Abstract
Abstract
Background
Ambulance services often represent a key entry point to healthcare for older adults, and subsequent care pathways have important implications for outcomes and resource use. Direct admission to geriatric units may offer benefits for certain patients, but access varies. This study aims to examine whether geographic area of residence is associated with ambulance conveyance to geriatric units as an alternative to emergency departments among older adults and to explore patient- and system-level factors associated with observed variation.
Methods
A cross-sectional, register-based study of ambulance assignments of adults ≥ 65 years in Stockholm, Sweden (2017–2019) was conducted. Eight geriatric units, each representing a distinct hospital service area, were included. The outcome was direct admission to a geriatric unit versus to an emergency department. Geographic variation was assessed using random intercept multilevel logistic models, adjusting for demand-side and supply-side factors, with measures of variation including intraclass correlation coefficient (ICC), median odds ratio (MOR), and proportional change in variance (PCV).
Results
Among 8208 patients included in the study, 603 (7%) were directly admitted to a geriatric unit. Crude conveyance rates across hospital service areas ranged from 4% to 21%. The unadjusted between-area variance (τ²) was 0.3510 (ICC = 0.096). When supply-side factors were assessed individually while controlling for demand, proximity to a geriatric unit, geriatric unit ownership, and number of staffed beds resulted in estimated reductions in the between-area variance by 26%, 27%, and 38%, respectively, suggesting that these factors may contribute to the observed area-level heterogeneity. In the fully adjusted model, τ² decreased by 41% to 0.2076 (ICC = 0.059), although considerable residual heterogeneity remained (MOR = 1.54), indicating that otherwise similar patients may face different probabilities of direct geriatric admission depending on their hospital service area. Robustness analysis with a Bayesian approach revealed large uncertainty around the estimated variance.
Conclusions
Direct ambulance conveyance to geriatric units varies substantially across hospital service areas. While proximity to units and greater geriatric bed capacity likely had an association with reduced between-area variation, meaningful residual differences persist, suggesting that unmeasured factors beyond patient need and measured supply characteristics may influence access to acute geriatric care.