of the skin has been suspected as having a role in the cause of eczema (Horrobin 2000; Melnik 1992). It has been suggested that eczema could be caused by a (possibly inherited) defect or deficiency of the enzyme delta‐6‐desaturase and possibly also the enzyme delta‐5‐desaturase (Manku 1982; Melnik 1992). These enzymes are responsible for converting linoleic acid to gamma‐linolenic acid, a substance with anti‐inflammatory properties (Diezel 1993). Reduced conversion may result in a relative skin deficiency of gamma‐linolenic acid, alpha‐linolenic acid, and stearidonic acid.

Epidemiology

Eczema is a common problem "affecting 10 to 20% of school age children in developed countries" (Johnson 1984; Rajka 1989) and 1% to 3% of the general adult population (Larsen 2002). The prevalence of the disease has increased by two‐ to three‐fold in the developed world during the last three decades (Saito 2005). "Its prevalence in the United States alone has nearly tripled in the past thirty to forty years" (Schram 2010). "However, it remains much lower in agricultural regions, such as China, Eastern Europe and rural Africa" (Leung 2003). The reasons for the rising prevalence are not clear; however, prevalence seems to be increasing as populations become increasingly urbanised (Leung 2008).

Impact

At a personal level, eczema has a significant effect on life quality and affects sleep (Reid 1995; Terreehorst 2002), school (Lewis Jones 1995), employment, relationships, and family life (Lewis Jones 1995; Su 1997).

For people of all ages, there is the physical impact of eczema ‐ the constant itch and the unbearable need to scratch, even though this exacerbates the condition. Chronic scratching causes the skin to take on a leathery texture because the skin has thickened (lichenification). The skin becomes hot, tight, and sore, and it is often worse at night, resulting in sleepless nights for those with eczema and their families.

The skin can become cracked, weeping, or even infected, making it painful to wash, swim, or perform everyday tasks. Psychologically, the red, damaged skin is embarrassing and difficult to hide. Promotion at work may be blocked for people who do not "look good". Children are taunted at school. They often have low self‐esteem. Their relationships can be difficult to initiate and sustain.

The extra time needed for bathing and skin‐care makes dealing with work, studies, and social life more stressful. Cracked skin makes routine everyday tasks, such as housework, gardening, childcare, and food preparation, difficult. Careers such as catering, nursing, hairdressing, horticulture, and many manual trades are unsuitable.

There are additional financial costs incurred in the purchase of items designed for sensitive skin, such as special clothing, bedding, detergents, and make‐up, etc. In 1996, a British study reported that the annual costs to the National Health Service (NHS) exceeded £125 million, and personal costs, including lost salary, were around £300 million (Herd 1996). Economic data in the USA have shown that the cost of illness from atopic dermatitis and eczema ranged from $0.9 billion to $3.8 billion per year in people younger than 65 years (Ellis 2002).

Description of the intervention

There are at least 47 different types of interventions for eczema, which can be classified according to 10 major groupings (Hoare 2000). The management of eczema requires adequate time allocation for doctors and nurses to explain to the person with it how to deal with the disease. Management involves a discussion of the avoidance of soaps and detergents, extremes of temperatures, and clothing that may irritate, such as wool. In one trial "patients attending the eczema workshop had a greater improvement in eczema severity than patients attending a dermatologist‐led clinic", supporting collaborative models of service provision (Moore 2009).

The mainstays of the management of eczema are the use of emollients (creams, ointments, and bath additives). Recently, manufactured "pseudo‐ceramides" (Draelos 2011) have joined topical corticosteroids and immune modulators (Thestrup‐Pedersen 2003; Williams 2008) as treatments for eczema. There is no evidence that probiotics have an effect on established eczema (Williams 2008).

Severe eczema, as well as nummular or hand forms, are often resistant to topical treatment and a variety of oral treatments are used. Sedating antihistamines are widely used to reduce itching and sleep disturbance. Other systemic treatments (Plötz 2010) have included cyclosporin, azathioprine (Poyner 2001), methotrexate (Roberts 2010), alitretinoin (Kubica 2011), adalimumab (Beswick 2011), cyclosporin A, or mycophenolate mofetil (Belch 2000). Phototherapy and photochemotherapy have also been used. When first‐ and second‐line therapies, i.e. topical steroids and ultraviolet light treatments, prove inadequate, Schmit suggested that until there are more adequate randomised controlled trials comparing systemic treatments for severe eczema (Schmitt 2007), it might be better to consider short‐term systemic steroids and avoid cyclosporin A because of its renal toxicity. He suggested: "Lefluonomide might be such a therapeutic alternative, but much larger trials are required."

In the USA in 2007, 38% of adults and 11% of children used some form of complementary or alternative medicine (CAM) (NHIS 2007). Some individuals use complementary therapies, including evening primrose oil supplements, borage oil, homeopathy, and Chinese herbal medicine (Ernst 2000; Sheehan 1992), in the treatment of eczema.

Natural sources of gamma‐linolenic acid (GLA) include the following: borage seed oil, blackcurrant seed oil, evening primrose seed oil, and human milk (0.6%). Evening primrose (Oenothera biennis) is a tall plant with pretty yellow flowers blooming in the evening. Evening primrose oil, which contains 8% to 10% gamma‐linolenic acid, has been widely used for treating eczema, both as a prescription item and an over‐the‐counter remedy. Borage (Borago officinalis) oil contains at least 23% gamma‐linolenic acid.

Adverse effects

An earlier concern that evening primrose oil (EPO) might cause epilepsy or seizures has been shown to be incorrect. In fact, the "EPO‐derived omega‐6 fatty acid arachidonic acid inhibits sodium ion currents and synaptic transmission, while the EPO‐derived eicosanoid prostaglandin E(1) appears to have anticonvulsant activity. In light of these findings, formularies should now remove seizures or epilepsy as a side‐effect of EPO, and should remove a history of seizures or epilepsy as a contraindication to taking EPO" (Puri 2007).

There has been one case report of lipoid pneumonia secondary to long‐term use of EPO in a person with gastroesophageal reflux disease (GERD) (Rabahi 2010).

Another case report warns that if EPO is taken for a prolonged period of time (more than one year), there is a potential risk of inflammation, thrombosis, and immunosuppression due to slow accumulation of EPO in the tissues (Phinney 1994).

Animal research suggests that EPO has important anticoagulant and antiplatelet activity, which may last 60 days (Riaz 2009). Anticoagulant drugs warn of interaction with EPO (Jellin 2005). Although suggested as helping heart attacks, essential fatty acids (EFAs) cause increased bleeding, which was documented in a study that showed no benefit to rheumatoid arthritis (Nordstrom 1995).

Ingesting EPO during pregnancy "may be associated with an increase in the incidence of prolonged rupture of membranes, oxytocin augmentation, arrest of descent, and vacuum extraction" (Dove 1999).

Borage oil (BO) has a similar profile of headache and minor gastrointestinal adverse effects. A recent single case report (Al‐Khamees 2011) brings up the same concern for seizures with BO, which was shown to be a spurious concern for EPO.

How the intervention might work

"The possible mechanisms by which EFAs might improve inflammatory skin disease include modification of eicosanoid metabolism so as to favour synthesis of relatively non‐inflammatory prostaglandins and leukotrienes (n6, n7). Since both n6 and n3 essential fatty acids may possess this property, it is possible that giving both together will have a synergistic effect" (Berth‐Jones 1993). Antioxidant properties have been studied in EPO (Shahidi 1997).

Why it is important to do this review

Many people with eczema are reluctant to use the most commonly recommended treatments, i.e. topical corticosteroids, because of fear of local side‐effects, such as allergies, skin thinning, and fear of other long‐term health effects, such as growth retardation and adrenal gland suppression (Charman 2000). People with eczema often use complementary medicine (Eisenberg 1998). Some evidence suggests that people with eczema seek treatment with complementary medicine because they are highly satisfied with complementary medicine (Augustin 1999; Jensen 1990; Triebskorn 1989). Another study showed that "(68%) of patients expressed a positive difference in the relationship they had with their practitioner at the Center compared to their own GP" (Finnigan 1991).

There is a lay perception that complementary treatments, such as EPO and BO, treat the skin 'from within' and are 'gentler' when compared to physician‐prescribed topicals.

Previous studies of supplementation of gamma‐linolenic acid for eczema have shown conflicting results. Thus, there is uncertainty whether EPO and BO are effective and whether it is safe to prescribe these oils. This systematic review aims to address these questions.

Objectives

To assess the effects of oral evening primrose oil or borage oil for treating the symptoms of atopic eczema.

Methods

Criteria for considering studies for this review

Types of studies

All randomised controlled trials (RCTs) with parallel or cross‐over design, investigating oral intake of EPO or BO for eczema or synonyms including atopic eczema (AE), atopic dermatitis (AD), atopic eczema/dermatitis syndrome (AEDS), and atopiform dermatitis.

Types of participants

We included adults and children diagnosed with eczema by a medical practitioner. We accepted diagnostic criteria such as the Hanifin and Rajka definition (Hanifin 1980) or the UK modification (Williams 1994), as well as atopic eczema diagnosed by a dermatologist using the terms 'atopic eczema' or 'atopic dermatitis'.

Types of interventions

Oral intake of EPO or BO compared against placebo (usually liquid paraffin, olive oil, or sunflower oil) or other active eczema therapies.

Types of outcome measures

Primary outcomes

  • Global degree of improvement in symptoms and signs as rated by participant or medical doctor.

  • Improvement in quality of life.

Secondary outcomes

  • Severe adverse events (caused by EPO or caused by BO) requiring withdrawal from the trial, including severe skin irritation.

  • Minor participant‐reported adverse events, including mild skin irritation not sufficient to require cessation of treatment.

  • Systemic or topical treatments taken concurrently with EPO or BO.

  • Skin irritation (such as redness, itch, scaling).

  • Change of score of any validated scoring system for atopic eczema.

  • Time off school/work.

  • Need for topical steroids and emollients.

Search methods for identification of studies

We aimed to identify all relevant randomised controlled trials (RCTs) regardless of language or publication status (published, unpublished, in press, or in progress).

Electronic searches

We searched the following databases up to 29 August 2012.

  • the Cochrane Skin Group Specialised Register using the following terms: (((atopic or infantile or childhood) and eczema) or ((atopic and eczema and dermatitis and syndrome) or AEDS) or ((atopic or atopiform) and dermatitis) or (neurodermatitis or neurodermitis) and (primrose or herb* or plant* or efamol or epogam or epo or borag*));

  • the Cochrane Central Register of Controlled Trials (CENTRAL) in The Cochrane Library using the search strategy in Appendix 1;

  • MEDLINE via OVID (from 1946) using the strategy in Appendix 2;

  • EMBASE via OVID (from 1974) using the strategy in Appendix 3;

  • AMED via OVID (Allied and Complementary Medicine, from 1985) using the strategy in Appendix 4; and

  • LILACS (Latin American and Caribbean Health Science Information database, from 1982) using the strategy in Appendix 5.

Trials registers

We searched for reports of trials in the following trials databases up to 29 August 2012 using the terms "eczema", "primrose", and "borage" or "borago":

Searching other resources

Reference lists

We checked the bibliographies of included studies for further references to relevant trials.

Grey literature

We identified unpublished and ongoing trials by correspondence with trial authors. We contacted pharmaceutical companies such as Nutricia, Canada (former Epogam® producer Scotia Holdings) who produce relevant products.

Correspondence

We (KS in a past version of the authoring team, and JB in the present team) sent emails and letters to investigators to ask for more information about their trials where there were insufficient documentation within reports of the studies (Bahmer 1992; Bamford 1985; Biagi 1994; Don 2003; Lovell 1981; Wright 1982).

Adverse effects

Safety data were as reported in the RCTs (Data and analyses). Additionally, we contacted manufacturers for long‐term surveillance data on EPO and BO safety. We made additional searches (4 November 2011) for non‐randomised EPO and BO studies, revealing more adverse events. We searched the following:

  • the World Health Organization;

  • the Medicines and Healthcare products Regulatory Agency's (MHRA's) Yellow Card reporting Scheme for adverse drug reactions;

  • the WHO database for adverse events; and

  • the German database 'medcertain.org', a collection of published and unpublished cases of adverse events.

Data collection and analysis

Selection of studies

Two review authors (JB and SR) independently screened title/abstract results of the searches. Where title/abstract did not provide clear direction, we used full‐text articles and a predesigned eligibility form based on the inclusion criteria to complete the selection of included articles. If details were unclear, we sought clarification from the trial authors, and we reassessed the corresponding articles. We resolved any difference between the eligibility results by consensus. We excluded studies that did not meet the inclusion criteria and stated the reasons in the 'Characteristics of excluded studies' section of the review.

Data extraction and management

Two review authors (JB and SR) independently extracted data on methods, participants, interventions, and outcomes from each included trial, using a specially designed prepiloted data extraction form. One review author (AM) also independently extracted data on outcomes for the purpose of data analysis. For each dichotomous outcome, we extracted the number of participants experiencing the event and the number randomised in each treatment group. For each continuous outcome, we extracted the mean, standard deviation (or information to estimate the standard deviation), and the number randomised in each treatment group. We contacted the trial authors in the case of unclear or missing data. We resolved any disagreements by consensus.

Assessment of risk of bias in included studies

Two review authors (JB and SR) independently assessed the risk of bias of the included studies using the guidelines provided in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We followed guidance given in the handbook when assessing whether adequate steps were taken to reduce the risk of bias across six components:
(a) sequence generation;
(b) allocation concealment;
(c) blinding (of participants, personnel, and outcome assessors);
(d) incomplete outcome data;
(e) selective outcome reporting; and
(f) other sources of bias.

We categorised judgements in order to indicate a low, high, or unclear risk of bias. The results were summarised using the 'Risk of bias' graph (Figure 1) and the 'Risk of bias' summary (Figure 2), in addition to the 'Risk of bias' table for each included study. Where necessary, we contacted trial authors for clarification. We resolved any disagreements by consensus with CvG.

1.

1

'Risk of bias' graph: review authors' judgements about each 'Risk of bias' item presented as percentages across all included studies

2.

2

'Risk of bias' summary: review authors' judgements about each 'Risk of bias' item for each included study

Measures of treatment effect

We calculated risk ratios (RR) for dichotomous outcomes and mean differences (MD) for continuous outcomes. We presented these results with 95% confidence intervals (CI).

Unit of analysis issues

We noted the unit of analysis at the level of randomisation (individual or group) and analysed accordingly.

Dealing with missing data

We contacted original authors of studies for clarification of study methodology and results. When we required further information to assess risk of bias, we contacted trial investigators of studies to obtain clarifications. We were unable to obtain information from many trial investigators because of the age of the study (the information was unavailable or the trial investigator could not be contacted). We carried out analyses according to the intention‐to‐treat (ITT) principle where there was no missing data. In the case of missing data, we carried out an available case analysis.

Assessment of heterogeneity

We assessed statistical heterogeneity by applying the Chi² test (P value < 0.10 considered statistically significant) and the I² statistic (Higgins 2011), where an I² statistic of greater than 50% represented substantial heterogeneity. The Chi² test (also called the Cochran's Q test) is a test of significance of heterogeneity, and the I² statistic describes the percentage of the variability in effect estimates that is due to heterogeneity rather than chance (Higgins 2011).

Assessment of reporting biases

We did not construct a funnel plot to assess the possibility of publication bias because there were few trials per meta‐analysis (less than 10). The Cochrane Handbook of Systematic Reviews recommends at least 10 studies in a meta‐analysis in order to construct a funnel plot. However, we employed the following criteria to assess reporting bias:

  1. authors and institutions involved;

  2. publication of articles;

  3. funding;

  4. sponsorship of commercially available supplements; and

  5. affiliation of the authors with the manufacturer.

We recorded this information in the 'Risk of bias' tables.

Data synthesis

We analysed data separately for EPO and BO using Review Manager 5 (RevMan 5). We examined the data from included studies for descriptive synthesis and pooled data where trials were sufficiently homogeneous in design, methodology, and outcomes. When we pooled data, we used the DerSimonian and Laird random‐effects model because of anticipated heterogeneity between the included studies.

Subgroup analysis and investigation of heterogeneity

The original authors of the protocol did not refer to an intention to carry out subgroup analyses, but if they found statistical heterogeneity, they did plan to explore further ‐ using the Chi² test (P < 0.1) ‐ possible reasons, such as participants, study design, and intervention. We anticipated we would investigate potentially influential study characteristics by conducting subgroup analysis with respect to the following (where sufficient studies existed):

  • RCTs reporting the use of oral placebo capsules with vitamin E versus oral placebo capsules without vitamin E.

  • Results in infected atopic dermatitis (bacterial, viral, and fungal) versus results in non‐infected atopic dermatitis.

  • Immunoglobulin E (IgE)‐associated atopic eczema versus non‐IgE‐associated atopic eczema.

  • Results in infants/children versus results in other (adults only or adults and children).

However, we only managed to carry out subgroup analyses with respect to the last comparison.

Sensitivity analysis

The protocol prespecified the following sensitivity analyses:

  • IgE‐associated atopic eczema versus non‐IgE‐associated atopic eczema;

  • protein deficiency versus no protein deficiency;

  • participants who continued with conventional treatment versus participants who stopped with conventional treatment;

  • studies of different levels of gamma‐linolenic acid ‐ borage seed oil 60 to 120 mg GLA, evening primrose oil 40 to 68 mg GL;

  • studies with parallel design versus studies with cross‐over design;

  • studies with oral intake of EPO or borage oil versus studies with oral intake of EPO plus vitamin E or oral intake of borage oil plus vitamin E;

  • results in infants/children versus results in adults; and

  • RCTs reporting the use of oral placebo capsules that additionally contain vitamin E or other substances "will initially be included and a sensitivity analysis will be carried out to see whether Vitamin E on its own could have had an effect on the improvement of eczema symptoms".

We could not perform sensitivity analyses with respect to any of the above factors because of the non‐availability of the relevant data from the trial reports, although the review considered the following two subgroups:

  • participants who continued with conventional treatment versus participants who stopped with conventional treatment; and

  • studies with parallel design versus studies with cross‐over design.

Most studies failed to report on whether conventional treatment was continued or stopped during the study; therefore, we could not factor this information in sensitivity analysis, and we did not include any cross‐over studies in the meta‐analyses.

We did carry out some sensitivity analysis for one meta‐analysis where there was significant heterogeneity by excluding an outlying study and simultaneously examining the effect that this had on both the treatment effect and heterogeneity.

We did not look at heterogeneity related to presence or absence of vitamin E in placebo capsules.

Other

The systematic review team included a consumer (RM) to ensure readability and clarity of the terminology of the final review.

Three studies had a third arm, which was not relevant to this review; therefore, we excluded the irrelevant arm from analysis.

Results

Description of studies

Results of the search

The study flow diagram is displayed in Figure 3. The search yielded a total of 179 citations, of which 167 came from electronic databases, 11 were unpublished 'in house' studies, and 1 was referenced in www.clinicaltrials.gov. A total of 3 duplicate records were removed, leaving 176 citations, which we assessed for eligibility. After reading titles and abstracts, we excluded a total of 142 studies. We obtained full texts for the remaining 34 citations and included 27 studies in the review. We excluded seven studies after reading the full‐texts, as well as one further study (NCT00878670), which was ongoing at the time we were preparing this review, when www.clinicaltrials.gov reported it was an open trial. No other ongoing trials were registered as of 16 December 2012.

3.

3

Study flow diagram

Included studies

A total of 27 studies (1596 participants) met the inclusion criteria. The characteristics and risk of bias have been presented figuratively and descriptively in the 'Characteristics of included studies' tables.

Design

Of the 27 included studies, 21 were single centre (Bahmer 1992; Bamford 1985; Berth‐Jones 1993; Biagi 1994; Borrek 1997; Buslau 1996; Don 2003; Harper 1990; Heddle 1990; Kenicer 2001; Kiehl 1994; Kovar 1992; Lovell 1981; MacKie Adult 1990; MacKie Child 1990; MacLeod 2001; Schalin‐Karrila 1987; Senapati 2008; Takwale 2003; Valsecchi 2006; Wright 1982), and 6 were multicentre with 2 to 12 sites (Finlay 2001; France 1988; Fraser 2001; Hederos 1996; Henz 1999; Wishart 1992).

Five studies were cross‐over (Bamford 1985; Borrek 1997; Heddle 1990; Lovell 1981; Wright 1982), and the remaining 22 were parallel in design (Bahmer 1992; Berth‐Jones 1993; Biagi 1994; Buslau 1996; Don 2003; Finlay 2001; France 1988; Fraser 2001; Harper 1990; Hederos 1996; Henz 1999; Kenicer 2001; Kiehl 1994; Kovar 1992; MacKie Adult 1990; MacKie Child 1990; MacLeod 2001; Schalin‐Karrila 1987; Senapati 2008; Takwale 2003; Valsecchi 2006; Wishart 1992).

Three studies had a third arm that was not relevant to this review (Berth‐Jones 1993; Biagi 1994; MacLeod 2001).

In all articles, the whole person was the unit of analysis.

Sample sizes in studies

We included 27 studies, with a total of 1596 randomised participants (65 are not counted here as they were on other trial arms in 3 studies not fitting this review criteria). Details of all the studies are included in the Characteristics of included studies. Sample sizes ranged from 12 (Bahmer 1992) to 160 (Henz 1999).

Setting

The included studies were carried out in 10 countries: UK (13), Germany (5), Italy (3), India (1), Finland (1), New Zealand (1), Sweden (1), USA (1), and Switzerland (1).

