Get access

Delivering Preventive Health Services for Breast Cancer Control: A Longitudinal View of a Randomized Controlled Trial


  • Phyllis A. Gimotty, Robert C. Burack, and Julie A. George

Phyllis A. Gimotty Ph.D Department of Biostatistics and Epidemiology, University of Pennsylvania School of Medicine, 631 Blockley Hall, 423 Guardian Drive, Philadelphia, PA 19104.

Abstract

Objective. To develop a longitudinal model to characterize the delivery of mammography services using repeated observations of mammography referral rates during a randomized controlled trial (RCT) of physician mammography reminders.

Data Sources/Study Setting. Administrative records of a health department and observational data on mammography appointment scheduling.

Study Design. The design was a longitudinal study of month-specific referral rates during a 1-year RCT. A retrospective case-control study was used to investigate differences between women with timely and delayed (or absent) mammography referral assessed at the end of the intervention year.

Data Collection/Extraction Methods. Month-specific indicators for referrals and missed clinical opportunities, that is, months when clinic visitors were due for a mammogram and not referred, were constructed using administrative and observational data.

Findings. In the unadjusted analysis, the effectiveness of the reminder declined over time. However, in a multivariate analysis that controlled for the number of missed opportunities, the effectiveness was constant over time. On a monthly basis, physician reminders were significantly associated with higher referral rates among clinic visitors newly due for mammography (adjusted OR=2.8, 95 percent CI=1.3, 5.8) or who had one previously missed clinical opportunity (adjusted OR=3.0, 95 percent CI=1.6, 5.3) but were not for those with two or more missed clinical opportunities (adjusted OR=1.2, 95 percent CI=0.7, 2.3). Factors independently associated with delayed referral were age over 65, presence of more than one chronic illness, and the absence of a physician mammography reminder.

Conclusions. Longitudinal models that examine rates of referral over time and include information about outcomes on previous visits can enhance our understanding of how intervention strategies work in practice.

Ancillary