Overview

It is estimated that 45 million people in the United States have acne vulgaris, with a prevalence of approximately 85 percent in the population 15-24 years of age. The disease is more common and more severe in males than in females. Morbidity primarily comes from the lesions themselves, which may be painful and tender, as well from scars left by nodules and cysts. Morbidity may be generated by adverse effects of treatments as well. Psychological morbidity has been noted. It is estimated that consumers spend $100 million per year in over-the-counter remedies. Coupled with loss of productivity and unemployment, the direct cost of acne may exceed $1 billion per year in the United States.

Reporting the Evidence

Three questions were addressed in the literature abstraction:

1. What treatments are effective in the management of acne?
2. What is the responsiveness of acne to first-line, second-line, and third-line (referral) treatments?
3. What are the side effects of treatment?

Evidence was sought for several subpopulations by acne severity, gender, and race/ethnicity; by outcome (e.g., response to treatment and acne sequelae); and cost.

Methodology

Working with our partners, the American Academy of Dermatology and the American Academy of Pediatrics, as well as with the American Pharmaceutical Association, the Departments of Dermatology and Pediatrics at Johns Hopkins School of Medicine, the Cochrane Collaboration, and the Society for Investigational Dermatology, we assembled a team of technical experts.

The questions for consideration were developed through discussions with our technical experts, as well as through preliminary searches of MEDLINE^® and MicroMedex and a review of consumers' questions submitted to a major consumer health Internet Web site (InteliHealth™).

The following reference databases were searched: CENTRAL database of the Cochrane Collaboration, MEDLINE^® (accessed through PubMed), OLDMEDLINE, PsycINFO^®, and CINAHL^®. The CENTRAL search strategy was (acne-vulgaris*.me OR acne*.tw). Similar strategies were used for the other databases. The search strategy for PubMed combined, using a Boolean AND, a topic-specific strategy (acne vulgaris[mh] OR acne*[tw]) with a strategy designed to retrieve all controlled trials. Reference lists from key articles were reviewed by hand. All unique controlled trials of acne therapy were sought that presented original, human data and were written in English.

Identified articles were screened by independent reviewers in an abstract review process to determine eligibility for full article review. Eligible articles were reviewed serially by two or more abstracters who sought methodological and outcome data.

Studies were qualitatively assessed in terms of methodological strengths and weaknesses. Studies were also classified by the level of acne severity in study patients at enrollment. A combined acne severity classification was created that integrated the 25 schemes found in the literature.

The unit of analysis was the comparison between pairs of treatment arms. Comparisons were qualitatively assessed by addressing methodological issues, by addressing commonality of outcomes across studies, and by addressing results and their statistical significance between arms. The evidence was assessed in terms of strength at Levels A, B, or C as follows:

• A -- At least two trials of acceptable quality show moderate to strong statistical evidence for a clinically meaningful endpoint and effect.
• B -- Evidence is of modest strength, such as when only one trial addresses a comparison, several trials have different qualitative conclusions, large differences are not statistically significant, or poor trial quality prevents accepting strong statistical evidence at face value.
• C -- Evidence does not permit coming to any scientific conclusion because of a paucity of the evidence, conflicting evidence, or very poor quality evidence.

Findings

• The literature is extremely heterogeneous, severely limiting the number of meaningful conclusions that can be drawn. The 274 trials included reports from 31 countries. There were 140 treatments. There were 505 distinctly named acne outcomes over four periods of followup (usually <3 months). There were 250 pairwise comparisons.
• Evidence for subpopulations is either not available or reported in such a varied manner as to prevent meaningful integration. Of the trials, 74 percent reported data on the age of the subjects, 13 percent reported phase of care at study entry, 8 percent reported race, and none reported on sexual maturity rating. Only eight trials stratified their results: four by gender, three by acne type or severity, and one by study location.
• Evidence for quality-of-life outcomes and cost is not available. Two trials assessed psychological effects and 43 trials provided anecdotal data on treatment compliance. None reported on cost.
• Of 250 comparisons, only 14 had evidence of Level A. These comparisons demonstrated the efficacy over vehicle or placebo control of aluminum chlorhydroxide/sulphur, topical clindamycin, topical erythromycin, benzoyl peroxide, topical isotretinoin, tretinoin, oral tetracycline, and norgestimate/ethinyl estradiol. Level A conclusions demonstrating equivalence include: Benzoyl peroxide at various strengths was equally efficacious in mild/moderate acne; adapalene and tretinoin were equally efficacious in unspecified severity; motretinide and tretinoin were equally effective; adding vitamin A to oxytetracycline conferred no added efficacy; cyproterone was equally effective at two different doses. There were 102 comparisons with Level B evidence and 134 with Level C evidence.
• The literature is not organized to differentiate responses to first-line, second-line, or third-line (referral) therapy. Investigators were not systematic in evaluating subjects' prior therapy. There is no basis from controlled trials to judge the efficacy of any treatment in relation to the sequence of therapy.
• Regarding side effects, evidence was gleaned for 88 primary comparisons. Of these, 10 provided evidence at Level A. These involved, primarily, adverse local reactions of retinoids. Because of appropriate exclusions, important side effects, like the teratogenicity of isotretinoin, were not documented in these trials.

Future Research

There is still much work to be done to define the best approach to management of acne in an individual patient based upon his or her specific characteristics. Until this work is done, the process of care for patients with acne vulgaris will remain highly individualized, based upon the experience of the patient and the treating physician. The effectiveness of a more evidence-based, stepwise approach that incorporates several therapies at different points in time still needs to be documented.

Future research on the treatment of acne vulgaris would benefit from:

• Population-based studies with detailed information about subjects including age, race, gender, sexual maturity rating, and previous treatment of acne.
• Consistent case definitions and methods for assessing acne type and severity.
• Greater consistency in outcomes across studies so that trials are comparable.
• Standards for assessment of clinical outcomes.
• Assessment of outcomes at time points distant from treatment.
• Explicit assessment of patients' acceptability of treatment.
• More information about psychological outcomes and costs related to treatment.
• Attention to the issue of bacterial resistance.

This report was developed using accepted evidence-based techniques; however, as in other areas of research, the methodology was necessarily constrained by issues of feasibility and practicality so that, for example, the analysis was limited to clinical trials reported in the English language. Nonetheless, this report is a systematic effort to collect and summarize the available evidence on the management of acne and represents a potentially powerful resource for the field. As with any systematic review, to be accurate and comprehensive, the generation of guidelines requires additional information, such as expert opinion, to put the evidence into context.