A novel approach to calculate disease incidence for hospital-based health events in England. (November 2021)

• Record Type: Journal Article
• Authors: Campling, James; Begier, Elizabeth; Vyse, Andrew; Hyams, Catherine; Heaton, Dave; Southern, Jo; Danon, Leon; Finn, Adam; Madhava, Harish; Gessner, Bradford D; Ellsbury, Gillian
• Is Part Of: Lancet. Volume 398:Supplement 2(2021)
• Page Start: S30
• Publication Date: 2021-11
• Subjects: Medicine -- Periodicals; Medicine -- Periodicals; Medicine; Medicine; Electronic journals; Periodicals; 610.5
• DOI: 10.1016/S0140-6736(21)02573-3
• Languages: English
• ISSNs: 0140-6736

Abstract

Abstract: Background: Disease incidence is a key indicator for the assessment of population health and to guide public health policies. Defining accurate population denominators is challenging for incidence studies of hospital-based events, for which hospital catchment areas depend on patient choice and geographic proximity and often do not match established town or city boundaries. We describe a new method to accurately define a hospital's population denominator, which we developed for a study designed to estimate the incidence of hospitalisation for adult acute lower respiratory tract disease in Bristol, UK. Methods: We examined hospital data for adult acute lower respiratory tract disease to identify general practitioner (GP) practices (n=82) whose patients were treated for adult acute lower respiratory tract disease at study hospitals and confirmed that nearly all patients (9230 [94·5%] of 9764) were registered with practices within the National Health Service's Bristol, North Somerset, and South Gloucestershire Clinical Commissioning Group (CCG). Hospital admission data were linked to aggregated GP practice patient registration data within this CCG for April 1, 2017, to March 31, 2020. The proportion of GP practices' patients who were hospitalised for adult acute lower respiratory tract disease that occurred at a specific study hospital was multiplied by their patient registration count for six age groups to obtain the practices' contribution to that hospital's denominator (eg, if 50% of GP practice admissions were at a specific study hospital among patients aged 50–64 years, the practice contributed half of their patients aged 50–64 years to the denominator). Alternate denominators were calculated using population census data matched to 20-minute drivetime estimates for study hospitals (the method used to define hospital catchment for local health service purposes). Findings: The percentage of admissions at study hospitals varied substantially by GP practice (0–100%). 231 342 (86·3%) of 268 093 of the CCG's GP practice registrants aged 18–34 years received treatment at study hospitals (compared with 19 229 [80%] of 23 938 registrants aged or older than 85 years); some received treatment at another local hospital to the south. With 20-minute drivetime as an estimate measure, hospital catchment would be 550 155 (66·5%) of 827 304 individuals in the CCG population. Interpretation: Using health-care data stratified by local CCG enables the estimation of local population denominators to support accurate incidence ascertainment within hospital-based surveillance studies. Funding: This work was funded by Pfizer.