The 11 house studies (Finlay 2001; France 1988; Fraser 2001; Harper 1990; Heddle 1990; Kenicer 2001; Kovar 1992; MacKie Adult 1990; MacKie Child 1990; MacLeod 2001; Wishart 1992) were never published outside the pharmaceutical company, and we use the term "house studies" to designate those studies.

Peer‐reviewed journals published 16 included studies, which had lead authors outside the pharmaceutical industry.

All the studies were carried out in a secondary care setting (referral or dermatology clinic).

Participants

The majority of articles described recruitment from those who were attending dermatology or paediatric clinics, but no article described advertising or stipends paid to participants. One article (Bamford 1985) detailed the method of participant recruitment.

Nine studies had participants who were mixed groups of adults and children (Bamford 1985; Berth‐Jones 1993; Fraser 2001; Kiehl 1994; Lovell 1981; Senapati 2008; Takwale 2003; Valsecchi 2006; Wright 1982). There were nine trials with only children enrolled (Biagi 1994; Borrek 1997; Don 2003; Finlay 2001; Harper 1990; Heddle 1990; Hederos 1996; Kovar 1992; MacKie Child 1990). The remaining nine studies had only adult participants (Bahmer 1992; Buslau 1996; France 1988; Henz 1999; Kenicer 2001; MacKie Adult 1990; MacLeod 2001; Schalin‐Karrila 1987; Wishart 1992).

In the (16) studies that stated the sex of the individuals, there were 344 men or boys and 384 women or girls (Biagi 1994; Borrek 1997; Buslau 1996; Don 2003; Fraser 2001; Harper 1990; Hederos 1996; Henz 1999; Kovar 1992; MacKie Adult 1990; MacKie Child 1990; MacLeod 2001; MacLeod 2001; Schalin‐Karrila 1987; Senapati 2008; Wishart 1992).

Interventions

The included trials fell into two broad therapeutic categories (EPO or BO). Further information about interventions, results, and outcomes can be found in the 'Characteristics of included studies' tables.

Major or primary outcomes

Evening primrose oil

  • Participant‐reported global improvement in 10 studies

  • Physician‐reported global improvement in 14 studies

  • Only two studies used quality of life measures (Don 2003; Finlay 2001)

Borage oil

  • Participant‐reported global improvement in two studies

  • Physician‐reported global improvement in five studies

Minor or secondary outcomes

  • There was only one report involving severe allergic reaction to penicillin, causing the subject on EPO to withdraw from the study (MacKie Adult 1990).

  • Four studies reported adverse effects.

  • Nine studies described adverse effects as minor side‐effects: "none noted", "too few to evaluate statistically", or "not stated".

  • Eighteen cases mentioned minor side‐effects. In these studies, there was a range of minor adverse effects: 17% (Hederos 1996) to 50% (Finlay 2001). Minor gastrointestinal symptoms (nausea, vomiting, diarrhoea, colic) were the most common (Bamford 1985; Finlay 2001; France 1988; Fraser 2001; Hederos 1996; Henz 1999; Kenicer 2001; Takwale 2003; Valsecchi 2006). Headache was equal in active and placebo groups in one study (Henz 1999).

  • No study mentioned protein deficiency versus no protein deficiency.

  • The trials used global measures of change more commonly than SCORAD (SCORing Atopic Dermatitis) (SCORAD index 1993), Atopic Dermatitis Area and Severity Index (ADASI) (Bahmer 1991), or Costa 1989's scoring methods.

With regard to adverse effects, in our search of four databases mentioned, we found headaches and minor gastrointestinal adverse effects, including abdominal pain, nausea, increased bowel movement, and diarrhoea, had been attributed to oral EPO. Most adverse effects have been mild gastrointestinal symptoms/signs (Bayles 2009).

With regard to subgroup analysis, firstly, we could not perform subgroup analysis with respect to whether placebos had vitamin E or not because there was only one study included in the second meta‐analysis with vitamin E in the placebo. Secondly, with respect to infected versus non‐infected eczema, most of the studies did not report the information. Thirdly, most studies also did not report information regarding IgE‐associated eczema.

Excluded studies

We excluded eight studies: Melnick 1995 did not have a placebo group; Biagi 1988 was an open study; Courage 1991 and Coskery 1988 were found to be non‐randomised, as was the study NCT00878670, which was initially listed as an RCT on the www.clinicaltrials.gov website (the lead author then confirmed it was not an RCT). In the study by Swapan 2008, even though randomisation was stated, we found that consecutive participant allocation had been carried out. The results of one study (Guenther 1987) were not available, as correspondence with the trial investigator confirmed that the study results had been destroyed. For the study Bordoni 1988, we received no reply to queries addressed to the lead author.

Risk of bias in included studies

We provide summaries of the risk of bias across the included studies in Figure 1, Figure 2, and the 'Characteristics of included studies' tables.

Sequence Generation (selection bias)

We judged bias related to randomisation to be low in two thirds (18) of the studies (Bamford 1985; Berth‐Jones 1993; Borrek 1997; Don 2003; Finlay 2001; France 1988; Harper 1990; Hederos 1996; Kenicer 2001; Kovar 1992; Lovell 1981; MacKie Adult 1990; MacKie Child 1990; MacLeod 2001; Senapati 2008; Takwale 2003; Wishart 1992; Wright 1982). Randomisation methods included computer‐generated methods (Bamford 1985; Berth‐Jones 1993; Don 2003; Finlay 2001; France 1988; Harper 1990; Hederos 1996; Kenicer 2001; Lovell 1981; MacKie Adult 1990; MacKie Child 1990; MacLeod 2001; Senapati 2008; Takwale 2003; Wishart 1992; Wright 1982). We assumed the block randomisation by draw was computer‐generated (Borrek 1997).

The remaining studies were unclear. No study had inadequate methods of randomisation leading to high risk of bias.

Allocation

We judged less than half of the studies (12) to have reported or carried out adequate allocation concealment. Only Bamford 1985; Don 2003; Finlay 2001; France 1988; Harper 1990; Kenicer 2001; Kovar 1992; MacKie Adult 1990; MacKie Child 1990; MacLeod 2001; Takwale 2003; and Wishart 1992 concealed the allocation sequence, and we judged them at low risk of bias. In the remaining studies, the method of allocation concealment was unclear in 12 studies and at high risk in the remaining 3 studies.

With Harper 1990, although we recorded "listing of randomization schedule and code is not available", we felt the other details (blinding, ITT analysis, randomisation) were enough to deem it as low bias for randomisation and concealment.

Blinding

Adequate blinding of practitioners and participants was present in 60% (16/27) of studies. We noted possible bias due to failure to blind or inadequate reporting of blinding (both participant and personnel, including outcome assessors) in 33% (9/27) of the included studies, which we judged as 'unclear' (Berth‐Jones 1993; Buslau 1996; Don 2003; Fraser 2001; Heddle 1990; Henz 1999; Schalin‐Karrila 1987; Senapati 2008; Takwale 2003). Masking of personnel and participants were not separated in Kenicer 2001 and Valsecchi 2006, and we judged them to be at high risk of bias; the latter was not blinded.

Incomplete outcome data

We evaluated the majority of studies (20) as having a low risk of attrition bias, and we judged 2 studies, MacLeod 2001 and Buslau 1996, to be at high risk of bias as a large proportion of the participants dropped out, and this was not balanced between the groups. With regard to dealing with missing data, we judged five studies as unclear. We stated where no details could be obtained, so judged these studies as 'unclear' in terms of risk of bias.

Selective reporting

Nine studies did not provide enough information to determine if the prespecified outcomes had been adequately reported; therefore, we judged these to be at an unclear risk of bias (France 1988; Henz 1999; Kiehl 1994; Kovar 1992; Lovell 1981; MacLeod 2001; Takwale 2003; Wishart 1992; Wright 1982). Reporting of outcomes was adequate in the remaining 18 studies.

Other potential sources of bias

Bias associated with other causes was rated as high in 10 studies (8 of which were unpublished, house studies), unclear in 7 studies, and low in 10 studies. The study Kenicer 2001 had participants who did not meet their criteria, and the statistical methods were made after all the data were revealed (and used in published meta‐analysis), so we classed this domain as at 'high risk of bias'.

We noted sources of medication and funding as possible other sources of potential bias, but the following information has not been used to make judgements about the risk of bias.

Effects of interventions

Comparison 1: EPO versus placebo

A total of 19 studies investigated the effect of EPO versus placebo for the treatment of eczema, of which 11 were unpublished house studies, and 8 were published studies. The house studies reported results both before and after trimming. The study reports describe trimming as a statistical method of removing participants with extreme baseline values from analysis. We used only untrimmed results (before trimming) in the analysis of this review.

All of the 11 house studies reported results for the analysis using models A, B, and C:

  • model A compared final values without adjustment for baseline values;

  • model B used change from baseline values (adjusted for baseline); and

  • model C used change from baseline values but not adjusted for baseline.

We used only results from model B in the analyses in this review, because they were the best statistically unbiased results among the three models. The results reported and therefore used in this review are the least squares means. The treatment duration for most of the house studies was 24 weeks, although some studies had a treatment duration of 16 or 20 weeks. For meta‐analysis, we used results comparing week 16 and baseline. We chose week 16 because for most studies there would be very few participants left in the study by week 24 due to attrition. Where standard errors (SEs) were given, we calculated the corresponding standard deviations (SDs) by multiplying SEs with the square root of the corresponding sample sizes. Since the primary outcome was change from baseline, the fact that the mean is less than SD does not imply skewed data. Most of the results for the house studies were obtained from additional tables in the reports.

Primary outcomes
(a) Global degree of improvement in symptoms and signs as rated by participant or medical doctor
(i) Participant‐reported

Ten studies reported results for overall improvement in symptoms as rated by the participants. All of these were house studies. Nine studies reported using the VAS (0 to 100) scores. Seven of these house studies reported the participants' assessment of overall impression as reduction or improvement of VAS scores in terms of means (SE), and their results were pooled in a meta‐analysis that showed that treatment with EPO failed to significantly improve eczema symptoms compared to placebo treatment, as measured by participants (MD ‐2.22, 95% CI ‐10.48 to 6.04, 176 participants, 7 studies, Analysis 1.1, Figure 4). However, we detected statistically significant heterogeneity between the studies (Chi² statistic = 12.08, df = 6, P = 0.06, I² statistic = 50%, Analysis 1.1).

1.1. Analysis.

1.1

Comparison 1: EPO versus placebo, Outcome 1: Participant‐reported global improvement in symptoms (0 to 100 VAS reduction)

4.

4

Forest plot of comparison: 1 EPO versus placebo, outcome: 1.1 Participant‐reported global improvement in symptoms (0 to 100 VAS reduction).

An inspection of the forest plot reveals that confidence intervals from six studies were overlapping, but there was one outlying study (MacKie Child 1990). While the results for the six studies showed non‐statistically significant difference, the results from MacKie Child 1990 showed statistically significant differences in favour of placebo. A sensitivity analysis through excluding this outlying study yielded statistically non‐significant heterogeneity (I² statistic = 0%). This may indicate that this study is the source of heterogeneity. We also noted that exclusion of the study did not change the conclusion regarding the treatment effect. Subgroup analysis with respect to children only versus other (adults or adults and children) showed no significant subgroup differences (Chi² = 0.81, df = 1, P = 0.37, I² statistic = 0%, Analysis 1.1). Although significant heterogeneity was evident between the 2 children‐only studies in the first subgroup (Chi² = 3.95, df = 1, P = 0.05, I² statistic = 75%, Analysis 1.1), there was no significant heterogeneity detected in the second subgroup (Chi² = 4.35, df = 4, P = 0.36, I² statistic = 8%, Analysis 1.1). This supports the conclusion above that the significant heterogeneity observed in this meta‐analysis is due to the one outlying study (MacKie Child 1990) that investigated children only. Also, we observed no significant difference in symptoms between the EPO and placebo groups in both subgroups.

One study (Hederos 1996) reported the parents' global assessment of eczema (VAS 0 to 100) at the end of treatment as mean difference (MD) with corresponding 95% confidence interval. We used the confidence interval to calculate the standard error (SE), then we used the MD and SE in calculating a treatment effect through the generic inverse‐variance method in RevMan 5. Treatment with EPO failed to significantly reduce the symptoms of eczema at the end of treatment as compared to placebo treatment (MD 2.90, 95% CI ‐8.80 to 14.60, 60 participants, 1 study, Analysis 1.2).

1.2. Analysis.

1.2

Comparison 1: EPO versus placebo, Outcome 2: Parent assessment at end of treatment (VAS)

One house study (Kovar 1992) reported parents' global assessment of eczema (VAS 0 to 100) at the end of treatment. We used data at 8 weeks, because all children were given the active treatment from 8 weeks to 16 weeks. The calculated treatment effect showed that treatment with EPO failed to significantly reduce eczema symptoms in children compared to the placebo treatment as assessed by the children's parents (MD ‐0.80, 95% CI ‐17.76 to 16.16, 32 participants, 1 study, Analysis 1.3).

1.3. Analysis.

1.3

Comparison 1: EPO versus placebo, Outcome 3: Parents global improvent in eczema symptoms at 8 weeks

Bamford 1985 also reported the overall severity of eczema as assessed by participants. However, they reported the results for all the participants irrespective of their treatment groups. Therefore, we could not calculate a treatment effect, and attempts to obtain the individual participant data were unsuccessful. The statistician of the study informed us that the original data had been destroyed. Although some results were presented in the form of some figures, it was not possible to accurately extract data from them since the figures were presented separately by age and dose and also because standard deviations of change were not graphed.

Wright 1982 reported the overall improvement in eczema symptoms as assessed by the participants but reported the results in the form of a figure from which values of means (SD) could not be accurately extracted for the calculation of a treatment effect.

Lovell 1981 reported the overall participants' assessment of eczema using a clinical grading scale (0 = no eczema, 10 = most severe), but the results were only given as MD of 0.52. Since the SE was not reported, we could not calculate a treatment effect.

(ii) Physician‐reported

Fourteen studies reported results for the overall improvement in symptoms as rated by the physician or doctor. Ten of these were house studies. Eight house studies reported the physicians' assessment of overall impression as reduction or improvement of 0 to 100 VAS scores in terms of means (SE) and were pooled in a meta‐analysis, which showed that treatment with EPO failed to significantly reduce the physician‐reported eczema symptoms compared to placebo treatment (MD ‐3.26, 95% CI ‐6.96 to 0.45, 289 participants, 8 studies, Analysis 1.4, Figure 5). There was no statistically significant heterogeneity detected between the studies (Chi² statistic = 4.45, df = 7, P = 0.73, I² statistic = 0%, Analysis 1.4). Subgroup analysis with respect to children only versus other (adult or adult and children) showed no significant subgroup differences (Chi² statistic = 0.01, df = 1, P = 0.91, I² statistic = 0%, Analysis 1.4), and there were no significant differences in symptoms between the EPO and placebo groups in both subgroups.

1.4. Analysis.

1.4

Comparison 1: EPO versus placebo, Outcome 4: Physician‐reported global improvement in symptoms (0 to 100 VAS reduction)

5.

5

Forest plot of comparison: 1 EPO versus placebo, outcome: 1.4 Physician‐reported global improvement in symptoms (0 to 100 VAS reduction).

One study (Hederos 1996) reported the physician's global assessment of eczema (VAS 0 to 100) at the end of treatment as mean difference (MD) with corresponding 95% confidence interval. We used the confidence interval to calculate the standard error (SE), then we used the MD and SE in calculating a treatment effect through the generic inverse‐variance method in RevMan 5. Treatment with EPO failed to significantly reduce the physicians' reported symptoms of eczema at the end of treatment as compared to placebo treatment (MD ‐2.80, 95% CI ‐14.50 to 8.90, 60 participants, 1 study, Analysis 1.5).

1.5. Analysis.

1.5

Comparison 1: EPO versus placebo, Outcome 5: Physician assessment at the end of treatment (VAS ‐ MD)

One house cross‐over study (Heddle 1990) reported the overall physician assessment of change or improvement in VAS (0 to 50) as means (SD). We only extracted first‐period data as we suspected a carry‐over effect. The treatment effect calculated from these first‐period data showed that treatment with EPO failed to significantly reduce the physician‐reported eczema symptoms compared to placebo treatment (MD ‐0.43, 95% CI ‐1.65 to 0.79, 25 participants, 1 study, Analysis 1.6).

1.6. Analysis.

1.6

Comparison 1: EPO versus placebo, Outcome 6: Physician‐reported improvement in symptoms (VAS 0 to 50)

One study (Biagi 1994) reported the physician assessment of overall severity of eczema using a simplified rating scale from 0 to 3 (0 = absent, 1 = slight, 2 = moderate, 3 = severe). The results at the end of treatment were reported as means (SD) for the three treatment arms (low‐dose EPO, high‐dose EPO, and placebo). We calculated treatment effects separately for high‐dose EPO versus placebo and for low‐dose EPO versus placebo. The calculated treatment effect supports a statistically significant improvement of symptoms in favour of 'high dose' EPO compared to placebo (MD ‐0.56, 95% CI ‐1.09 to ‐0.03, 32 participants, 1 study, Analysis 1.7), but there were no significant differences between the 'low dose' EPO versus placebo (MD ‐0.51, 95% CI ‐1.05 to 0.03, 32 participants, 1 study, Analysis 1.8).

1.7. Analysis.

1.7

Comparison 1: EPO versus placebo, Outcome 7: Physician‐reported assessment at end of treatment (0 to 3 scale): high‐dose

1.8. Analysis.

1.8

Comparison 1: EPO versus placebo, Outcome 8: Physician‐reported assessment at end of treatment (0 to 3 scale): low‐dose

One study (Senapati 2008) reported the overall physician assessment at the end of treatment using some total disease intensity scores (extent, intensity, itching, dryness), where lower values indicate less symptoms. The treatment effect from this study showed a statistically significant improvement in eczema symptoms in favour of EPO as compared to placebo treatment (MD ‐3.52, 95% CI ‐4.42 to ‐2.62, 50 participants, 1 study, Analysis 1.9).

1.9. Analysis.

1.9

Comparison 1: EPO versus placebo, Outcome 9: Physician‐reported assessment at end of treatment (total disease scores)

One study (Berth‐Jones 1993) reported improvement in the overall clinical severity score as measured by Leicester scores using three treatment arms (placebo, EPO, and EPO + fish oil). We used results for EPO versus placebo groups and excluded results for the other treatment EPO‐fish oil arm. The calculated treatment effect showed that treatment with EPO failed to significantly reduce the physician‐reported eczema symptoms compared to placebo treatment (MD 1.33, 95% CI ‐6.61 to 9.72, 67 participants, 1 study, Analysis 1.10).

1.10. Analysis.

1.10

Comparison 1: EPO versus placebo, Outcome 10: Physician‐reported improvement in Leicester scores

Schalin‐Karrila 1987 reported the overall severity of symptoms (0 = no symptoms and 100 = worst possible) at the end of treatment (12 weeks) in the form of a figure from which values of means (SE) were extracted. The SDs were calculated from the SEs and totals. The calculated treatment effect showed that treatment with EPO failed to significantly reduce the physician‐reported eczema symptoms compared to placebo treatment (MD 1.30, 95% CI ‐13.64 to 16.24, 25 participants, 1 study, Analysis 1.11).

1.11. Analysis.

1.11

Comparison 1: EPO versus placebo, Outcome 11: Physician‐reported assessment at end of treatment (VAS 0 to 100)

Bamford 1985 also reported the overall severity of eczema as assessed by physicians. However, they reported the results for all the participants irrespective of their treatment groups. Therefore, we could not calculate any treatment effect, and attempts to obtain the individual participant data were unsuccessful. Although some results were presented in the form of some figures, it was not possible to accurately extract data from them since they presented figures separately by age and dose and also because they did not graph standard deviations of change.

Wright 1982 reported the overall improvement in eczema symptoms as assessed by doctors, but reported the results in the form of a figure, from which values of means (SD) could not be accurately extracted for the calculation of a treatment effect.

Lovell 1981 reported the overall physicians' assessment of eczema using a clinical grading scale (0 = no eczema; 10 = most severe), but the results were only given as MD of 0.37. Since the report trial did not report the SE, we could not calculate a treatment effect.

(b) Improvement in quality of life

Only one house study (Finlay 2001) reported the Dermatology Life Quality Index (DLQI). The calculated treatment effect showed that treatment with EPO failed to significantly improve quality of life compared to placebo treatment (MD 0.08, 95% CI ‐1.26 to 1.42, 36 participants, 1 study, Analysis 1.12).

1.12. Analysis.

1.12

Comparison 1: EPO versus placebo, Outcome 12: Dermatology Life Quality Index (DLQI) at the end of treatment

Secondary outcomes  
(a) Adverse events
(i) Severe adverse events (caused by EPO) requiring withdrawal from the trial, including severe skin irritation

Only two house studies (Finlay 2001 and France 1988) reported this outcome. From both studies, there were too few adverse events to warrant any formal statistical analysis.

(ii) Minor participant‐reported adverse events, including gastrointestinal irritation not sufficient to require cessation of treatment

Four studies reported the number with adverse events without specifying the actual adverse events. We pooled these in a meta‐analysis, and there was no difference in the proportion of participants experiencing adverse events between the EPO‐ and placebo‐treated groups (RR 0.97, 95% CI 0.71 to 1.33, 264 participants, 4 studies, Analysis 1.13). We did not detect any statistically significant heterogeneity between the studies (Chi² statistic = 1.97, df = 3, P = 0.58, I² statistic = 0%, Analysis 1.13).

