Case-matching methodology as an adjunct to trauma performance improvement for evaluating lengths of stay and complications.

Objective: To apply case-matching methodology to a statewide trauma registry to identify for peer review one trauma center's patients with "unexpected" survival deaths, complications or prolonged length of stay in hospital (H-LOS) or in Intensive Care Unit (ICU-LOS).

Methods: Matching patients were defined by mechanism of injury, age, physiology and intubation status on emergency department admission, preexisting conditions, and the pattern and severity of anatomic injury. Matches for one trauma center's 1997 patients (TCP) were sought from other centers' patients (SWP) submitted to the statewide registry from October 1, 1993, though December 31, 1997. A minimum of 20 SWP matches was required for creating a matching group. TCP were compared with matching SWP for survival, H-LOS, and ICU-LOS and the occurrence of complications. Unexpected survivors and deaths identified by matching (by using the standard 50% cutpoint) were also evaluated by TRISS and the American College of Surgeon's Committee on Trauma. A patient whose complications occurred in less than 10% of matching patients or whose H-LOS or ICU-LOS exceeded the 90th percentile of its matching SWP were also designated for peer review.

Results: The potential matching pool included 69,660 SWP. At least one SWP match was found for 565 (92.9%) of the 608 TCPs. The average, median, and maximum number of matches were 917.6, 235, and 6,504 patients, respectively. The 451 patients with 20 or more SWP matches were further studied. Case matching and TRISS agreed on the identification of five unexpected deaths; the American College of Surgeon's Committee on Trauma agreed on four of five. Peer review included, however, that none of the deaths were clinically unexpected. Case matching identified 14 TCP with 29 complications; only 4 of those patients had undergone concurrent peer review. Retrospective review revealed two preventable and potentially seven preventable complications. There were 35 TCP patients with prolonged H-LOS and 24 with prolonged ICU-LOS. Peer review of those cases identified that complications and problems related to discharge planning contributed to the prolonged ICU-LOS (20.8% and 25.0%) and the H-LOS (20.% and 48.6%), respectively.

Conclusion: Peer review of patients identified by case-matching methodology uncovered opportunities for system improvement that were missed by the concurrent performance improvement process. This method may also allow identification of anticipated H-LOS and ICU-LOS to promote earlier discharge.