Antibiotic Resistance in a Hospital ICU
An infection-control committee reviews 5 years of data from a 20-bed ICU. Carbapenem-resistant Klebsiella pneumoniae (CRKP) rose from 4% to 18% of K. pneumoniae isolates. Carbapenem use rose 30% in the same period; hand-hygiene audits ranged 55–80%; cleaning compliance was not measured. A new rapid test could identify carriers on admission, but it costs $45/test and has 85% sensitivity. One physician says, 'The bacteria figured out the drug; we need a stronger carbapenem.' Another says, 'This is just better detection.' The committee must choose one intervention for the next fiscal year: stricter antibiotic stewardship, hand-hygiene coaching, admission screening, or environmental cleaning.
1. Evaluate the two physicians' claims. What evidence would distinguish increased resistance from increased detection?
Read the worked answer
Both claims outrun the data. The 'bacteria figured out the drug' statement is teleological; resistance arises by selection of pre-existing variants, not learning. The 'better detection' claim is possible but unsupported without method stability data. The rise could reflect true selection from increased carbapenem use, clonal spread, or surveillance artifacts. To distinguish:
- Confirm constant culture methods, breakpoints, and denominators (incidence per 1,000 patient-days, not percent of isolates).
- Use molecular typing to see whether cases are one expanding clone or many unrelated strains.
- Compare admission screening results with later cultures to separate imported carriage from ICU acquisition.
- Link antibiotic exposure to resistance at the patient level.
A 'stronger carbapenem' is risky: many CRKP strains produce carbapenemases, so higher dose may add toxicity and selection without overcoming resistance. The committee should not act on either claim alone.
2. Choose one intervention for the next fiscal year and justify it, including likely effect, feasibility, and one unintended consequence.
Read the worked answer
Choose hand-hygiene coaching with audit and feedback. It targets transmission directly, is feasible in a 20-bed ICU, and does not require new laboratory infrastructure. Baseline compliance of 55–80% leaves substantial room for improvement; multimodal coaching can plausibly reduce healthcare-associated infections by 10–20%, though the effect on CRKP specifically will be smaller and slower than the effect on all hospital infections. It also addresses gaps left by unmeasured cleaning compliance. Unintended consequence: audit fatigue or gaming of observations, so use direct observation plus product-use data and avoid punitive reporting. Admission screening would identify carriers but at 85% sensitivity would miss 15%, and it would require isolation capacity. Stewardship is important but acts more slowly on resistance. Hand hygiene is the best single first-year investment because it is cheap, measurable, and broadly protective.
3. How does the rapid test's 85% sensitivity affect interpretation of a negative admission screen, and how should it shape isolation policy?
Read the worked answer
A negative screen does not rule out carriage. If ICU prevalence is 10% and specificity is 99%, the negative predictive value is approximately (0.99 × 0.90) / [(0.99 × 0.90) + (0.15 × 0.10)] ≈ 0.983, so about 1.7% of negative patients would still be carriers. At 18% prevalence, the missed fraction is larger. Policy should therefore not cohort test-negative high-risk patients with low-risk patients or relax contact precautions solely on a negative rapid result. Use the test to flag positives promptly, but retain precautions for high-risk patients until a second culture or molecular test is negative. Document the test's limits so clinicians do not overinterpret a negative result.