Impact of intraday temperature variation on acute hospital admissions and its spatial pattern in Hong Kong older adults

Background Evidence on the association between diurnal temperature range (DTR) and acute hospital admissions among older adults remains limited, especially in subtropical urban settings. Moreover, no study has quantified the spatial distribution of DTR-related admissions or translated such findings into actionable hospital capacity planning. Methods We conducted a territory-wide study of over 3.5 million acute hospital admissions among Hong Kong residents aged ≥ 65 years (2012–2021). Using a two-stage quasi-Poisson time-series regression with distributed lag nonlinear models, we estimated the association between DTR and all-cause acute admissions, overall and by season (cool: Nov–Apr; warm: May–Oct), and pooled district-level estimates via random-effects meta-analysis. Attributable burden was computed under two district-specific reference scenarios (25th and 50th percentiles of DTR). Spatial variation was decomposed using ecological regression with multiple district-level characteristics, and district-level burdens were mapped to six hospital clusters using patient-flow matrices as a planning scenario. Results Each 1 °C higher DTR was associated with a 0.748% linear increase in all-cause acute admissions over lag 0–4 days (95% CI: 0.634–0.862%). The association was concentrated in the cool season (excess relative risk percentage, ERR% = 0.998%; 95% CI: 0.858–1.139%) and near null in the warm season (ERR% = 0.014%; 95% CI: −0.186 to 0.214%). The cool-season association was stronger among adults aged ≥ 85 years and those residing in old-age homes. Respiratory admissions showed a more pronounced cool-season association and remained elevated in the warm season. In cool-season attributable burden analyses, the highest attributable numbers occurred in densely populated urban districts (Kwun Tong, Kwai Tsing, Wong Tai Sin), while DTR values were highest in the northern New Territories. Variance decomposition identified median per capita floor area and district-specific ERR as the dominant contributors. When mapped to hospital clusters, the largest burdens fell disproportionately on three of the six clusters. Conclusions DTR is independently associated with acute hospital admissions among older adults during the cool season. The association is broadly linear without a clear threshold; scenario-based attributable burden estimates provide acontinuous measure and spatial variation for resource planning, supporting proactive capacity management and targeted community outreach.