Toward optimal recording of surgical complications: concurrent tracking compared to the discharge data set

BACKGROUND: Information extracted from the hospital discharge data set is used increasingly for outcomes research and for benchmarking hospital and provider performance. The accuracy of these data in detecting vascular complications has never been validated. METHODS: We compared morbidity and mortality data derived from the hospital discharge data set to similar data recorded concurrently by our Surgical Activity Tracking System (SATS) for 1 year on the vascular surgery service. RESULTS: Of 798 total admissions, no complications were detected by either system in 598 admissions (75%). In 200 admissions (25%), there were 335 complications, including 24 deaths (3.0%), that occurred either in-hospital or within 30 days of the date of operation or the date of discharge for nonoperative admissions. Of the 335 complications, 180 (53.7%) were recorded by both systems; the SATS missed 59 complications recorded in the hospital discharge data set (17.6%), whereas the hospital discharge data set missed 96 complications recorded in the SATS (28.7%, P = .003). Of the 289 in-hospital complications, the SATS recorded 230 (79.5%), whereas the hospital discharge data set recorded 229 (79.2%). Of the 24 deaths, the hospital discharge data set missed 6 that occurred after discharge but within the 30-day reporting period CONCLUSIONS: Both systems are not completely accurate for tracking inpatient complications. The SATS was more representative than the hospital discharge data set in capturing 30-day morbidity and mortality. An amalgamation of the 2 systems would provide more optimal tracking of complications.