When QI Data Tell a Convenient Story: CAUTI Reduction and Ethical Pressure
You lead a quality improvement project in a 32-bed ICU. Baseline catheter-associated urinary tract infection (CAUTI) rate was 2.4 per 1,000 catheter-days. After a nurse-driven removal protocol, the rate fell to 1.1 per 1,000 catheter-days over 8 months. While preparing a grant application for $200,000, you discover that 3 of the 12 baseline months lack catheter-day denominators. The infection preventionist believes those months had higher census and more catheter use. A senior physician suggests excluding the three missing months from the baseline because “the improvement will look cleaner and the grant is important for the unit.” The CMO wants to submit the project for a state quality award next month. A staff nurse privately tells you she felt pressured to document removal criteria as met even when they were not.
1. Analyze the ethical and professional issues in the missing-data suggestion and the documentation concern. What actions would you take, and why?
Read the worked answer
The suggestion to exclude the three missing months is data manipulation, not cleaning. Missing catheter-day denominators are unlikely to be missing completely at random; the infection preventionist’s explanation that those months had higher census and more catheter use means exclusion could bias the baseline downward or upward and make the improvement look cleaner than it is. Ethically, this violates integrity, transparency, and stewardship of public funds; it could mislead the state award and grant reviewers and erode trust. I would not exclude the months silently. I would document the missingness, attempt to recover denominators from billing or device logs, and if unavailable, prespecify a sensitivity analysis with multiple imputation or report a range. The documentation pressure is a separate patient-safety and professionalism issue: I would thank the nurse, protect her from retaliation, review the charts, and escalate through the compliance or quality-reporting channel. I would inform the CMO that the award submission must include limitations and that the grant budget can support data infrastructure, not just the intervention.
2. How would you design a credible program evaluation that addresses the data-quality problem? Include measures, analysis, and communication with stakeholders.
Read the worked answer
Program evaluation should use a pre–post design with a comparison ICU if available, and report both CAUTI rate per 1,000 catheter-days and catheter utilization ratio (catheter-days/patient-days). Data-quality plan: define denominators, assign ownership, audit 10% of charts monthly, and use run charts or SPC charts with annotated interventions. Outcomes: CAUTI rate, catheter days, appropriate removal, and balancing measures such as urinary retention, incontinence-associated dermatitis, and reinsertion. Process measures: documented removal criteria, nurse-driven removal compliance, and daily necessity review. Analysis: report missing data transparently, use interrupted time series if enough points, and avoid claiming causation from a simple before–after drop. Stakeholders-infection prevention, nursing, physicians, finance, and patients-should review the dashboard monthly. Communication: a one-page summary with the improvement, the uncertainty, and the data limitations; the grant application should propose a data-quality officer and a prospective evaluation. This makes the project credible and protects the unit from a backfire if the award reviewers find the missing months.