The study was designed to evaluate the efficacy of dimeticone 4% lotion against head louse infestation in a different study population from previously. It was also designed to compare dimeticone 4% lotion with a potential alternative silicone formulation. It was anticipated that for both objectives the risk of reinfestation would be minimised, compared with previous investigations, by offering treatment to the whole community.
The study was based around two rural village schools in Manisa province, western Turkey. The villages were selected because the schools were large enough to provide a sufficient number of participants but small enough for the investigation team to screen all students for head lice during a single day. One school in the village of Osmancali serves approximately 20 small villages and communities in the locality and children are bussed in daily. The other school in Maldan village serves only that community and a few adjacent farms. Children in each of the schools were screened for presence of live lice using plastic detection combs ("PDC", KSL Consulting, Denmark) . Infested children were provided with an information sheet to take home and subsequently the family of every child found positive for head lice was visited at home by at least one investigator and the community nurse. A standard consent and assent procedure was followed. All infested family members were able to participate in the study.
All enrolled participants provided baseline data on age, gender, hair characteristics, and previous pediculicide use. The lower age limit was 4 years; there was no upper limit. Treatments and assessments were conducted in the school, which ensured access to all participants. When children were not available for assessment on the assigned school day a follow up was arranged by the locality nurse to visit them at home. No payment was offered for participation.
Inclusion and exclusion criteria for this study were essentially the same as we have used previously [1–3], excluding potential participants with: known sensitivity to any component of treatment; secondary bacterial infection or other chronic scalp condition; bleached, coloured, or permanently waved hair within the previous four weeks; use of pediculicide within two weeks or co-trimoxazole or trimethoprim antibiotics within four weeks (either of which could affect the viability of any lice remaining on the head). Pregnancy and breast feeding were also exclusion factors, as were having participated in another clinical study within 1 month or prior participation in this study.
The study was granted ethical approval by the ethics committee of the Medical School of Celal Bayar University, Manisa, Turkey (Study 025, 14/02/2007). Approval was also obtained from the Board of Education covering that part of Manisa province as well as individual school authorities prior to commencement. None of the children attending the two schools was older than 14 years of age so parents signed a form giving written consent for them to participate and stating they understood the purpose of the study as set out in the information leaflet. All the children also gave written assent prior to enrolment.
The study was conducted in conformity with the principles of the Declaration of Helsinki and of European Directive 2001/20/EC.
Two physically similar silicone lotions were used for this study. Both were based on 4% dimeticone in a cylcomethicone solvent. The original dimeticone 4% lotion (Hedrin® 4% lotion, Thornton & Ross Ltd, Huddersfield, UK) was compared with essentially the same mixture into which 2% nerolidol had been incorporated. Nerolidol is a linear sesquiterpene alcohol that has properties that may allow it to modify the surface tension of the lotion and facilitate its penetration into the breathing apparatus of the louse and its egg. Tests in vitro indicted that it had greater ability to enter the breathing system of the louse eggs, resulting in reduced risk of hatching by a factor of about 30%.
Participants were randomised to receive either dimeticone 4% lotion or dimeticone 4% lotion with nerolidol 2%, each with two applications 7 days apart. As in previous studies of dimeticone 4% lotion the products were applied by members of the investigation team. Sufficient product was applied to dry hair to moisten the whole scalp and length of the hair. Treatment was applied at the end of the school day, left to dry naturally, and washed off the following morning at home, using a non-medicated shampoo supplied by the investigation team. Family members who were found with lice but were unable to participate for any reason were provided with dimeticone 4% lotion for self administration in order to reduce the risk of reinfestation to the study group. Only sixteen people reported having ever used a pediculicide, six had used permethrin about 2 months previously with varying results; the others had not been treated for between 6 months and 6 years.
