Hospital Efficiency in Sub-Saharan Africa Evidence from South Africa

This study evaluates the technical efficiency and productivity of a sample of public sector hospitals in three provinces of South Africa using the non- parametric techniques of data envelopment analysis (DEA) and DEA-based Malmquist productivity index (MPI). A tobit regression is also estimated to identify some factors that may be associated with (in)efficiency. The sample consists of 86 hospitals classified into three levels: community hospitals with emergency services only (Level I), community hospitals with outpatient services (Level II) and non-academic secondary and tertiary hospitals (Level III). Recurrent expenditure and bed-size are used as inputs (but only the first one in Level III hospitals because of their small number). Outputs include inpatient days and outpatient visits - these are expected to capture the bulk of the activity of non-academic hospitals.The findings indicate that there is a marked variation of performance among hospitals within each group. An average overall technical efficiency of 0.74, 0.68 and 0.70 is computed for Level I, II and III hospitals respectively. This implies that when compared to their peers on the frontier, the inefficient hospitals on average consume 35-47 per cent more resources. The number of hospitals on their respective group's frontier is 6 in Level I (n=55), 3 in Level II (n=19) and 2 in Level III (n=12). Most hospitals operate at a non-optimal scale, with decreasing returns to scale dominating in Level II and III hospitals.If the inefficient hospitals were to operate as efficiently as their peers on the frontier, efficiency gains in terms of reduction in recurrent expenditure would amount to about R 279 million (about US$ 47 million) - an amount which can cover the costs of constructing a sizeable number of clinics or upgrading service quality where necessary. This would offset the need to raise user charges, and would potentially be more equitable. Furthermore, the number of hospital beds could also be reduced significantly.The occupancy rate affects the technical efficiency positively in all three Levels of hospitals. The average length of stay, however, seems to have an adverse effect only in Level III hospitals. The tobit estimates for Level I hospitals indicate that the number of outpatient visits as a proportion of inpatient days affects technical efficiency positively. This might be an indication of the presence of economies scope between the two outputs. The MPI for a sample of Western Cape hospitals exhibits a decrease in total factor productivity of about 12.5 per cent for the period 1992/93-1997/98. This is a result of both a decline in efficiency and a technical regress.These results indicate the potential to improve access and/or quality of care without injecting additional resources into the health sector. This is important given the financial constraints on social sector investment in South Africa. It is also concluded that inefficiency levels of such a magnitude are likely to undermine the government's initiatives to redress inequity in a sustainable manner. Finally, it is desirable to replicate this study on a large scale, covering all types of hospitals (public and private) so as to assess the gravity of the problem and its causes, and thereby maximize possible efficiency savings.