Enumerative, Bayesian, and analytic statistical analysis of nosocomial infection for quality improvement: first of two parts

Health Matrix. 1989 Fall;7(3):18-22.

Abstract

Readmission rates are used by the Health Care Financing Administration (HCFA) as an indicator of a hospital's quality of care. Specifically, HCFA seeks to relate readmission to complications of the primary admission. A recent study at the Cleveland Clinic Foundation examined patients readmitted within 30 days of dismissal during two non-consecutive months to the cardiovascular surgery, cardiology, and gastroenterology services at this 1,008 bed hospital, accounting for 31% (1640-5342) of all hospital discharges. Only 17% (25/149) of readmissions were due to a complication of the previous hospital admission. Examining those who were readmitted for complications of their primary admission, we found that 36% (9/25) of such readmitted patients had infections that occurred during their primary admission. Otherwise stated, 5.4% (9/149) of readmissions were for nosocomial infections. These patients were compared with 20 retrospectively matched, non-readmitted patients chosen randomly from a group of 100 patients retrospectively matched for gender, age, type, and date of surgery. Using these controls, a computer-based expert system was used to help identify patient variables that are associated with readmission due to nosocomial infection. We found if a patient had (1) a weight less than 66 Kilograms or (2) had a first postoperative total serum protein greater than 6.1 or (3) was reoperated, we could predict with 100% specificity and 60% sensitivity that the patient would have an unplanned readmission. Total readmission rates do not accurately reflect complications of the primary admission.(ABSTRACT TRUNCATED AT 250 WORDS)

MeSH terms

  • Adult
  • Bayes Theorem
  • Blood Proteins
  • Body Weight
  • Cardiovascular Diseases / surgery
  • Cross Infection / transmission*
  • Gastrointestinal Diseases / surgery
  • Hospital Bed Capacity, 500 and over
  • Hospital Departments / statistics & numerical data*
  • Humans
  • Ohio
  • Patient Readmission / statistics & numerical data*
  • Probability
  • Reoperation
  • Research Design
  • Surgery Department, Hospital / statistics & numerical data*
  • Utilization Review

Substances

  • Blood Proteins