Between May 1, 2006, and Jan. 26, 2007, 30 family practices from the Leiden Primary Care Research Network in the Netherlands invited all patients aged four years and older who attended the clinic with one or more new cutaneous warts to participate. We defined new cutaneous warts as those on the skin that were diagnosed in family practice and had not been treated by a physician or dermatologist in the previous year, regardless of previous self-treatment with over-the-counter medication. We excluded immunocompromised patients and patients with genital warts, seborrheic warts or warts larger than 1 cm in diameter. Patients who fulfilled the inclusion criteria and agreed to participate were visited at home by a trained research nurse, who confirmed their eligibility. Informed consent (child as well as parental informed consent for participants less than 18 years of age) was obtained, and baseline characteristics were collected.
Study design and randomization
We stratified patients by location of warts: plantar (warts on the soles of the feet) or common (warts on the hands or other locations).26
Participants who had both plantar and common warts were stratified according to where the majority of their warts were located. We used opaque, sealed envelopes that were numbered based on a computerized randomization list delivered by an independent statistician to conceal allocation. After stratification by location of warts and by number of warts (< six warts v. ≥ six warts), random allocation of participants to treatment groups was done without blocking. The study protocol was approved by the medical ethical committee of the Leiden University Medical Center.
One of us (K.Z.) trained all participating family physicians and assistants working in their practices in the three 13-week treatment protocols, which were designed to reflect best practice.10,24
Training consisted of a one-hour interactive practical session, during which all tools and techniques were demonstrated; real warts were not used in the demonstrations.
For cryotherapy, we used a high-intensity regimen of one session every two weeks until all warts were completely gone. During each session, the participant received three serial applications in which a wad of cotton wool saturated with liquid nitrogen was moved around on the wart. Each application was executed until a frozen halo of 2 mm around the base of the wart appeared (usually after 2–10 seconds).
For the topical application of salicylic acid, we used a white petroleum jelly containing 40% salicylic acid. We chose this concentration to provide a stronger treatment than over-the-counter products, which usually contain 17% salicylic acid. Participants assigned to this group were asked to apply the salicylic acid every day until the warts were completely gone. They were instructed to cover the surrounding skin with tape to protect healthy skin and apply the salicylic acid on top of the wart with another piece of tape. Before each subsequent daily application, they used a file to pare the softened surface area of the wart.
Participants assigned to the wait-and-see group were informed about the benign natural course of warts and were advised not to undergo treatment (apart from over-the-counter medication) for at least 13 weeks.
After the 13-week treatment period, all participants who still had warts could switch to another treatment according to their own preferences. Participants were free to use over-the-counter medication during the entire follow-up period but were asked to report all usage.
Trained research nurses assessed outcomes during home visits at 4, 13 and 26 weeks of follow-up, independently of the treating physicians. A wart was considered cured if it was no longer visible (skin colour and skin lines were reestablished) and could not be palpated anymore by hand. The primary outcome measure was the proportion of participants whose warts were all cured at 13 weeks. Research nurses assessed side effects, newly developed warts (which were not included in the primary outcome assessment) and adherence to treatment. Treatment adherence was considered adequate if participants had received cryotherapy at least every three weeks, had self-administered salicylic acid at least four days per week and had not undergone any co-intervention (treatment of warts other than over-the-counter medication).
In addition, participants were asked to rate treatment burden using a 10-point scale (1 = no burden, 10 = the worst imaginable burden). A scores of six or higher was considered to reflect a substantial burden. Participants rated treatment satisfaction using a five-point scale (one = very unsatisfied, five = very satisfied); those with a score of four or five were considered to be satisfied.
Research nurses, family physicians and participants were not blinded to treatment allocation. For quality control, 5% of the assessments were directly supervised by experienced family physicians (J.E. and K.Z.).
We chose a sample size that would provide 80% power, at a significance level of 5%, to detect an absolute increase in the cure rate of 20% between the two active treatment groups. Based on a literature review, we expected salicylic acid to be most effective, with a 70% cure rate.10,11
A total of 91 patients were required per treatment arm.
We used the χ2 test for all comparisons of cure rates and percentages. In our primary analysis, we compared cure rates between the three treatment arms on an intention-to-treat basis. We also calculated relative risks, risk differences and numbers needed to treat for cryotherapy versus salicylic acid, cryotherapy versus wait-and-see approach, and salicylic acid versus wait-and-see approach.
In secondary analyses, we compared cure rates between the three study arms (a) with patients lost to follow-up considered not cured, (b) after excluding patients who had both plantar and common warts, (c) at 26 weeks’ follow-up, (d) using individual warts as the unit of analysis instead of patients and (e) per protocol cure rates based on reported treatment adherence.
Subgroup analyses were pre-planned for location of warts (common wart group v. plantar wart group), age clusters (4–12 years v. ≥ 12 years), number of warts per participant, and duration of warts (≤ six months v. > six months). We formally tested for effect modification of treatment by location of warts using a logistic regression model.
Lastly, we compared the percentages of patients with side effects and considerable treatment burden between the two active treatment arms, and the percentages of patients satisfied with treatment between the three arms.