Insight in non-participation and drop-out among coronary artery disease patients in a prospective cohort study – the THORESCI study

Background: Non-random study non-participation and loss to follow-up are problematic in most cohort studies, and frequently impair data quality and representativeness. The current study examined demographic, disease-related and psychological correlates of patient non-participation (“refusers”), and loss to follow-up (“drop-outs”) in a large cohort of patients with coronary artery disease. We also examined whether, in comparison to participants, refusers and drop-outs differed in their likelihood to start cardiac rehabilitation and their survival status. Method: Participants who dropped out (n=376; mean age=67.5±122; 70% men), and refusers (n=148; mean age=71.8±12.6; 69% men) were compared with active participants (n=865; mean age=65.1±11.8; 79% men). Demographics, clinical risk factors, medical history, participation to cardiac rehabilitation (CR), and mortality were obtained from patients’ medical records and baseline questionnaires. Information about psychological symptoms (GAD-7, PHQ-9, DS14) was derived from medical records for patients who started CR in hospital. For patients who did not participate to CR but who participated in the study, psychological data were completed with study questionnaire data. Results: Compared to participants, refusers were older (OR= 1.04; 95% CI: 1.02-1.07; p = .002), had a lower socio-economic status (OR= 2.78; 95% CI: 1.52-5.07; p = .001) were more often smokers (OR= 1.64; 95% CI: 1.04-2.58; p = .03), and experienced a larger burden of illness (cardiac history: OR= 1.56; 95% CI: 1.03-2.37; p = .04; non-cardiac history: OR= 1.97; 95% CI: 1.24-3.13; p = .004). Refusers also participated less often in CR, independent of age and sex (OR= .69; 95% CI: .48-.99, p = .04). They also showed increased mortality as compared to patients who initially decided to participate (10% vs. 5%; OR= 2.09; 95% CI: 1.14-3.85, p = .02). Compared to participants, drop-outs were more often women (OR= 1.48; 1.08-2.03, p = .02). In drop-outs, increased levels of smoking (OR = 1.51; 1.13-2.02, p = .01), and a larger burden of illness were found (non-cardiac history: OR= 1.31; 95% CI: 1.00-1.71; p = .05). We did not find any differences between the groups regarding psychological symptoms. Conclusion: Refusers and drop-outs differed systematically on some demographics and disease-related factors, which was translated into poorer health outcomes, as there were differences in the mortality risk of refusers versus participants. Therefore, we conclude that results from survey studies examining the association between psychosocial factors and health outcomes may underestimate the effects, as a healthier sample is participating than is present in the real life population.