Presentation on theme: "Estimating Prevalence of Diabetes and Other Chronic Diseases for Small Geographic Areas Peter Congdon, Geography, QMUL."— Presentation transcript:
Estimating Prevalence of Diabetes and Other Chronic Diseases for Small Geographic Areas Peter Congdon, Geography, QMUL
Major chronic diseases are leading source of morbidity and health service costs in developed societies with ageing populations Age Effect compounded by high diabetes rate among relatively young immigrant/ethnic group populations (e.g. South Asian community in England) Diabetes also increasing in developing societies (e.g. China)
Trends §Prevalence Trends have to be distinguished from Trends in Mortality (e.g. Important for CHD) §Diabetes relatively small as direct cause of death but important as risk factor for CHD, stroke, etc §Trends in Diabetes Prevalence: US Data shows upward trend and this also applies for England §HSE 1993 to 2003 (ages 16+) l Males 3% 4.3% l Females 2% 3.4%
Risk Factors for Diabetes §Diabetes type 2 (adult onset) incidence varies considerably by age, ethnic group and income group §Diabetes linked to socio-economic deprivation: diabetic patients with lower education level less likely to follow advice on lifestyle/medicines; less likely to attend their GP for a review of their condition. §HSE also shows higher prevalence for males
Indirect Area Estimates §Can model gradients over these demographic categories using Health Survey for England; use logit regression to develop prevalence rate profile by age, ethnicity, and gender. §Apply prevalence rate profile to census populations disaggregated according to these categories §Census table ST101 cross-tabulates age, ethnicity and gender down to electoral ward level (approx. 8000 wards in England) §Deprivation gradient also applied. Gradient in prevalence over IMD quintiles, as ratio to average prevalence, is (0.76, 0.84, 0.91, 1.13, 1.37) for males, (0.80, 0.84, 0.93, 1.07, 1.36) for females.
Relevance of Quality Outcomes Framework (QOF Registers) §QOF Registers also supply estimates of prevalence but geographic profile limited (PCTs only, not local authorities or electoral wards - at least in terms of publicly available data) §Also subject to registration biases-work in Jrnl of Public Health (Sigfrid et al, Vol 28(3)) shows exception reporting increased for deprived practices - means prevalence gradient by deprivation flatter than should be
Other Applications § Same principles can be applied to other major chronic diseases (CHD, Serious Mental Illness) §Potential to link prevalence estimation to development of IMD indicator domains §Prevalence forecasts linked to mortality forecasts §Prevalence forecasts linked to ethnic population estimates & projections §Compare indicative prevalence (indirect approaches) with QOF registrations - assess under or over-registration