Spatial analysis of geographic variation in inpatient rehabilitation provision under Japan’s public health insurance system: a nationwide ecological study using national open data

Background Geographic variation in inpatient rehabilitation provision may identify regions requiring further assessment of healthcare resources and service delivery. This study aimed to quantify age- and sex-standardized geographic variation in inpatient rehabilitation provision across Japan and identify regional factors associated with this variation. Methods We analyzed fiscal year 2019 data from all 335 secondary medical service areas in Japan. We calculated the standard rehabilitation provision ratio (SRPR) using indirect age- and sex-standardization based on national rehabilitation provision rates. We assessed geographic clustering of SRPRs using Moran’s I. We estimated smoothed SRPRs (sSRPRs) using a Bayesian Poisson Leroux conditional autoregressive model that incorporated the expected number of rehabilitation units. We then used an adjusted version of this spatial model to examine associations between rehabilitation provision and regional healthcare-resource and socioeconomic characteristics. Bayesian posterior inference was performed using integrated nested Laplace approximation. Results The median (interquartile range) of the SRPR for total rehabilitation units was 0.762 (0.466–1.035) and showed weak positive spatial autocorrelation (Moran’s I = 0.207, p < 0.001). Relatively high sSRPRs were observed in several areas of western Japan, whereas relatively low sSRPRs were more frequently observed in parts of eastern and northern Japan. In the spatial model, a higher number of therapists per 100,000 population was associated with greater rehabilitation provision (rate ratio [RR], 2.40; 95% credible interval [CrI], 1.97–2.93), as was the presence of at least one physiatrist (RR, 5.80; 95%CrI, 3.50–9.66). In contrast, a higher aging rate was associated with lower rehabilitation provision (RR, 0.53; 95%CrI, 0.41–0.68). Conclusions Our spatial analysis highlights potential geographic variation in Japan’s inpatient rehabilitation provision and weak spatial autocorrelation as assessed descriptively using Moran’s I. Rehabilitation provision was also associated with rehabilitation workforce availability and population aging. These findings identify regional patterns that warrant further assessment of patient-level data, regional need indicators, and cross-area utilization to better inform specific policy decisions.