Reprodutibilidade do índice de risco para infecção em sítio cirúrgicodo National Nosocomial Infections Surveillance System (CDC, USA) emhospitais de Belo Horizonte

Abstract

To verify the reproducibility of National Nosocomial Infections Surveillance System (NNISS) Surgical Site Infections Risk Index (SSIRI) of USA Centers for Disease Control and Prevention, one historical cohort of 244,879 surgical procedures carried out, from 1993 to 2006, in threehospitals of Belo Horizonte, Minas Gerais, Brazil was analyzed (median beds: 120; two gynecoobstetrics and one clinical-surgical hospitals; none with academic affiliation). Its data had been compared with the ones from 2004 surgical site infections (SSI) report from NNISS (period: 1992 to 2004), using Ó 2 and Bonferroni¡¦s correction for multiple comparisons statistical methods. In Brazilian hospitals active SSI surveillance was carried out, including telephone contact with the patient at the thirtieth postoperative day (success in 60.1% of cases). In local database it hasan evident predominance of gyneco-obstetric procedures (47.8%). It has, also, a very great occurrence (97.8%) of procedures in the two index strata of lesser risk of SSI. In local database, it found a global SSI rate in endoscopic access lesser than in non-endoscopic one in five proceduresgroups. The SSIRI did not stratify risk of infection in 25 procedures groups with 46.7% of procedures in local base. In 19 procedures groups, with 45.8% of local procedures, it had occurred index groupings different of ones in NNISS. And only in 3 procedures groups, with 7.5% of local database procedures, it had index groupings similar to those in NNISS. The causes for the differences between local results and those from NNISS can be the SSIRI inadequate power for: a) detect differences in intrinsic risk (nutritional status, race, etc) that would not be measured adequately by the score of the American Society of Anesthesiologists used in the index to evaluate the intrinsic risk for SSI; b) detect the differences in quality of preventive cares; c) stratify risk of SSI in all kinds of population; d) control differences in active SSI surveillance,especially in extent of post-discharge follow-up; or e) control quality differences in information collection. It concludes that, in studied database, SSIRI is not reproducible. Hospitals of this base are very different from the NNISS ones. Comparisons of SSI rates can not be made with the reports from NNISS, but only within the local database. Efforts should be made to develop a more appropriate SSI risk index.