1.13. Analysis.

1.13

Comparison 1: EPO versus placebo, Outcome 13: Adverse events

Three house studies reported minor signs or symptoms as an adverse event. There was no statistically significant difference in the proportion of participants with minor signs or symptoms between the EPO‐ and placebo‐treated groups (RR 1.00, 95% CI 0.64 to 1.57, 172 participants, 3 studies, Analysis 1.14). We did not detect any statistically significant heterogeneity between the studies (Chi² statistic = 0.85, df = 2, P = 0.66, I² statistic = 0%, Analysis 1.14).

1.14. Analysis.

1.14

Comparison 1: EPO versus placebo, Outcome 14: Adverse event (minor signs or symptoms)

There were other minor adverse events reported by other studies, including headache, infection, diarrhoea, nausea and bloating, and hyperactivity, but their frequencies were too small to warrant any formal statistical analysis. Other studies either reported that there were no adverse events observed or did not report any information regarding adverse events.

(b) Systemic or topical treatments taken concurrently with EPO or BO

Only two studies reported this outcome. Hederos 1996 reported steroid usage classified into four categories of the most potent class of steroids used, and we could not calculate a treatment effect. Senapati 2008 reported the amount (in kg) of emollient cream and topical steroid consumed during the 12‐week period of treatment. We calculated treatment effects for these 2 outcomes, and there was no significant difference in mean amount of emollient cream (MD ‐0.02, 95% CI ‐0.42 to 0.38, 24 participants, 1 study, Analysis 1.15) used between the 2 groups, but there was significantly more topical steroid used by participants in the placebo group compared to the EPO group (MD ‐0.14, 95% CI ‐0.26 to ‐0.02, 24 participants, 1 study, Analysis 1.16).

1.15. Analysis.

1.15

Comparison 1: EPO versus placebo, Outcome 15: Concurrent treatment (emollient cream) (kg)

1.16. Analysis.

1.16

Comparison 1: EPO versus placebo, Outcome 16: Concurrent treatment (topical steroid) (kg)

Berth‐Jones 1993 reported topical steroid requirement for all the participants, irrespective of their treatment groups; therefore, we could not calculate a treatment effect.

Comparison 2: borage oil versus placebo

A total of eight studies investigated the effect of BO versus placebo for the treatment of eczema.

Primary outcomes
(a) Global degree of improvement in symptoms and signs as rated by participant or medical doctor
(i) Participant‐reported

Only two studies reported overall improvement in symptoms by participants. Bahmer 1992 reported the number with improvement in severity of skin changes. The calculated treatment effect showed that treatment with BO failed to significantly reduce the participant‐reported eczema symptoms compared to placebo treatment (RR 3.57, 95% CI 0.58 to 21.92, 12 participants, 1 study, Analysis 2.1). The confidence interval is wide because of the very small sample sizes and number of events. Borrek 1997 reported the participants' subjective complaints at the end of treatment in terms of means (SD). Again, a calculated treatment effect showed that treatment with BO failed to significantly reduce the participant‐reported eczema symptoms compared to placebo treatment (MD 0.10, 95% CI ‐1.57 to 1.77, 22 participants, 1 study, Analysis 2.2).

2.1. Analysis.

2.1

Comparison 2: Borage oil versus placebo, Outcome 1: Participant‐reported improvement in symptoms

2.2. Analysis.

2.2

Comparison 2: Borage oil versus placebo, Outcome 2: Participant‐reported complaints at end of treatment

(ii) Physician‐reported

Five studies reported this outcome, but in different ways. Borrek 1997 reported the overall physician assessment using Costa (Costa 1989) scores at the end of treatment in terms of means (SD), from which we calculated a treatment effect. Borage oil failed to significantly reduce the physician‐reported eczema symptoms compared to placebo treatment (MD ‐1.00, 95% CI ‐13.18 to 11.18, 22 participants, 1 study, Analysis 2.3).

2.3. Analysis.

2.3

Comparison 2: Borage oil versus placebo, Outcome 3: Physician‐reported Costa scores at end of treatment

Buslau 1996 reported the number showing significant improvement in the Atopic Dermatitis Area and Severity Index (ADASI score) as assessed by a physician at the end of treatment. A calculated treatment effect showed that treatment with BO failed to significantly reduce the physician‐reported eczema symptoms compared to placebo treatment (RR 1.81, 95% CI 0.94 to 3.49, 32 participants, 1 study, Analysis 2.4).

2.4. Analysis.

2.4

Comparison 2: Borage oil versus placebo, Outcome 4: Physician‐reported number showing improvement (ADASI score)

Don 2003 reported the assessment by physicians of clinical severity of eczema using the SCORAD index at the end of treatment. A calculated treatment effect showed that treatment with BO failed to significantly reduce the physician‐reported eczema symptoms compared to placebo treatment (MD 4.53, 95% CI ‐10.34 to 19.40, 18 participants, 1 study, Analysis 2.5).

2.5. Analysis.

2.5

Comparison 2: Borage oil versus placebo, Outcome 5: Physician‐reported severity of eczema at end of treatment (SCORAD Index)

Takwale 2003 reported the assessment by physicians of severity of eczema using the overall Six Area, Six Sign Atopic Dermatitis severity score (SASSAD score). Again, a calculated treatment effect showed that treatment with BO failed to significantly reduce the physician‐reported eczema symptoms compared to placebo treatment (MD 4.00, 95% CI ‐1.56 to 9.56, 140 participants, 1 study, Analysis 2.6).

2.6. Analysis.

2.6

Comparison 2: Borage oil versus placebo, Outcome 6: Physician‐reported SASSAD score at end of treatment

Valsecchi 2006 reported the clinical score improvement in eczema symptoms from baseline. A calculated treatment effect showed that treatment with BO failed to significantly reduce the physician‐reported eczema symptoms compared to placebo treatment (MD 22.31, 95% CI ‐8.97 to 53.59, 28 participants, 1 study, Analysis 2.7).

2.7. Analysis.

2.7

Comparison 2: Borage oil versus placebo, Outcome 7: Physician‐reported clinical score improvement from baseline

(b) Improvement in quality of life

Only one study (Don 2003) reported the quality of life at the end of treatment using a four‐point scale (four = calm, three = slightly distressed, two = moderately distressed, one = very distressed). The mean scores were 3.44 and 3.55 for the BO and placebo groups, respectively. Since the study did not provide the corresponding standard deviations, we could not calculate a treatment effect.

Secondary outcomes  
(a) Adverse events
(i) Severe adverse events caused by BO requiring withdrawal from trial, including severe skin irritation

None of the studies in this comparison reported this outcome.

(ii) Minor participant‐reported adverse events, including gastrointestinal irritation not sufficient to require cessation of treatment

Only two studies (Henz 1999 and Takwale 2003) reported minor adverse events. Henz 1999 reported adverse events (not specific), but there were similar proportions of adverse events in the BO and placebo groups (RR 1.07, 95% CI 0.70 to 1.64, 158 participants, 1 study, Analysis 2.8). There was also no difference in the proportion of participants experiencing flu‐like symptoms by the end of treatment between the BO and placebo groups (RR 1.11, 95% CI 0.48 to 2.59, 158 participants, 1 study, Analysis 2.9) (Henz 1999).

2.8. Analysis.

2.8

Comparison 2: Borage oil versus placebo, Outcome 8: Adverse events

2.9. Analysis.

2.9

Comparison 2: Borage oil versus placebo, Outcome 9: Adverse event (flu‐like symptoms)

Takwale 2003 reported upper respiratory tract infection, but there was no difference in the prevalence of this adverse event between the BO and placebo groups (RR 0.68, 95% CI 0.41 to 1.11, 140 participants, 1 study, Analysis 2.10). The other adverse events reported by Takwale 2003, albeit with too few numbers to warrant any statistical analysis, included diarrhoea, nausea and vomiting, abdominal pain, asthma, allergic rhinitis, urticaria, new rash, musculoskeletal pains, skin sepsis, glandular fever, and headache.

2.10. Analysis.

2.10

Comparison 2: Borage oil versus placebo, Outcome 10: Adverse event (upper respiratory tract infection)

Other studies either reported that they did not observe adverse events, or they did not report any information regarding adverse events.

(b) Systemic or topical treatments taken concurrently with EPO or BO

Don 2003 reported the number of children treated with topical steroids, oral antihistamine drugs, and moisturising lotion, but there were similar numbers in the borage group compared to the placebo group. The actual values reported for BO versus placebo were 4/9 versus 2/9, 4/9 versus 5/9, and 7/9 versus 7/9 for topical steroids, oral antihistamine drugs, and moisturising lotion, respectively.

Discussion

Summary of main results

The objective of this review was to assess the efficacy of oral evening primrose oil (EPO) and borage oil (BO) for treating the signs and symptoms of eczema. A comprehensive search of studies yielded a total of 27 studies (1596 participants), which we included in this review, where 19 studies assessed EPO against placebo, and the remaining 8 studies assessed BO against placebo. Eleven studies out of the 27 included studies were unpublished house pharmaceutical studies. The included studies varied in participants (adults, children, adults and children), duration of follow‐up, and outcomes assessed. In some trials, the sample size was as small as 12, and the largest trial had a sample size of 160.

A meta‐analysis of results from seven EPO studies demonstrated that the difference in global eczema symptoms between the EPO and placebo groups was not statistically significant as assessed by the participants. A second meta‐analysis of results from eight studies also yielded no significant improvement of global eczema symptoms from EPO compared to placebo treatment, as measured by doctors.

Only two studies measured quality of life. Only a single study reported the results for quality of life improvement data, but no significant benefit in favour of EPO was evident. A few studies also reported adverse effects, but there was no significant difference between the EPO and placebo groups. Borage oil also did not show any significant benefit in terms of improving global eczema symptoms or reducing adverse events compared to the placebo treatment, although we could not carry out meta‐analysis because the few studies assessing BO reported results in different ways. None of the studies reported costs.

We could not do sensitivity analyses to assess the effect of a large number of house studies on the outcomes, because we only included house studies in the meta‐analyses.

Overall completeness and applicability of evidence

A major limitation of the current evidence is the variation in how outcomes were reported, thus, making it impossible to pool all studies measuring the same outcome in meta‐analyses. There was also variation in the duration of the included studies, from 6 weeks to 24 weeks. While some studies assessed children or adults separately, other studies evaluated both. Furthermore, we did not find any studies from developing countries except one study from India, but there were no studies from Africa or Asia. Therefore, we are unable to generalise our results, which show a lack of a significant effect of EPO and BO in reducing eczema symptoms to all populations and all age groups.

The available studies did not allow us to evaluate several secondary end points, which we hoped to cover: IgE, protein, cost, and quality of life. We encountered missing data problems, and attempts to obtain the missing data were not successful. Some of the reasons for this were due to failed execution of trial protocols, delayed evaluation and analysis, changing authors, and destruction of data after publication. We were also unable to carry out intention‐to‐treat (ITT) analyses, because of lack of data. We therefore carried out available case analyses instead. We could have investigated the dose‐response effect of the interventions on the outcomes, but this was not necessary because the effect of the interventions were not statistically significant.

There were no studies completed within the last eight years that met the inclusion criteria. One study (clinicaltrials.gov identifier: NCT00878670) was completed during the period of writing this review; it was not a RCT but an open trial, which showed no benefit of EPO for eczema.

Both EPO and BO are commercially advertised on the internet as treatment for eczema.

Quality of the evidence

The number of studies and their consistency in showing no benefit of EPO and BO over placebo oils seems persuasive. When we compared studies with paraffin (which contains no essential fatty acids) as placebo (Bamford 1985; Berth‐Jones 1993; France 1988; Lovell 1981; MacLeod 2001; Schalin‐Karrila 1987; Takwale 2003; Valsecchi 2006; Wright 1982) with those using various other placebos (palm, olive, corn, coconut, sunflower, glandol, or miglyol oils) (Biagi 1994; Borrek 1997; Buslau 1996; Don 2003; Finlay 2001; Fraser 2001; Harper 1990; Heddle 1990; Hederos 1996; Henz 1999; Kenicer 2001; Kiehl 1994; Kovar 1992; MacKie Adult 1990; MacKie Child 1990; Senapati 2008; Wishart 1992), the results were the same.

With regard to the risk of bias, the majority of studies were of low risk of bias; we judged two thirds of the included studies as having low risk of bias for random sequence generation, 40% for allocation concealment, 30% for other biases, and almost 60% for blinding. However, the sample sizes for the majority of the studies were small, and there were limitations regarding missing data in meta‐analyses, mainly due to the longitudinal nature of the studies.

Potential biases in the review process

We conducted a comprehensive search of electronic databases to ensure that we identified all relevant studies. We also scanned through the reference lists of all the identified studies to look for any relevant studies. There were no language restrictions applied to studies for this review, and in fact, we translated four German articles and included them in the review. It is unlikely that we missed any relevant RCTs. We reduced potential bias in the conduct of this review by having two of the review authors independently scan through the search output, extract data, and assess the methodological quality of each study. However, we were unable to formally assess the likelihood of publication bias (selective reporting of positive findings) using funnel plots because of the small number of studies per meta‐analysis. Although we did not carry out a Grading of Recommendations Assessment, Development and Evaluation (GRADE) appraisal for the review, we could comment on the quality of evidence using the magnitude of sample sizes, the results from meta‐analyses, and the assessment of risk of bias.

The authors of this systematic review have had the following involvement with the subject of this review: The lead author, Mr Joel Bamford, MD, studied paraffin versus EPO in adults and children with atopic eczema (Bamford 1985). Christel van Gool authored a trial of EPO to prevent eczema in children of mothers who had atopic dermatitis (van Gool 2003). Edzard Ernst has published widely on complementary therapies. Rosemary Humphreys has had personal experience of the use of EPO for atopic dermatitis.

We felt the delay in data becoming accessible and never being subject to peer review were sources of reporting bias. Defining statistical methods used six years after the last participant completed the study was also a source of concern (France 1988). Finally, it may be that there are articles that were never indexed and therefore were not available to our search methods.

Agreements and disagreements with other studies or reviews

This review found no significant improvement in global eczema symptoms from both EPO and BO, as measured by both participant and doctors. Our results are consistent with results from the included studies. We also assessed quality of life and adverse events, but few of the included studies reported this in sufficient detail to allow firm conclusions regarding these. Our review disagrees with an earlier meta‐analysis of RCTs comparing EPO and placebo, which found a significant improvement in both participant‐ and doctor‐reported global eczema symptoms in favour of EPO (Morse 1989). This meta‐analysis used two of our included studies in print (Bamford 1985; Wright 1982) and referenced nine unpublished RCTs from eight centres. A subsequent meta‐analysis (Morse 2006) of 26 RCTs (1207 participants) found a beneficial effect of EPO on itch, crusting, oedema, and erythema. This meta‐analysis also found a significant association of the treatment effect with increasing frequency of potent steroid use, but we could not assess this in our review because most studies did not properly report results on steroid use. Although we did not assess the specific symptoms, we included overall effects in the global measures used in this review.

Authors' conclusions

Implications for practice.

Oral borage oil and evening primrose oil appear to be equivalent in their lack of effect on eczema. They had the same effect on global ratings as the placebos used. They do not seem to add any benefit to eczema as measured in this systematic review.

They seem to have the same fairly common mild transient adverse effects, which are mainly gastrointestinal (ranging from 7% to 15% in‐house studies) (France 1988). There is evidence that EPO may increase bleeding for people on Coumadin® (warfarin) medication.

Over the years since the last study closed 10 years ago, there have been new developments in the use of systemic treatments for eczema: methotrexate for children, biologics, selective neuroepinephrine re‐uptake inhibitors, serotonin uptake inhibitors, μ‐ and κ‐opioid receptor agonists, and substance P antagonists, which are being evaluated (Patel 2011).

During the years since the protocol to this review was published, there have been changes in practice:

1. the term 'atopic eczema/dermatitis syndrome' (AEDS) to summarise all forms of eczema that have previously been called 'prurigo Besnier' or 'atopic dermatitis' (we did not find AEDS used in any included study);
2. the introduction of pseudo‐ceramide, immunomodulator and biologic agents which came to the market after all our included studies;
3. disclosure of authorship and sponsorship requirements; and
4. the preregistration of protocols and the possibility of public posting of results on web sites such as www.clinicaltrials.gov.

The latter two will bring greater transparency and reproducibility to clinical studies as well as helping to combine study data in meta‐analyses.

Implications for research.

Noting that the confidence intervals between the active and placebo treatments are so narrow to exclude the possibility of any clinically useful difference, we conclude that further studies EPO or BO for eczema would be hard to justify.

In the future, as initial publication of trial protocols and early publication of detailed results become standard practice, all data will be available for evaluation by journal peer reviewers, readers, and researchers to those wishing to confirm or extend their investigations or carry out their own meta‐analyses.

What's new

Date Event Description
7 December 2020 Amended Typos fixed in 'Declarations of interest': should have read Miller‐Dwan Foundation and Efamol Pharmaceutical company.

History

Protocol first published: Issue 4, 2003
Review first published: Issue 4, 2013

Date Event Description
15 May 2014 Review declared as stable The conclusion is so certain that the addition of new information will not change it, and there is unlikely to be any more research done on this topic. There were no ongoing studies or studies awaiting classification listed in the last published review.
13 November 2013 Amended Author information (affiliation) updated.
25 June 2013 Amended Contact author's out‐of‐date email address removed and current email address added

Notes

We have deemed this review as stable and no longer in need of updating because the conclusion is so certain that the addition of new information will not change it, and there is unlikely to be any more research done on this topic. There were no ongoing studies or studies awaiting classification listed in the last published review.

Acknowledgements

We should like to thank Professor Hywel Williams for his useful suggestions on the draft protocol. Katja Bhoem (formerly Katja Schmidt: previous author) translated and extracted German articles for this review. Max Pittler participated as an author in earlier versions of this review. Diane Wennberg and Anna Robbins, Medical Librarians with Essentia Health Library, Duluth, Minnesota, secured the PDFs of articles for this review. Jo Leonardi‐Bee extracted needed data from graphs. Cathy Bennett (Systematic Research Ltd.) provided consultancy advice about the format of the review. We also thank Anke Rohwer, a lecturer at Stellenbosch University, Faculty of Medicine and Health Sciences, in South Africa, for her assistance in translating the results sections of the four German studies to the English language.

We would like to thank the Trials Search Co‐ordinator of the Skin Group editorial base, Liz Doney, for developing the search strategies and running the searches for the review.

The Cochrane Skin Group editorial base wishes to thank Ching‐Chi Chi who was the Key Editor for this review; Matthew Grainge and Philippa Middleton who were the Statistical and Methods Editors, respectively; the clinical referees, Mandy Elvira Schram and Mary Glover; and the consumer referee, Amanda Roberts.

Appendices

Appendix 1. CENTRAL (Cochrane Library) search strategy

#1 (atopic dermatitis) or (atopic eczema)
#2 (neurodermatitis or (atopic ekzema))
#3 MeSH descriptor Dermatitis, Atopic explode all trees
#4 MeSH descriptor Neurodermatitis explode all trees
#5 MeSH descriptor Eczema explode all trees
#6 (#1 OR #2 OR #3 OR #4 OR #5)
#7 MeSH descriptor Borago explode all trees
#8 MeSH descriptor Oenothera biennis explode all trees
#9 (evening next primrose)
#10 (borag* or efamol or epogam or epo)
#11 (oenothera biennis)
#12 (#7 OR #8 OR #9 OR #10 OR #11)
#13 (#6 AND #12)

Appendix 2. Medline (OVID) search strategy

1. randomised controlled trial.pt.
2. controlled clinical trial.pt.
3. randomized.ab.
4. placebo.ab.
5. clinical trials as topic.sh.
6. randomly.ab.
7. trial.ti.
8. 1 or 2 or 3 or 4 or 5 or 6 or 7
9. (animals not (human and animals)).sh.
10. 8 not 9
11. exp Eczema/ or eczema.mp.
12. atopic dermatitis.mp. or exp Dermatitis, Atopic/
13. atopic eczema.mp.
14. neurodermatitis.mp. or exp Neurodermatitis/
15. exp Dermatitis/ or dermatitis.mp.
16. 11 or 12 or 13 or 14 or 15
17. 10 and 16
18. borage.mp. or exp Borago/
19. primrose.mp.
20. epo.mp.
21. exp Oenothera biennis/
22. exp Plant Oils/
23. epogam.mp.
24. efamol.mp.
25. 18 or 19 or 20 or 21 or 22 or 23 or 24
26. 17 and 25

Appendix 3. EMBASE (OVID) search strategy

1. random$.mp.
2. factorial$.mp.
3. (crossover$ or cross‐over$).mp.
4. placebo$.mp. or PLACEBO/
5. (doubl$ adj blind$).mp. 
6. (singl$ adj blind$).mp. 
7. (assign$ or allocat$).mp.
8. volunteer$.mp. or VOLUNTEER/
9. Crossover Procedure/
10. Double Blind Procedure/
11. Randomized Controlled Trial/
12. Single Blind Procedure/
13. 1 or 2 or 3 or 4 or 5 or 6 or 7 or 8 or 9 or 10 or 11 or 12
14. eczema.mp. or ECZEMA/
15. atopic dermatitis.mp. or exp atopic dermatitis/
16. atopic eczema.mp.
17. neurodermatitis.mp. or exp NEURODERMATITIS/
18. exp DERMATITIS/ or dermatitis.mp.
19. borage.mp. or exp Borago/
20. exp PRIMROSE OIL/ or exp EVENING PRIMROSE/ or primrose.mp.
21. epo.mp.
22. efamol.mp.
23. oenothera biennis.mp
24. epogam.mp
25. 14 or 15 or 16 or 17 or 18
26. 19 or 20 or 21 or 22 or 23 or 24
27. 13 and 25 and 26

Appendix 4. AMED (OVID) search strategy

1. randomised controlled trial.pt.
2. controlled clinical trial.pt.
3. randomized.ab.
4. placebo.ab.
5. clinical trials as topic.sh.
6. randomly.ab.
7. trial.ti.
8. 1 or 2 or 3 or 4 or 5 or 6 or 7
9. (animals not (human and animals)).sh.
10. 8 not 9
11. exp Plant extracts/ or exp Plant oils/ or borage.mp. or exp Plants medicinal/
12. primrose.mp.
13. exp Efamol/ or efamol.mp.
14. epo.mp.
15. oenothera biennis.mp.
16. epogam.mp.
17. 11 or 13 or 14 or 16
18. exp Eczema/ or eczema.mp.
19. exp Dermatitis atopic/ or atopic dermatitis.mp.
20. atopic eczema.mp.
21. neurodermatitis.mp. or exp Neurodermatitis/
22. dermatitis.mp. or exp Dermatitis/
23. 18 or 19 or 20 or 21 or 22
24. 10 and 17 and 23

Appendix 5. LILACS search strategy

((Pt RANDOMIZED CONTROLLED TRIAL OR Pt CONTROLLED CLINICAL TRIAL OR Mh RANDOMIZED CONTROLLED TRIALS OR Mh RANDOM ALLOCATION OR Mh DOUBLE‐BLIND METHOD OR Mh SINGLE‐BLIND METHOD OR Pt MULTICENTER STUDY) OR ((tw ensaio or tw ensayo or tw trial) and (tw azar or tw acaso or tw placebo or tw control$ or tw aleat$ or tw random$ or (tw duplo and tw cego) or (tw doble and tw ciego) or (tw double and tw blind)) and tw clinic$)) AND NOT ((CT ANIMALS OR MH ANIMALS OR CT RABBITS OR CT MICE OR MH RATS OR MH PRIMATES OR MH DOGS OR MH RABBITS OR MH SWINE) AND NOT (CT HUMAN AND CT ANIMALS)) [Palavras] and ((atopic and dermatitis) or (atopic and eczema) or ((atopic and ekzema) or neurodermatitis) or ((atopica or atopia) and eccema)) [Palavras]

Appendix 6. Abbreviations used

Abbreviation Full term
ADRsMHRA adverse drug reaction noted by the World Health Organization
AEDS atopic eczema/dermatitis syndrome
BO borage oil
CAM complementary and alternative medicine
E‐BM evidence‐based medicine
EFA essential fatty acid
EPO evening primrose oil
GLA gamma‐linoleic acid
MIMS Monthly Index of Medical Specialties severity classificaton
Costa, ADASI, SASSAD, and SCORAD these are scales for measuring eczema severity

Data and analyses

Comparison 1. EPO versus placebo.

