Background: This analysis was conducted to identify the participant characteristics associated with noncompliance in an oral contraceptive (OC) clinical trial.
Study design: We studied ovarian suppression among normal-weight and obese women during the use of levonorgestrel (LNG)-containing combination OCs. Participants underwent twice weekly phlebotomy during the study cycle and received up to $360 for participation. Along with other study assays, we analyzed 903 specimens from 181 women to measure LNG to assess OC compliance. Consistently undetectable LNG levels indicated noncompliance. To evaluate predictors of OC noncompliance during this study, we compared the characteristics of compliant and noncompliant participants using multivariable logistic regression. We assigned each participant to a relative poverty level based on US census data; all other individual characteristics came directly from participant responses during the baseline interview.
Results: One hundred eighty-one women completed the study; 31 were noncompliant (17%). In multivariable analyses, poverty level was the strongest predictor of noncompliance. Compared with those women in the quartile with the lowest level of residential poverty, other women were far more likely to be noncompliant, especially women in the quartile with the greatest prevalence of poverty (adjusted odds ratio, 8.4; 95% confidence interval, 1.5-46.1). Additional factors associated with noncompliance were education level less than a bachelor's degree and Hispanic ethnicity. Other demographic and psychometric measures were not associated with compliance.
Conclusions: We found that noncompliance was strongly associated with residential poverty level, an indirect measure of individual income. In the United States, poverty is associated with female obesity, Hispanic ethnicity and low education, which were also associated here with noncompliance. Study compensation may motivate poor individuals to participate in clinical trials for income. Noncompliance in clinical trials, particularly differential noncompliance, jeopardizes study validity.
Copyright © 2012 Elsevier Inc. All rights reserved.