Clinical Incidence of Methicillin‐Resistant Staphylococcus aureus (MRSA) Colonization or Infection as a Proxy Measure for MRSA Transmission in Acute Care Hospitals
The incidence of methicillin‐resistant Staphylococcus aureus (MRSA) colonization or infection has been used as a proxy measure for MRSA transmission, but incidence calculations vary depending on whether active surveillance culture (ASC) data are included.
To evaluate the relationship between incidences of MRSA colonization or infection calculated with and without ASCs in intensive care units and non–intensive care units.
A Veterans Affairs medical center.
From microbiology records, incidences of MRSA colonization or infection were calculated with and without ASC data. Correlation coefficients were calculated for the 2 measures, and Poisson regression was used to model temporal trends. A Poisson interaction model was used to test for differences in incidence trends modeled with and without ASCs.
The incidence of MRSA colonization or infection calculated with ASCs was 4.9 times higher than that calculated without ASCs. Correlation coefficients for incidences with and without ASCs were 0.42 for intensive care units, 0.59 for non–intensive care units, and 0.48 hospital‐wide. Trends over time for the hospital were similar with and without ASCs (incidence rate ratio with ASCs, 0.95 [95% confidence interval, 0.93–0.97]; incidence rate ratio without ASCs, 0.95 [95% confidence interval, 0.92–0.99]; P = .68). Without ASCs, 35% of prevalent cases were falsely classified as incident.
At 1 Veterans Affairs medical center, the incidence of MRSA colonization or infection calculated solely on the basis of clinical culture results commonly misclassified incident cases and underestimated incidence, compared with measures that included ASCs; however, temporal changes were similar. These findings suggest that incidence measured without ASCs may not accurately reflect the magnitude of MRSA transmission but may be useful for monitoring transmission trends over time, a crucial element for evaluating the impact of prevention activities.
Patients who acquire methicillin‐resistant Staphylococcus aureus (MRSA) while hospitalized are at increased risk for subsequent infections and potentially increased morbidity, mortality, and healthcare costs.1‐3 Preventing the transmission of MRSA among patients in healthcare settings has become a priority for United States healthcare facilities, regulators, and government agencies. The Centers for Medicare and Medicaid Services Quality Improvement Organization program has dedicated its Ninth Scope of Work to the reduction of the rate of healthcare‐associated MRSA infection; >200 hospitals are currently participating in more than 50 Quality Improvement Organization collaborations to reduce the number of healthcare‐associated MRSA infections.4 In addition, the United States Health and Human Services Healthcare‐Associated Infection Action Plan has as one of its 5‐year goals a 50% reduction in the rate of invasive healthcare‐associated MRSA infections.5
Having a reliable and well‐characterized measure of MRSA transmission is important for evaluating the effectiveness of MRSA prevention programs targeting healthcare‐associated transmission. The ideal measure of MRSA transmission would be both easy to calculate and accurate; however, to date, a number of different MRSA transmission measures have been used, leading to inconsistencies in data collection, interpretation, and comparison.6 In response to this lack of standardization, the Society for Healthcare Epidemiology of America (SHEA) and the Healthcare Infection Control Practices Advisory Committee published recommendations outlining the use of either of 2 incidence rates as a proxy for quantifying healthcare‐associated MRSA acquisition. The first incidence rate is based on clinical culture data alone, whereas the second, considered to be a more advanced metric, is based on both clinical culture and active surveillance culture (ASC) data.7
Although the addition of ASCs significantly increases the accuracy of identifying MRSA carriers, it remains unknown whether using ASCs is necessary for incidence estimates to be a useful proxy for MRSA transmission or whether incidence measures based solely on clinical cultures are sufficient.8,9 A previous multicenter study, which involved 12 intensive care units (ICUs) from facilities involved in the Centers for Disease Control and Prevention (CDC) Epicenters Program, found that by excluding ASC data, the magnitudes of the clinical incidences were substantially underestimated, compared with estimates that included ASC data. However, this study also established a strong linear correlation between the 2 incidence measures.10 This suggests that, despite the underestimation, estimates of the incidence of MRSA colonization or infection without ASCs may be a useful indicator of MRSA transmission when ASC data are not available. To describe this relationship more fully, we sought to examine the relationship between measurements of the incidence of MRSA colonization or infection calculated with and without ASC data in both ICU and non‐ICU settings at 1 hospital and to assess the usefulness of incidence estimates based solely on clinical culture data as a surrogate measure for MRSA transmission trends over time.
In October 2001, a Veterans Affairs medical center, in collaboration with the CDC, initiated a multifaceted MRSA prevention program in both ICUs (n = 4) and non‐ICU settings (n = 3). The intervention was implemented in 3 phases, starting in a non‐ICU surgical ward in 2001, expanding to a surgical ICU in late 2003, and then extending to all acute care units in 2005. The MRSA prevention intervention consisted of the following 3 elements: (1) use of systems, cultural, and behavioral change strategies to promote adherence to infection control protocol; (2) enhanced emphasis on standard precautions, including hand hygiene and environmental disinfection; and (3) performance of ASCs (with samples obtained from nares and open wounds, without broth enrichment) within 48 hours of admission and within 48 hours of discharge from a unit, to identify patients asymptomatically colonized with MRSA for prompt initiation of contact precautions (overall study adherence rate, 81%). These 3 components were implemented in parallel.