Outcome or subgroup title No. of studies No. of participants Statistical method Effect size
1.1 Participant‐reported global improvement in symptoms (0 to 100 VAS reduction) 7 176 Mean Difference (IV, Random, 95% CI) ‐2.22 [‐10.48, 6.04]
1.1.1 Children only 2 30 Mean Difference (IV, Random, 95% CI) ‐11.92 [‐39.34, 15.49]
1.1.2 Other (adults only or adults and children) 5 146 Mean Difference (IV, Random, 95% CI) 0.98 [‐5.47, 7.42]
1.2 Parent assessment at end of treatment (VAS) 1   Mean Difference (IV, Random, 95% CI) Totals not selected
1.3 Parents global improvent in eczema symptoms at 8 weeks 1   Mean Difference (IV, Random, 95% CI) Totals not selected
1.4 Physician‐reported global improvement in symptoms (0 to 100 VAS reduction) 8 289 Mean Difference (IV, Random, 95% CI) ‐3.26 [‐6.96, 0.45]
1.4.1 Children only 3 101 Mean Difference (IV, Random, 95% CI) ‐2.94 [‐9.18, 3.30]
1.4.2 Other (adults only or adults and children) 5 188 Mean Difference (IV, Random, 95% CI) ‐3.39 [‐8.12, 1.33]
1.5 Physician assessment at the end of treatment (VAS ‐ MD) 1   Mean Difference (IV, Random, 95% CI) Subtotals only
1.6 Physician‐reported improvement in symptoms (VAS 0 to 50) 1   Mean Difference (IV, Random, 95% CI) Totals not selected
1.7 Physician‐reported assessment at end of treatment (0 to 3 scale): high‐dose 1   Mean Difference (IV, Random, 95% CI) Totals not selected
1.8 Physician‐reported assessment at end of treatment (0 to 3 scale): low‐dose 1   Mean Difference (IV, Random, 95% CI) Totals not selected
1.9 Physician‐reported assessment at end of treatment (total disease scores) 1   Mean Difference (IV, Random, 95% CI) Totals not selected
1.10 Physician‐reported improvement in Leicester scores 1   Mean Difference (IV, Random, 95% CI) Totals not selected
1.11 Physician‐reported assessment at end of treatment (VAS 0 to 100) 1   Mean Difference (IV, Random, 95% CI) Totals not selected
1.12 Dermatology Life Quality Index (DLQI) at the end of treatment 1   Mean Difference (IV, Random, 95% CI) Totals not selected
1.13 Adverse events 4 264 Risk Ratio (M‐H, Random, 95% CI) 0.97 [0.71, 1.33]
1.14 Adverse event (minor signs or symptoms) 3 172 Risk Ratio (M‐H, Random, 95% CI) 1.00 [0.64, 1.57]
1.15 Concurrent treatment (emollient cream) (kg) 1   Mean Difference (IV, Random, 95% CI) Totals not selected
1.16 Concurrent treatment (topical steroid) (kg) 1   Mean Difference (IV, Random, 95% CI) Totals not selected

Comparison 2. Borage oil versus placebo.

Outcome or subgroup title No. of studies No. of participants Statistical method Effect size
2.1 Participant‐reported improvement in symptoms 1   Risk Ratio (M‐H, Random, 95% CI) Totals not selected
2.2 Participant‐reported complaints at end of treatment 1   Mean Difference (IV, Random, 95% CI) Totals not selected
2.3 Physician‐reported Costa scores at end of treatment 1   Mean Difference (IV, Random, 95% CI) Totals not selected
2.4 Physician‐reported number showing improvement (ADASI score) 1   Risk Ratio (M‐H, Random, 95% CI) Totals not selected
2.5 Physician‐reported severity of eczema at end of treatment (SCORAD Index) 1   Mean Difference (IV, Random, 95% CI) Totals not selected
2.6 Physician‐reported SASSAD score at end of treatment 1   Mean Difference (IV, Random, 95% CI) Totals not selected
2.7 Physician‐reported clinical score improvement from baseline 1   Mean Difference (IV, Random, 95% CI) Totals not selected
2.8 Adverse events 1   Risk Ratio (M‐H, Random, 95% CI) Totals not selected
2.9 Adverse event (flu‐like symptoms) 1   Risk Ratio (M‐H, Random, 95% CI) Totals not selected
2.10 Adverse event (upper respiratory tract infection) 1   Risk Ratio (M‐H, Random, 95% CI) Totals not selected

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Bahmer 1992.

Study characteristics
Methods Design: randomised, placebo‐controlled, parallel
Duration: 3 months
Interval of assessment: baseline, 6 weeks, 3 months
This was a randomised controlled trial
Participants Number randomised: 12 (7 in the active group and 5 in the placebo group)
Sex (M/F): 6/6 total (3/4 in the active group and 3/2 in the placebo group)
Age of participants: active group = 31 (mean), placebo group = 27 (mean)
Unit of allocation: whole person
Country and setting: Saarlandes, Germany; single university dermatology clinic
Inclusion criteria of the study (specified)

  • Participants having atopic dermatitis

  • Having used topical steroids

  • Diagnostic criteria: not reported

  • Severity of condition: "mild to moderate", glandol group ADASI score = 1.63, control group ADASI score = 1.89


Exclusion criteria of the study

  • Participants using systemic steroids or antimetabolites
Interventions

  • Treatment group: glandol (BO), 3 capsules twice daily, each capsule 500 mg

  • Placebo group: palm kernel oil, 3 capsules twice daily, each capsule 500 mg
Outcomes

  1. ADASI score by clinician (method of scoring by mapping)

  2. Skin condition by participant (method of assessment subjective: participant diary)

  3. Serum lipids (lab)
Notes Previous treatment was not stopped ("no limitations regarding other treatment were required")
Assessment of compliance was not undertaken
Risk of bias
Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Unclear risk The trial mentioned randomisation, but did not give a specific method
Allocation concealment (selection bias) Unclear risk The trial did not give a method for allocation concealment
Incomplete outcome data (attrition bias)
All outcomes Low risk There were no dropouts or losses to follow up (7/7 glandol and 5/5 placebo completed)
Selective reporting (reporting bias) Low risk The trial reported all outcomes
Other bias Low risk All randomised participants were included in the analysis in the groups to which they were randomised. We contacted the author
Blinding of participants and personnel (performance bias)
All outcomes Low risk This was adequate as both the participant and assessor were blinded

Bamford 1985.

Study characteristics
Methods Design: randomised, double‐blind, placebo‐controlled, cross‐over
Duration: 6 months (3 months then 3 months reversed)
Interval of assessment: 3 months
Participants Number randomised: 154
Sex (M/F): not stated
Age of participants: completers were 49 children aged 2 to 16 years (mean age = 9.1) and 74 adults aged 16 to 66 years (mean age = 37.7)
Unit of allocation: single participant
Country and setting: USA, single dermatology outpatient department
Inclusion criteria of the study

  • Participants having atopic dermatitis

  • Having used topical steroids

  • Diagnostic criteria: clinical

  • Severity of condition: mild to moderate


Exclusion criteria of the study

  • Participants using systemic steroids or antimetabolites
Interventions

  • Treatment group divided as follows:

    • children < 15 years of age, who received 2 or 4 capsules twice daily each with 500 mg of EPO for 3 months, then crossed over;

    • children > 15 years of age, who received 6 or 8 capsules twice daily each with 500 mg of EPO

  • Placebo group divide in the same way as above and same dosage schedule with each capsule having 500 mg of paraffin oil
Outcomes

  1. Symptoms and signs of eczema (erythema, scaling, excoriation, weeping, etc) by physician (method = 0 to 10 scale, where 0 is none and 10 is most severe)

  2. Symptoms and signs of eczema (erythema, scaling, excoriation, weeping, etc) and effect on daily living (level of discomfort, inability to sleep, etc) by participant (method = 0 to 10 scale, where 0 is none and 10 is most severe)


Interval of assessment: unclear
Notes Concomittent treatment: permitted emollients, topical steroids, and antihistaminics
Compliance to treatment: undertaken
Previous treatment: continued
The trial report did not state the numbers randomised to each group (further information obtained from trial investigator)
Risk of bias
Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Low risk The trial used block randomisation
Allocation concealment (selection bias) Low risk Specific study numbers were used on labels, which did not contain medication/placebo name. (The trial investigator provided this further information)
Incomplete outcome data (attrition bias)
All outcomes Low risk 14/77 participants in the EPO group dropped out; 17/77 participants in the placebo group dropped out. They gave reasons
Selective reporting (reporting bias) Low risk The trial reported all prespecified outcomes
Other bias Unclear risk We were unsure whether all participants were included in the analysis in the groups to which they were randomised. We contacted the biostatistician author. A non‐profit foundation funded the trial
Blinding of participants and personnel (performance bias)
All outcomes Low risk This was adequate as both the participant and assessor were blinded. The placebo capsules were identical

Berth‐Jones 1993.

Study characteristics
Methods Design: double‐blind, randomised, parallel with 3 treatment arms
Duration: 24 weeks
Interval of assessment: Assessment was done at 4, 8, and 16 weeks, and then 8 weeks after wash‐out
Participants Number randomised: 123 (41 in the active EPO group, 41 in the active EPO and Marine group, and 41 in the placebo group)
Sex (M/F): 25/16 in the active group, 22/19 in the active EPO and Marine group, and 22/19 in the placebo group
Age of participants: well‐matched for age above and over 12 years
Unit of allocation: whole person
Country and setting: UK, single centre
Inclusion criteria of the study

  • Outpatients attending the dermatology department for treatment of AD, of either sex and all ages

  • Diagnostic criteria: Hanifin and Rajka


Exclusion criteria of the study

  • Not specified
Interventions

  • 2 treatment groups:

    • first treatment group: 6 capsules twice daily, each capsule contained 500 mg of evening primrose oil, largely made up of the n6 series EFAs linoleic (321 mg) and gamma‐linolenic acid (40 mg)

    • second treatment group: 6 capsules twice daily, each contained 430 mg of evening primrose oil and 107 mg of marine fish oil. This fish oil included the n3 series EFAs eicosapentaenoic acid (17 mg) and docosahexaenoic acid (11 mg)

  • Placebo group: 6 capsules: Each capsule contained liquid paraffin for adults and olive oil for children
Outcomes

  1. Disease activity monitored by clinical severity scores (2 types of scoring systems were used) as recorded by the investigator (methods for scoring: Leicester Score, the body was divided into 10 zones: face, neck, abdomen, back, elbows, antecubital fossae, dorsa of hands, palms and wrists, popliteal fossae, and feet. Each zone was scored on a scale: 0 (absent) to 3 (severe) for erythema, excoriation, dryness, cracking, and lichenification. A second, simpler scoring system for clinical severity, as described by Costa 1989, was also used. In this system, only the most severely affected site was assessed. Erythema, oedema, vesiculation, crusting, excoriation scaling, lichenification, pigmentation, pruritus, and sleep disturbance were each graded on a 7‐point scale. This score was only briefly reported)

  2. Topical steroid requirement (usage was assessed by weighing returned tubes of medication)

  3. Symptom scores recorded by participants. Every participant was given a diary containing a page of 10 cm visual analogue scales for baseline and each of the next 24 weeks. Scales were provided for itch, dryness, scaling, redness, and overall impression. All scales were labelled "none" at the left‐hand end and "worst ever" at the right. Scores were added to give a maximum of 50 cm


Interval of assessment: 4, 8, and 16 weeks on treatment, and again after an 8‐week wash‐out
Notes Concomittent treatment: allowed continued emollients and low‐dose steroids
Compliance to treatment: unsure
Risk of bias
Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Low risk Quote: "Randomization in blocks of three..."
Comment: The trial used block randomisation
Allocation concealment (selection bias) Unclear risk The trial did not state the method by which it concealed allocation
Incomplete outcome data (attrition bias)
All outcomes Low risk The trial reported all outcomes
Selective reporting (reporting bias) Low risk There was no selective reporting
Other bias Unclear risk We were unsure whether all participants were included in the analysis in the groups to which they were randomised
Blinding of participants and personnel (performance bias)
All outcomes Unclear risk Quote: "The study had a double‐blind design"
Comment: The trial stated blinding, but did not specify the method

Biagi 1994.

Study characteristics
Methods Design: double‐blind, placebo‐controlled using 2 different dosages
Duration: 8 weeks
Interval of assessment: start and end of study
Participants Number randomised: 48 (16 in each group)
Sex (M/F): 24/27 total (this applied to the whole group; there was no separation for the active and placebo groups)
Age of participants: children average 4.2 years (2.2 to 8.5 years)
Unit of allocation: whole person
Country and setting: Italy, single referral outpatient department
Inclusion criteria of the study

  • Diagnostic criteria: Hanifin and Rajka


Exclusion criteria of the study

  • Not specified
Interventions

  • 2 treatment groups:

    • first treatment group: high‐dose group, EPO capsules (dose = 0.5 mg/kg body weight)

    • second treatment group: low‐dose group, 50% EPO capsule and 50% placebo capsule (dose = 0.5 mg/kg body weight)

  • Placebo group: capsules of olive oil and vitamin E (0.5 mg/kg body weight)
Outcomes

  1. Severity of symptoms (method: clinical assessment by scoring 10 symptoms. Scoring from 0 to 3, 0 = absent, 1 = mild, 2 = moderate, 3 = severe)

  2. Red cell membrane fatty acid concentration

  3. Red cell membrane viscosity
Notes Concomittent treatment: allowed continued emollients and topical steroids
Compliance to treatment: unsure
Risk of bias
Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Unclear risk Quote: "Children were randomised to receive either..."
Comment: The trial gave no method of generation
Allocation concealment (selection bias) High risk The trial gave no details
Incomplete outcome data (attrition bias)
All outcomes Unclear risk 3/48 participants did not attend for follow up. The dropouts or losses to follow up by assignment group were unclear, as were the reasons
Selective reporting (reporting bias) Low risk The trial reported all prespecified outcomes
Other bias Low risk We contacted the author; there was no other bias
Blinding of participants and personnel (performance bias)
All outcomes Low risk Quote: "...all assessments were made double blind"
Comment: This was adequate, but details were not given of who was blinded and how. This was probably done

Borrek 1997.

Study characteristics
Methods Design: randomised by draw, double‐blind, placebo‐controlled, cross‐over
Duration: 20 to 24 weeks
Interval of assessment: 4 to 3 weeks
Participants Number randomised: 24 children and teenagers from 3 to 17 years of age
Sex (M/F): 14/10 total
Age of participants: 3 to 17 years (mean age = 10 years)
Unit of allocation: single person
Country and setting: Germany, 1 university pediatric clinic
Inclusion criteria of the study

  • Chronic recurrent atopic dermatitis at the time of enrolment in the study

  • Was clinically symptomatic within the last 6 months

  • Diagnostic criteria: Hanifin Rajka's criteria

  • Severity of condition: mild to moderate


Exclusion criteria of the study

  • Epilepsy

  • Fat storage disease

  • Participating in another study
Interventions

  • Treatment group: 3 capsules daily, each with 60 mg borage seed oil and 5 mg vitamin E

  • Placebo group: 3 capsules daily, each with 60 mg corn oil and 5 mg vitamin E
Outcomes

  1. Costa (Costa 1989) score by clinician (method: this simple scoring system (SSS) scores 10 severity criteria (0 to 7) and 10 topographic sites (0 to 3) giving a maximum score of 100)

  2. Acute, chronic, and subjective symptoms by participants (method: diaries)

  3. Use of steroidal creme and antihistamines (method: questionnaires)
Notes Concommitant treatment: permitted
Assessment of compliance: unsure
Risk of bias
Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Low risk Randomisation was by draw (entrance to study)
Allocation concealment (selection bias) Unclear risk The trial did not specify this
Incomplete outcome data (attrition bias)
All outcomes Low risk 2 participants from each group withdrew due to non‐compliance
Selective reporting (reporting bias) Low risk The trial reported all prespecified outcomes
Other bias Low risk Quote (page 101): "No difference between the two groups in the most important variables"
Blinding of participants and personnel (performance bias)
All outcomes Low risk Quote (page 101): "Patient and examiner were blinded with respect to after completion of the capsule contents"
Comment: This was adequate (both blinded)

Buslau 1996.

Study characteristics
Methods Design: randomised, double‐blind, placebo‐controlled, parallel
Duration: 12 weeks
Interval of assessment: daily by participant
Participants Number randomised: 50 (25 each in the active and placebo groups)
Sex (M/F): 18/32 total (8/17 in the active group and 10/15 in the placebo group)
Age of participants: not mentioned
Unit of allocation: whole person
Country and setting: Germany, not noted
Inclusion criteria of the study

  • Severity of condition: mild to moderate

  • Diagnostic criteria: Hanifin Rajka's criteria


Exclusion criteria of the study

  • Not specified
Interventions

  • Treatment group: 2 capsules daily each with 500 mg of BO

  • Placebo group: 2 capsules daily each with 500 mg of palm kernel oil
Outcomes

  1. ADASI score (method of assessment: mapping with the help of coloured markers)
Notes Concomitant treatment: not permitted
Compliance to treatment: not undertaken
Previous treatment: unsure
Risk of bias
Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) High risk The trial did not specify the method of sequence generation
Allocation concealment (selection bias) High risk The trial did not specify the method of allocation concealment
Incomplete outcome data (attrition bias)
All outcomes High risk 7/25 participants in the active group withdrew, and 11/25 in the placebo group withdrew. The trial undertook ITT analysis, but dropouts were not balanced between the groups
Selective reporting (reporting bias) Low risk The trial reported all prespecified outcomes
Other bias Unclear risk No detail of funding or support was listed
Blinding of participants and personnel (performance bias)
All outcomes Unclear risk This was adequate for participants, but it was unclear if they blinded the clinician outcome assessor

Don 2003.

Study characteristics
Methods Design: double‐blind, randomised, placebo‐controlled, parallel
Duration: 12 weeks
Interval of evaluation: 4 weeks and 12 weeks
Participants Number randomised: 20 (10 each in the active and placebo groups)
Sex (M/F): 8/1 in the active group and 5/4 in the placebo group
Age of participants: 20 children aged 7 to 48 months (mean age = 22.9)
Unit of allocation: whole person
Country and setting: Italy, 1 centre
Duration of eczema: placebo = 9.44 years, borage = 9.66 years (averages)
Inclusion criteria of the study

  • Severity of condition: SCORAD figures at start = 39.44 for borage group, 44.93 for placebo group

  • Diagnostic criteria: Hanifin Rajka's criteria


Exclusion criteria of the study

  • Not specified


Number: 20
Interventions

  • Treatment group: 60 drops daily of BO

  • Control group: 60 drops daily of olive oil
Outcomes

  1. SCORAD (method of assessment: standard method)

  2. Quality of life (parents' evaluation) (method of assessment: scores of calm = 4, slightly distressed = 3, moderately distressed = 2, very distressed = 1)
Notes Concomittent treatment: allowed continued emollients and low‐dose steroids
Compliance to treatment: undertaken by parental information and serum level changes
Risk of bias
Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Low risk Quote (page 428): "Patients were assigned to two groups on the basis of a randomised list created by suitable statistical software"
Allocation concealment (selection bias) Low risk The BO group was identified by letter A; the olive oil (placebo) group, by letter B. Investigators were unaware of this classification before and during the study. Comment: It is unclear if allocation was adequately concealed
Incomplete outcome data (attrition bias)
All outcomes Low risk 9/10 participants in the borage oil group completed the study, and 9/10 in the olive oil group completed the study
Selective reporting (reporting bias) Low risk The trial reported all prespecified outcomes
Other bias Low risk We contacted the author; there was no other bias
Blinding of participants and personnel (performance bias)
All outcomes Unclear risk Oils were dispensed in dark unlabelled bottles. Interventions had a different taste, smell, and colour
Quote: "None of the investigators was aware of (...the A vs B...) classification before or during the study." (Methods section)
Comment: Attempts were made to blind both participants and personnel, but it was unclear if this was adequate

Finlay 2001.

