Measurement and determinants of public hospital efficiency in West Bengal, India

Abstract

This article focuses on analysing the efficiency of secondary-level government-run hospitals in West Bengal, a medium performer state in terms of health indicators, in India. Two main objectives of this article are (1) to measure the technical efficiency of the same and (2) to identify the factors determining their inefficiency. For the first purpose, output-oriented data envelopment analysis (DEA) under variable returns to scale has been used. For the second purpose, we have used two-part regression models, first showing why some hospitals are efficient while others are not and, secondly, what are the factors that determine the relative efficiency of inefficient hospitals. We have used different forms of regression models for both types. The results show that the overall mean efficiency of all hospitals is 0.728, suggesting that on average the hospitals could produce at least 37 per cent more of output with same input volume if they had produced efficiently. The results also reveal that the highest contributor to slack is the group D staff, followed by doctors. It is found that increase in average length of stay, availability of free medicine from the hospital, ratio between doctor and non-doctor staff strength, outpatient bed day and share of emergency admission significantly decrease the efficiency of a hospital, while the ratio of nurses to non-nurses affects it positively. The article concludes with policy suggestions based on the analysis that decrease in average length of stay (ALOS) by better medical intervention and improved quality of care would definitely allow for more patients to be treated, while increase in nursing staff vis-à-vis other staff would also result in better care. Furthermore, as far as availability of medicines is concerned, the policy should not be to reduce access and coverage, but should rather centre round improving the quality of the drugs procured and distributed through the hospitals. The article concludes that the main source of inefficiency in a specialist hospital in a typical developing country is not just resource crunch, but huge gaps in planning and implementation by the central authorities as well as managerial inefficiency of the local hospital establishment. The determinants that play a vital role in increasing efficiency levels also indicate that most of the inefficiencies can be tackled either with long-term planning or by quick intercession by the hospital authority itself.

Keywords:

• hospital efficiency
• data envelopment analysis
• two-part regression models
• West Bengal
• India

Acknowledgements

The study was funded by Department of Health and Family Welfare, Government of West Bengal, India. The fund providers also helped us in preparing the survey questionnaires.

Notes

^1. In fact, West Bengal is the second most densely populated state of India, as per Census (2011).

^2. The reason behind choosing secondary hospitals as this study is the fact that secondary hospitals are said to be the ‘face of the health system’ in West Bengal as lying between primary care and super-speciality tertiary care and playing an important role for the proper functioning of the referral system.

^3. We had to leave out certain hospitals in Darjeeling and Bankura due to political unrest.

^4. The survey was funded by the Department of Health and Family Welfare, Government of West Bengal. The survey period was from July 2010 to November 2010. The executive summary of the report is available at http://www.wbhealth.gov.in/notice/summary.pdf.

^5. The study showed that only 44 per cent of total OPD doctor hours were actually held in West Bengal during the surveyed period.

^6. We here considered perception of quality from the patients only in inpatient services because only an admitted patient can have a clear perception about the whole gamut of services in the hospital, namely, diet, cleanliness, security, manpower efficiency and behaviour.

^7. We could not use the data of one hospital as it failed the consistency check.

^8. There are press reports that doctors do not offer regular ward visits and they do not meet the patient parties either.

^9. There are already media reports claiming that the quality of government procured and distributed drugs is highly questionable (Anandabazar Patrika, Kolkata Edition, 1 January 2012). Anticipating any endogeneity problem, the variable was dropped and the regression was run again, but with almost no change in the significance of other variables.

^10. If we run the DEA programme once again with just one output as inpatient numbers, the VRSTE scores obtained are found to be negatively associated with OPBD, which supports the existing literature too. This supports our earlier view that actually in public hospitals in India all the inputs are provided according to the bed size.

^11. Formally, the partial correlation between X and Y given a set of n controlling variables Z = {Z₁, Z₂, …, Z_n}, written as ρ_XY·Z, is the correlation between the residuals R_X and R_Y resulting from the linear regression of X with Z and of Y with Z, respectively.

^12. Many directions towards policy of the report have already been considered by the state government.