Association of ambient temperature and sun exposure with hip fractures in Japan: A time-series analysis using nationwide inpatient database

Abstract

Background: Evidence on whether meteorological conditions affect hip fractures (HFs) is limited. This study aimed to clarify the associations between ambient temperature and sun exposure and HFs in Japan.

Methods: Record of daily hospital admissions for HFs between 2015 and 2018 were extracted from a Japanese nationwide inpatient database. We conducted a time-series quasi-Poisson regression analysis using a distributed lag non-linear model with lag 0-39 days to estimate prefecture-specific relative risks (RRs) of HFs. We also estimated pooled RRs using random-effects meta-analysis.

Results: We identified 355,563 HFs. For mean temperature, immediate RRs (lag 0-2 days) were 1.349 (95% confidence interval (CI): 1.305, 1.395) and 0.754 (95% CI: 0.727, 0.782) for low (mean of the 2.5th percentile) and high (mean of the 97.5th percentile) mean temperature, respectively, relative to the reference (mean of medians). For sunshine duration, immediate RRs were 0.929 (95% CI: 0.913, 0.946) and 1.056 (95% CI: 1.029, 1.085) for short (mean of the 2.5th percentile) and long (mean of the 97.5th percentile) sunshine duration, respectively, and delayed RRs (lag 3-39 days) was 0.770 (95% CI: 0.696, 0.851) for long sunshine duration relative to the reference (mean of medians). Immediate RRs were larger for both exposures in patients admitted from home than in those from care facilities.

Conclusions: Lower mean temperature and longer sunshine duration were associated with immediate higher HF risks. Higher mean temperature and shorter sunshine duration were associated with immediate lower HF risks. These associations were modified by admission routes. Longer sunshine duration was also associated with delayed lower HF risks.

Keywords: Hip fractures; Meta-analysis; Observational study; Sunlight; Temperature; Weather.