Study characteristics
Methods Design: double‐blind, placebo‐controlled, randomised
Duraton: 3 months
Interval of assessment: 3 weeks
Participants Number randomised: 60 (30 each in the active and placebo groups)
Sex (M/F): not stated
Age of participants: 5 to 12 years for both groups
Unit of allocation: whole person
Country and setting: UK, 12 sites
Duration of condition: problems for over a year, some intermittent
Inclusion criteria of the study

  • Children under 10 years of age from 12 sites in UK (1 to 17 cases per site) with AD

  • Diagnostic criteria: UK working party criteria and requiring treatment (though ~24% used none during the 30 days prior to joining)

  • Severity of condition: not specified but stated equal in both groups


Exclusion criteria of the study

  • Any other itch condition

  • Active infection

  • Use of phenothiazines

  • Severe illness

  • Seizures

  • Use of anything stronger than 1% hydrocortisone
Interventions

  • Treatment group: 8 capsules daily each with 40 mg EPO and vitamin E

  • Placebo group: 8 capsules daily each with 40 mg coconut oil and vitamin E
Outcomes

  1. Pruritis by physician (method: VAS 0 to 100, where 0 = none and 100 = severe)

  2. Itch, erythema, edema, dryness, infection, scale, overall by physician (method: VAS 0 to 100, where 0 = none and 100 = severe)

  3. Overall by medical doctor (method: 1, 2, 3 = mild, moderate, severe)

  4. Quality of life (method: DLQI)

  5. Hydrocortisone use (method: weight of tubes)

  6. Area (method: rule of 9)
Notes Pretreatment: stated needed treatment, but 25% not doing any treatment
Compliance: stated and undertaken
The results were evaluated ‐ by other individuals ‐ 6 years after the study completed
Risk of bias
Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Low risk The trial used block randomisation (blocks of 4). Searle held the randomisation code, which was not identified to the investigators unless in an emergency
Allocation concealment (selection bias) Low risk The trial used sealed envelopes
Incomplete outcome data (attrition bias)
All outcomes Low risk 2/30 participants in the EPO group withdrew due to adverse events, as did 2/30 in the placebo group
Selective reporting (reporting bias) Low risk The trial reported all prespecified outcomes
Other bias Low risk There was no other bias
Blinding of participants and personnel (performance bias)
All outcomes Low risk There was adequate blinding of both investigators and participants

France 1988.

Study characteristics
Methods Design: double‐blind, placebo‐controlled, randomised, parallel
Duraton: 16 weeks
Interval of assessment: 4 weeks
Participants Number randomised: 45 (25 in the active group and 20 in the placebo group)
Sex (M/F): numbers for sites were not recorded
Age of participants: not recorded (page 24)
Unit of allocation: whole person
Country and setting: UK in 10 centres
Inclusion criteria of the study

  • Either sex

  • Age 5 to 75 years

  • Written informed consent

  • Diagnosed with severe atopic eczema, as defined by extensive body involvement with a regular need for Betnovate or equivalent steroid, frequent need for more potent steroid or for systemic steroids, or both; cytotoxic therapy; combined with other measures (page 13)

  • Must begin out of hospital, continue into hospital. May start immediately after discharge from a hospital

  • Diagnostic criteria: not specified


Exclusion criteria of the study

  • Other itch cause, inflammatory, malignant, severe (liver/renal)

  • Pregnant/lactating

  • Needing cytotoxic currently
Interventions

  • Treatment group: EPO, 12 capsules daily, each capsule 500 mg

  • Placebo group: paraffin oil, 12 daily, each capsule 500 mg
Outcomes

  1. Participant perception of itch (method: VAS, from none to worst ever)

  2. Other Skin condition, such as itch, dryness, etc, by participant (method of assessment: VAS)
Notes Previous treatment not stopped
Concomittant treatment permitted
Assessment of compliance undertaken but incomplete
The results were evaluated ‐ by other individuals ‐ 6 years after the study completed
Risk of bias
Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Low risk The trial used computer‐generated block randomisation
Allocation concealment (selection bias) Low risk A Securitainer® was used for each participant and at each visit
Incomplete outcome data (attrition bias)
All outcomes Unclear risk Of 25 participants in the active group, 4 withdrew and 3 were lost to follow up. Of 20 participants in the placebo group, 2 withdrew and 2 were lost to follow up
Selective reporting (reporting bias) Unclear risk Quote: "Primary and secondary efficacy variables not stated in the study..."
Other bias High risk Case report forms (CRFs) varied at different sites. Ages were not recorded (page 24)
The results were evaluated 6 years after study completion and evaluated by other individuals
Quote (page 4): "Efficacy variables were not stated in the protocol...for the purpose of this report the patient assessment of itching was the primary efficacy variable"
3 participants were included in analyses that were not accounted for in the randomisation schedule (page 14)
Blinding of participants and personnel (performance bias)
All outcomes Low risk This was adequate

Fraser 2001.

Study characteristics
Methods Design: randomised, double‐blind, parallel, placebo‐controlled
Duration: 3 months
Interval of assessment: 3 weeks
Unit of randomisation: whole person
Unit of analysis: whole person
Duration of trial: 12 weeks
Interval of assessment: 4 weeks
Participants Number randomised: 90 (43 in the active group and 47 in the placebo group)
Sex (M/F): 17/28 in the active group and 23/24 in the placebo group
Age of participants: 10 to 29 years old in the majority; both groups had 4 participants under 10 years old
Country and setting; UK, 4 sites
Duration of condition: average duration = 20.8 years for Efamol® and 16.7 years for placebo
Inclusion criteria of the study

  • Patients over the age of 3, attending dermatology clinic with a "history of signs and symptoms suggestive of a diagnosis of atopic eczema"

  • Diagnostic criteria: not stated

  • Severity of condition: moderate: 44.4% Efamol®, 57.9 % placebo group, rest not stated


Exclusion criteria of the study

  • Taking systemic steroids, NSAIDS, beta blockers, or phenothiazine

  • Having Epilepsy
Interventions

  • Treatment group: 8 capsules daily, each with 500 mg EPO

  • Placebo group: 8 capsules daily, each with 500 mg Olive oil
Outcomes

  1. Participant assessment of itch (method: VAS 1 to 100, where 0 = none and 100 = worst ever)

  2. All other measures by doctor and participant: redness, scale, dryness, and overall improvement (method: VAS, as above)
Notes Pretreatment in both groups: 4 weeks' placebo
The results were evaluated ‐ by other individuals ‐ 6 years after the study completed
Risk of bias
Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Unclear risk The trial was randomised, but did not mention a method of generation
Allocation concealment (selection bias) Unclear risk There was no statement about allocation concealment
Incomplete outcome data (attrition bias)
All outcomes Low risk No participants dropped out in the EPO group. 4 participants dropped out in the placebo group
Selective reporting (reporting bias) Low risk There was no selective reporting
Other bias Low risk There was no other bias
Blinding of participants and personnel (performance bias)
All outcomes Unclear risk This was stated, but details were not specified

Harper 1990.

Study characteristics
Methods Design: randomised, double‐blind, placebo‐controlled, parallel
Duration of trial: 16 weeks
Interval of assessment: first weekly, then 4 weeks
Participants Number randomised: 20 (9 in the active group and 11 in the placebo group)
Sex (M/F): 1/8 in the active group and 4/7 in the placebo group
Age of participants: active group = 7.0, placebo group = 4.3
Unit of analysis: whole person
Country and setting: patients from single centre dermatology in UK, no other specifications
Inclusion criteria of the study

  • "Moderate to severe atopic eczema requiring constant treatment with mild to moderate potent steroids, emollients or antihistamines"

  • Aged 6 months to 12 years

  • No trial treatment to be started in hospital setting, none to be discontinued, if hospitalised. May enter study immediately after leaving hospital

  • Diagnostic criteria: clinical


Exclusion criteria of the study

  • Other cause of itch

  • Any other inflammatory disorder

  • Any severe intercurrent illness, including renal liver failure or cancer

  • Epilepsy

  • Phenothiazines
Interventions

  • Treatment (active) group:

    • 8 capsules daily, each with 500 mg EPO, for children between 6 months to 2 years;

    • 12 capsules daily, each with 500 mg EPO, for children between 2 years to 12 years

  • Placebo group: capsules each with 500 mg olive oil, divided into 2 groups as above


Duration: 16 weeks' treatment and 8 weeks' off therapy
Outcomes

  1. Parent evaluation of itch (method: VAS, from none to worst ever)

  2. Parental assessments of the participant's skin dryness, scaling, redness, and overall impression (method of assessment: VAS)

  3. Dermatologist assessments of the patient's skin dryness, scaling, redness, and overall impression (method of assessment: VAS)

  4. Adverse events
Notes Previous treatment not stopped ("no limitations regarding other treatment were required")
Asessment of compliance: "Insufficient data were available to evaluate compliance"
Compliance treatment of the children's eczema was permitted to continue as usual
The results were evaluated ‐ by other individuals ‐ 6 years after the study completed
Risk of bias
Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Low risk The trial used computer‐generated randomisation
Allocation concealment (selection bias) Low risk Quote (page 15): "A listing of the randomization schedule and code is not available. Appendix 4 contains a listing of children with their randomised assigned treatment"
Comment: This was adequate
Incomplete outcome data (attrition bias)
All outcomes Low risk The trial undertook ITT analysis
Selective reporting (reporting bias) Low risk There was no selective reporting
Other bias High risk Pharmacia sponsored the trial
Blinding of participants and personnel (performance bias)
All outcomes Low risk Quote (page 15): "Securicontainer #1 was given out by physician or pharmacist to the first patient entered in the trial"
Comment: This was adequate

Heddle 1990.

Study characteristics
Methods Design: randomised, double‐blind, placebo‐controlled, single‐centre, cross‐over
Unit of randomisation: whole person
Unit of analysis: whole person
Duration: There was a 4‐week run‐in period during which all participants received olive oil. Participants then received treatment with either EPO or safflower oil (SAFF) for 8 weeks (period 1)
Following a wash‐out period of 4 weeks, during which participants again received olive oil, participants received the opposite treatment to that which they had received in period 1 for 8 weeks (period) (page 4)
"Number of patients (planned and analysed): Planned: not known. Enrolled and analysed: 25 patients." (page 3)
Use of concomitant treatment: not specified/mentioned
Previous treatment stopped/continued: not specified
Participants Number randomised: 25 (12 received EPO first and 13 received SAFF first)
Sex (M/F): not specified
Age of participants: 8 years (mean)
Unit of allocation/randomisation/analysis: whole person
Country/setting: participants from single centre in UK, no other specifications
Inclusion criteria of the study

  • Although no inclusion criteria are listed in the final report, we choose to believe that a physician made the diagnosis of eczema clinically and that was used as an inclusion criteria

  • Diagnosis: "dermatologist's assessment"

  • Severity of condition: not specified


Exclusion criteria of the study

  • Not specified
Interventions

  • Treatment group: 500 mg EPO in each capsule, total dose 1000 mg/kg body weight/per day

  • Placebo group: 500 mg olive in each capsule, total dose 1000 mg/kg body weight/per day


"Each capsule had 20 mg/gm vitamin E as an antioxidant"
Follow‐up: treatment schedule = 8 weeks' treatment, 4 weeks' wash‐out, then 2 groups switched over for the last 8 weeks
Outcomes Timing of outcome assessment: baseline and 8 weeks for each period of treatment (periods 1 and 2)
Primary outcomes

  1. Participant perception of itch (method: VAS, from none to worst ever)

  2. Participant/parent assessments of sleep loss due to itch and changes in general health and eczema (method of assessment: VAS)

  3. Dermatologist assessments of the participant's skin dryness, scaling, redness, and overall impression of all symptoms and signs of AD (method of assessment: VAS)


Secondary outcomes

  1. "EFA levels and cellular determinations were considered secondary efficacy endpoints"

  2. Asessment of compliance undertaken: "At each clinic visit patients returned all unused capsules. The returned capsules were used to assess treatment compliance"


Adverse events: none listed
Notes Previous treatment stopped/continued: not specified
Asessment of compliance undertaken: "At each clinic visit patients returned all unused capsules. The returned capsules were used to assess treatment compliance"
Concommitant treatment permitted/not permitted: not specified
The sponsor was contacted and had no further information about the investigator or study
The results were evaluated ‐ by other individuals ‐ 6 years after the study completed
Risk of bias
Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Unclear risk A randomisation table was given, but the method of sequence generation was not stated
Quote: "This was a single center, double‐blind, randomised, placebo‐controlled, crossover study of EPO versus safflower oil (SAFF, Placebo)"
Allocation concealment (selection bias) Unclear risk The trial did not mention a method of concealment
Incomplete outcome data (attrition bias)
All outcomes Low risk All (25/25) participants completed, and an ITT model was used
Selective reporting (reporting bias) Low risk The trial reported all prespecified outcomes
Other bias High risk The final write‐up and statistical evaluation was done at least 6 years after the last participant was in the study
Blinding of participants and personnel (performance bias)
All outcomes Unclear risk The study did not state the method, but it was performed under double‐blind conditions

Hederos 1996.

Study characteristics
Methods Design: randomised, double‐blind, parallel
Duration: 16‐week
Interval of assessment: 4‐week
Participants Number randomised: 60 Active (30 each in the active and placebo groups)
Sex (M/F): 13/17 in the active group and 13/17 in the placebo group
Age of participants: active group = 7.5, placebo group = 8.6
Unit of allocation: whole person
Country and setting: Karlstad, Sweden, paediatric department of central hospital ‐ 2 clinics
Inclusion criteria of the study

  • Outpatients

  • Either sex

  • 1 to 16 years of age

  • Diagnosis: Hanifin and Rajka


Exclusion criteria of the study

  • Not specified
Interventions

  • Treatment group:

    • < 12 years, = 8 Epogam® capsules daily (each containing 500 mg evening primrose oil providing 40 mg GLA with 10 mg vitamin E)

    • > 12 years = 12 Epogam® capsules daily (same composition)

  • Placebo group: same number of identical gelatin capsules as for the treatment group according to age (containing 500 mg of sunflower oil with 10 mg vitamin E)
Outcomes

  1. Clinician rating of redness, dryness, crusts, excoriation, itch, scaling, lichenification, fidget, overall impression of condition, area of involvement (method of assessment: VAS, scale = 0 to 10)

  2. Global assessment by parents (method: diary booklet)

  3. Total IgE and essential fatty acids (laboratory)

  4. Skin prick test with allergens (laboratory)

  5. DGLA levels (laboratory)
Notes N/A
Risk of bias
Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Low risk Quote (page 484): "A randomisation list was provided that allocated patients in blocks of two"
Allocation concealment (selection bias) Unclear risk No details of concealment given
Incomplete outcome data (attrition bias)
All outcomes Low risk 2/30 participants in the EPO group withdrew before the end of therapy, as did 0/30 in the placebo group
Selective reporting (reporting bias) Low risk The trial reported all prespecified outcomes
Other bias Low risk There was no other bias
Blinding of participants and personnel (performance bias)
All outcomes Low risk Quote (page 494): "The study was a double blind, randomised, placebo controlled trial"
Comment: Participants were blinded as "identical" capsules were given

Henz 1999.

Study characteristics
Methods Design: randomised, parallel, double‐blind, placebo‐controlled
Duration: ‐
Interval of assessment: ‐
Unit of randomisation: whole person
Unit of analysis: whole person
Duration of trial: 24 weeks
Interval of assessment: at 1, 2, 6, 12, 18, and 24 weeks
Participants Number randomised: 160 (80 each in the active and placebo groups)
Sex (M/F): 66% women in the borage oil group and 54% women in the miglyol group
Age of participants: 14 to 65 years
Unit of allocation: whole person
Country and setting: Switzerland, multicentre
Duration of condition: average duration = 22 years for BO and 20 years for miglyol
Inclusion criteria of the study

  • Severity of condition: moderate

  • Diagnostic criteria: Hanifin Rajka's criteria


Exclusion criteria of the study

  • Not specified
Interventions

  • Treatment group: BO, 6 capsules per day, each capsule 500 mg

  • Placebo group: miglyol, 6 capsules per day, each capsule 500 mg
Outcomes

  1. Improvement of individual eczema symptoms

  2. Method of assessment by 50% reduction in Costa score
Notes Previous treatment: stopped at the beginning of study
Concommitant treatment: allowed and assessed
Asessment of compliance undertaken by measuring blood levels of drug and use of corticosteroids
"The primary efficacy of the treatment was assessed by counting the number of patients achieving a 50% reduction in Costa score (as defined as clinical response) as a function of corticosteroid use"
Risk of bias
Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Unclear risk This was provided from external sources, which were not specified
Allocation concealment (selection bias) Unclear risk This was unclear
Incomplete outcome data (attrition bias)
All outcomes Unclear risk This was unclear
Selective reporting (reporting bias) Unclear risk This was unclear
Other bias Unclear risk This was unclear
Blinding of participants and personnel (performance bias)
All outcomes Unclear risk This was unclear

Kenicer 2001.

Study characteristics
Methods Design: randomised, double‐blind, placebo‐controlled
Duration: 16 weeks
Interval of assessment: start, 4 weeks
Participants Number randomised: 66 (32 in the active group and 34 in the placebo group)
Sex (M/F): both
Age of participants: between 10 and 70 years of age
Unit of allocation: whole body
Country and setting: UK at Ninewells Hospital, Dundee or Hull Royal Infirmary
Inclusion criteria of the study

  • Diagnosis: satisfying previously defined and published criteria for atopic eczema (Hanifin 1980)

  • Informed consent

  • Active disease

  • No severe intercurrent disease


Exclusion criteria of the study

  • Not pregnant, lactating

  • Child‐bearing age and not on contraception

  • Non‐compliant

  • Epileptic

  • Taking phenothiazines or immunosuppressive drugs

  • Receiving UV treatments
Interventions

  • Treatment group: 8 capsules daily, with 400 mg GLA and 11 mg vitamin A in each capsule

  • Placebo group: 8 capsules daily, with 500 mg olive oil in each capsule
Outcomes

  1. Participant assessment of itch

  2. Participant assessment of dermatology (global score)

  3. Physician assessment of dermatology (global score)

  4. Physician assessment of redness (method of assessment: all participant and dermatological assessment were done on a VAS and scoring system used as 0 = no lesion to 100 = worse lesion)
Notes Concomittent treatment: previous treatment with moderate to potent topical steroids, emollients continued for first 8 weeks of study, or both Next 8 weeks' steroids discontinued
Compliance to treatment: Although the protocol called for pill counting and cream weighing to determine use of treatments, the final report did not include any evaluation of these measures
Quote (page 29): "34% of Active and 32% of placebo participants did not satisfy the inclusion criterion"
The results were evaluated ‐ by other individuals ‐ 6 years after the study completed
There were a large number of dropouts
Risk of bias
Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Low risk Randomisation was by computer‐generated sequence in blocks of 4, which was adequate
Allocation concealment (selection bias) Low risk This was adequate
Incomplete outcome data (attrition bias)
All outcomes Low risk The trial undertook ITT analysis
Selective reporting (reporting bias) Low risk There was no selective reporting
Other bias High risk Quote (page 29): "34% of Active and 32% of placebo participants did not satisfy the inclusion criterion"
Quote (page 3): "For the purpose of this report the primary and secondary objectives have been re‐defined"
The results were evaluated 6 years after the completion of the study; other individuals evaluated the results
Blinding of participants and personnel (performance bias)
All outcomes High risk Quote (page 19): "There was no evidence in the protocol that the study was performed under blind conditions"

Kiehl 1994.

Study characteristics
Methods Design: double‐blind, placebo‐controlled
Duration of trial: 3 months
Interval of assessment: at baseline, after 4 weeks, and at end of study (12 weeks)
Participants Number randomised: not reported
Sex (M/F): not reported
Age of participants: not reported
Unit of allocation: whole person
Country and setting: single centre in Germany
Duration of disease: not reported
Inclusion criteria of the study

  • Mild to moderate eczema

  • Diagnostic criteria: clinical

  • Severity of disease: not reported


Exclusion criteria of the study

  • If participants were not corticoid‐free 1 month before start of the trial, they were excluded

  • Participants using corticosteroids during trial
Interventions

  • Treatment: 6 capsules twice daily for 4 weeks, each with 40 mg of glandol, followed by 3 capsules twice daily for 10 days

  • Placebo group: 4 capsules daily, each with 480 mg GLA for 4 weeks, followed by 2 capsules daily for children 1 to 3 years of age, and 3 capsules daily for children 4 to 12 years of age
Outcomes

  1. Immune profile (method of assessment: blood)

  2. ADASI score (method: not reported)

  3. Plasma concentrations of linolenic acid (method of assessment: blood)

  4. Plasma gamma‐linolenic acid concentration (method of assessment: blood)
Notes Previous treatment stopped for 4 weeks
Concomittent treatment: antihistaminics permitted
Compliance to treatment: unstated
Risk of bias
Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Unclear risk The trial did not state the method of sequence generation
Allocation concealment (selection bias) Unclear risk The paper stated "double‐blind", but did not specify the method
Incomplete outcome data (attrition bias)
All outcomes Low risk There were no dropouts (39/39 in the glandol group; 23/23 in the EPO group)
Selective reporting (reporting bias) Unclear risk The trial reported all prespecified outcomes
Other bias Low risk There was no other bias
Blinding of participants and personnel (performance bias)
All outcomes Low risk Both participants and personnel were blinded; it is unclear if the outcome assessors were blinded
Comment: This was probably done

Kovar 1992.

