The sensitivity of CPE surveillance tests from stool or rectal swabs was found to be 87.3% in our present study. That is, 12.7% of patients with CPE colonization would be missed if only stool surveillance tests were used. The CPE clearance rate was shown in our present analysis to be lower in the stool CPE-positive group than the stool CPE-negative group. The transmission rates were similar however when compared according to the results of the initial surveillance tests for CPE.
Because of the high morbidity and mortality associated with CPE infection, multimodal interventions for infection control and prevention in relation to this pathogen are strictly recommended in many countries [4]. In addition to hand hygiene, contact precautions, environmental cleaning, and antimicrobial stewardship, active surveillance can also be performed to control transmission in patients at a high risk of CPE colonization [7, 8]. However, the optimal anatomic sites for CPE surveillance in addition to stool sampling have not yet been determined [9, 10]. WHO and CDC guidelines currently recommend CPE surveillance testing using stools or rectal swabs. However, as indicated in a previous study, the sensitivity for detecting CPE in this way is still only about 88% (95% CI 68–97), which will lead to a relatively high proportion of false-negative test results [5]. In our current study, we confirmed this sensitivity level (87.3%), and that it is suboptimal for definitively ruling out CPE colonization. Notably however, we found that a combination of stool and sputum testing will detect CPE colonization with a better sensitivity of 93.7%.
To the best of our knowledge, there remains little available evidence whether additional surveillance tests at sites other than stool could usefully improve the sensitivity of CPE detection [5]. In our current study cohort, 10 patients in the initial stool CPE-negative group showed a conversion to CPE positivity in a subsequent stool sample at a median of 6.0 days, and 9 of these patients had initial CPE-positive results from a sputum test. Although these patients would still have been confirmed with a CPE infection after about one week if surveillance tests had not been conducted at other sites, the transmission rate from close contact with other individuals was not significantly different among our initial stool CPE-positive, the initial sputum CPE-positive, and the initial CPE-positive from other sites groups. Hence, the risk of CPE transmission might be increased in situations where only stool testing is done and contact precautions are not taken in the period before the index patient is confirmed as being positive. Additional surveillance tests at sites other than the stool would thus have value in terms of control and prevention of CPE transmission. However, this would necessarily increase medical costs and the clinical burden in hospitals. We thus suggest that stool and sputum testing for CPE may be reasonable in resource-limited settings and that additional sites such as sputum, urine, and drained fluid can be tested in more well-resourced settings. Because of the small sample sizes of the close contacts with our index patients in the initial CPE-positive at other sites group, further studies with a larger cohort are needed to evaluate the transmission profile for cases such as these.
We found from our present analysis that the clearance of CPE was more common among the initial stool CPE-negative cases that had a positive isolation of these bacteria from urine, bile, and JP or pigtail drained fluid than from the sputum (9/10 vs 1/10 patients). We speculated that a subsequent CPE positive result from a stool sample would be likely in the patients who showed a positive result in an initial sputum sample because sputum can be ingested into the gastrointestinal tract if swallowed. Conversely, this may be less likely in patients showing CPE positivity at other sites such as urine that are not directly connected with the gastrointestinal tract.
This study had some limitations of note. First, the analysis was performed in a single tertiary care hospital, and most of the enrolled patients had undergone a liver transplantation. Because such cases involve the use of several different medical devices, require mechanical ventilation, and have a higher risk of exposure to antimicrobial agents and thus they have a greater risk of CPE colonization than the general population [11, 12]. Second, we did not perform genetic analyses of the CPE strains to compare the isolates from close contact patients and index cases. In the absence of any bacterial sequencing data, and given that the prevalence of CPE colonization has been increasing in both hospital and community settings [13], it was therefore not definitive that all of the transmission cases we examined had arisen due to exposure to one of our index patients. Third, because we defined CPE clearance as three consecutive negative surveillance tests in one week, patients who did not complete this rate of testing could not be regarded as cleared and this may have biased our data. Finally, a substantial portion of the close contact hospital patients had already been discharged and did not undergo surveillance testing for CPE.