Assessment of Outcome
At initial diagnosis an assessment of intensity of infestation was based on the rapidity with which lice were found during the initial combing. This measure was used previously [1–3] as a guideline during post-study evaluations of outcome: heavy = >1 louse with the first stroke of the comb, medium = 1 louse with the first stroke of the comb, light = 1 louse found only after 5-6 strokes of the comb.
Each participant was assessed for presence of lice on five occasions, days 1, 2, 6, 9, and 14 after the first application of treatment, using the "PDC" comb on dry hair in the same way as the children were initially screened in the schools. The purpose of the additional combing immediately following the first treatment (i.e. a combing on days 1 and 2 rather than on day 1 or day 2 as have been used in other studies) was to enable the investigators a better opportunity to find any newly hatched nymphs in relation to evaluation of the first primary outcome measure, the ovicidal effect of the treatments, defined as no nymphal lice emerging from eggs between treatments. the other primary outcome measure, elimination of live lice following the second application of treatment, that is, no lice present on days 9 and 14, was monitored by combing on days 9 and 14 as in previous studies [1–3].
Interim combing assessments employed the same approach as previously, drawing the comb 2-3 times through each section of hair. This level of dry combing has minimal intervention effect because, if no lice are found the combing cannot affect the outcome of treatment in any way. If lice are found the non-intensive combing removes only a proportion of any lice present, but allows identification of treatment failure or possible reinfestation at the earliest opportunity. The day 14 assessment involved more extensive combing than on other days to ensure no live lice were present. Non-participating family members with lice who self treated were also checked if they assented. This allowed us to confirm whether their treatments had been successful and to understand the circumstances that could lead to reinfestation. Reinfestation was defined using an algorithm described previously [2, 3].
Sample size and Randomisation
Previous studies conducted in the UK have demonstrated efficacy rates for dimeticone 4% lotion of around 70% and no newly emerged nymphs have been found on approximately 50% of participants after first treatment [1, 3]. Based on group sizes used previously a sample size of 31 per group was considered to have at least 80% power to detect (with 95% confidence) a difference of 35% between the products in not finding newly hatched nymphs or in efficacy rates based on elimination of infestation. The actual sample sizes of 36 per group made allowance for dropout. Treatment allocation was in balanced blocks of 12 predetermined using a computer generated list (http://www.randomization.com, 17 March 2008). Allocation at the point of delivery was made from instruction sheets enclosed in opaque, sealed, sequentially numbered envelopes distributed to the investigators in batches of 12. As each participant was enrolled the investigator on site selected the next available numbered envelope from the allocation and the treatment was then applied from one of two series of numbered bottles.
Randomisation was by individual so if more than one member of a family was enrolled it was possible for them to receive different treatments. This approach for studies on head louse infestation has been the accepted randomisation methodology by the Medicines and Healthcare products Regulatory Agency (MHRA) in the UK, whereas the US Food and Drug Administration (FDA) have been inclined towards cluster randomisation by family. However, the latter methodology is generally not practical for a small proof of concept study as there is a need to stratify households according to number of participants, and possibly even intensity of infestation of individuals within the households, and the working assumption is generally that all the household units are independent and the participants within a household are correlated. A statistical difficulty arises on how to deal with within-household correlation because some of the evidence (unpublished) from previous studies indicates that for some households there is no within-household correlation. Consequently, it was decided at an early stage to continue to use randomisation by individual.
As the two products were physically similar the allocation was essentially blinded to participants at the point of delivery. In order to avoid contamination bias all study treatments were applied by investigators and the bottles of study medication returned to the centre for weighing after single use. Assessments were performed by different investigators from those who conducted the initial screening, enrolment, and treatment of each child in order to reduce the possibility of bias.
For presence/absence variables a Fisher exact test was used. Differences in success rates between the treatment outcomes for the two preparations were quantified by the 95% confidence interval, calculated using a normal approximation to the binomial distribution. For graded or semi-continuous variables, Kruskal-Wallis analysis of variance was used, which as only two groups were tested was equivalent to using the Mann-Whitney U test.