Study characteristics
Methods Design: 2 separate phases: first phase was double‐blind and placebo‐controlled with a trial duration of 8 weeks; the second phase was single‐blind ‐ with no placebo control ‐ with a trial duration of 8 weeks
Duration: 8 weeks each phase
Interval of assessment: 4 weeks by physician and weekly by parent
Participants Number randomised: 48 (24 each in the active and placebo groups)
Sex (M/F): 16/8 in the active group and 12/12 in the placebo group
Age of participants: active group = 1.9 years, placebo group = 1.7 years
Duration of eczema before trial: 1.3 years for both groups
Unit of allocation: whole person
Country and setting: single centre in Germany
Inclusion criteria of the study

  • Children of either sex

  • Children up to 5 years of age

  • Children with moderate to severe atopic eczema requiring constant treatment with mild to moderately potent topical steroids, emollients, antihistamines, or a combination of the aforementioned

  • Diagnostic criteria: clinical, Hanifin


Exclusion criteria of the study

  • Children with pruritus from any other cause

  • Children with an inflammatory skin disorder other than atopic eczema

  • Children with systemic inflammatory disorders.

  • Children with severe intercurrent illnesses, including liver failure, renal failure, or malignancy

  • Children with epilepsy or on phenothiazines
Interventions

  • Treatment group: 8 capsules per day, each containing 500 mg of EPO (total dose of 4000 mg per day)

  • Control group: 8 capsules per day, each containing 500 mg of paraffin oil (total dose of 4000 mg per day)
Outcomes

  1. Parent assessment of itching (for assessment, the status of itch was recorded by placing a vertical mark in the diary card on a 100 mm VAS), ranging from none/no symptoms to worst ever for itch and from worst ever to best ever for overall impression

  2. Parent assessment of overall symptoms and clinical picture

  3. Physician assessment of itch, redness, and overall symptoms (method used again was VAS ‐ 2 types: 1 as by parent, other as mild, moderate, or severe)

  4. Laboratory assessments of IgE, RAST (radioallergosorbent test) essential fatty acid (EFAs) in plasma phospholipids and in red blood cells

  5. Profilometric (surface roughness or texture) measurements of silicon casts were performed
Notes The results were evaluated ‐ by other individuals ‐ 6 years after the study completed
Risk of bias
Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Low risk Quote: "Computerized, block method for the 8 week double blind, parallel, placebo controlled study"
Allocation concealment (selection bias) Low risk Codes were kept in sealed envelopes, which were only to be opened if needed
Incomplete outcome data (attrition bias)
All outcomes Low risk There were no withdrawals. 3/24 participants in the active group were lost to follow up, as were 2/24 in the placebo group
Selective reporting (reporting bias) Unclear risk This was unclear
Other bias High risk Quote (page 22): "No statistical methodology was planned in the protocol..."
Comment: A statistical analysis plan was developed in June 2001 (some years after the study was completed)
Quote (page 23): "A primary objective for this study was not stated within the protocol"
Blinding of participants and personnel (performance bias)
All outcomes Low risk This was adequate: Pill appearance was the same, and the medication was dispensed in a blind fashion

Lovell 1981.

Study characteristics
Methods Design: randomised, double‐blind, placebo‐controlled, cross‐over
Duration: 6 weeks (2 periods of 3 weeks each)
Interval of assessment: 4 weeks
Participants Number randomised: 64 (32 each in the active and placebo groups)
Sex (M/F): not listed
Age of participants: 17 children = 3 months to 13 years, 15 adults = 24 to 32 years
Unit of allocation: whole person
Country and setting: Bistol, United Kingdom; single centre
Duration of eczema: 6 months and more
Inclusion criteria of the study

  • Diagnosis: "had atopic dermatis"

  • Severity of condition: not stated


Exclusion criteria of the study

  • Not specified
Interventions

  • Treatment group: 8 capsules daily for adult and 4 capsules daily for child, each with 500 mg EPO

  • Placebo group: 8 capsules daily for adult and 4 capsules daily for child, each with 500 mg paraffin oil
Outcomes

  1. Severity of eczema rated by physician (method of assessment: severity as measured on a continuous 10 cm linear scale ranging from no eczema (zero) to most severe (10 cm))

  2. Severity of eczema rated by participant (method of assessment: severity as measured on a continuous 10 cm linear scale ranging from no eczema (zero) to most severe (10 cm))
Notes Previous treatment: unsure
Concommitant treatment: mild topical steroids and emollient allowed
Asessment of compliance: not undertaken
No power calculation
Risk of bias
Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Low risk Quote (page 278): "...received on a randomised, double‐blind crossover basis..."
Allocation concealment (selection bias) Unclear risk The paper did not state this (a short 5‐paragraph report)
Incomplete outcome data (attrition bias)
All outcomes Low risk All randomised participants were included in the results
Selective reporting (reporting bias) Unclear risk This was unclear
Other bias High risk 1 author (DFH) worked full time for the company. We contacted the author
Blinding of participants and personnel (performance bias)
All outcomes Low risk The first phase was double‐blind
The second phase was only participant‐blinded

MacKie Adult 1990.

Study characteristics
Methods Design: randomised, double‐blind, placebo‐controlled
Duration: 24 weeks (16 weeks' therapy and 8 weeks off therapy)
Interval of assessment: 4 weeks
Participants Number randomised: 20 (10 each in the active and placebo groups)
Sex (M/F): 11/9 total (5/5 in the active group and 6/4 in the placebo group)
Age of participants: active group = 26.8, placebo group = 25.0 (means)
Unit of allocation: whole body
Country and setting: Glasgow, UK; single dermatology centre
Inclusion criteria of the study

  • Diagnosis: patients with "moderate to severe atopic eczema, as defined by extensive body involvement with a regular need for moderately potent or potent steroids (MIMS (Monthly Index of Medical Specialties) classification) or frequent need for very potent steroid or for systemic steroids, cytotoxic therapy, combined with other measures, particularly antipruritic agents and antibiotics as require" (page 15)

  • Participants of either sex

  • Participants between the ages of 12 and 75 years

  • Outpatients. No one could start in the hospital; no treatment was to be stopped if going into hospital. On hospital discharge they could immediately enter the study


Exclusion criteria of the study

  • Pruritis of any other cause

  • Any other inflammatory skin disorder or any systemic disorder

  • Any severe intercurrent illness, including liver failure, renal failure, malignancy

  • Participants pregnant or trying to conceive

  • Participants with epilepsy

  • Participants on phenothiazines
Interventions

  1. Treatment group: 8 capsules daily, each with 40 mg GLA and 11 mg vitamin E

  2. Placebo group: 8 capsules daily, each with 500 mg olive oil
Outcomes

  1. Participant assessment of itching (method of assessment: VAS 0 to 100; none to worst ever)

  2. Participant assessment of all other signs and symptoms of eczema except itch (method of assessment: VAS)

  3. Skin assessments by the dermatologist (method of assessment: Glasgow method and skin profilimetry)

  4. EFA levels

  5. Adverse events
Notes Asessment of compliance: "There were no details relating to any assessments of compliance in the protocol. All unused capsules were returned to Scotia Pharmaceuticals at the end of the study"
Continued or new concomitant treatment of eczema was permitted
The results were evaluated ‐ by other individuals ‐ 6 years after the study completed
Risk of bias
Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Low risk Randomisation was computer‐generated
Allocation concealment (selection bias) Low risk Allocation concealment was done through sealed coded envelopes
Incomplete outcome data (attrition bias)
All outcomes Low risk This was low risk
Selective reporting (reporting bias) Low risk This was low risk
Other bias High risk The results were evaluated 6 years after the completion of the study; other individuals evaluated the results
Blinding of participants and personnel (performance bias)
All outcomes Low risk Blinding was adequate

MacKie Child 1990.

Study characteristics
Methods Design: double‐blind, randomised, placebo‐controlled
Duration: 24 weeks (16 weeks' intervention, 8 weeks off treatment)
Interval of assessment: 4 weeks
Participants Number randomised: 19 (10 in the active group and 9 in the placebo group)
Sex (M/F): 10/10 total (3/7 in the active group and 7/3 in the placebo group)
Age of participants: active group = 6.2, placebo group = 5.8 (means)
Unit of allocation: whole person
Country and setting: Glasgow, UK; single dermatology centre
Inclusion criteria of the study

  • Participants of either sex

  • Participants < 12 years

  • Participants with moderate to severe atopic eczema, as defined by extensive body involvement

  • Diagnosis: severity of condition: moderate to severe eczema defined as extensive body involvement with regular need for moderately potent or potent steroid or frequent need for systemic steroids, cytotoxic therapy, combined with other measures, particularly antipruritic agents or antibiotics as required


Exclusion criteria of the study

  • Pruritis of any other cause

  • Any other inflammatory skin disorder or any systemic disorder

  • Any severe intercurrent illness, including liver failure, renal failure, malignancy

  • Participants pregnant or trying to conceive

  • Participants with epilepsy

  • Participants on phenothiazines
Interventions

  1. Treatment group: 8 capsules daily, each with 40 mg GLA and 11 mg vitamin E

  2. Placebo group: 8 capsules daily, each with 500 mg olive oil
Outcomes

  1. Participant assessment of itching (method of assessment: VAS 0 to 100; none to worst ever)

  2. Participant assessment of all other signs and symptoms of eczema except itch (method of assessment: VAS)

  3. Skin assessments by the dermatologist (method of assessment: Glasgow method and skin profilimetry)

  4. EFA levels

  5. Adverse events (mild, moderate, severe)
Notes Previous treatment not stopped
Assessment of compliance: "There were no details relating to any assessments of compliance in the protocol. All unused capsules were returned to Scotia Pharmaceuticals at the end of the study"
Concommitant treatment of the children's eczema was permitted to continue as usual
The results were evaluated ‐ by other individuals ‐ 6 years after the study completed
Risk of bias
Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Low risk Randomisation was computer‐generated
Allocation concealment (selection bias) Low risk The trial used sealed envelopes
Incomplete outcome data (attrition bias)
All outcomes Low risk There were no withdrawals, but defaults: 4/10 in the active group and 2/9 in the placebo group
Selective reporting (reporting bias) Low risk This was low risk
Other bias High risk The results were evaluated 6 years after the completion of the study; other individuals evaluated the results. 1 child was included against protocol (age 12)
Blinding of participants and personnel (performance bias)
All outcomes Low risk This was adequately blinded

MacLeod 2001.

Study characteristics
Methods Design: randomised, double‐blind, placebo‐controlled
Duration: 16 weeks
Interval of assessment: 4 weeks
Participants Number randomised: 19 (10 in the active group and 9 in the placebo group)
Sex (M/F): 4/4 in the active group, 3/5 in the EPO/Marine group, and 6/2 in the placebo group
Age of participants: EPO group = 29.1, EPO/Marine group = 25.3, placebo group = 26.4 years
Unit of allocation: whole person
Country and setting: Glasgow, UK; single dermatology centre
Inclusion criteria of the study

  • Participants of either sex

  • Participants 16 to 30; and, yet, the range of participants' ages was from 17 to 35

  • Participants with moderate to severe atopic eczema, as defined by extensive body involvement

  • Diagnosis: severity of the condition: moderate to severe atopic eczema, as defined by extensive body involvement with a regular need for moderately potent or potent steroid or frequent need for systemic steroids; cytotoxic therapy, combined with other measures, particularly antipruritic agents and antibiotics as required


Exclusion criteria of the study

  • Pruritis of any other cause

  • Any other inflammatory skin disorder or any systemic disorder

  • Any severe intercurrent illness, including liver failure, renal failure, malignancy

  • Participants pregnant or trying to conceive

  • Participants with epilepsy

  • Participants on phenothiazines
Interventions

  • Treatment group ‐ 2 groups:

    • Efamol® ‐ 12 capsules daily, each with 40 mg GLA and 11 mg vitamin E

    • Efamol® ‐ marine, 12 capsules, each 500 mg, with 80% Efamol® and 20% fish oil

  • Placebo: 12 capsules daily each with 500 mg paraffin oil
Outcomes

  1. Participant assessment of itching (method of assessment: VAS 0 to 100; none to worst ever)

  2. Participant assessment of all other signs and symptoms of eczema except itch (method of assessment: VAS and diary)

  3. Dermatologist assessment of all other signs and symptoms of eczema (method of assessment: VAS)
Notes Prior and concomitant treatment: allowed
Compliance assessment: undertaken by counting the returned capsules
The results were evaluated ‐ by other individuals ‐ 6 years after the study completed
Risk of bias
Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Low risk A computer‐generated sequence was used
Allocation concealment (selection bias) Low risk The trial used sealed opaque envelopes
Incomplete outcome data (attrition bias)
All outcomes High risk There were 3/8 withdrawals in the active group, and there were 4/8 withdrawals in the placebo group
Selective reporting (reporting bias) Unclear risk This was unclear
Other bias High risk The results were evaluated 6 years after completion of the study; other individuals evaluated the results
Blinding of participants and personnel (performance bias)
All outcomes Low risk This was adequate

Schalin‐Karrila 1987.

Study characteristics
Methods Design: double‐blind, placebo‐controlled, parallel
Duration: 12 weeks
Interval of assessment: 3 weeks
Participants Number randomised: 25 (14 in the active group and 11 in the placebo group)
Sex (M/F): 9/16
Age of participants: 19 to 31 years
Unit of allocation: whole person
Country and setting: Finland, Turku Pediatrics single centre
Inclusion criteria of the study

  • Maintain normal diet

  • No other specifications

  • Diagnosis: clinical appearance. Many also had personal history of respiratory atopy with family history of atopy

  • Severity of condition: moderate to severe


Exclusion criteria of the study

  • Not specified
Interventions

  • Treatment group: 8 capsules daily, each with 500 mg of EPO

  • Placebo group: 8 capsules daily, each with 500 mg paraffin oil
Outcomes

  1. Consumption of emollient cream and topical steroids

  2. Overall severity and grade of inflammation

  3. Response to treatment (method of assessment of above 3 by discussion between doctor and participant)

  4. Fatty acid composition of plasma phospholipids (method of assessment was by laboratory testing)

  5. Blood levels of prostaglandins and thromboxane (method of assessment was by laboratory testing)
Notes Prior treatment: 2 weeks off topical steroids and systemic medications
Concomitant treatment: permitted
Compliance undertaken by recording of mild topical cortisone, antihistamines, emollient, and blood assays
Risk of bias
Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Unclear risk The trial was described as 'randomly divided', but the method was not stated
Allocation concealment (selection bias) High risk The method was not stated
Incomplete outcome data (attrition bias)
All outcomes Low risk 1 participant out of 14 in the EPO group dropped out due to allergy to concomitant medication (unrelated to EPO); there were no dropouts (out of 11 participants) in the placebo group
Selective reporting (reporting bias) Low risk The trial reported all prespecified outcomes
Other bias Low risk There was no other bias. A non‐profit foundation provided funding
Blinding of participants and personnel (performance bias)
All outcomes Unclear risk Quote (page 12): "...14 patients receiving evening primrose oil (EPO) and 11 patients receiving placebo, in a double‐blind trial. EPO was provided in capsules containing 360 mg linoleic acid, 50 mg oleic acid and 45 mg GLA (Efamol®). The placebo capsules contained 500 mg of liquid paraffin"
Comment: This was probably identical and adequate, but it was unclear if assessors were blinded (described as 'double blind')

Senapati 2008.

Study characteristics
Methods Design: randomised, double‐blind, parallel
Duration: 5 months
Interval of assessment: 4 weeks
Participants Number randomised: 50 (25 each in the active and placebo groups)
Sex (M/F): 11/14 in the active group and 7/18 in the placebo group
Age of participants: detailed to decade, similar in the active and placebo groups
Unit of allocation: whole person
Country and setting: Kolkata, India; single centre, referral hospital's outpatient department
Inclusion criteria of the study

  • Diagnostic criteria: Hanifin and Rajka

  • Severiity: "all degrees from mild to severe using 'Intensity Item Score Aggregate (IISA)"


Exclusion criteria of the study

  • Pregnant and lactating women

  • Epilepsy patients

  • Those with history of peptic ulceration, intake of phenothiazines

  • Patients who received UVB phototherapy or photochemotherapy in the last month

  • Patients who received systemic steroid or other immunosuppressive drugs in the last 3 months
Interventions

  • Treatment (active) group:

    • 1 to 4 capsules/day for participants up to 1 year of age;

    • 5 to 6 capsules/day for those aged 2 to 5 years;

    • 7 to 8 capsules/day for those aged 6 to 10 years;

    • 9 to 10 capsules/day for those 11 to 16 years;

    • 12 capsules/day for participants above 16 years of age (each capsule had 500 mg of EPO and vitamin E 10IU)

  • Placebo: dose the same way according to age groups as the treatment arm (each capsule had 300 mg of sunflower oil and vitamin E 10IU)
Outcomes

  1. Extent (method: extent of the disease: if only 1 area of predilection or less than 20% of body surface area (BSA) was involved, score was as follows: 1. If 2 to 3 areas or < 40% of BSA was involved, the participant scored 2. For any involvement more than the aforementioned, the score was 3)

  2. Intensity (method: individual scores from 6 separate clinical items ‐ erythema, edema/papulation, vesiculation/oozing/crusting, excoriation, scaling, and lichenification (each graded as 0, 1, 2, and 3 for absence, mild, moderate, and severe degrees) ‐ were summated to develop 'Intensity Item Score Aggregate (IISA)', which reflects the actual intensity of the disease. From this IISA, 'intensity score' was derived. For IISA 0, intensity score was 0. For IISA 1 to 6, score was 1. For IISA 7 to 12, score was 2)

  3. Itching (method: Itching was also graded as 0, 1, 2, and 3 for absence, mild, moderate, and severe degrees, respectively)

  4. Dryness (method: Dryness was evaluated only on uninvolved areas as follows: absence = score 0; mild = score 1; moderate = score 2; and, severe = score 3)

  5. Total score (method: adding all 4 major parameter scores. The case was labelled as mild, moderate, or severe when the total score ranged from 1 to 4, 5 to 8, and 9 to 12, respectively)
Notes Concomittent treatment: allowed continued emollients
Compliance to treatment: unsure
Risk of bias
Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Low risk Quote: "...using a random number table"
Comment: This trial did not specify the method of sequence generation
Allocation concealment (selection bias) High risk The paper did not specify this
Incomplete outcome data (attrition bias)
All outcomes Low risk This was low risk
Selective reporting (reporting bias) Low risk This was low risk
Other bias Low risk This was low risk
Blinding of participants and personnel (performance bias)
All outcomes Unclear risk The study title did not state this: "Evening primrose oil is effective in atopic dermatitis: A randomised placebo‐controlled trial". The study did not mention blinding

Takwale 2003.

Study characteristics
Methods Design: double‐blind, placebo‐controlled, parallel
Participants stratified into 2 groups of equal size: those aged up to 12 and those aged over 12
Duration: 12‐week
Interval of assessment: 2, 4, 8, and 12 weeks
Participants Number randomised: 140
Sex (M/F): not mentioned other than (page 1) "...evaluable population 140 (including 69 children)"
Age of participants: not listed
Unit of allocation: whole person
Country and setting: UK, acute district general hospital in Nuneaton, England; single centre
Inclusion criteria of the study

  • Participants of either sex aged over 2 years attending the hospital for treatment of atopic dermatitis

  • Diagnosis: Hanifin and Rajka criteria

  • Severity of condition: not specified


Exclusion criteria of the study

  • Pregant, lactating, and fertile women not using effective contraception
Interventions

  • Treatment groups:

    • 4 capsules of 500 mg of BO each for adults

    • 2 capsules of 500 mg of BO each for adults

  • Placebo group: same dosage as above with liquid paraffin oil for adults and olive oil for children
Outcomes

  1. Mean change in SASSAD score by physicians (method: SASSAD 6 signs (erythema, exudation, excoriation, dryness, cracking, and lichenification) at 6 sites (hands, feet, arms, legs, head and neck, and trunk). Each sign is graded at each site on a 4‐point scale (0 to 3, representing grades of none, mild, moderate, and severe)

  2. Overall assessment of treatment response by participant with 5‐point visual analogue scale: worse, same, improved, much improved, or cleared

  3. Assessment of tolerability by participant (4‐point scale: very good, good, fair, or poor)
Notes Concomitant treatment: allowed participants to use conventional treatment for atopic eczema throughout the study
Prior treatment: previous treatment continued, except for "potent" topical steroids, which were prohibited for children. Systemic antihistamines were allowed, but no other systemic treatment or ultraviolet light treatment was permitted
A minimum 4‐week wash‐out period was needed for participants who had received systemic steroids, psoralen plus ultraviolet A (PUVA), or other oral immunosuppressive treatment
Assessment of compliance: not specified
Risk of bias
Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Low risk Quote: "...using computer generated random numbers"
Allocation concealment (selection bias) Low risk The study used sealed containers
Incomplete outcome data (attrition bias)
All outcomes Low risk 1 participant withdrew, out of 14, in the active group due to severe allergic reaction; there were 0 dropouts in the placebo group
Selective reporting (reporting bias) Unclear risk ‐
Other bias Unclear risk ‐
Blinding of participants and personnel (performance bias)
All outcomes Unclear risk The study stated "blinded", with no details

Valsecchi 2006.

