infection control and hospital epidemiology april 2007, vol. 28, no. 4 original article Risk Adjustment for Surgical Site Infection After Median Sternotomy in Children Jessica Kagen, BA; Warren B. Bilker, PhD; Ebbing Lautenbach, MD, MPH, MSCE; Louis M. Bell, MD; Susan E. Coffin, MD, MPH; Keith H. St. John, MT(ASCP), MS, CIC; Eva Teszner, RN, CIC; Troy Dominguez, MD; J. William Gaynor, MD; Samir S. Shah, MD objective. To determine whether the National Nosocomial Infections Surveillance (NNIS) System risk index adequately stratified a population of pediatric patients undergoing cardiac surgery according to the risk of developing surgical site infection (SSI). design. A retrospective, case-control study. setting. An urban tertiary care children’s hospital. patients. Patients who had a median sternotomy performed between January 1, 1995, and December 31, 2003, were eligible for inclusion in the study. For all case patients, medical records were reviewed to verify that all patients met the case definition for SSI. Control subjects were chosen randomly from among all patients who underwent median sternotomy during the study period who did not develop SSI. results. Thirty-eight patients with SSI and 172 patients without SSI were included. One hundred six patients (50%) were male. The median patient age was 4 months. The sensitivity of the NNIS risk index with cutoff scores of 0 to 1 and 2 to 3 was 20%. The distribution of patients with SSI for an NNIS risk index score of 0 was 0%; for a score of 1, 80%; for a score of 2, 20%; and for a score of 3, 0%. The distribution of patients without SSI for a scores of 0 was 4%; for a score of 1, 87%; for a score of 2, 9%; and for a score of 3, 0%. The area under the receiver–operating characteristic curve (AUC) of the original NNIS risk index was 0.57. The modified risk indices did not perform significantly better, with an AUC range of 0.58 to 0.73. conclusions. The NNIS risk index did not adequately stratify pediatric patients undergoing median sternotomy according to their risk of developing an SSI. Various modifications to the risk index yielded only slightly higher AUC values. Infect Control Hosp Epidemiol 2007; 28:398-405 From the Divisions of Infectious Diseases (J.K., L.M.B., S.E.C., S.S.S.) and Cardiothoracic Surgery (J.W.G.), and the Department of Infection Prevention and Control ( S.E.C., K.H.S.J., E.T.), the Children’s Hospital of Philadelphia, and the Center for Clinical Epidemiology and Biostatistics (W.B.B., E.L.,S.S.S.), the Departments of Pediatrics (L.M.B., S.E.C., S.S.S.), Medicine (E.L., J.W.G.), Anesthesia and Critical Care (T.D.), and Biostatistics and Epidemiology (W.B.B., E.L., S.S.S.), and the Centers for Education and Research on Therapeutics (W.B.B., E.L., S.S.S.), University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania. Received May 5, 2006; accepted June 29, 2006; electronically published March 9, 2007.  2007 by The Society for Healthcare Epidemiology of America. All rights reserved. 0899-823X/2007/2804-0006$15.00. DOI: 10.1086/513123 Postoperative surgical site infection (SSI) complicates up to 22% of all operative procedures 1 and is a major source of morbidity, increased healthcare costs, and prolonged hospital stay for patients who undergo surgical procedures. 2 The in- crease in mandated reporting of hospital infection rates has created a growing need for risk adjustment for postoperative infections. To use infection rates as a basis for measuring and improving quality of care, the rates must be meaningful for interhospital comparison. Hospitals that perform advanced surgery on patients with greater severity of illness might be reasonably expected to have higher rates of nosocomial in- fection. Adjusting infection rates to account for potentially important confounding variables, such as severity of illness and the complexity of the surgical procedure, may allow for more accurate interhospital comparisons. The National Nosocomial Infections Surveillance (NNIS) System (now the National Healthcare Safety Network [NHSN]), a performance measurement system originally de- veloped in the early 1970s and devoted to hospital-acquired infections, created a risk index to stratify patients based on the likelihood of SSI. The index was devised to account for differences in intrinsic patient risk, thereby allowing a more accurate comparison of SSI rates across different institutions. The NNIS risk index scores each operation by counting the number of risk factors present among these 3: (1) a patient with an American Society of Anesthesiologists (ASA) pre- operative assessment score of 3, 4, or 5; (2) an operation classified as either contaminated or dirty-infected; and (3) an operation with duration of surgery of more than T hours, where T depends on the operative procedure being per-