Perioperative local anaesthesia for reducing pain following tonsillectomy

Background

Pain relief after tonsillectomy is an important part of post‐operative management. Sometimes local anaesthetics are administed to the tonsillar region, but their effectiveness in relieving pain has not been formally assessed.

Objectives

To assess the effects of pre‐ and post‐operative local anaesthesia for pain reduction following tonsillectomy.

Search methods

Cochrane Controlled Trials Register, Medline, Embase and the Oxford Pain Database. Reference lists from identified publications, including those in non‐English language publications, were scanned. Date of the most recent search was September 1998.

Selection criteria

Randomised controlled trials of adults and/or children undergoing tonsillectomy alone with local anaesthetic 
 (a) injected into the tonsillar region immediately prior to removal of the tonsils 
 (b) injected into the tonsillar region after removal of the tonsils 
 (c) sprayed or otherwise applied to the tonsillar region after removal of the tonsils 
 Outcome measures included the reduction in demand for post‐operative analgesia, both in terms of time elapsed to request for first analgesia and of the total amount of analgesia used.

Data collection and analysis

All three reviewers extracted data independently and assessed trials for quality. Four authors were contacted to obtain information on either the method of randomisation or the type of intervention used.

Main results

Thirty trials were initially identified but only six trials met the inclusion criteria. These were all published in the English language literature between 1989 and 1997. Trials were excluded because procedures in addition to tonsillectomy were undertaken, inadequate randomisation techniques were used or because group sizes were inadequate. Of the six included studies, five involved local anaesthetic injection and one application of local anaesthetic spray. These studies used validated pain scores and measurement of supplemental analgesic intake for assessment of pain. However, some studies also included inappropriate, possibly post‐hoc, outcome measures such as 'global pain scores' which may have resulted in bias. No included study showed a significant difference between intervention and control, other than for outcome measures which were felt to be inappropriate, such as the one mentioned above.

Authors' conclusions

There is no evidence that the use of perioperative local anaesthetic in patients undergoing tonsillectomy improves post‐operative pain control. The trials identified were of small size and several involved the perioperative co‐administration of intravenous opiates which may have masked any beneficial effect of the local anaesthetic. Further randomised controlled trials are necessary.

Criteria for considering studies for this review

Search methods for identification of studies

Randomised controlled trials of tonsillectomy were identified. The optimum search strategy for detecting controlled trials formulated by the Cochrane Collaboration was combined with the term TONSIL* and used to search Medline from 1966 onwards. Embase, the Pain Database (Jadad 1996) and the Cochrane Controlled Trials Register were also searched using the term TONSIL*. For Embase, this was combined with appropriate terms to identify clinical trials. The initial search results were scanned by one reviewer to identify trials which loosely met the inclusion criteria, i.e. which included patients undergoing tonsillectomy with perioperative local anaesthetic administration. Reference lists from identified publications were scanned and authors contacted if details regarding methods of randomisation and patient inclusion were unclear (Valijan 1989; Arcioni 1995; Ørntoft 1994; Molliex 1996). The full text articles of the retrieved trials were then reviewed by two reviewers and the inclusion criteria applied independently. Any differences in opinion about which studies to include in the review were resolved by the third reviewer. The reviewers were blinded to the names of the journals, the authors' names and the study results when they applied the criteria for determining which studies to include in the review.

Data collection and analysis

QUALITY ASSESSMENT 
 The quality of all included trials was assessed blindly and independently by two reviewers. A third reviewer was asked to resolve differences in opinion. A modification of the method used by Schulz et al (Schulz 1995) was used. The selected studies were assessed for the following characteristics: 
 1. the adequacy of the randomisation process 
 2. the adequacy of study group size (more than ten patients in each group) 
 3. the potential for selection bias after allocation to study group, i.e. losses to follow‐up and whether analysis was by intention to treat 
 4. whether there was blinding of the outcome assessors to the patients' study group 
 5. the quality of the outcome assessment, e.g. use of well validated visual analogue pain scores for adults and children.

Studies were graded A, B or C for their overall methodological quality. 
 A: minimisation of bias in all five categories above, i.e. (1) adequate randomisation; (2) adequate group size; (3) few losses to follow‐up and intention to treat analysis; (4) blinding of outcome assessors; (5) high quality outcome assessment 
 B: criteria in A partially met for each category 
 C: one or more of the criteria in A not met

Only studies classed as 'A' were included. However, we had planned to use study quality for sensitivity analysis, i.e. to re‐include studies classed as 'B' or 'C' to assess how this affected the overall result.

DATA EXTRACTION 
 Data from the studies were independently extracted by two reviewers using standardised data forms. Data were extracted so as to allow an intention to treat analysis. Where data were missing, the reviewers contacted the authors of the study requesting the missing data.

DATA ANALYSIS 
 We planned to analyse data by intention to treat.

Description of studies

Thirty studies were identified but only six studies fulfilled the entry criteria for study design and the measurement of clinically appropriate outcome measures.