Study characteristics
Methods Design: randomised, placebo‐controlled, parallel
Duration: 16‐week
Interval of assessment: 4 weeks
Participants Number randomised: 31 (15 in the active group and 16 in the placebo group)
Sex (M/F): not mentioned
Age of participants: 18 children = 2 to 14 years, 13 adults = 15 to 38
Unit of allocation: whole person
Country and setting: Bergamo, Italy; single clinic
Inclusion criteria of the study

  • "Having active lesions" (page 77)

  • Diagnosis: Hanifin and Rajka

  • Severity of condition: mild to severe


Exclusion criteria of the study

  • Not taken systemic corticosteroids for the past 3 weeks
Interventions

  • Treatment group: 5 capsules daily, each with 500 mg EPO for adults, and 3 capsules daily each with 500 mg of borage for children

  • Placebo group: 5 capsules, each with 500 mg paraffin oil, number of capsules daily for adult and child same as treatment group
Outcomes

  1. Quantitative measurement, which included both area involved and severity of signs and symptoms of eczema


The Atherton area and severity scoring system was used to measure eczema symptoms and signs. Laboratory investigations were performed
Notes Previous treatment: if using systemic steroids, stopped at 3 weeks prior to inclusion in study. If using topical steroid, stopped at start Emollients continued
Compliance assessment: not specified
Concomitant treatment: only emollient
Risk of bias
Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Unclear risk The paper stated 'randomly divided', but no details were given
Allocation concealment (selection bias) Unclear risk The paper did not state the method of allocation concealment
Incomplete outcome data (attrition bias)
All outcomes Unclear risk 2/15 participants in the GLA group dropped out, and 1 participant, out of 16, dropped out in the placebo group; the reasons were unclear
Selective reporting (reporting bias) Low risk The trial reported all prespecified outcomes
Other bias Unclear risk There was no other bias
Blinding of participants and personnel (performance bias)
All outcomes High risk The trial report did not mention blinding

Wishart 1992.

Study characteristics
Methods Design: randomised, double‐blind, placebo‐controlled
Duration: 16 weeks
Interval of assessment: 4 weeks
Participants Number randomised: 75 (38 in the active group and 37 in the placebo group)
Sex (M/F): 43/32 total (22/16 in the active group and 21/16 in the placebo group)
Age of participants: active group = 25.5 years, placebo group = 22.0 years
Unit of allocation: whole person
Country and setting: New Zealand, multicentre
Inclusion criteria of the study

  • Participants of either sex

  • Participants between the ages of 16 and 75 years

  • Treatment was only to be commenced on outpatients. Participants who needed hospitalisations for treatment of eczema were not permitted to enter the study during their stay in the hospital. They were permitted to be entered into the study immediately upon discharge from hospital. If they were in the study, they were to continue the study regimen in hospital

  • Diagnosis: severity of the condition 'moderate to severe atopic eczema, as defined by extensive body involvement with a regular need for high potency or equivalent steroid, frequent need for more potent steroid (Dermovate or equivalent) or for systemic steroids, or both; cytotoxic therapy, combined with other measures, particularly antipruritic agents and antibiotics as required


Exclusion criteria of the study

  • Participants with pruritus from any other cause other than atopic eczema

  • Participants with an inflammatory skin disorder, other than atopic eczema, or other systemic inflammatory disorders

  • Participants with severe intercurrent illnesses, including liver failure, renal failure, malignancy

  • Participants who were pregnant or actively trying to conceive

  • Participants with epilepsy

  • Participants on phenothiazines
Interventions

  • Treatment group: 12 capsules of Efamol® daily, each with 40 mg GLA and 11 mg vitamin E

  • Placebo group: 12 capsules daily, each with 500 mg sunflower oil
Outcomes

  1. Dermatologist assessment of excoriation/itch (VAS)

  2. Dermatologist assessment of other parameters of eczema except itch, individually and overall impression (VAS)
Notes Previous treatment not stopped
Concommitant treatment of eczema was permitted to continue as usual
The results were evaluated ‐ by other individuals ‐ 6 years after the study completed
Risk of bias
Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Low risk Randomisation was computer‐generated
Allocation concealment (selection bias) Low risk The trial used sealed opaque envelopes
Incomplete outcome data (attrition bias)
All outcomes Unclear risk This was unclear
Selective reporting (reporting bias) Unclear risk This was unclear
Other bias High risk The results were evaluated 6 years after the completion of the study; other individuals evaluated the results
Blinding of participants and personnel (performance bias)
All outcomes Low risk The trial used Securitainers® for medication handling

Wright 1982.

Study characteristics
Methods Design: randomised, double‐blind, placebo‐controlled, cross‐over
Duration: 24 weeks (12 and 12)
Interval of assessment: 3 weeks
Participants Number randomised: 99 (60 in the active group and 39 in the placebo group)
Sex (M/F): not mentioned
Age of participants: 60 adults = 15 to 58 years, 39 children = 4 months to 14 years
Unit of allocation: whole person
Country and setting: UK, Bristol; single dermatology department
Inclusion criteria of the study

  • Diagnosis: clinical "atopic eczema" with personal or family history of atopic conditions and using mild topical steroid, emollients, and oral antihistamine

  • Severity: moderate or severe (no criteria)


"None was receiving potent topical steroids or systemic steroids" (page 1121)
Interventions

  • Adults = 3 groups:

    • 4 Efamol® capsules

    • 8 Efamol® capsules daily

    • 12 Efamol® capsules daily

  • Children = 2 groups:

    • 2 Efamol® capsules daily

    • 4 Efamol® capsules daily


Each Efamol® capsule had 500 mg of Efamol® (LA and GLA)
Each group had a corresponding placebo control taking the same number of capsules daily with 500 mg of liquid paraffin
Outcomes

  1. Doctor's assessment of degree of scaling, redness, and overall severity (method: 10 cm linear scale)

  2. Participant's assessment of same as above, plus severity of itch (method: 10 cm linear scale)
Notes Previous treatment not stopped
Asessment of compliance: unsure
Concommitant treatment of eczema was permitted to continue as usual
Author S Wright did not respond to email. Dr Burton said he had did not have the records and was retired
Risk of bias
Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Low risk The trial used block randomisation
Allocation concealment (selection bias) Unclear risk The paper did not describe allocation concealment
Incomplete outcome data (attrition bias)
All outcomes Low risk 8/60 participants in the active group dropped out, as did 8/39 in the placebo group
Selective reporting (reporting bias) Unclear risk This was unclear there was no ITT statement
Other bias Unclear risk This was unclear. We contacted the second author, who replied: "...retired, I do not know how to contact first author or [know the] whereabouts of records"
Blinding of participants and personnel (performance bias)
All outcomes Low risk This was adequate

Characteristics of excluded studies [ordered by study ID]

Study Reason for exclusion
Biagi 1988 This was an open study
Bordoni 1988 The paper did not give mean data, and the lead author did not reply to inquiry
Coskery 1988 This was an abstract only (no detailed data). The corresponding author had no details about the clinical course. It was a double‐blind, placebo‐controlled study (EPO/placebo), but not randomised. It measured the texture of skin
Courage 1991 This was not randomised
Guenther 1987 The results of the study were shredded (author, 2011). 17 children in a double‐blind cross‐over study of the sign of figiting in people with atopic dermatitis. There was no significant difference in atopic dermatitis between placebo and EPO using this measure
Melnick 1995 This had no placebo; both had the same amount of GLA/weight
NCT00878670 This was initially found as an RCT on www.clinicaltrials.gov. After contacting the author, we checked the web site again and learned that it was not an RCT but an open study. The study has now closed
Swapan 2008 Even though randomisation was stated, consecutive participant selection was carried out

Differences between protocol and review

We updated the Background section because of the date of protocol publication (2003).

As per CSG policy, we changed odds ratios to risk ratios in this review.

We used The Cochrane Collaboration's tool for assessing risk of bias (Higgins 2011), rather than quality assessment. (Our protocol specified quality assessment, but 'Risk of bias' assessment superceded this.)

We added the secondary outcome 'Systemic or topical treatments taken concurrently with EPO or BO', which was not in our published protocol, because we thought it was important as most of the included studies allowed continued use of topical steroids.

Although not prespecified by the original authors of the protocol, we anticipated it would be beneficial to the review to carry out some subgroup analyses where there were sufficient studies.

Contributions of authors

The following contributions were made by the authors stated.

JB updated the background section.
JB was the contact person with the editorial base, co‐ordinated contributions from the co‐authors, and wrote the final draft of the review.
JB and SR screened papers against eligibility criteria.
JB obtained data on ongoing and unpublished studies.
JB and SR appraised the quality of papers.
JB, SR, AM, and CvG extracted data for the review.
JB sought additional information about papers.
SR, AM, and JB entered data into RevMan.
AM analysed and interpreted data.
JB and SR worked on the methods sections.
JB drafted the clinical sections of the background and responded to the clinical comments of the referees.
JB, SR, AM, and RM responded to the methodology and statistics comments of the referees.
RH was the consumer co‐author and checked the review for readability and clarity, as well as ensuring outcomes are relevant to consumers.
JB, SR, AM, CvG, RH, and EE drafted and approved the review.
JB is the guarantor of the update.

Sources of support

Internal sources

  • Cathy Bennett, Other

External sources

  • No sources of support supplied

Declarations of interest

Joel Bamford's research was funded with a grant from the Miller‐Dwan Foundation and repaid by the Efamol Pharmaceutical company, which supplied medication, placebo, and blood testing. Joel Bamford was neither involved in the selection nor data extraction of the Bamford 1985 included study for which he is a first author. Instead, the third review author (AM) took up these roles in the case of this included study.

Edited (no change to conclusions)

References

References to studies included in this review

Bahmer 1992 {published data only (unpublished sought but not used)}

  1. Bahmer FA, Schafer J. Treatment of Atopic Dermatitis with Borage Seed Oil (Glandol) - a Time Series Analysis Study [Die Behandlung der atopischen Dermatitis mit Borretschsamen-Ol (Glandol)--eine zeitreihenanalytische Studie]. Kinderarztliche Praxis 1992;60(7):199-202. [MEDLINE: ] [PubMed] [Google Scholar]

Bamford 1985 {published and unpublished data}

  1. Bamford JT, Gibson RW, Renier CM. Atopic eczema unresponsive to evening primrose oil (linoleic and gammma-linolenic acids). Journal of the American Academy of Dermatology 1985;13(6):959-65. [PMID: ] [DOI] [PubMed] [Google Scholar]

Berth‐Jones 1993 {published data only}

  1. Berth-Jones J, Graham-Brown RA. Placebo-controlled trial of essential fatty acid supplementation in atopic dermatitis. Lancet 1993;341(8860):1557-60. [MEDLINE: ] [DOI] [PubMed] [Google Scholar]

Biagi 1994 {published data only (unpublished sought but not used)}

  1. Biagi PL, Bordoni A, Hrelia S, Celadon M, Ricci GP, Cannella V, et al. The effect of gamma-linolenic acid on clinical status, red cell fatty acid composition and membrane microviscosity in infants with atopic dermatitis. Drugs Under Experimental & Clinical Research 1994;20(2):77-84. [MEDLINE: ] [PubMed] [Google Scholar]

Borrek 1997 {published data only}

  1. Borrek S, Hildebrandt A, Forster J. Gamma-linolenic-acid-rich borage seed oil capsules in children with atopic dermatitis. A placebo-controlled double-blind study [Gammalinolensaure-reiche Borretschsamenol-Kapseln bei Kindern mit Atopischer Dermatitis. Eine placebo-kontrollierte Doppelblind-Studie]. Klinische Padiatrie 1997;209(3):100-4. [DOI] [PubMed] [Google Scholar]

Buslau 1996 {published data only}

  1. Buslau M, Thaci D. Atopic dermatitis: Borage oil for systemic therapy [Atopische Dermatitis: Borretschol zur systemischen Therapie]. Zeitschrift fur Dermatologie 1996;182(3):131-6. [Google Scholar]

Don 2003 {published data only (unpublished sought but not used)}

  1. Don M, Melli P, Braida F, Comici A, Barbi E, Spano C, et al. Efficacy of essential fatty acids in the treatment of atopic dermatitis and correlations of their changes with clinical response [Efficacia degli acidi grassi essenziali nel trattamento della dermatite atopica e correlazione de; loro cambiamenti con la risposta clinica]. Italian Journal of Pediatrics 2003;29(6):427-32. [Google Scholar]

Finlay 2001 {unpublished data only}

  1. Finlay AY, Marks R, Wilson AM, Huntley C. Epogam in Children with Atopic Eczema. Pharmacia (Formerly Searle Division of Monsanto) 2001:1-473.

France 1988 {unpublished data only}

  1. France et al. Epogam in Severe Atopic Eczema. Pharmacia 2001(18 October). [ISN 880002/3/4/24/25/128/130/890141]

Fraser 2001 {unpublished data only}

  1. Fraser. Efamol in Atopic Eczema. Pharmacia 2001 (22 October). [ISSN 86004/5/6/7]

Harper 1990 {unpublished data only}

  1. Harper J statician authors: Jaderberg M and Wilson A. Efamol in Moderate to Severe Atopic Eczema. Pharmacia 2001. [SN 890136/7]

Heddle 1990 {unpublished data only}

  1. Heddle RJ, Forsyth K, Gordon LA, TAO BK, Gibson RA. Oral Evening Primrose Oil (EPO) in Atopic Eczema. Pharmacia 2001. [ISN 910033]

Hederos 1996 {published data only}

  1. Hederos CA, Berg A. Epogam evening primrose oil treatment in atopic dermatitis and asthma. Archives of Diseases in Childhood 1996;75(6):494-7. [PMID 9014601] [DOI] [PMC free article] [PubMed] [Google Scholar]

Henz 1999 {published data only}

  1. Henz BM, Jablonska S, de Kerkhof PC, Stingl G, Blaszczyk M, Vandervalk PG, et al. Double-blind, multicenter analysis of the efficacy of borage oil in patients with atopic eczema. British Journal of Dermatology 1999;140(4):685-8. [MEDLINE: ] [DOI] [PubMed] [Google Scholar]

Kenicer 2001 {unpublished data only}

  1. Kenicer K, Keczkes K, Green C, George S. Evening Primrose Oil in the Treatment of Atopic Dermatitis: a Placebo-controlled Study. Pharmacia Company 2001. [ISN: 900210/910098]

Kiehl 1994 {published data only}

  1. Kiehl R, Ionescu G, Manuel P, Stern LP, Peters G, Niemann A, et al. Clinical, Immuno- and Lipid Modulating Effects of Unsaturated Fatty Acids in Patients with Atopic Eczema [Klinische, immun- und lipidmodulatorische effekte einer behandlung mit ungesattigten fettsauren bei atopischer dermatitis]. H+G Zeitschrift fur Hautkrankheiten 1994;69(1):42-8. [Google Scholar]

Kovar 1992 {unpublished data only}

  1. Kovar IZ, Wright S. An assessment of efamol evening primrose oil in moderate to severe atopic eczema in young children. Pharmacia Corporation 1992. [ISN 900158]

Lovell 1981 {published and unpublished data}

  1. Lovell CR, Burton JL, Horrobin DF. Treatment of atopic eczema with evening primrose oil. Lancet 1981;1(8214):278. [PMID: ] [DOI] [PubMed] [Google Scholar]

MacKie Adult 1990 {unpublished data only}

  1. MacKie R. A Clinical and Mechanistic Assessment of Eamol Evening Primrose Oil in Atopic Eczema (Adults). Pharmacia 2001. [ISN 890054]

MacKie Child 1990 {published data only}

  1. MacKie R. A clinical and mechanistic assessment of efamol evening primrose oil in atopic eczema in Children. Pharmacia 2001. [ISN 890067]

MacLeod 2001 {unpublished data only}

  1. MacLeod A, Misch KJ. A Comparative Assessment of Efamol Evening Primrose Oil and Efamol Marine in Severe Atopic Eczema. Pharmacia 2001. [ISN 89100]

Schalin‐Karrila 1987 {published data only}

  1. Schalin-Karrila M, Mattila L, Jansen CT, Uotila P. Evening primrose oil in the treatment of atopic eczema: effect on clinical status, plasma phospholipid fatty acids and circulating blood prostaglandins. British Journal of Dermatology 1987;117(1):11-9. [PMID: ] [DOI] [PubMed] [Google Scholar]

Senapati 2008 {published data only}

  1. Senapati S, Banerjee S, Gangopadhyay DN. Evening primrose oil is effective in atopic dermatitis: A randomized placebo-controlled trial. Indian Journal of Dermatology, Venereology and Leprology 2008;74(5):447-52. [PMID: ] [DOI] [PubMed] [Google Scholar]

Takwale 2003 {published data only}

  1. Takwale A, Tan E, Agarwal S, Barclay G, Ahmed I, Hotchkiss K, et al. Efficacy and tolerability of borage oil in adults and children with atopic eczema: randomized, double blind, placebo controlled, parallel group trial. BMJ 2003;327(7428):1385. [PMID: ] [DOI] [PMC free article] [PubMed] [Google Scholar]

Valsecchi 2006 {published data only}

  1. Valsecchi R, Di Landro A, Pansera B, Reseghetti A. Gamma-lenolenic acid in the treatment of atopic dermatitis. Journal of European Academy of Dermatology & Venereology 2006 July;7(1):77-9. [Google Scholar]

Wishart 1992 {unpublished data only}

  1. Wishart J, Macdonald K, Oakley A. A Multi-center Assessment of Efamol Evening Primrose Oil in Severe Atopic Eczema. Pharmacia October 2001. [ISN 900196/197/198]

Wright 1982 {published data only (unpublished sought but not used)}

  1. Wright S, Burton JL. Oral evening-primrose-seed oil improves atopic eczema. Lancet 1982;2(8308):1120-2. [6128449] [DOI] [PubMed] [Google Scholar]

References to studies excluded from this review

Biagi 1988 {published data only (unpublished sought but not used)}

  1. Biagi PL, Bordoni A, Masi M, Ricci G, Fanelli C, Patrizi A, et al. A long-term study on the use of evening primrose oil (Efamol) in atopic children. Drugs Under Experimental & Clinical Research 1988;14(4):285-90. [MEDLINE: ] [PubMed] [Google Scholar]

Bordoni 1988 {published data only}

  1. Bordoni A, Biagi PL, Masi M, Ricci G, Fanelli C, Patrizi A, et al. Evening primrose oil (Efamol) in the treatment of children with atopic eczema. Drugs Under Experimental & Clinical Research 1988;14(4):291-7. [PMID: ] [PubMed] [Google Scholar]

Coskery 1988 {published data only (unpublished sought but not used)}

  1. Coskery G, Cowley N, Allen R. The Effect of Dietary Supplementation with Evening Primrose Oil on Skin Surface Texture in Atopic Dermatitis, A quantitative method of assessment. [Abstract 216]. The 18th Annual Meeting of the European Society for Dermatologic Research 19-22 June 1988, Munich, Germany. Journal of Investigative Dermatology 1988;91(4):413.

Courage 1991 {published data only}

  1. Courage B, Nissen HP, Wehrmann W, Biltz H. Influence of polyunsaturated Fatty Acids on the Plasma Catecholamines of Patients with Atopic Eczema [Veranderungen des Katecholaminspiegels bei Neurodermatitis - Beeinflussung durch ungesattigte Fettsauren]. H+G Zeitschrift fur Hautkrankheiten 1991;66(6):509-10. [EMBASE: 1991249008] [Google Scholar]

Guenther 1987 {published data only (unpublished sought but not used)}

  1. Guenther L, Wexler D. Efamol in the treatment of atopic dermatitis. Journal of the American Academy of Dermatology 1987;17(5 Pt 1):860. [DOI] [PubMed]

Melnick 1995 {published data only}

  1. Melnik BC, Bahmer FA. Treatment of Atopic Dermatitis with Glandol and Epoleum - A Comparative Study [Die behandlung des atopischen ekzems mit glandol und epoleum - eine vergleichende studie]. Aktuelle Dermatologie 1995;21(7):215-9. [Google Scholar]

NCT00878670 {unpublished data only}

  1. NCT00878670. Investigation of Efficacy and Safety of EPOGAM. clinicaltrials.gov/ct2/show/NCT00878670 (accessed 16 December 2012).

