Positive predictive value of the Agency for Healthcare Research and Quality Patient Safety Indicator for central line-related bloodstream infection ("selected infections due to medical care")

J Healthc Qual. Mar-Apr 2011;33(2):29-36. doi: 10.1111/j.1945-1474.2010.00114.x. Epub 2011 Jan 11.


As part of the Agency for Healthcare Research and Quality Patient Safety Indicator (PSI) Validation Pilot Project, we evaluated the criterion validity of PSI 7. At the time of this study, PSI 7 was entitled "Selected Infections Due to Medical Care" and targeted catheter-related infections and inflammatory reactions. We conducted a retrospective cross-sectional study of 23 volunteer U.S. hospitals, where trained abstractors reviewed a sample of records that met PSI 7 criteria from October 1, 2005 to March 31, 2007. Of the 191 cases that met PSI 7 criteria, 104 (positive predictive value = 54%, 95% confidence interval: 40-69%) represented true infections. Of these cases, 77 (74%) were associated with central venous catheters, 15 (15%) were associated with peripheral intravenous (n=13) and or or arterial catheters (n=6), and 12 (11%) were associated with unknown catheters. Of the 87 (46%) false-positive cases, 41 (47%) did not have a qualifying infection identified by the abstractor, 38 (44%) had an infection present on admission, and 8 (9%) had an exclusionary diagnosis. PSI 7 has a low positive predictive value compared with other PSIs recently studied. Present on admission diagnoses and improved coding for infections related to central venous catheters (implemented October 2007) may improve validity.

MeSH terms

  • Catheter-Related Infections / epidemiology*
  • Catheterization, Central Venous / adverse effects*
  • Catheterization, Peripheral / adverse effects*
  • Confidence Intervals
  • Cross-Sectional Studies
  • False Positive Reactions
  • Female
  • Humans
  • Male
  • Middle Aged
  • Pilot Projects
  • Predictive Value of Tests
  • Quality Indicators, Health Care*
  • Retrospective Studies
  • United States / epidemiology
  • United States Agency for Healthcare Research and Quality