- Robert H Vander Stichele, general practitionera,
- Els M Dezeure, school health services physiciana,
- Marc G Bogaert, clinical pharmacologista
- Correspondence to: Dr Vander Stichele.
- Accepted 14 July 1995
Abstract Objectives: To collect and evaluate all trials on clinical efficacy of topical treatments for head lice.
Design: Systematic review of randomised trials identified from following data sources: Medline, International Pharmaceutical Abstracts, Science Citation Index, letters to key authors and companies, and hand search of journals.
Setting: Trials in schools or communities.
Subjects: Patients infested with lice.
Main outcome measure: Cure rate (absence of live lice and viable nits) on day 14 after treatment.
Results: Total of 28 trials were identified and evaluated according to eight general and 18 lice specific criteria. Of the 14 trials rated as having low to moderate risk of bias, seven were selected as they used the main outcome measure. These seven trials described 21 evaluations of eight different compounds and placebo (all but two evaluations were of single applications). Only permethrin 1% creme rinse showed efficacy in more than two studies with the lower 95% confidence limit of cure rate above 90%.
Conclusions: Only for permethrin has sufficient evidence been published to show efficacy. Less expensive treatments such as malathion and carbaryl need more evidence of efficacy. Lindane and the natural pyrethrines are not sufficiently effective to justify their use.
In this systematic review we found only 28 randomised trials on clinical efficacy of topical treatments for head lice
Of these trials, only seven were of acceptable methodological quality and measured outcome at 14 days after treatment (the optimum time to assess clinical efficacy)
Of the eight different compounds evaluated, only permethrin 1% creme rinse showed efficacy in more than two studies with a lower 95% confidence limit of cure rate above 90%
Only for permethrin has sufficient evidence been published to show efficacy: less expensive treatments such as malathion and carbaryl need more evidence of efficacy, while lindane and the natural pyrethrines are not sufficiently effective to justify their use
Head lice are among the most common of human ectoparasites, though they are not vectors of serious diseases and in many cases do not cause symptoms.1 2 3 Treatment with natural pyrethrines has been known for more than 100 years, and lindane has been used since the second world war. The synthetic pyrethrines were marketed in the 1950s, malathion and carbaryl in the ‘60s, and permethrin in the ‘80s. Although products abound, the prevalence of head lice remains high and epidemics occur regularly despite all efforts at control.3 4 5 Problems such as fear of insects (entomophobia), fear of stigmatisation, and denial of infection by patients and schools may cause under-treatment, overtreatment, and unnecessary prophylaxis, which can lead to development of resistance and insufficient control of epidemics.6 Furthermore, many of the commercially available treatments might be underdosed, incorrectly labelled, or ineffective.
We searched for trials of topical treatments for people infested with head lice (Pediculus humanus capitis) in which the outcome was measured clinically by inspection of the scalp to determine cure rate (absence of live lice and viable nits).
We searched the medical literature in Medline (1966 to March 1995 using the MESH keywords “Pediculosis,” “Lice,” “Pediculus”), in International Pharmaceutical Abstracts, and in the Science Citation Index without restriction for language of publication. We scanned the references of all identified clinical trials. We sent letters requesting information about unpublished studies to seven key authors in the subject, to the pharmaceutical companies active in this subject, and to the World Health Organisation centre Vector biology Control. We hand searched journals in which key references were published for comments, letters, or corrections in the year after publication of the key reference.
We focused on clinical efficacy--the result of pediculicidity, ovicidity, and residual activity--and so we chose cure rate as the main outcome measure for clinical evaluation. The cure rate is the percentage of patients cured after application of the treatment (the 95% confidence interval=p +/- 1.96 (square root p (100-p)/n), where p is the sample percentage and n the number of subjects in the study). Determination of the cure rate by experienced evaluators--on the basis of visual inspection for viable nymphs in nits, hatching nymphs, and adult lice (with a x 10 magnifying lens)--has an acceptable specificity and sensitivity.9 We considered an interval of 14 days between treatment and evaluation to be optimal as this would allow evaluation of the combined effect of pediculicidity (on living lice), ovicidity (on ripening eggs), and residual activity (on hatching nymphs and reinfesting lice).
