Abstract
Background and Objectives: Comparing outcome measures in health care is a relatively common practice. Reports are designed to compare hospital infection rates in an accurate and fair manner. The current methodology used by New York State (NYS) has some limitations and flaws. This research provides a methodology that overcomes these limitations and flaws.
Methods: The methodology is a replication study using data from NYS and includes the use of the binomial and Poisson distribution to calculate upper tail (UTP) and lower tail probabilities (LTP). The UTP is used to screen for poor performers, and the LTP is used to screen for good performers.
Results: The results we obtained using the tail probability method compared with NYS's confidence interval approach are similar across all health care-associated infection (HAI) categories but have the benefit of allowing for the analysis of any hospital regardless of the number of procedures, number of central line-days, or number of patient-days. In addition, we provide an evaluation of a hospital's performance over time where we identified hospitals that were consistently performing poorly and others consistently performing well.
Conclusion: Identifying hospitals that are consistently performing poorly and hospitals consistently performing well will allow administrators and clinicians to focus their efforts including budgetary to where improvements are needed. Patient care and the reduction of HAIs are a priority for health care institutions. While the results are similar to those reported by NYS, this approach can be used more comprehensively and can be interpreted more easily by administrators and practitioners. Health care administrators and clinicians may find the information useful to address infection rates. Hospitals consistently performing well may be used as a benchmark.