Funding Hospital Services: A Critical Analysis and Feasibility Study of the Casemix Funding Model in Iran

Background Hospitals in Iran have mainly been managed in a centralised system and funded historically through annual budgeting with little autonomy at hospital level. The current annual budgeting system is inequitable and is not reflective of hospital activity. Hospital resources are not distributed with regard to efficiency indicators and lobbying and political power of the managers are common issues influencing budget. Evidence suggests that hospitals in Iran will be even further challenged due to the growing and aging population. Reform of funding policy, particularly in hospitals, is now being considered as a critical step to improvement of Iran’s health system. Objectives This is a study of the theoretical and practical aspects of the implementation of casemix funding of hospitals in Iran. It aims to identify the knowledge and attitude of hospital managers and staff about the feasibility of casemix; to investigate availability, reliability and completeness of hospital discharge and financial data; to measure the appropriateness of the Australian Refined Diagnosis Related Groups (AR-DRGs); to build up a basis for further studies on casemix funding of hospitals; and, to assist the efficient use of scarce resources among and between hospital systems. Methods First, a descriptive survey, using an eleven-item questionnaire, was conducted to assess the level of knowledge and attitudes of hospital managers and key staff about casemix funding and its appropriateness. Second, patients’ clinical and demographic information were collected from the discharge system of a single study hospital, to evaluate the accuracy and completeness of these data for adopting casemix in Iran’s hospitals. This information was used to classify patient episodes into DRG classes using the LAETA Grouper and AR-DRGs. Third, DRG cost weights were calculated based on the internationally accepted principles of 'activity-based' cost accounting and cost-modelling, taking into account current realities of hospital accounting structures, availability of data, as well as time and budget constraints. To identify whether there is any association between modelled cost weights and length of stay at the DRG level, two statistical measures, the Pearson correlation coefficient and regression coefficient were calculated using the STATA statistical package. Finally, a total of 465,531 acute inpatient separations, from 35 hospitals, was used to examine the performance of AR-DRGs in the study environment. L3H3; IQR; and 10th- 95th percentile methods were used for excluding extreme cases. The coefficient of variation (CV) and reduction in variance (R2) were used to measure the degree of homogeneity achieved by the classification system and the extent to which the dispersion of lengths of stay could be explained by grouping the cases into the discrete DRG classes. Results The staff survey results showed that 75% and 58% of the participants had not ever heard of the terms casemix and DRGs, respectively. The majority of the participants described casemix and DRGs as a cost allocation and/or funding tool rather than a classification system useful for management and performance measurement. The most common barriers to casemix implementation outlined by the participants included: the lack of good foundation knowledge; difficulty in data access; and lack of or incomplete knowledge of the chief managers and staff about the casemix. The data quality study findings suggest that the accuracy and completeness of the available data in the study hospital is variable and not highly reliable. The grouper identified invalid records of principal diagnosis, age, sex, and length of stay for 4% of total separations. No complication and comorbidity effects were recorded for 93% of cases. Although general practitioners are employed as gate keepers to control coding accuracy, there is no standard quality control to secure the accuracy and consistency of coding either at the physician or coder level. Coders, except in a few cases, have not been formally trained. According to the data study, the estimation of DRG cost weights using a clinical costing approach is almost impossible due to inadequate financial and utilisation information at the patient level, poorly computerised 'feeder systems', and low quality data. In contrast, the cost modelling approach, using Australian service weights resulted in the average DRG cost weight of 2.723 million Iranian Rials (equal to US $295). A regression coefficient of 0.14 (CI = 0.12 − 0.16) suggests that the average cost weight increases by 14% for every one day increase in average length of stay. Classifying a total of 465,531 acute inpatient separations using AR-DRG resulted in 579 DRG classes. Although reduction in variance (R2) for untrimmed data was low (R2 = 0.17) for LOS, trimming by L3H3, IQR, and 10th-95th percentile method improved the value of R2 to 0.53, 0.48, and 0.51, respectively. Low values of R2 for DRGs within several MDCs such as MDC 02, 05, 10, 15, and MDC 20 were identified. Conclusion This study concludes that the implementation of the casemix funding of hospitals in the Iranian health system and in Iranian Social Security Organisation in particular, is quite feasible and that AR-DRGs would provide a useful basis for introducing casemix in the system. However, the effective implementation of casemix in Iran would depend on a number of factors including: active cooperation and contribution of hospital staff at all levels and in all departments in the implementation process and provision of reliable data; updating hospital information systems; improving the quality of costing information; adopting an appropriate classification system, and, finally, adequate scrutiny of health care providers’ behaviours through the regular assessment of hospital performance and quality of care.