Swapan 2008 {published data only}

  1. Swapan S, Banerjee S, Gangopadhyay DN. Evening primrose oil is effective in atopic dermatitis: A randomized placebo-controlled trial. Indian Journal of Dermatology, Venereology & Leprology 2008;74(5):447-52. [PMID: ] [DOI] [PubMed] [Google Scholar]

Additional references

Al‐Khamees 2011

  1. Al-Khamees WA, Schwartz MD, Alrashdi S, Algren DA, Morgan BW. Status epilepticus associated with borage oil ingestion. Journal of Medical Toxicology 2011;7(2):154-7. [PMID 21387119] [DOI] [PMC free article] [PubMed] [Google Scholar]

Augustin 1999

  1. Augustin M, Zschocke I, Buhrke U. Attitudes and prior experience with respect to alternative medicine among dermatological patients: the Freiburg Questionnaire on attitudes to naturopathy (FAN). Forschende Komplementarmedizin 1999;6(Suppl 2):26-9. [10352379] [DOI] [PubMed] [Google Scholar]

Bahmer 1991

  1. Bahmer FA, Schubert PH-J. Quantification of the extent and the severity of atopic dermatitis: the ADASI score. Archives of Dermatology 1991;127(8):1239-1240. [PubMed] [Google Scholar]

Barnetson 2002

  1. Barnetson RS, Rogers M. Childhood atopic eczema. BMJ 2002;324(7350):1376-9. [12052810] [DOI] [PMC free article] [PubMed] [Google Scholar]

Batchelor 2010

  1. Batchelor JM, Grindlay DJ, Williams HC. What's new in atopic eczema? An analysis of systemic reviews published in 2008 and 2009. Clinical & Experimental Dermatology 2010;35(8):823-7. [PMID: 20649899] [DOI] [PubMed]

Bayles 2009

  1. Bayles B, Usatine R. Evening Primrose Oil. American Family Physician 2009;80(12):1405-1408. [MEDLINE: ] [PubMed] [Google Scholar]

Belch 2000

  1. Belch JJ, Hill A. Evening primrose oil and borage oil in rheumatologic conditions. American Journal of Clinical Nutrition 2000;71(1 Suppl):352S-6S. [MEDLINE: ] [DOI] [PubMed] [Google Scholar]

Beswick 2011

  1. Beswick T, Morton J, Allen P. Report of two cases of chronic atopic dermatitis treated with adalimubab [P1318]. Journal of American Academy of Dermatology 2011;64(2 Suppl 1):AB60. [PIIS19096221001354X] [Google Scholar]

Bos 2010

  1. Bos JD, Brenninkmeijer EE, Schram ME, Middelkamp-Hup MA, Spuls PI, Smitt JH. Atopic eczema or atopiform dermatitis. Experimental Dermatology Apr 2010;19(4):325-31. [20100192] [DOI] [PubMed] [Google Scholar]

Charman 2000

  1. Charman CR, Morris AD, Williams HC. Topical corticosteroid phobia in patients with atopic eczema. British Journal of Dermatology 2000;142(5):931-6. [10809850] [DOI] [PubMed] [Google Scholar]

Cork 2009

  1. Cork MJ, Danby SG, Vasilopoulos Y, Hadgraft J, Lane ME, Moustafa M, et al. Epidermal Barrier Dysfunction in Atopic Dermatitis. Journal of Investigative Dermatology 2009;129(8):1892-908. [MEDLINE: ] [DOI] [PubMed] [Google Scholar]

Costa 1989

  1. Costa C, Rilliet A, Nicolet M and Surat JH. Scoring Atopic Dermatitis: The Simpler the Better? Acta Dermato-Venereologica 1989;69:41-45. [PMID: PMID: 2563607 ] [PubMed] [Google Scholar]

Diezel 1993

  1. Diezel W, Schulz E, Shanks M, Heise H. Plant oils: topical application and anti-inflammatory effects (Croton oil test) [Pflanzliche ole: ausserliche anwendung und antientzundliche wirkungen (crotonol-test)]. Dermatologische Monatsschrift 1993;179(5):173-6. [Google Scholar]

Dove 1999

  1. Dove D, Johnson P. Oral evening primrose oil: its effect on length of pregnancy and selected intrapartum outcomes in low-risk nulliparous women. Journal of Nurse-Midwifery 1999;44(3):320-4. [MEDLINE: 10380450 ] [DOI] [PubMed] [Google Scholar]

Draelos 2008

  1. Draelos ZD. The effect of ceramide-containing skin care products on eczema resolution duration. Cutis 2008;81(1):87-91. [PubMed] [Google Scholar]

Draelos 2011

  1. Draelos ZD, Koutatsu B, Butcher D, Billhimer W. Delivery of a pseudoceramide to atopic skin is related to objective and subjective improvement [P1306]. Journal of the American Academy of Dermatology 2011;64(2 Suppl 1):AB57. [Google Scholar]

Eisenberg 1998

  1. Eisenberg DM, David RB, Ettner SL, Appel S, Wilkey S, Van Rompay M, et al. Trends in alternative medicne use in the United States 1990-1997: results of a follow-up national survey. Journal of the American Medical Association 1998;280(18):1569-75. [EMBASE: 1998386275] [DOI] [PubMed] [Google Scholar]

Ellis 2002

  1. Ellis CN, Drake LA, Prendergast MM, Abramovits W, Boguniewitcz M, Daniel CR, et al. Cost of atopic dermatitis and eczema in the United States. Journal of the American Academy of Dermatology 2002;46(3):361-70. [11862170] [DOI] [PubMed] [Google Scholar]

Ernst 2000

  1. Ernst E. The usage of complementary therapies by dermatological patients: a systematic review. British Journal of Dermatology 2000;142(5):857-61. [MEDLINE: ] [DOI] [PubMed] [Google Scholar]

Finnigan 1991

  1. Finnigan MD. The Centre for the Study of Complementary Medicine: an attempt to understand its popularity through psychological, demographic and operational criteria. Complementary Medical Research 1991;5(2):83-8. [Google Scholar]

Friedmann 2002

  1. Friedmann PS. The pathogenesis of atopic eczema. Hospital Medicine 2002;63(11):653-6. [DOI] [PubMed] [Google Scholar]

Gfesser 1997

  1. Gfesser M, Abeck D, Rugemer J, Schreiner V, Stab F, Disch R, et al. The early phase of epidermal barrier regeneration is faster in patients with atopic eczema. Dermatology 1997;195(4):332-6. [MEDLINE: ] [DOI] [PubMed] [Google Scholar]

Hanifin 1980

  1. Hanifin J, Rajka G. Diagnostic features of atopic dermatitis. Acta Dermato-Venereologica 1980;92:44-7. [Google Scholar]

Herd 1996

  1. Herd RM, Tidman MJ, Prescot RJ, Hunter JA. The cost of atopic eczema. British Journal of Dermatology 1996;135(1):20-3. [PMID: ] [PubMed] [Google Scholar]

Higgins 2011

  1. Higgins JPT, Green S (editors). Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0 [updated March 2011]. The Cochrane Collaboration, 2011. Available from www.cochrane-handbook.org.

Hoare 2000

  1. Hoare C, Li Wan Po A, Williams H. Systematic review of treatments for atopic eczema. Health Technology Assessment 2000;4(37):1-191. [MEDLINE: ] [PMC free article] [PubMed] [Google Scholar]

Horrobin 2000

  1. Horrobin DF. Essential fatty acid metabolism and its modification in atopic eczema. American Journal of Clinical Nutrition 2000;71(1 Suppl):367S-72S. [MEDLINE: ] [DOI] [PubMed] [Google Scholar]

Imokawa 1991

  1. Imokawa G, Abe A, Jin K, Higaki Y, Kawashima M, Hidano A. Decreased level of ceramides in stratum corneum of atopic dermatitis: an etiologic factor in atopic dry skin. Journal of Investigative Dermatology 1991;96(4):523-6. [MEDLINE: ] [DOI] [PubMed] [Google Scholar]

Jellin 2005

  1. Jellin JM, editorial staff at Therapeutic Research Center. Natural Medicines: Comprehensive Database (Natural Medicines). 7th edition. Stockton CA: Pharmacist's Letter / Prescriber's Letter, 2005. [ISBN 0974706221] [Google Scholar]

Jensen 1990

  1. Jensen P. Use of alternative medicine by patients with atopic dermatitis and psoriasis. Acta Dermato-Venereologica 1990;70(5):421-4. [MEDLINE: ] [PubMed] [Google Scholar]

Johansson 2001

  1. Johansson SG, Hourihane JO, Bousquet J, Bruijnzeel-Koomen C, Dreborg S, Haahteela T, et al. A revised nomenclature for allergy. An EAACI position statement from the EAACI nomenclature task force. Allergy 2001;56(9):813-24. [11551246] [DOI] [PubMed] [Google Scholar]

Johnson 1984

  1. Johnson ML, Johnson KG, Engel A. Prevalence, morbidity and cost of dermatological diseases. Journal of the American Academy of Dermatology 1984;11(5 Pt 2):930-6. [MEDLINE: ] [DOI] [PubMed] [Google Scholar]

Kramer 2012

  1. Kramer MS, Kakuma R. Maternal dietary antigen avoidance during pregancy or lactation, or both, for preventing or treating atopic disease in the child. Cochrane Database of Systematic Reviews 2012;9:published on line. [DOI: 10.1002/14651858.CD000133.pub3] [DOI] [PMC free article] [PubMed] [Google Scholar]

Kubica 2011

  1. Kubica E, Ezzedine K, Lalanne N, Dartial Y, Taieb A, Milpied B. Oral alitretinoin in chronic refactory hand eczema: a "real life" case-series of 12 patients. European Journal of Dermatology 2011;21(3):454-6. [MEDLINE: ] [DOI] [PubMed] [Google Scholar]

Larsen 2002

  1. Larsen FS, Hanifin JM. Epidemiology of atopic dermatitis. Immunology & Allergy Clinics of North America 2002;22(1):1-24. [EMBASE: 2002061960] [Google Scholar]

Leung 2003

  1. Leung DY, Bieber T. Atopic dermatitis. Lancet 2003;361(9352):151-60. [MEDLINE: ] [DOI] [PubMed] [Google Scholar]

Leung 2008

  1. Leung DYM, Eichenfield LF and Bogunlewicz M. Atopic dermatitis (atopic eczema). In: Wolff K, Goldsmith LA, Katz SI, Gilchrest BA, Paller A and Leffel DJ, editors(s). Fitzpatrick's Dermatology in General Medicine. 7th edition. Vol. 1. NYC: McGraw=Hill, 2008:146-158. [ISBN = 9780071466905] [Google Scholar]

Lewis Jones 1995

  1. Lewis-Jones MS, Finlay AY. The Childrens' Dermatology Life Quality Index (CDLQI): initial validation and practical use.. British Journal of Dermatology 1995;132(6):942-9. [MEDLINE: ] [DOI] [PubMed] [Google Scholar]

Manku 1982

  1. Manku MS, Horrobin DF, Morse NL, Kyte V, Jenkins K, Wright S, et al. Reduced levels of prostaglandin precursors in the blood of atopic patients: defective delta-6-desaturated function as a biochemical basis for atopy. Prostaglandins Leukotrienes & Medicine 1982;9(6):615-28. [MEDLINE: ] [DOI] [PubMed] [Google Scholar]

Matsumoto 2000

  1. Matsumoto M, Sugiura H, Uehara M. Skin barrier function in patients with completely healed atopic dermatitis. Journal of Dermatological Science 2000;23(3):178-82. [MEDLINE: ] [DOI] [PubMed] [Google Scholar]

McNally 1998

  1. McNally NJ, Williams HC, Smallman-Raynor M, Lewis S, Venn A, Britton J. Atopic eczema and domestic water hardness. Lancet 1998;352(9127):527-31. [MEDLINE: ] [DOI] [PubMed] [Google Scholar]

McNally 2001

  1. McNally NJ, Williams HC, Phillips DR. Atopic eczema and the home environment. British Journal of Dermatology 2001;145(5):730-6. [MEDLINE: ] [DOI] [PubMed] [Google Scholar]

Melnik 1992

  1. Melnik B, Plewig G. Are disturbances of omega-6-fatty acid metabolism involved in the pathogenesis of atopic dermatitis? Acta Dermato-Venereologica 1992;176:77-85. [MEDLINE: 1476044 ] [PubMed] [Google Scholar]

Moore 2009

  1. Moore EJ, Williams A, Manias E, Varigos G, Donath S. Eczema workshops reduce severity of childhood atopic eczema. Australasian Journal of Dermatology 2009;50(2):100-6. [MEDLINE: ] [DOI] [PubMed] [Google Scholar]

Morse 1989

  1. Morse PF, Horrobin DF, Manku MS, Stewart JC, Allen R, Littlewood S, et al. Meta-analysis of placebo-controlled studies of the efficacy of Epogam in the treatment of atopic eczema. Relationship between plasma essential fatty acid changes and clinical response. British Journal of Dermatology 1989;121(1):75-90. [MEDLINE: ] [DOI] [PubMed] [Google Scholar]

Morse 2006

  1. Morse NL, Clough PM. A meta-analysis of randomized, placebo-controlled clinical trials of Efamol evening primrose oil in atopic eczema. Where do we go from here in light of more recent discoveries? Current Pharmaceutical Biotechechnology 2006;7(6):503-24. [MEDLINE: ] [DOI] [PubMed] [Google Scholar]

Nankervis 2010

  1. Nankervis H, Smith EV, Boyle RJ, Rushton L, Williams HC, Hewson DM, et al. House dust mite reduction and avoidance measures for treating eczema. Cochrane Database of Systematic Reviews 2010, Issue 3. Art. No: CD008426. [DOI: 10.1002/14651858.CD008426] [DOI] [PMC free article] [PubMed] [Google Scholar]

NHIS 2007

  1. National Health Interview Survey (NHIS). 2007 CAM Use Data. http://nccam.nih.gov/news/camstats/ (accessed 28 August 2012).

Nordstrom 1995

  1. Nordstrom DC, Honkanen VE, Nasu Y, Antila E, Friman C, Konttinen YT. Alpha-linolenic acid in the treatment of rheumatoid arthritis. A double-blind, placebo-controlled and randomized study: flaxseed vs. safflower seed. Rheumatology International 1995;14(6):231-4. [MEDLINE: ] [DOI] [PubMed] [Google Scholar]

Pastar 2005

  1. Pastar Z, Lipozencic J, Ljubojevic S. Etiopathogenesis of atopic dermatitis-an overview. Acta Dermatovenerologica Croatica 2005;13(1):54-62. [MEDLINE: ] [PubMed] [Google Scholar]

Patel 2011

  1. Patel T, Gosipovitch G. Therapy of Pruritus. Expert Opinion in Pharmacotherapy 2011;11(10):1673-82. [PMID: ] [DOI] [PMC free article] [PubMed] [Google Scholar]

Phinney 1994

  1. Phinney S. Potential risk of prolonged gamma-linolenic acid use. Annals of Internal Medicine 1994;120(8):692. [MEDLINE: ] [DOI] [PubMed] [Google Scholar]

Plötz 2010

  1. Plötz SG,  Ring J. What's new in atopic eczema? Expert Opinion on Emerging Drugs 2010;15(2):249-67. [MEDLINE: ] [DOI] [PubMed] [Google Scholar]

Poyner 2001

  1. Poyner T. PCDS atopic eczema guidelines optimise GP management. Guidelines in Practice 2001;4(1):33-43. [Google Scholar]

Puri 2007

  1. Puri BK. The safety of evening primrose oil in epilepsy. Prostaglandins Leukotrienes & Essential Fatty Acids 2007;77(2):101-3. [PMID: ] [DOI] [PubMed] [Google Scholar]

Rabahi 2010

  1. Rabahi MF, Ferreira AA, Madeira JG, Galvao PM, Pinto SA. Lipoid pneumonia secondary to long-term use of evening primrose oil. Jornal Brasileiro De Pneumologia 2010;36(5):657-61. [PMID: ] [DOI] [PubMed] [Google Scholar]

Rajka 1989

  1. Rajka G. Essential aspects of atopic dermatitis. Berlin: Springer Verlag, 1989. [Google Scholar]

Reid 1995

  1. Reid P, Lewis-Jones MS. Sleep difficulties and their management in preschoolers with atopic eczema. Clinical & Experimental Dermatology 1995;20(1):38-41. [7671394] [DOI] [PubMed] [Google Scholar]

Riaz 2009

  1. Riaz A, Khan RA, Ahmed SP. Assessment of anticoagulant effect of evening primrose oil. Pakistan Journal of Pharmaceutical Sciences 2009;22(4):355-9. [MEDLINE: ] [PubMed] [Google Scholar]

Roberts 2010

  1. Roberts H, Orchard D. Methotrexate is a safe and effective treatment for paediatric discoid (nummular) eczema: a case series of 25 children. Australasian Journal of Dermatology 2010;51(2):128-30. [PMID: ] [DOI] [PubMed] [Google Scholar]

Saito 2005

  1. Saito, H. Much Atopy about the Skin: Genome-Wide Molecular Analysis of Atopic Eczema. International Archives of Allergy & Immunology 2005;137(4):319-25. [15970641] [DOI] [PubMed] [Google Scholar]

Sampson 1992

  1. Sampson HA. Atopic dermatitis. Annals of Allergy 1992;69(6):469-79. [1471777] [PubMed] [Google Scholar]

Schmitt 2007

  1. Schmitt J, Schakel K, Schmitt N and Meurer M. Systematic Treatment of Severe Atopic Eczema: A Systematic Review. Acta Dermato-Venereologica 2007;87:100-111. [DOI: 10.2340/00015555-0207] [ISSN 0001-5555] [DOI] [PubMed] [Google Scholar]

Schram 2010

  1. Schram ME, Tedja AM, Spijker R, Bos JD, Williams HC, Spurls PI. Is there a rural/urban gradient in prevalence of eczema? A systematic review. British Journal of Dermatology 2010;162(5):964-73. [PMID: ] [DOI] [PubMed] [Google Scholar]

Schultz‐Larsen 1993

  1. Schultz Larsen F. Atopic dermatitis: a genetic-epidemiological study in a population-based twin sample. Journal of the American Academy of Dermatology 1993;28(5 Pt 1):719-23. [8496415] [DOI] [PubMed] [Google Scholar]

SCORAD index 1993

  1. Severity scoring of atopic dermatitis: the SCORAD index. Consensus Report of the European Task Force on Atopic Dermatitis. Dermatology 1993;186(1):23-31. [DOI] [PubMed] [Google Scholar]

Shahidi 1997

  1. Shahidi F, Amarowicz R, He Y, Wettasinghe M. Antioxidant activity of phenolic extracts of evening primrose (Oenothera biennis): a preliminary study. Journal of Food Lipids 1997;4:75-86. [Google Scholar]

Sheehan 1992

  1. Sheehan MP, Rustin MH, Atherton DJ, Buckley C, Harris DW, Brostoff J, et al. Efficacy of traditional Chinese herbal therapy in adult atopic dermatitis. Lancet 1992;340(8810):13-7. [MEDLINE: ] [DOI] [PubMed] [Google Scholar]

Su 1997

  1. Su JC, Kemp AS, Varigos GA, Nolan TM. Atopic eczema: its impact on the family and financial cost. Archives of Diseases of Childhood 1997;76(2):159-62. [PMID: ] [DOI] [PMC free article] [PubMed] [Google Scholar]

Terreehorst 2002

  1. Terreehorst I, Duivenvoorden HJ, Temples-Pavlica Z, Oosting AJ, Monchy JG, Bruijnzeel-Koomen CA, et al. The unfavourable effects of concomitant asthma and sleeplessness due to the atopic eczema/dermatitis syndrome (AEDS) on quality of life in subjects allergic to house-dust mites. Allergy 2002;57(10):919-25. [MEDLINE: ] [DOI] [PubMed] [Google Scholar]

Thestrup‐Pedersen 2003

  1. Thestrup-Pedersen, K. Tacrolimus treatment of atopic eczema/dermatitis syndrome. Current Opinion in Allergy & Clinical Immunology 2003;3(5):359-62. [14501435] [DOI] [PubMed] [Google Scholar]

Tofte 2001

  1. Tofte SJ, Hanifin JM. Current management and therapy of atopic dermatitis. Journal of the American Academy of Dermatology 2001;44(1 Suppl):S13-16. [MEDLINE: ] [DOI] [PubMed] [Google Scholar]

Triebskorn 1989

  1. Triebskorn A, Drosner M. Methods of alternative medicine treatment assessed by patients with allergies and patients with chronic skin diseases ["Alternativ-medizinische" Behandlungsmethoden in der Beurteilung vonAllergikern und chronisch Hautkranken]. H+G Zeitschrift für Hautkrankheiten 1989;64(6):487-94. [MEDLINE: ] [PubMed] [Google Scholar]

van Gool 2003

  1. Gool CJ, Thijs C, Henquet CJ, Houwelingen AC, Dagnelie PC, Schrander J, et al. Gamma-linolenic acid supplementation for prophylaxis of atopic dermatitis - a randomized controlled trial in infants at high familial risk. American Journal of Clinical Nutrition 2003;77(4):943-51. [PMID: ] [DOI] [PubMed] [Google Scholar]

van Odijk 2003

  1. Odijk J, Kull I, Borres MP, Brandtzaeg P, Edberg U, Hanson LA, et al. Breastfeeding and allergic disease: a multidisciplinary review of the literature (1966-2001) on the mode of early feeding in infancy and its impact on later atopic manifestations. Allergy 2003;58(9):833-43. [MEDLINE: ] [DOI] [PubMed] [Google Scholar]

Wang 2007

  1. Wang IJ, Guo YL, Weng HJ, Hsieh WS, Chuang YL, Lin SJ, Chen PC. Environmental risk factors for early infantile atopic dermatitis. Pediatric Allergy & Immunology  2007;18(5):441-7. [PMID: ] [DOI] [PubMed] [Google Scholar]

Weller 2009

  1. Weller RPJB, Hunter JAA, Savin JA, Dahl MV. Eczema and Dermatitis. chapter in Clinical Dermatology 2009;Chapter 7:79-103. [DOI: ] [SN: 9781444300086.ch7] [Google Scholar]

Williams 1994

  1. Williams HC, Strachan DP, Hay RJ. Childhood eczema: disease of the advantaged? BMJ 1994;308(6937):1132-5. [PMID: ] [DOI] [PMC free article] [PubMed] [Google Scholar]

Williams 2000

  1. Williams HC, Wüthrich B. The natural history of atopic dermatitis. In: Williams HC, editors(s). Atopic Dermatitis. Cambridge University Press, 2000:49-59. [978-0521570756] [Google Scholar]

Williams 2008

  1. Williams HC, Grindlay DJ. What's new in atopic eczema? An analysis of the clinical significance of systemic reviews on atopic eczema published in 2006 and 2007. Clinical & Experimental Dermatology 2008;33(6):685-8. [PMID: ] [DOI] [PubMed]