Background Traditional classifications of the metabolic syndrome (MetS) rely on measurements of central obesity and clinical diagnoses of diabetes, hypercholesterolaemia and hypertension – criteria that are rarely available in large-scale epidemiological surveys. The aim of the present analysis was to assess whether questionnaire-derived, self-reported information on clinical diagnoses might offer a valid assessment of MetS in populations with good access to health care and health surveillance.

Methods The analyses used data from Understanding Society’s (USoc) Wave 2 Health Assessment – a subsample of n = 6853 men and n = 8793 women, with a mean (SD) age of 50.5 (17.8) years that included data collated from previous questionnaires with measurements taken during home visits by trained research nurses. These were used to generate self-reported and objectively-assessed classifications of MetS. ‘Central obesity’ (CO) was defined as participants: who were obese (body mass index 30 kg/m²); and whose waist circumference exceeded WHO cut-offs (102 cm for men; 88 cm for women). Objective measurements of blood pressure exceeding NICE hypertension severity stage 2 cut-offs (Systolic blood pressure (SBP) 160 mm Hg; Diastolic blood pressure (DBP) 100 mm Hg), with/without prescribed antihypertensive medication, were used to classify ‘observed hypertension’ (O-HBP). Participant-reported clinical diagnoses of hypertension were used to classify ‘self-reported hypertension’ (SR-HBP). ‘Objective Hypertensive MetS’ (O-HBPMetS) was defined as CO plus O-HBP; ‘Self-reported MetS’ (SR-HBPMetS) was defined as CO plus SR-HBP.

Results Over a quarter of participants (28.7%) were obese and almost half (44.2%) exceeded waist circumference cut-offs; women being more likely to have obesity (57.5%) or CO (59.0%) than men (42.5% and 41.0%, respectively). While 21.8% had a self-reported clinical diagnosis of hypertension, only 8.9% had been prescribed hypertensive medication and/or had SBP/DBP readings that met NICE stage 2 severity levels. Unsurprisingly, substantially more participants were estimated to have SR-HBPMetS (8.7%) than O-HBPMetS (3.5%), such that the sensitivity and specificity of SR-HBPMetS for O-HBPMetS was 81% and 94%, respectively. And while the negative predictive value of SR-HBPMetS was 99% its positive predictive value was just 32%.

Conclusion These findings indicate that substantially more participants recall a clinical diagnosis of hypertension than those with NICE stage 2 severity hypertension and/or prescribed antihypertensive medication – suggesting that clinicians diagnose and/or treat hypertension at lower blood pressure levels than those recommended by NICE. For this reason SR-HBPMetS provides a less conservative assessment of MetS than O-HBPMetS when the latter are based on NICE severity criteria. Context-specific sensitivity analyses are therefore required to optimise the utility of SR-HBPMetS as a measure of O-HBPMetS in epidemiological surveys.