We reviewed the published literature on the natural history of MRSA and VRE colonization. There is substantial heterogeneity among the included studies. This heterogeneity both identifies a clear need for further investigation and tempers our interpretation of the pooled estimates of time to clearance of colonization. Based on the studies meeting our inclusion criteria, our systematic review demonstrates that persistent colonization decreases over time, with clearance of colonization in half of patients at 88 weeks for MRSA and 26 weeks for VRE.
While the weight of the existing data, and clinical experience, suggest that clearance of MRSA and VRE colonization increases over time, precision around the time to clearance is not possible due to the major limitations of the studies in this domain. These limitations not only hinder our interpretation of the pooled findings, but also highlight the need for additional studies. Lack of a consistent definition of clearance, uncertainty regarding the time of initial colonization, variation in frequency of sampling for persistent colonization, and variation in duration of follow-up and loss to follow-up all impose substantial constraints on our interpretation of the median time to clearance.
Because subjects are not screened prospectively and continuously, the initial date of colonization is assumed to be the time MRSA or VRE were discovered and documented (based on either a clinical infection or a surveillance screening). There is almost certainly variation in the lag time between actual colonization and the identification of a subject as being colonized. Under these inevitable circumstances, calculations of duration of colonization may underestimate the true duration of colonization and the pattern of clearance. On the other hand, if the time interval between initial documentation of colonization and re-screening is prolonged, duration of colonization may also be overestimated. A striking observation from the combined reviews is the relatively short time to clearance for VRE, as compared to MRSA. The longer median time to clearance estimated for MRSA is in part a reflection of the length of follow-up, as evidenced by the reduction from 88 weeks to 41 weeks observed when follow-up was restricted to the same duration as the VRE studies.
We did not impose a universal definition of clearance as a prerequisite for inclusion in our review, and across studies, the definition of clearance varied in part due to the absence of guidelines that define clearance of colonization . The diversity of results that were considered evidence of clearance across the studies is a reflection of the lack of consensus on this point, and limits our interpretation of the pooled estimates. The frequency of re-sampling and duration of follow up varied as well. These factors would be expected to affect reported time to clearance and thus add reasons to be cautious in the interpretation of the systematic review. Several of the MRSA studies [11, 14, 18, 22, 23, 25] showed an initial brisk decline in colonization followed by a stabilization of the pool of colonized (in those studies following patients for extended periods), supporting the general consensus that there are likely sub-groups among colonized patients including those who are transiently, intermittently or persistently colonized. While some of the studies did make such distinctions, again, the definition of each carrier-state varied.
Beyond the concepts of transient, intermittent or persistent colonization, isolates identified in the screening studies may represent an initial colonizing strain or a second (or third) isolate. Some studies performed additional analysis to identify strain types. In the absence of strain-typing, it is not possible to conclude that a patient who remains persistently colonized is in fact colonized with the endogenous strain, or intermittently colonized with different strains. From the perspective of infection control implementation, such distinctions may not be meaningful in terms of the practical implementation of CP measures, and those cases in which cohorting is permitted [40, 41]. Given currently available assays, documented clearance of colonization may in fact represent a level of colonization below the limits of detection (with the same strain or different strain).
A conceptual limitation of our review relates to colonization dynamics and specifically the clearance of the colonizing strain or re-colonization with a new strain, which may be particularly relevant in the setting of selective antibiotic pressure. In the VRE analysis, one clinical variable, prior antibiotic use, was associated with a trend toward early clearance of colonization, supporting the observation that concurrent antibiotic therapy affects the sensitivity of surveillance cultures for VRE [42, 43]. The issue of test sensitivity is particularly relevant in the setting of selective antibiotic pressure, as has been demonstrated in the case of VRE [44–47]. A detailed analysis of the impact of antibiotics on colonization dynamics was not directly within the scope of the study, however, is an important area for further investigation, especially with respect to VRE. In the VRE studies included in this review, the lack of universal definitions of clearance and carriage-type again limit our interpretation. Finally, the years included span close to 20 years for both MRSA and VRE, during which changes in epidemiology, screening practices, and technology have taken place, raises challenges for comparison across studies and interpretation of the pooled clearance estimates.
Additional limitations are second-order compared to the fundamental deficiencies described above. Most studies provided insufficient details regarding the demographic characteristics of subjects who cleared versus those who remained persistently colonized. However, one demographic variable for MRSA-colonized patients, ambulatory patient status versus inpatient or LTCF patient status, was associated with a trend toward early clearance of colonization in three studies. It is possible that the persistence of colonization in hospitalized and LTCF patients may be the result of re-colonization. Based on our quality assessment of cohort studies, the majority were of moderate quality.
Our analysis was also limited by the use of aggregate data rather than patient-level data, which would have permitted a survival analysis approach. Most studies screened using microbiological methods that relied on standard culture techniques. Although molecular assays are more costly on a per-test basis, their greater sensitivity may allow reduction in repetitive testing and more effective implementation . Additional research employing molecular methods in studies of the natural history of colonization is needed. Even with the more widespread use of molecular assays however, it is still possible that colonized patients may fail to be identified, if the level of colonization is below the limit of the sampling methods. In terms of practical implications of false negative screens, patients with low levels of colonization may pose a lower transmission risk. Finally, we are not able to address the risk of publication bias in the inclusion of studies.