amount of outputs. Further, large hospitals have, on average, higher IT investment. The analysis starts with calculating the three efficiency measures for each hospital using the linear programming problem (Eq. (1)). These measures include the efficiency measure for the sample as a whole (Pooled), the efficiency measure when the sample is partitioned (Separate), and a comparison of the pooled and separate efficiency measures (Pooled/Separate). Statistics of these efficiency measures
2 A measure of patient acuity reflecting different patients’ needs for hospital resources. There are many ways of measuring case-mix; some are based on patients’ diagnoses or the severity of their illnesses, and some on their utilization of services. A high case-mix index refers to a patient population more ill than average. 3 A system for classifying hospital patients based on their clinical condition (diagnosis or surgical procedure), age, and whether they had any other illnesses (complications or comorbidities); a predetermined price is set for each of over 500 DRGs. DRGs are used by the federal government for Medicare prospective pricing system.
B. Watcharasriroj, J.C.S. Tang / Journal of High Technology Management Research 15 (2004) 1–1610
are shown in Table 2. The efficiency measures can indicate how much input usage could be proportionally reduced and still produce the same amount of outputs. For example, an efficiency score of 0.8 or 80% of a hospital implies that input consumption of this hospital could be reduced by 20% (1– 0.8 or 100%–80%) to produce the same amount of outputs. The results of the comparison between the mean values of pooled and separate efficiency measures (Pooled/Separate) for both hospital groups in Table 2 suggest that large and small hospitals may have different frontiers. This may be because, at any level of outputs, large hospitals may practice more sophisticated production technology by using advanced medical equipment or experienced specialists. The frontier difference is assessed using a nonparametric test, the Mann–Whitney test. Because the DEA efficiency measures are calculated by the nonparametric method that has no underlying error structure, the Mann–Whitney test, which is distribution free, is considered as a suitable statistical test in this case (Grosskopf & Valdmanis, 1987).
Table 1 Descriptive statistics of input, output, and IT variables
Variable Mean (S.D.) Min Max
Large hospitals (n=29) Inputs Physicians 68 (36.2022) 19 159 Nurses 318 (98.4153) 177 557