This paper reviews Pronovost et al. (2006), a landmark study examining the reduction of central catheter-related bloodstream infections in intensive care units (ICUs) across Michigan. The study implemented five patient safety measures — including hand hygiene, barrier precautions, and chlorhexidine use — across 103 ICUs in 67 hospitals over 18 months. The paper summarizes the study's design, data collection methods, statistical analysis, and outcomes, which showed a 66% decline in infection rates. It also evaluates the practical implications for individual clinicians and healthcare systems, noting the essential role of the National Nosocomial Infections Surveillance infrastructure in enabling the intervention.
The paper demonstrates effective use of evidence synthesis: it traces the five foundational studies that informed Pronovost et al. (2006) and explains how they collectively constitute the evidence base for the intervention. This technique — explicitly situating a study within its supporting literature — strengthens the critical analysis and shows the reader how knowledge builds cumulatively in clinical research.
The paper opens by introducing the intervention and its evidence base, then transitions to study design and methodology, followed by data collection procedures and statistical methods. Outcomes are reported with specific numerical results before a numbered list of five conclusions drawn from the data. A final section evaluates broader implications, including infrastructure requirements and what individual clinicians can act on independently. The structure mirrors a standard research critique format appropriate for undergraduate health sciences coursework.
Teams of clinicians at five hospitals across Michigan reported their efforts to reduce the frequency of catheter-related bloodstream infections in intensive care units (ICUs) through the implementation of five patient safety measures: (1) hand washing, (2) full barrier precautions during insertion of central venous catheters, (3) use of chlorhexidine to clean the insertion site, (4) avoiding the femoral site for catheter insertion, and (5) never leaving unnecessary catheters in the patient (Pronovost et al., 2006).
The intervention was based on published research findings from five groups: Berenholtz et al., 2004; Cohran et al., 1996; Eggimann et al., 2000; Warren et al., 2004; and Warren et al., 2006. Three of these studies investigated the efficacy of interventions on catheter-related bloodstream infections in an ICU setting, while the other two evaluated the effect of interventions more generally. All emphasized clinician education in the prevention of these infections, and together this group of studies represented the evidence base upon which Pronovost and colleagues (2006) designed the intervention they tested.
The study design involved designating ICU team leaders — consisting of a physician and a nurse — who would be trained in the intervention strategies (Pronovost et al., 2006). Through a partnership with hospital-based infection-control professionals, the team leaders implemented the intervention and collected infection data. In addition to the five patient safety measures, a number of other measures were implemented, including checklists, the creation of a central-line cart with all necessary supplies, best practice procedures that prevented clinicians not adhering to guidelines from continuing, and reminders and feedback at daily rounds and meetings.
The independent variable in this study was the intervention, and the dependent variable (outcome) was the incidence of central catheter-related ICU infections occurring over time. The study period ran from March 2004 to September 2005 — a full 18 months — and took place in 103 ICUs across 67 Michigan hospitals. About half of the ICUs, however, did not contribute to baseline data.
Central catheter-related infections were defined according to National Nosocomial Infections Surveillance (NNIS) guidelines (Pronovost et al., 2006). Infections were determined by culture and validated by hospital infection-control officers. Data were collected by these officers and submitted to researchers once a month. One catheter day was defined as the use of one or more central lines per patient per day. For the purpose of data analysis, three months of sequential data were grouped to form quarterly data.
Statistical analysis was based on calculations of median values and interquartile ranges, because the distribution of the data was non-normal — that is, the data did not form the bell-shaped curve necessary for the use of more common and powerful statistical tools. Significance between baseline and outcome data was determined using a two-sample Wilcoxon rank-sum test and was based on a two-sided alpha of 0.05. Possible interactions between hospital status as teaching or non-teaching, bed size, and geographic location were explored using a generalized linear latent and mixed model with a Poisson distribution.
You’re 53% through this paper. Sign up to read the remaining 3 sections.
Sign Up Now — Instant Access Already a member? Log inAlways verify citation format against your institution’s current style guide requirements.