Determining "Like" Hospitals for Benchmarking

Purpose:
This paper reports the examination of hospital size as a proxy characteristic to define like hospitals for the purpose of benchmarking nurse sensitive outcomes.

Background/Significance:
Selection of appropriate “like” hospitals is critical for benchmarking performance. Based on 10 years of observed outcome differences between small and large hospitals, CALNOC sought to empirically define small hospitals and determine if they were similar to current administrative categories; and to determine if there were statistical differences between small and large hospitals for selected nursing sensitive outcome indicators (falls, falls with injury, and HAPU 2+).

Methods:
We looked for an optimal classification of hospitals into small and large size so that the resulting groups were the best outcome predictors. In the optimization search, we varied the hospital size from 30 to 310 and calculated the sample median rate of outcome and classified rates as low or high. This resulted in contiguous size and homogeneous hospital groups(resulted in a two-by-two table of hospital size by outcome level). Accuracy of prediction was measured by the logistic regression c-statistic (using small or large hospital size as a predictor). Outcomes, patient characteristics, and hospital variables were then compared across hospitals using t-tests for differences in means.

Results:
Findings suggest that optimal classifications into small and large hospital size based on the outcome indicators of falls, falls with injury, and HAPU 2+ were not consistent with historical administrative categories based on average daily census and not consistent by outcome. Statistical differences were only found with HAPU 2+ in critical care units, with no differences in the fall outcomes. These analyses suggest that comparison of like-sized hospitals may have limited value for the purpose of benchmarking performance for hospitals on the quest for high reliability and top performance.

Conclusions and Implications for Practice:
Further research is needed to continue to explore data-based hospital size comparisons with other outcomes or in states where staffing has not been held constant and equalized by legislated nurse-patient ratios.