Participation bias assessment in a community-based study of myocardial infarction, 2002-2005

To compare the characteristics and survival of participants and nonparticipants in a community-based study of myocardial infarction (MI).

Residents of Olmsted County, MN, who presented with elevated cardiac troponin T levels from September 1, 2002, through December 31, 2005, were prospectively enrolled and classified with standardized criteria for MI. With specific Institutional Review Board approval, the medical records of patients with MI who did not provide consent but who had given general research authorization were reviewed, as was done for their consenting peers.

During the study period, 2277 individuals with elevated cardiac troponin T levels were approached, of whom 1863 (82 percent) consented to participate. Among the 414 nonparticipants, 375 (91 percent) had general research authorization. Of the 558 with general research authorization who met the criteria for incident (ie, first-ever) MI, 67 (12 percent) refused to participate. These participants tended to be older (mean plus or minus SD age, 71 plus or minus 14 vs 67 plus or minus 15 years; P equals .04), were more likely to be of races other than white (9 percent vs 2 percent; P equals .01), and had more comorbidities, including peripheral vascular disease (P equals .02), chronic pulmonary disease (P equals .06), heart failure (P equals .07), and impaired creatinine clearance (P equals .02). No significant differences were detected in cardiovascular risk factors or MI characteristics. During a median follow-up of 517 days, nonparticipants experienced increased mortality rates compared with participants (hazard ratio, 1.97; 95 percent confidence interval, 1.21 to 3.20), which was largely attributable to their older age and excess comorbidities (adjusted hazard ratio, 1.43; 95 percent confidence interval, 0.86 to 2.35).

In this community-based study of MI, nonparticipants experienced worse survival rates than participants largely because of differences in demographic and clinical characteristics. These differences should be kept in mind when interpreting study results, particularly if participation is low.