EFFECTS OF CLIMATIC FACTORS ON SPATIAL-TEMPORAL DISTRIBUTION OF HIP FRACTURE IN PORTUGAL
METHODS: From the National Hospital Discharge Register we selected admissions of patients ≥50 years-old, with diagnosis of HF (codes 820.x, ICD9.CM) caused by low energy traumas. We exclude cases of bone cancer and readmissions for after-care. Data from meteorological stations were obtained from “Instituto Português do Mar e da Atmosfera”. An exposure assessment inference for Portugal was made by a geostatistical procedure: with the geographic coordinate stations and year/monthly mean of mean temperature (MT;ºC), rainfall (RF;mm), relative humidity (RH;%) and sunshine daily (SSD;hours), a suitable semivariogram model was selected to characterize the spatial variability and interpolation was performed using kriging. A Poisson general additive model (GAM) was used to estimate the relative risk (RR) of HF associated with changes in CF, adjusting for space, time trends, seasonal variations, socioeconomic status, rural condition and age-group structure. A model without (model1) and with (model2) CF were assessed.
RESULTS: We selected 96,905 HF, 77.3% being in women. On average, women were older than men (81.1±8.5 vs 78.1±10.1 years; p<0.001) at admission. Except for RF, a significant inverse association between CF and HF was found (women: RRMT0.988(p<0.001), RRHR0.993(p<0.001), RRSSD0.986(p<0.001) and men: RRMT0.994(p<0.05), RRHR0.993(p<0.001), RRSSD0.979(p<0.001)). Model2 had a significantly better performance and small changes between model1 and model2 in the prediction of the spatial distribution of the RR were observed, especially in regions with higher climate variation between seasons.
CONCLUSIONS: Meteorological variables seem to explain part of the seasonal effect on HF incidence and these findings may have an implication on the heath management in the context of climate changes.