We evaluated all identified clinical trials with regard to eight general criteria of quality in clinical trials (adapted from Chalmers et al10) and 18 criteria specific for head lice treatment (see table I). We developed these specific criteria after studying the literature to systematically screen the trials for flaws in design, execution, or reporting. Firstly, we made a structured abstract of each clinical trial according to recommended guidelines.11 Each of us then independently assessed the trials. Trials were rejected if four or more flaws in general criteria or 12 or more flaws in treatment specific criteria were found. The remaining trials were rated as having a low risk of bias if less than eight specific criteria were flagged, or a moderate risk of bias if otherwise. Again, we set these cut off points for rating of quality after studying the literature. The structured abstracts, assessment scores, and overall ratings were submitted to an advisory panel (four physicians and a community pharmacist) and discussed until consensus was reached. Of the trials of acceptable methodological quality, we selected those in which the main outcome measure was the cure rate at 14 days after treatment.
SELECTION OF TRIALS
We identified 28 trials of clinical efficacy, 27 from the computer databases12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 and one supplied by an author in reply to our request.39 We also identified an internal document of the Wellcome company describing a series of 11 small (11<n<74) unpublished comparative trials of permethrin and malathion performed between 1983 and 1986 and apparently yielding high cure rates for permethrin and malathion.40 As we decided not to accept the company's demands for confidentiality, the full texts of these studies were not made available to us and these trials were not included in the analysis. Hand searching did not reveal additional relevant reports of trials. The search in the Science Citation Index showed only limited citing activity in this field. An official of WHO confirmed that the debate on the choice of head lice treatment for the list of essential drugs had been based on expert opinion without a formal literature review. Narrative reports of studies and older reports of treatment campaigns41 42 43 44 45 were not included in the analysis.
Four studies were not controlled,12 18 26 39 three studies (two of malathion and one of permethrin) were placebo controlled,15 21 23 and the remaining 21 studies were comparisons between two or more active substances. We rated the quality of the identified trials according to the criteria in table I and rejected 14 of the trials because of an excess of general or treatment specific flaws (table II). Seven trials were excluded from the analysis, although their methodological quality was acceptable, because the cure rate was not determined on day 14 after treatment (three measured cure rate on day 7,15 21 38 and four measured it on day 21 or later16 29 30 32). The characteristics of the excluded trials are listed in table III.
The remaining seven trials were of acceptable methodological quality and had the cure rate on day 14 as main outcome measure.14 22 23 24 27 28 36 Five of the trials had an overall quality rating of low risk of bias, while two had a moderate risk of bias (table II). Three trials were conducted in an area with high background prevalence of head lice (>50% of screened population).14 23 24
RESULTS OF SELECTED TRIALS
In the seven selected trials 21 individual evaluations of topical treatments were performed, comparing placebo and eight compounds (lindane, bioresmethrin, chlorphenamide, (delta)-phenothrin, pyrethrin, malathion, carbaryl, and permethrin), and all but two evaluations27 36 were of single applications (table IV). We juxtapositioned the results obtained in different trials for the same compounds, and the figure shows the cure rates for each compound. The cure rate with placebo was 6%,23 showing the lack of placebo effect and spontaneous remission with this condition. We found six evaluations of lindane; in none of them did the lower confidence limit of the cure rate exceed 90%, and in two trials even the upper confidence limit was below 90%. For the natural pyrethrines (pyrethrin, bioresmethrin, chlorphenamide, and (delta)-phenothrin), all the evaluations resulted in cure rates with lower confidence limits below 90%. Only one evaluation was available for carbaryl 0.5% lotion and malathion 0.5% lotion, both giving cure rates with lower confidence limits above 90%. We found five evaluations of permethrin 1% creme rinse (in a single application of 10 minutes) with cure rates of nearly 100% and lower 95% confidence limits above 90%. Two of these studies were high quality studies in populations with a high background prevalence of head lice (>50% of screened population infested).23 24
We also made an intrastudy comparison of the two largest trials.22 24 both trials were of high quality and compared single applications of permethrin 1% creme rinse with lindane 1% shampoo. The odds ratio of treatment failure for lindane versus permethrin was 15.11 (95% confidence interval 4.60 to 49.62) in one study22 and was 15.28 (5.13 to 45.52) in the second study.24 After performing a breslow-Day test for homogeneity of odds ratios (P=0.99), we obtained the Mantel-Haenszel summary odds ratio of 15.18 (7.99 to 28.84). Hence, the risk of treatment failure was likely to be at least eight times higher with lindane than with permethrin.
