Estimating influenza-related hospital admissions in older people from GP consultation data
This study seeks to determine the association between general practice (GP) consultations for influenza-like illness (ILI) and hospital admissions, and to estimate the number of excess hospitalisation due to ILI epidemics, for two age-groups: 51-65 years and >65 years. Methods: Structural time series models (STSM) with stochastic trend and seasonal components were developed. Weekly data from Switzerland for 1987 to 1996/1997 were used. Hospital admission rates for pneumonia and influenza (P&I) and other respiratory conditions (ORC) were regressed against GP consultation rates. Excess hospitalisations were calculated as the difference between predicted hospital admissions during influenza epidemics and predicted hospital admissions using a baseline GP consultation rate. Results: Excess admission rates were substantially higher in the oldest age-group (51-65 years: P&I 4.42, ORC 2.14; >65 years: P&I 6.30, ORC 4.74 per 10,000 population per year). The models explained between 43 and 76% of the variation in hospital admissions. The seasonal patterns were stable over the 10 years modelled and the variances of the trends were small.