Both observed and modelled COPD prevalence varies considerably across England. Cost-effective case-finding strategies should be evaluated, especially in areas where the ratio of observed to expected cases is low.

BackgroundThere is under-diagnosis of cardiovascular disease (CVD) in the English population, despite financial incentives to encourage general practices to register new cases. We compared the modelled (expected) and diagnosed (observed) prevalence of three cardiovascular conditions- coronary heart disease (CHD), hypertension and stroke- at local level, their geographical variation, and population and healthcare predictors which might influence diagnosis.MethodsCross-sectional observational study in all English local authorities (351) and general practices (8,372) comparing model-based expected prevalence with diagnosed prevalence on practice disease registers. Spatial analyses were used to identify geographic clusters and variation in regression relationships.ResultsA total of 9,682,176 patients were on practice CHD, stroke and transient ischaemic attack, and hypertension registers. There was wide spatial variation in observed: expected prevalence ratios for all three diseases, with less than five per cent of expected cases diagnosed in some areas. London and the surrounding area showed statistically significant discrepancies in observed: expected prevalence ratios, with observed prevalence much lower than the epidemiological models predicted. The addition of general practitioner supply as a variable yielded stronger regression results for all three conditions.ConclusionsDespite almost universal access to free primary healthcare, there may be significant and highly variable under-diagnosis of CVD across England, which can be partially explained by persistent inequity in GP supply. Disease management studies should consider the possible impact of under-diagnosis on population health outcomes. Compared to classical regression modelling, spatial analytic techniques can provide additional information on risk factors for under-diagnosis, and can suggest where healthcare resources may be most needed.