Our aim was to collect details of all trials on the clinical efficacy of treatment for head lice and to describe the results of the trials that were not invalidated by too many flaws. To our surprise, we found only 28 published studies.
METHODS OF REVIEW
We cannot exclude the possibility that important research findings were missed by our method of retrieval. We did not engage in a hand search of core journals, as recommended for structured reviews,46 because there did not seem to be a core group of journals, where research publications are concentrated,47 for this subject. Our failure to obtain the full text of unpublished trials comparing permethrin with malathion excluded evidence for the efficacy of malathion.
In this first systematic approach to the treatment of head lice it was not possible to determine the acceptance criteria a priori. The rating system was instead developed after study of the identified trials. We chose cure rate at 14 days as the main outcome measure since it was the most commonly used criterion for efficacy and is, in our opinion, the most appropriate outcome measure. We preferred not to consider the cure rate at seven days, as hatching of nymphs can take longer than this48 and as a week is too short to evaluate the effect of residual activity on reinfestation (which is important in stopping transmission during epidemics).49 We have, however, presented cure rates at day 7 and beyond day 14 (tables III and IV), but these results do not challenge the overall conclusions of our review.
We presented the efficacy data of each active ingredient by juxtaposition of treatment groups from the seven selected trials. This procedure is methodologically weak but has the advantage of extracting at least some of the limited knowledge from the clinical studies. The representation of the confidence intervals in the figure should be interpreted with caution, as many of the results fall in the extreme end of the range of cure rates, where confidence intervals based on the sample percentage tend to shrink to zero.
Within the seven selected trials it was difficult to make sensible intrastudy comparisons. The sample size of two of the trials27 36 was insufficient for testing relevant differences. Two trials were selected despite flaws in randomisation, one because of the comparison with placebo23 and one because of the range of products tested.14 The relevance of testing a product against itself in an underdosed formulation in one study28 can be questioned. Hence, we limited ourselves to the comparison of the two bigger high quality trials, which compared permethrin with lindane.22 24
OTHER ASPECTS OF EVALUATING TREATMENT
We made no attempt to formally weigh criteria such as side effects, toxicity, and cost. In the course of reviewing the literature, we found only one large scale postmarketing surveillance of safety, which provided evidence for the safety of permethrin.50
Theoretically, products with residual activity might facilitate selection of resistant strains of head lice, and proposals have been made for a rotational or mosaic treatment strategy.49 51 There is no convincing evidence for a need for such a strategy, but development of resistance should be monitored if the therapeutic arsenal diminishes to a few products of proved efficacy.
RECOMMENDATIONS FOR FUTURE TRIALS
The number of well conducted trials of clinical efficacy in this subject of medical and economic importance is limited, and recommendations for treatment published in the medical literature are not sufficiently sustained by the results of research. Our list of treatment specific criteria could be a starting point for evaluation of new trials in the subject. Moreover, inspection of the figure leads us to recommend that only products with an expected cure rate of over 90% should be tested and that this should be done in trials with sufficient power to establish cure rates with a lower confidence limit above 90%. We propose the use of equivalence testing (testing for non-null hypothesis) and of the odds ratio of treatment failure with its 95% confidence interval, as the χ2 test might not be valid for testing differences in cure rates near 100%. Future research on head lice treatments should preferably test single applications of compounds, as was the case in almost all the acceptable clinical trials in this study. Any treatment for lice might be effective if it is applied repeatedly over a short interval.52 We found many examples of instructions on labels encouraging multiple application, especially with products that lacked well documented efficacy. In some cases these instructions clearly played on people's entomophobia to stimulate consumption, as has been stated by others.6
We are indebted to the members of the advisory panel: C Carton, army pest control physician; J M Kaufman, endocrinologist; G Laekeman, professor of pharmacy; K Seynaeve, school health services physician; and E Van Hecke, dermatologist. We thank D De bacquer for statistical advice and A Herxheimer for critical comments on the manuscript.
Funding This study was supported by a grant from the Belgium Ministry of Health.
Conflict of interest None.