Abstract
BACKGROUND: Population health behaviour and risk factor surveys most often rely on self-report but there is a lack of studies assessing the validity of self-report using Australian data. This study investigates the sensitivity, specificity and agreement of self-reported hypertension and hypercholesterolaemia with objective measures at standard and more stringent diagnostic thresholds; and factors associated with sensitivity and specificity of self-report at different thresholds. METHODS: This study was a secondary analysis of a representative community-based cross-sectional sample of 5,092 adults, aged 45-69 years, residing in Busselton, Western Australia, surveyed in 2010-2015. Participants completed a self-administered questionnaire. Blood pressure and serum cholesterol levels were measured. RESULTS: At currently accepted diagnostic thresholds, sensitivities of self-reported hypertension and hypercholesterolaemia were 58.5% and 39.6%, respectively and specificities were >90% for both. Agreement using Cohen's kappa coefficient was 0.562 and 0.223, respectively. At two higher diagnostic thresholds, sensitivities of self-reported hypertension and hypercholesterolaemia improved by an absolute 14-23% and 15-25%, respectively and specificities remained >85%. Agreement was substantial for hypertension (kappa = 0.682-0.717) and moderate for hypercholesterolaemia (kappa = 0.458-0.533). Variables that were independently associated with higher sensitivity and lower specificity of self-report were largely consistent across thresholds and included increasing age, body mass index, worse self-rated health, diabetes and family history of hypertension. CONCLUSIONS: Self-reported hypertension and hypercholesterolaemia often misclassify individuals' objective status and underestimate objective prevalences, at standard diagnostic thresholds, which has implications for surveillance studies that rely on self-reported data. Self-reports of hypertension, however, may be reasonable indicators of those with blood pressures ≥160/100 mmHg or those taking anti-hypertensive medications. Self-reported hypercholesterolaemia data should be used with caution at all thresholds.