In this analysis of 144,051 hospital admissions a strategy of risk-based screening for MDRO carriage upon hospital admission identified previously unknown MDRO carriage in 0.06% (95% CI: 0.04%-0.07%) of all admissions and in 1.8% (95% CI: 1.4%-2.2%) of all patients considered to be at high risk of MDRO carriage. The numbers of ‘MDRO risk assessments needed to perform’ and individual ‘MDRO risk assessment questions needed to ask’ to detect one new MDRO carrier upon hospital admission were 1,778 and 10,420, respectively. Still, the vast majority of MDRO carriers most likely remained undetected.
The calculated numbers needed to ask are actually even underestimated as 16% of admissions had more than one MDRO risk assessment completed on the same day, and these copy-assessments were excluded from our analysis. If included, the numbers of “MDRO risk assessments actually performed” and “MDRO risk assessment questions actually asked” to detect one new MDRO carrier upon admission would have been 2,123 (171,974/81) and 12,440 (1,007,640/81), respectively. If we, conservatively, estimate one minute of labour time per MDRO risk assessment and one minute for administration, at least 160 36-hour working weeks were spent on performing assessments during these four and a half years. This reflects at least two working weeks spent per newly identifed MDRO carrier (160 weeks divided by 81 new carriers).
Newly identified carriers were most often colonized with ESBL-producing and/or Enterobacterales strains resistant to both an aminoglycoside and ciprofloxacin (70%), MDRO of which the value of screening upon admission for the prevention of transmission and hospital-acquired infections is not well-established (25–27). In our study, the prevalence of newly detected ESBL carriage upon admission was 0.03% (95% CI: 0.02–0.04), which is considerably lower than the prevalence of faecal ESBL carriage in the Dutch community; which was 5% in randomly selected subjects (19) and 6.4–7.0% upon admission to our hospital (16). As a result, in our hospital, the proportion of ESBL carriers that still remained undetected upon admission despite risk-based screening was probably more than 99%. For CRE and VRE the proportion of undetected carriers was equally high, being > 98% and > 99%, respectively.
The second most common MDRO in new carriers was MRSA (26%), which was identified in 0.02% (95% CI: 0.01%-0.03%) of all admissions. Screening and pre-emptive isolation of high-risk patients for MRSA has been an important part of the Dutch ‘search and destroy‘ policy for the prevention of MRSA transmission (6–8,17,28–31). In our study, positive predictive values to detect - among others - MRSA carriage ranged from 2.4% (95% CI: 1.2%-4.5%) (working with living pigs, veal calves or broilers) to 5.0% (95% CI: 3.9%-6.5%) (previous hospitalization in a foreign hospital). Still, presence of these risk factors was rare and even lowest for the question about being a household member of an MRSA carrier (0.1%), which needed to be asked 71,563 times in order to identify one new MRSA carrier upon hospital admission. In a recent analysis of routine universal pre-operative screening for nasal S. aureus carriage during a 7-year period in another Dutch hospital the prevalence of MRSA carriage was 0.13%, comparable to the reported prevalence of 0.11% upon admission in a study performed eight years earlier (17,18). Assuming a similar prevalence in patients admitted to our hospital would imply that the current screening strategy identified only 15% of all MRSA carriers upon admission, suggesting that 85% still remained undetected. This is in line with other studies that report that currently most MRSA carriers do not have the classical risk factors (i.e. as inquired with our risk assessment) for MRSA carriage (7,32–34).
The assessment question on known MDRO carriage had the highest yield, as it was answered positively in 2.8%. Indeed, 76% (n = 260) of all patients with an MDRO positive screening culture were already labelled in our EMR as a known MDRO carrier, of which 11% (n = 29) had this label due to previous risk-based screening. This implies that if the risk assessment would have been replaced by the use of existing MDRO labelling in the EMR 68% (231/341) of MDRO carriers – that were now identified by risk-based screening – would still be captured.
Typically, the unexpected identification of an MDRO carrier during admission (i.e. through a positive clinical culture) is associated with extra workload, for screening of exposed roommates or healthcare workers of the index patient. This is not needed if the carrier was already identified upon admission (and thus contact precautions had already been installed). In our study only 15% (n = 12) of detected carriers had a clinical culture positive for MDRO during admission, for which contact tracing would have been implemented if screening upon admission had not been applied. We estimated that abandoning risk-assessment based screening would have led to 566 patient days without protective measures for MDRO carriers in the 4.5 years of the observation period. As the vast majority of MDRO carriers remained undetected, these 566 days add little to the total number of patient days without protective measures for – unknown - MDRO carriers. The number of prevented episodes of cross-transmission due to the identification of new MDRO carriers upon admission is difficult to determine. Yet, the total number of hospital-acquired bacteraemia episodes caused by MDRO in our hospital during the 54 months of the study period was 44; 37 were caused by ESBL/MDR-Enterobacterales, four by multidrug-resistant P. aeruginosa, two by MRSA and one by VRE. Of these 44 patients eleven were known MDRO carriers at the time of admission and two were newly identified as MDRO carrier through risk assessment and screening. The remaining 31 patients (including those with hospital-acquired bacteraemia caused by MRSA and VRE) had no risk factors for MDRO carriage upon admission.
A strength of the current analysis was the combination of routine care data and medical microbiology information of 90% of all admissions during the predefined study period. There are also important limitations of this study that should be acknowledged. Firstly, retrograde manual changes to the MDRO risk assessment during hospital admission could not be retrieved. It is, therefore, not excluded that the MDRO risk assessment (e.g. the first question) was manually changed to ‘positive’ in case of MDRO positive cultures during admission. If so, the value of MDRO risk assessment would have been overestimated. Secondly, this was a real-life evaluation of clinical practice, without confirmation whether the individual questions of the MDRO risk assessment were answered correctly. Thirdly, this was a single-centre analysis in an academic medical centre in the Netherlands, with a well-developed system of identification, labelling and isolation of MDRO carriers as well as good adherence to standard precautions in routine care. Extrapolation of findings to other settings or countries should always occur in light of local epidemiology and established routine infection and prevention practices within a hospital.
The current low levels of AMR in hospitals in the Netherlands are partly explained by a restrictive use of antibiotics combined with the well-established ‘search and destroy’ policy over the last decades. Still, critical appraisal as well as continuous improvement is a fundamental part of infection prevention and control, considering that local epidemiology and target populations may change over time. Results of this study imply that the majority of MDRO carriers in the community remains undetected upon admission despite current risk-based screening. Combined with the low prevalence of risk factors and the types of MDRO that are most often identified, the question arises whether the number of newly identified MDRO carriers truly justifies the invested workload across all hospital wards in risk assessment upon admission. We propose a system in which risk-based screening is abandoned and instead, contact precautions are installed upon hospitalization of patients that are known (previous) carriers of MDRO. This captures the majority of MDRO carriers that would elsewise be identified through risk assessment.