OBJECTIVE:
Surgical-site infections are serious complications and thorough follow-up is important for accurate surveillance. We aimed to compare the frequency of complications recorded in a clinical quality database with those noted in a randomized clinical trial with follow-up visits.
DESIGN:
Evaluation study.
SETTING:
Danish Gynecological Cancer Database (DGCD) and the Danish multicenter trial on perioperative oxygen and surgical-site infections (PROXI).
SAMPLE:
Paired data from 222 patients who participated in the PROXI trial taking place at Copenhagen University Hospital, Rigshospitalet between November 2006 and October 2008 and data from the DGCD.
METHODS:
Outcomes within 30 days from the trial and the database were compared and levels of agreements were calculated with kappa-statistics.
MAIN OUTCOME MEASURES:
Primary outcome was surgical-site infection. Other outcomes included re-operation, urinary tract infection, pneumonia and sepsis.
RESULTS:
Surgical-site infection was found in 21 of 222 patients (9.5%) in the PROXI trial versus 6 of 222 patients (2.7%) in the DGCD (p < 0.01, kappa 0.42). Twelve of 15 superficial and three of six deep or organ-space surgical-site infections were registered in the PROXI trial, but not in the DGCD. Agreements between secondary outcomes were very varying (kappa-value 0.77 for re-operation, 0.37 for urinary tract infections, 0.19 for sepsis and 0.18 for pneumonia).
CONCLUSIONS:
The randomized trial reported significantly more surgical-site infections than the clinical database. The DGCD reported only 50% of the deep and organ-space infections, and hence, the low-overall agreement indicates a need for more careful database registration.