Exploring the efficiency of hospital and pharmacy utilizations in Taiwan: An application of dynamic network data envelopment analysis

Abstract

BackgroundBuilding continuity of care by enabling optimal collaboration between hospitals and community pharmacies is critical. However, existing research on healthcare efficiency has largely focused on hospital services, and few studies have evaluated hospital services alongside pharmaceutical services. ObjectiveTo set up a dynamic network production model to assess the divisional efficiency of hospital and pharmacy utilizations and to investigate the dynamic changes and spatial differences for a more productive utilization of healthcare services. MethodsA dynamic network data envelopment analysis was applied on two interdependent healthcare divisions, hospitals and community pharmacies, in order to assess utilization efficiencies in both divisions simultaneously. Subsequent analyses were performed with a set of decision making units consisting of Taiwan's cities. Data were gathered from the Ministry of Health and Welfare and the Bureau of National Health Insurance, spanning 2014–2019. ResultsThe mean scores for overall healthcare, hospital, and pharmacy among cities were 81.9%, 89.2%, and 76.7% respectively. It appears that pharmacy is not very efficient, possibly because of the low rate of prescription release. A 13.26% reduction in pharmacists in the pharmacy sector is suggested, but the number of community pharmacies remains the same in cities and counties in 2019, based on the results of the slack-based measure. The bootstrapped-truncated regression indicated the different business districts had statistically significant determinants of overall efficiency, including hospital efficiency and pharmacy efficiency. ConclusionsThis study illustrated the inefficient operations of most hospital and pharmacy divisions in Taiwan's southern cities. It is hoped to draw the government's attention to the existing inefficiency in the allocation of healthcare and pharmaceutical resources. For future studies, a dynamic network model can be adapted to introduce the Malmquist productivity index in order to compare divisional productivity changes over time. Other topics for future research include handling undesirable outputs (quality variables) and information on input/output prices.