On Contact Precautions: The Good, the Bad, and the Ugly

Abstract

Affiliations: 1. Division of Infectious Diseases, Department of Medicine, Duke University School of Medicine, Durham, North Carolina; 2. Duke Infection Control Outreach Network, Durham, North Carolina; 3. Department of Infectious Disease, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina. Received December 5, 2013; accepted December 8, 2013; electronically published February 3, 2014. 2014 by The Society for Healthcare Epidemiology of America. All rights reserved. 0899-823X/2014/3503-0002$15.00. DOI: 10.1086/675295 Contact precautions are designed to reduce opportunities for pathogen transmission by direct and indirect contact with the patient or the patient’s environment. Contact precautions typically include patient placement in a private room, the wearing of gloves and gowns by healthcare personnel while in the room, and proper hand hygiene before room entry and after gloves are removed. Contact precautions have been demonstrated to be effective during outbreaks and, as a result, are widely used as an intensified control measure for multidrug-resistant organisms (MDROs) during outbreak situations. Target pathogens for implementation of contact precautions include methicillin-resistant Staphylococcus aureus (MRSA), vancomycin-resistant Enterococcus (VRE), C. difficile, and multidrug-resistant gram-negative bacilli. There is no one way to implement contact precautions, however. The Healthcare Infection Control Practices Advisory Committee guideline for management of MDROs in healthcare settings does not specify precisely when to use contact precautions; the appropriate duration of contact precautions for patients who are colonized or infected with an MDRO is undefined; and the definition of an MDRO varies between facilities. Importantly, the use of contact precautions may lead to unintended consequences. Morgan et al identified 4 main adverse outcomes related to contact precautions: less patienthealthcare provider contact, changes in systems of care that produce delays and more noninfectious adverse events, increased symptoms of depression and anxiety, and decreased patient satisfaction with care. Since this review was published in 2009, other reports have confirmed these findings, while additional reports have added to the list of adverse outcomes due to contact precautions, including increased risk of delirium, hypoglycemia, hyperglycemia, and anticoagulant prescribing errors. In this issue, Dhar et al assessed 5 measures of compliance with contact precaution practices in 11 teaching hospitals. These measures included rates of hygiene prior to donning gloves/gowns, gowning, gloving, removal of gloves/gowns, and hand hygiene after gown/glove removal. We believe that readers should draw 2 main conclusions from their analysis. First, compliance with contact precautions was strikingly poor. Overall compliance with all 5 measures was only 28.9% after 1,103 observations. In a multivariate analysis, increasing burden of isolation was associated with (1) an increase in noncompliance with all 5 components of the recommended contact isolation practices and (2) an increase in noncompliance with hand hygiene prior to donning gloves and gowns. That is, not surprisingly, compliance with contact isolation decreased as the burden of isolation increased. Second, there was substantial heterogeneity in the different approaches to contact precautions among the study hospitals. For example, 4 sites required masks in addition to gowns and gloves, 4 sites included infectivity in their local criteria for use of contact precautions, and different hospitals required contact precautions for different organisms (eg, 7 hospitals required contact precautions for MRSA but only 4 for VRE). The multicenter approach and total number of observations are strengths of this study. Other strengths include the use of undercover observers and statistical controls to ensure that results were not driven by individual hospitals. Importantly, these data represent real-world research that confirms most epidemiologists’ fears. The authors acknowledge weaknesses of their study but fail to mention 4 additional problems. First, it is unclear how many observations occurred in the setting of a high (greater than 40%) burden of isolation. Given the wide confidence intervals of the estimate in their model, we presume that few observations occurred when the burden of isolation was high. Second, there was no assessment of interobserver reliability or description of the protocol used by the observers. Lack of such assessment may have biased the observed performance toward higher rates of compliance. Third, observations were excluded if one of the 5 activities were not observed, but it is unclear how many observations were excluded. Finally, overall compliance required