Experience With Two Validation Methods in a Prevalence Survey on Nosocomial Infections

Abstract Objective: To determine whether an investigator effect remained on the first German study on the prevalence of nosocomial infections Nosokomiale Infektionen in Deutschland Erfassung und Prävention (NIDEP), despite extensive validation efforts. Design: Two validation methods were applied: bedside validation and validation by case studies. In both cases, the results of the four investigators were compared with the diagnosis of gold standard observers. Setting: Validation measures were applied before, intermittently, during, and at the end of the surveillance period in 72 acute-care hospitals with 14,966 patients. Results: The overall sensitivity in the bedside-validation periods was 89.0%; the overall specificity was 99.5%. For validation by case studies, overall sensitivity was 95.6%, and overall specificity was 92.8%. At the end of the surveillance, a remarkable investigator effect was found. Conclusion: Despite validation results that were assessed as satisfactory, based on available literature, an investigator effect was observed. This underlines the need for data validation and the formulation of recommendations for data validation. Clarification of the Centers for Disease Control and Prevention criteria for pneumonia and primary bloodstream infection and the inclusion of some diagnostic test results may reduce or prevent an investigator effect in future studies.