Initial remarks: This blog, written by Lawrence F Muscarella, PhD,  lists several factors that are shared in common by a number of reviewed prospective cohort studies that evaluate the effectiveness of (e.g., the percentage by which) an intervention, bundles of best practices, or checklist for the reduction of central line-associated bloodstream infections (CLABSIs) in intensive care units (ICUs). The study upon which this blog is based and that provides the results of a review of these prospective cohort studies can be read by clicking here. (Also refer to this other blog.)

1. Limited to associations Many of the prospective cohort studies evaluating the impact of an intervention on the rate of CLABSIs in ICUs suggest, intimate, or conclude that the studied intervention resulted in, was responsible for, or caused a reduction in the CLABSI rate by, for example, 66%, 50% or 74%.17-24,35,43,44 In general, however, such conclusions are questioned (see: main article), because the limitations of the designs of these studies restrict them to observing instead only an association between the implementation of the studied intervention and a reduction in the rate of CLABSI (see: #6, below, on same page).

Click here to Dr. Muscarella’s related blog/letter “Dear Pediatrics: An Assessment of the Effectiveness of Bundles and Checklists.”

2. Adherence to the intervention not confirmed: Moreover, these studies may not verify adherence by staff members to the intervention, which raises fair questions about whether, not the intervention, but other factors, such as measurement, feedback, publication, or confounding biases might themselves have been responsible for the observed reduction in the CLABSI rate.35,44

3. Confounding factors: The reviewed prospective cohort studies do not control for every relevant confounding factor, which itself, independent of the studied intervention, might have had a substantive effect on the measured CLABSI rate reduction. The use during (but not prior to) the studied “intervention period” of well-trained attending physicians instead of less experienced medical residents to insert and maintain central lines is an example of such a confounding factor, which is typically unrecognized, un-controlled, not a component of the studied intervention, and could itself cause a reduction in the CLABSI rate that might be misattributed to the intervention (a “false-positive” effect or result).

Click here to read Dr. Muscarella’s blog that lists a number of biases and/or confounding factors that, if not controlled or whose effects are not negated, can introduce artifact and may themselves – not the intervention – be responsible for the observed reduction in the CLABSI rate.

4. Biases: The reviewed prospective cohort studies do not eliminate the effects that biases may have on the measured outcome. For example, because staff members are not blinded during these studies, but rather are usually told of the study’s intent to reduce the rate of CLABSIs and provided with “feedback” and progress reports about the effectiveness of the intervention, the measured infection rates may inadvertently and unwittingly become inaccurate due to feedback bias.21 If the ICUs were additionally provided with financial incentives to report reduced CLABSI rates, then the measured infection rates may be prone, too, to financial bias.22 Further, the use during the intervention period of a less sensitive surveillance method may introduce even more error into the reported rate of CLABSI due to measurement bias.10,11

Such biases can cause the observed incidence of CLABSIs in one or more ICUs to under-report the true infection rate—for example, bacteremia associated with a central line (a primary CLABSI) might not be counted, having been subjectively assigned in error to (or reclassified as) a secondary infection due to an unrelated source or site, such as an urinary tract infection.11,18,19,30,36,37 Consequence, these prospective cohort studies can be prone to over-exaggerating the percentage by which a studied intervention might have reduced the CLABSI rate.17-24,35,43

Again, this blog is based on a more complete and comprehensive article entitled “Published Infection Rates:  More Conjectural than Scientific?,” which can be read in its entirety by clicking here.

5. Lack of validation: These reviewed prospective cohort studies17-24,35,43 generally base their conclusions on CLABSI rates that are detected, interpreted, and reported by the participating hospitals themselves. In general, these studies do not independently validate the accuracy and completeness of these CLABSI rates to ensure that none were missed and every CLABSI was counted. This omission raises reasonable questions about the actual CLABSI rates and the true effectiveness of the studied intervention on them.32,40

Click here to read a peer-reviewed article by Dr. Muscarella entitled “Assessment of the Reported Effectiveness of Five Different Quality-Improvement Initiatives for the Prevention of Central Line-Associated Bloodstream Infections in Intensive Care Units.” This detailed article complements this blog’s discussion of the effectiveness of federal efforts to reduce the rate of CLABSIs in ICUs.

6. Immediate impact: It is recognized that the reviewed prospective cohort studies17-24,35,43 are generally quality improvement projects seeking to achieve prompt reduction in CLABSI rates for as many patients as possible. Rather than perform a controlled and randomized (and blinded) study that might admittedly benefit only the “treatment” (or “intervention”) group of patients (and not also the “control” or “non-intervention” group24), these studies expose all patients to the intervention, gauging the intervention’s effectiveness by comparing the respective CLABSI rates measured before and after the intervention’s implementation, instead of the more rigorous comparison of the treatment group’s CLABSI rate to that of the control group’s.

Click here to read a related blog by Dr. Muscarella entitled “Three Facts and Myths about Central Line-Associated Bloodstream Infections.”

Although these prospective cohort studies are most insightful, their goals most admirable, and their results potentially having immediate impact for the treatment group of patients, their designs—in addition to rendering them prone to under-reporting the true incidence of infection and to over-exaggerating an intervention’s true effectiveness on CLABSI rates—necessarily preclude them from concluding a causal relationship between the studied intervention and an observed reduction in CLABSI rates in ICUs, some of these studies’ conclusions notwithstanding (see: #1, above).17-24,35,43

References to this blog: Click here.

Blog by: Lawrence F Muscarella PhD posted on 12-13-2012; updated 3-22-2013

Leave a Reply

Your email address will not be published. Required fields are marked *