Date of Award


Document Type

Campus Access Dissertation


Epidemiology and Biostatistics



First Advisor

Robert E McKeown


Healthcare-associated infections (HAIs) are a cause of significant morbidity and mortality in U.S. hospitals. HAI prevention is receiving increased attention on account of recent legislative mandates related to public reporting of hospital infection rates and limitations in hospital reimbursement from Centers for Medicare and Medicaid Services (CMS) for patients with hospital-acquired conditions. In order to provide a historical context for the current factors influencing hospital infection control, a literature review was conducted. In this first study, we investigated HAIs from a policy perspective and discussed the implications of using HAI rates as quality measures. This project included a study of the development of the hospital infection control specialty. Internal and external influences that have directed the practice of infection control in recent years were considered. The majority of public reporting relies on infection surveillance data conducted in accordance with National Healthcare Safety Network (NHSN) protocols, while CMS definitions rely on administrative data. To better understand the relationship between infection control surveillance and hospital administrative data, we conducted a second study that compared the infection cases identified by each methodology. Three categories of infection were considered: urinary catheter-associated urinary tract infections (CAUTI), central line-associated bloodstream infections (CLABSI), and ventilator-associated pneumonia (VAP). The sensitivity and specificity of administrative data were measured, using infection surveillance conducted according to NHSN-defined protocols as the gold standard. Positive and negative predictive values were also calculated.

The results of this research indicated that the ability of administrative data to identify infections that also met surveillance criteria varies across infection type. When a single diagnosis code in combination with a negative present on admission indicator was used to identify infections according to a restricted definition, the sensitivity of administrative data ranged from 0% for CAUTI to 25% for VAP. Additional diagnosis codes were used for expanded infection definitions, and sensitivities increased to 61% for VAP, 62% for CLABSI, and 70% for CAUTI, with little effect on specificity. The positive predictive value of administrative data was at its lowest for the restricted definition of CAUTI (0%) and highest for the restricted definition of CLABSI (41%). Overall, we found administrative data to fall short as a screening tool for infections.

A third study focused specifically on ventilator-associated pneumonia. This pilot study investigated the feasibility of using a modified Clinical Pulmonary Infection Score (CPIS) for identifying VAP in the clinical care environment of a cardiovascular intensive care unit (ICU) and neuro/trauma ICU. In contrast to other studies of the CPIS, which have occurred in clinical trial settings, this study utilized the CPIS as a daily measurement tool in a clinical care setting without conducting any tests or measures beyond what was obtained as current standard of care. The investigation provided insight into barriers to implementing the CPIS in such a setting and discussed measures that would be necessary to pursue routine use of the CPIS in the clinical care of intensive care patients at risk of VAP.