Benchmarking hospital lengths of stay using histograms

The authors demonstrated that length of stay histograms can provide considerably more benchmark information concerning hospital lengths of stay than numerical benchmarks. Examples of histograms described the complete distribution of hospital stays, as well as levels of outliers, rather than simple numerical averages. Gathering such data led to a clearer understanding of the significant LOS impact of certain DRG outliers in the two different hospitals in Syracuse, NY. Given that the other two communities represented, Seattle, Washington and San Diego, California, were more influenced by extensive managed care penetration, variations in histogram data were less in evidence there. Histograms were designed with bars to show LOS distributions at the 50th, 75th, and 90th percentiles for each of the above DRGs. The greatest variations could be shown when comparing the 1997 LOS data on the various DRGs at the two Syracuse hospitals. At both hospitals the presence of a large contingent of outliers (for different types of mostly medical patients) could be seen as the major factor in driving up their overall LOS.