EXCLUDED STUDIES 
 Trials were excluded either because of inappropriate methods of randomisation (19 trials) (Davidson 1952; Reed 1955; Pande 1970; Castanos 1971; Raj 1986; Broadman 1989; Violaris 1989; Boliston 1990; Revilla Borjas 1990; Jebeles 1991; Nigam 1991; Candan 1992; Jebeles 1993; Melchor 1994; Elhakim 1995; Masuda 1995; Wong 1995; Goldsher 1996; Strub 1996) because procedures in addition to tonsillectomy were carried out (13 trials) (Reed 1955; Riddington‐Young1987; Broadman 1989; Jebeles 1991; Candan 1992; Jebeles 1993; Schoem 1993B; Melchor 1994; Stuart 1994; Elhakim 1995; Wong 1995; Goldsher 1996; Strub 1996) or because of inappropriate outcome measures (6 trials) (Pande 1970; Raj 1986; Bissonette 1990; Boliston 1990; Revilla Borjas 1990; Masuda 1995). One trial by Johansen (Johansen 1996) included only nine patients in the intervention group.

INCLUDED STUDIES

METHODS: All studies were randomised.

PARTICIPANTS: Most of the included studies involved adult patients undergoing tonsillectomy alone except for the study by Molliex 1996 which included adults and children, and the study by Arcioni 1995 which included children only. The study by Bell 1997 examined patients following uvulopalatopharyngoplasty as well as following tonsillectomy alone. Data were presented separately, thereby enabling us to include only those undergoing tonsillectomy.

INTERVENTIONS: Local anaesthetic infiltration was used in five studies but the specific interventions in each study varied widely in terms of local anaesthetic used, the dose and time of application. The only study to examine post‐operative application of local anaesthetic spray was that by Valijan 1989.

OUTCOMES: Post‐operative pain levels were assessed by: 
 (a) Visual analogue scales: pain intensity was measured using validated visual analogue scales (0‐100mm) in all studies except in the one by Schoem 1993A, in which a categorical scale was used. In the paediatric study by Arcioni 1995, pain intensity was assessed on a validated five point Smile analogue scale. These measurements were predominantly made during the first 24 hours after surgery. Only Molliex 1996 measured pain after 36 hours and Ørntoft 1994 described pain scores up to the eigth post‐operative day. 
 (b) Need for supplemental analgesia was assessed in all studies except that by Arcioni 1995. No study measured time to request for first analgesia, but instead total amounts of additional analgesic requirement were assessed. These additional agents included paracetamol, non‐steroidal anti‐inflammatory drugs, cocodamol (codeine phosphate and paracetamol) and morphine. 
 (c) morbidity and adverse events were assessed specifically in the studies by Bell 1997 and Schoem 1993A.

OUTCOMES ‐ NOTES: Visual analogue scales were used at different times during the post‐operative period in the various included trials precluding quantitative meta‐analysis. The injected anaesthetics differed in respect to the concentrations, volumes used, sites of injection and whether agents contained adrenaline.

Risk of bias in included studies

Appropriate allocation techniques were explicit in the methods sections of Bell 1997, Molliex 1996 and Schoem 1993A, but not in the studies by Arcioni 1995, Ørntoft 1994 and Valijan 1989. However, use of appropriate allocation techniques was confirmed by correspondence with the authors. All included studies stated that both patients and assessors of outcome were blinded to the treatment assignment.

Effects of interventions

Data extracted from the randomised controlled trials are shown in the table of included studies.

Inappropriate measures were also applied in some of the included trials, such as measurement of mean arterial blood pressure and heart rate in the study by Arcioni 1995, categorical sedation scores in the study by Molliex 1996 and pain on jaw opening at 10 hours in the study by Schoem 1993A. The measurement of a 'global pain score' in the study by Molliex 1996, which yielded a positive result, may have been biased. This 'global pain score' was calculated by averaging the measurements for the whole of the first post‐operative day ‐ results, which, when analysed by themselves, did not show a positive result.

PAIN ASSESSMENT: No statistically significant differences were shown between placebo and active interventions except in the study by Molliex 1996 in which a difference was shown in isolation for the 'global pain score' for the first 24 hours. It is our impression that this was a post‐hoc outcome measure and should be disregarded.

DEMAND FOR ANALGESIA: The need for supplemental analgesia was reported in all studies except in the study by Arcioni 1995. None of the studies showed a statistically significant difference between the treatment and placebo groups. Although results were not analysed formally in the study by Valijan 1989, no difference between control and intervention groups was apparent.

ADVERSE EFFECTS: These were monitored specifically in the studies by Bell 1997 and Schoem 1993A. Bell 1997 noted that one patient from the intervention group suffered nausea and vomiting which delayed discharge from the recovery room. There were no reported adverse events in Schoem 1993A. Valijan 1989 noted that one patient from each group suffered stinging and nausea. Two patients each in the studies by Ørntoft 1994 and Molliex 1996 suffered post‐operative bleeding, which is a recognised complication of the procedure of tonsillectomy.

Implications for practice.

Implications for research.

Likar 2000

Cyna 2000

Characteristics of included studies [ordered by study ID]

Characteristics of excluded studies [ordered by study ID]

References to studies included in this review

References to studies excluded from this review

References to studies awaiting assessment

Additional references