Measuring contraceptive efficacy and side effects.

Abstract

To assess effectiveness of contraceptives, one reports all pregnancies during use. Multiple and single decrement life-table pregnancy rates computed from these reports constitute the fundamental measures of effectiveness in clinical trials. Minimally biased evaluation of a new method requires a randomized, double-blind (or its nearest semblance), multicentered study of sexually active women. New contraceptives are studied in special, volunteer populations. Evaluation of marketed contraceptives implies broader-gauged representation, larger numbers and longer time frames. Prospective studies, like clinical trials, should include several methods, and may use subsamples for case-control analysis of risk. Well-designed and executed multicentered, prospective studies, nevertheless, may represent specially selected populations. Risk-benefit analyses based on findings from selected populations require careful scrutiny. Although the most persuasive estimates of effectiveness derive from national, representative samples, moderately large samples in the USA have not been able to distinguish among products within a class. Formidable problems of recall are present in household surveys. In the United Kingdom, Europe and North America, analysis of risks and benefits may be enriched by better utilization of the data collection and record linkage resources of national health statistics systems.