screening

Not content with a single well-planned study to provide information on what works to control multidrug-resistant organisms (MDROs) in the ICU, the MOSAR study group published an interrupted time series and a cluster randomized trial of various interventions in the Lancet ID. This makes the study rather complex to read and follow, but there are a number of important findings.

Interrupted time series – ‘hygiene’ intervention (chlorhexidine and hand hygiene)

Following a 6-month pre-intervention period, a 6-month interrupted time series of a ‘hygiene’ intervention (universal chlorhexidine bathing combined with hand-hygiene improvement) was performed. The key outcomes were twofold: whether there was a change in trend during each phase, and whether there was a step-change between the phases. The hygiene intervention effected a trend change reduction in all MDROs combined and MRSA individually, but not in VRE or ESBLs (Table). However, there was no step-change compared with the baseline period.

Table: Summary of reduced acquisition of all MDROs combined, or MRSA, VRE and ESBLs individually.

Cluster RCT – screening and isolation

In the 12-month cluster RCT of screening and isolation, the 13 ICUs in 8 European countries were randomized to either rapid screening (PCR for MRSA and VRE plus chromogenic media for ESBL-Enterobacteriaceae) or conventional screening (chromogenic media for MRSA and VRE only). When analysed together, the introduction of rapid or conventional screening was not associated with a trend or step-change reduction in the acquisition of MDROs (Table). In fact, there was an increase in the trend of MRSA acquisition. When comparing rapid with conventional screening, rapid screening was associated with a step-change increase in all MDROs and ESBLs.

Discussion

The study suggests, prima facie, not to bother with screening and isolation. Indeed, the authors conclude: “In the context of a sustained high level of compliance to hand hygiene and chlorhexidine bathing, screening and isolation of carriers do not reduce acquisition rates of multidrug-resistant bacteria, whether or not screening is done with rapid testing or conventional testing”. However, the major limitation here is that many of the ICUs were already doing screening and isolation during the baseline and hygiene intervention phases! I checked the manuscript carefully (including the supplemental material) to determine exactly how many units were, but it is not disclosed. To make this conclusion, surely the cluster RCT should have been ‘no screening and isolation’ vs. ‘screening and isolation’.
The increasing trend of MRSA associated with screening and isolation by either method, and step-change increases in all MDROs and ESBLs associated with rapid screening are difficult to interpret. Is an increase in acquisition due to screening and isolation plausible? Can more rapid detection of carriers really increase transmission (the turnaround time was 24 hours for rapid screening, and 48 hours for chromogenic screening)? The rapid screening arm also included chromogenic screening for ESBLs, whereas the conventional screening arm did not, so perhaps this apparent increase in acquisition is due to improved case ascertainment somehow?
Looking at the supplemental material, a single hospital seemed to contribute the majority of MRSA, with an increasing trend in the baseline period, and a sharp decrease during the hygiene intervention. There’s a suspicion, therefore, that an outbreak in a single ICU influenced the whole study in terms of MRSA. Similarly, a single hospital had a sharp increase in the ESBL rate throughout the screening intervention period, which may explain, to a degree, the increasing trend of ESBL in the rapid screening arm.
There was an evaluation of length of stay throughout the study phases, with a significant decrease during the hygiene intervention (26%), a significant increase during the rapid screening intervention, and no significant change during the conventional screening intervention. It seems likely that improved sensitivity of rapid screening identified more colonized patients who are more difficult to step down, resulting in an overall increase in length of stay.
The carriage of qacA and qacB was compared in the baseline and hygiene intervention phase, finding no difference in carriage rate (around 10% for both). This does not match our experience in London, where carriage rates of qacA increased when we introduced universal chlorhexidine bathing. However, this was restricted to a single clone; the acquisition of genes associated with reduced susceptibility to chlorhexidine seems to be clone-specific.
ICUs varied from open plan to 100% single rooms. Whilst the average proportion of patients in single rooms (15-22%) exceeded the average requirement of patients requiring isolation (around 10%), there was no measure of unit-level variation of single room usage. Since the study was analysed by cluster, the lack of single rooms on some units could have been more important than would appear from looking at the overall average. Put another way, a 100% open plan unit would have been forced to isolate all carriers on the open bay, and vice versa for a 100% single room unit.
The impact of the various interventions was moderate, even though a ‘high’ MRDO rate was necessary for enrollment (MRSA bacteraemia rate >10%, VRE bacteraemia rate >5%, or ESBL bacteraemia rate >10%). Would the impact of screening and isolation be different on a unit with a lower rate of MDROs? It’s difficult to tell.
Some of the microbiology is quite interesting: 8% of MRSA were not MRSA and 49% of VRE were not VRE! Also, 29% of the ESBLs were resistant to carbapenems (although it’s not clear how many of these were carbapenemase producers).

In summary, this is an excellent and ambitious study. The lack of impact on ESBL transmission in particular is disappointing, and may lead towards more frequent endogenous transmission for this group. The results do indicate screening and isolation did little to control MDRO transmission in units with improved hand hygiene combined with universal chlorhexidine. However, we need a ‘no screening and isolation’ vs. ‘screening and isolation’ cluster RCT before we ditch screening and isolation.

Article citation: Derde LP, Cooper BS, Goossens H et al. Interventions to reduce colonisation and transmission of antimicrobial-resistant bacteria in intensive care units: an interrupted time series study and cluster randomised trial. Lancet Infect Dis 2014; 14: 31-39.

The recent PHE CPE toolkit advocates implementing targeted screening and isolation of carriers. Reading the guidelines in a little more detail, the ‘triggers’ for screening a patient for CPE are, in the last 12 months: (a) an inpatient in a hospital abroad or (b) an inpatient in a UK hospital which has problems with spread of CPE (if known) or (c) a ‘previously’ positive case. Patients who screen positive should be placed in contact isolation; patients who screen negative should be placed in contact isolation until a further two consecutive negative screens have been taken. It’s important to note that the negative screens must be at least 48 hours apart. So, for patient who turns out to be negative will be in contact isolation for around 6 days (screens collection on days 0, 2 and 4, and a further day for the final negative screen result).

The number of patients who will meet the trigger for screening is currently unknown, but I have heard whispers of 25-50% of all admissions. This will place a considerable burden on already over-stretched isolation facilities, and bear substantial cost implications.

Single rooms in the NHS are already in very short supply. Indeed, recent press and commentary highlights the implications of running out of single rooms: patients shunted around “like parcels” in the middle of the night to relieve bed pressures.

Now, you could argue that patients who screen negative for CPE but are awaiting their confirmatory negative screens do not need to be isolated in a single room; they can be isolated in a bay. But if 25-50% of patients suddenly begin contact precautions, you’d quickly run into problems. Patients on contact precautions take longer to care for, and tend to get less attention than other patients resulting in more adverse events, as illustrated by a couple of recent Controversies blogs. Also, I fear you may begin to see ‘isolation fatigue’, where the procedure loses its impact if it has to be applied so broadly. And then there’s the cost. A recent US study calculated that contact precautions cost around £23 ($35) per patient day (not including the cost of disposal for all that additional waste!). If 25% of the 100,000 patients admitted to a London hospital Trust met the trigger for CPE screening and turned out to be negative, the price tag for isolation alone would be pushing £3.5m.

I support the PHE guidelines and agree that we need to “search and destroy” CPE above all else before it becomes endemic. However, are they feasible to implement in their current form?

Image: ‘Swabs’ by Frank Carey.