Use of exercise tests in primary care: importance for referral decisions and possible bias in the decision process; a prospective observational study

BACKGROUND The utility of clinical exercise tests depends on their support of treatment decisions. We sought to assess the utility of exercise tests for the selection of primary-care patients for referral to cardiologic care, and to determine whether referral decisions were biased by gender or socioeconomic status. We also evaluated referral rates and cardiovascular events in patients with positive exercise tests. METHODS We designed a prospective observational study of 438 men and 427 women from 28 Swedish primary-care clinics who were examined with exercise testing for suspected coronary disease. All participants were followed-up with respect to cardiologist referrals and cardiovascular events (hospitalisation for unstable angina, myocardial infarction, and cardiovascular death) within six months and revascularisation within 250 days. Variables associated with referral were identified by multivariable logistic regression. Socioeconomic status was determined by educational level and employment. RESULTS Positive/inconclusive exercise tests and exertional chest pain predicted referral in men and women. Of 865 participants, patients with positive, inconclusive, or negative exercise tests were referred to cardiologists in 67.3%, 26.1%, and 3.5% of cases, respectively. Overall, there was no significant difference in referral rates related to gender or socioeconomic level. Self-employed women were referred more frequently compared to other women (odds ratio (OR) 3.62, 95% confidence interval (CI) 1.19-10.99). Among non-manual employees, women were referred to cardiologic examination less frequently than men (OR 0.40, 95% CI 0.16-1.00; p = 0.049; ORs adjusted for age, exertional chest pain, and exercise test result). In patients with positive exercise tests, the referral rate decreased continuously with age (OR 0.48, 95% CI 0.23-0.97; adjusted for cardiovascular co-morbidity). Cardiovascular events occurred in 22.2% (4/18) of non-referred patients with positive exercise tests; 56% (10/18) of these patients were not considered for cardiologic care, with continuity problems in primary care as one possible contributing cause. CONCLUSIONS Exercise tests are important for selecting patients for referral to cardiologic care. Interactions related to gender and socioeconomic status affected referral rates. In patients with positive exercise tests, referral rates decreased with age. An increased awareness of possible bias regarding age, gender, and socioeconomic status, which may influence medical decisions, is therefore necessary.