Effect of an infection surveillance and control program on the accuracy of retrospective chart review

The primary analyses of the SENIC Project (Study on the Efficacy of Nosocomial Infection Control) will test the association between the presence of infection surveillance and control programs (ISCPs) and changes in nosocomial infection rates (NIRs) as measured by retrospective chart review (RCR). If the establishment of an ISCP affects the quality or completeness of information important for diagnosing infection by RCR, the analyses could be biased (i.e., there could be an increased chance of a Type I or Type II error). To determine whether this type of "ISCP effect" on the accuracy of RCR is likely to occur, the authors carried out a prospective intervention study in one hospital where 1) nosocomial infections among a pre-ISCP cohort of patients were detected by prospective data collection (PDC), 2) the hospital's first ISCP was instituted, and 3) infections were identically studied by PDC exactly two years later. Several months after the end of the second PDC, a team of trained chart reviewers read the medical records of the patients in both study cohorts and abstracted all clinical data bits used for diagnosing nosocomial infection. By a nonparametric matched correlation analysis, no significant change was found in the amount of relevant clinical information recorded in the medical records, and sensitivity and specificity did not change significantly. The authors conclude that, if an ISCP effect on RCR accuracy is present at all, it must be small.