Use of censored data to monitor surgical-site infections.

ABSTRACT

OBJECTIVE: To take into account the proportion of patients lost to follow-up when calculating surgical-site infection (SSI) rates. DESIGN: A multicenter SSI monitoring network in Basse-Normandie, France, using the definitions for SSI of the National Nosocomial Infections Surveillance System of the Centers for Disease Control and Prevention. PATIENTS Between January 1, 1998, and December 31, 1999, 3,705 patients were operated on in 25 units of 10 institutions. RESULTS: Of the patients, 41.2% (range, 5.1% to 95.5%) were seen 30 days or more after their operation. The global SSI attack rate was 2.19% (95% confidence interval, 1.72% to 2.66%). With the use of the Kaplan-Meier method, the incidence rate was 3.11% (95% confidence interval, 3.06% to 3.16%). The difference between the attack rate and the Kaplan-Meier incidence rate for each unit varied according to the percentage of patients seen on or after day 30 postoperatively and the number of SSIs diagnosed in patients seen on or after day 30. CONCLUSIONS: Practice guidelines are needed for the international monitoring for postdischarge SSIs and the calculation of SSI rates. The proportion of patients seen 30 days after their operation is a major quality criterion for SSI monitoring and should be routinely given in monitoring reports, oral communications, and publications to compare results obtained by different teams