Five CRKP Isolates in a Medical ICU After a Lab Method Change
The 24-bed medical ICU has had 5 patients with carbapenem-resistant Klebsiella pneumoniae (CRKP) isolated from clinical cultures between March 1 and April 12, compared with 1 such isolate in the previous 12 months. The ICU logged 1,850 patient-days during this period. Three isolates came from respiratory specimens, one from urine in a catheterized patient with no urinary symptoms, and one from blood. The microbiology laboratory switched from disk diffusion to a molecular carbapenemase panel (blaKPC/blaNDM/blaOXA-48) on February 1. Two of the five patients had respiratory specimens collected on the same morning by the same respiratory therapist, and both were colonized without clinical signs of pneumonia. The ICU medical director asks whether this is an outbreak and wants a plan by Friday.
1. Is this an outbreak? What data would you assemble to decide, and what findings would change your conclusion?
Read the worked answer
Start with rates, not counts. 5 isolates / 1,850 patient-days x 1,000 = 2.7 per 1,000 patient-days versus roughly 1 isolate in the prior 12 months. Get the prior-year denominator to make a real comparison, and add device-specific rates (VAP, CLABSI, CAUTI) since three of five are device-associated specimens.
Then test the ascertainment hypothesis before the transmission hypothesis. The lab changed methods on February 1. Molecular carbapenemase testing detects resistance that disk diffusion may have missed or under-reported, so part of the rise is likely increased detection. Ask the lab for: (a) the number of CRKP isolates per month for 18 months, not just the ICU; (b) whether the five isolates are confirmed by a reference method; (c) a comparison of pre- and post-February susceptibility profiles.
Distinguish colonization from infection. Four of five look like colonization (asymptomatic bacteriuria, respiratory colonization without pneumonia). Colonization still drives transmission, but it changes urgency and messaging.
Rule out pseudo-outbreak: same-therapist respiratory specimens raise the possibility of contaminated bronchoscope, shared suction equipment, or specimen contamination. Pull reprocessing logs for the bronchoscope used, and send the two respiratory isolates for whole-genome sequencing (WGS) - clonality supports true transmission; unrelated strains support lab or procedural artifact.
Build a line list: admission/discharge dates, room, roommates, devices, procedures, antibiotic days, prior facilities. Plot by week and by room to look for person-place-time clustering.
2. Given one IP and no extra staffing, what would you do in the first two weeks, and in what order?
Read the worked answer
Days 1-3 (low cost, high yield):
- Place all known CRKP patients on Contact Precautions; cohort with dedicated equipment if rooms allow.
- Direct observation of hand hygiene and PPE doffing on all shifts - report unit-level compliance, not hospital-level.
- Confirm that shared items (glucometers, ECG leads, ultrasound probes, mobile computers) are disinfected between patients; remove shared supplies from hallways.
- Review environmental cleaning with EPA-registered product against the organism; ask EVS to verify dwell time.
- Ask the lab to save all CRKP isolates for WGS and to notify IP in real time.
Days 4-14: 6. Initiate active surveillance (rectal or perirectal swab) on ICU patients at admission and weekly if the lab can absorb the volume - this is the only way to find the silent reservoir. If capacity is absent, prioritize patients in rooms near known cases and those with prior admissions. 7. Antimicrobial stewardship review of carbapenem days and unnecessary device days - the strongest modifiable driver of CRE selection. 8. CHG bathing if not already standard.
Do not spend the first two weeks writing a policy. Prioritize interventions that interrupt transmission today; document decisions and rationale as you go.
3. How would you decide, four weeks later, whether transmission is still occurring?
Read the worked answer
Use three independent signals rather than isolates alone:
- Incident clinical isolates plotted weekly on a statistical process control chart, with the pre-period median as the centerline. A run of points above the centerline, or any point above the upper control limit, indicates special-cause variation.
- Point-prevalence surveillance cultures at weeks 2, 4, and 8. New colonized patients with no prior positive identify ongoing acquisition; the same strain in a roommate is direct evidence.
- WGS relatedness. If new isolates are genetically distinct from the index cluster, you are seeing importation, not ward transmission - the response shifts toward admission screening and interfacility communication.
Also verify that the interventions were actually delivered (hand hygiene observations, cleaning audit ATP or fluorescent marker results, surveillance culture compliance). A flat epi curve with poor process measures means the intervention was not implemented, not that it failed.
Declare containment only after at least two consecutive surveillance periods with no new related isolates and documented process compliance. Then step down to enhanced surveillance rather than stopping entirely.