Incident cases were defined by a positive MRSA culture result obtained at least 48 hours after admission to a unit through 48 hours after discharge or transfer from a unit in a patient with no history of MRSA colonization or infection during the previous 12 months. Incidence was defined as the number of incident cases of MRSA colonization or infection per 1,000 patient‐days at risk. Patient‐days at risk, instead of total patient‐days, were used as the denominator to avoid underestimation of MRSA clinical colonization or infection incidence rates.10 Ineligible patient‐days included (1) the first 48 hours of any hospitalization and (2) all patient‐days within 1 year after a patient was newly identified as having a positive culture result for MRSA.
Two measures of the incidence of MRSA colonization or infection were calculated. The first was based on incident cases detected by clinical culture data alone, and the second was based on clinical cases detected by combining both clinical culture and ASC data.
Incident MRSA colonization or infection cases identified by one but not both incidence measures were grouped into 1 of 2 potential misclassification categories. The first included false incident cases, defined as MRSA prevalent cases of colonization or infection (based on a positive ASC result at the time of admission to the unit) incorrectly identified as an incident case in the absence of ASC data. The second included incident cases that were missed, defined as incident cases not detected by clinical cultures alone that would have been identified with ASCs.
Incidences were computed and analyzed using SAS software, version 9.2 (SAS Institute). Correlation analyses were performed to evaluate the possibility of a linear relationship between the 2 incidence measures (ie, incidence calculated with and without ASC data). Because of the lack of normality among the data, Spearman coefficients were calculated for each unit type and for all acute care units combined. The magnitudes of the Spearman coefficients and associated P values were used to determine the strength and significance of any linear relationships.
To detect temporal changes in the incidence of MRSA colonization or infection, we used a Poisson regression to model the slope of postintervention incidences of MRSA colonization or infection over time, by quarter, for both incidence measures: ln(λ) = β0 + β1(Quarter). In addition, an interaction model was used to assess whether the inclusion of ASC data resulted in a different MRSA colonization or infection incidence trend over time, compared with the exclusion of ASC data: ln(λ) = β0 + β1(Quarter) + β2(ASC) + β3(Quarter × ASC). A significant P value for the interaction term indicated a difference in variation over time in MRSA transmission characterized by the 2 proxies.
The number of incident cases of MRSA colonization or infection, the number of patient‐days at risk, and incidences of MRSA colonization or infection calculated with and without ASC data are shown in Table 1. The number of incident cases identified with use of ASCs was 5.1 times greater in the non‐ICU settings, 1.6 times greater in ICUs, and 4.1 times greater hospital‐wide than the number of cases identified without the use of ASCs. The number of patient‐days at risk were 14%, 24%, and 15% lower in non‐ICUs, in ICUs, and hospital‐wide, respectively, when ASC data were included. Without the use of ASCs, incidences of MRSA colonization or infection in non‐ICUs, in ICUs, and hospital‐wide were all lower than those calculated with ASC data. Incidence rate ratios (IRRs) comparing incidences calculated with ASC data to those calculated without ASC data were higher in non‐ICUs than in ICUs (non‐ICU IRR, 5.9; ICU IRR, 2.1; hospital‐wide IRR, 4.9).
(n = 3)
(n = 4)
|Patient‐days at risk||41,855||6,882||48,737|
|Incidence, cases per 1,000 patient‐days at risk||8.46||6.10||8.13|
|Patient‐days at risk||48,560||9,057||57,617|
|Incidence, cases per 1,000 patient‐days at risk||1.44||2.87||1.67|
The 2 quarterly incidence measures are shown, stratified by unit type, in Figure 1. Correlation analyses demonstrated a statistically significant Spearman correlation coefficient of 0.59 (P = .002) in the non‐ICUs and 0.48 (P = .046) hospital‐wide. The ICUs had a similar r value of 0.42, which approached statistical significance (P = .08).
The 2 hospital‐wide, postintervention quarterly measurements of the incidence of MRSA colonization or infection over time from November 2001 through February 2006 are shown in Figure 2. With use of Poisson regression, we found a hospital‐wide IRR of 0.95 for both the incidence calculated with ASCs and the incidence calculated without ASCs, corresponding to a 5% decrease in incidence each quarter. In addition, IRRs with and without ASC data were similar within non‐ICUs (IRR with ASCs, 0.96; IRR without ASCs, 0.94) and ICUs (IRR with ASCs, 0.87; IRR without ASCs, 0.85). However, despite these similarities, IRRs calculated with surveillance cultures were more precise, evidenced by narrower confidence intervals (Table 2). Results from the Poisson interaction model showed that there was no significant difference between hospital‐wide incidence slopes over time calculated with and without ASCs (
|Setting||Including ASCs||Excluding ASCs|
|Hospital‐wide||0.95 (0.93–0.97)||0.95 (0.92–0.99)|
|Non–intensive care units||0.96 (0.94–0.97)||0.94 (0.90–0.98)|
|Intensive care units||0.87 (0.78–0.97)||0.85 (0.74–0.98)|
A comparison between the incident cases of MRSA colonization or infection identified with and without ASCs revealed that misclassification and underestimation of the number of incident cases occurred frequently without the inclusion of ASCs (Figure 3). Of the 96 incident cases of MRSA colonization or infection identified hospital‐wide without ASC data, 34 were false incident cases, leaving only 62 cases correctly identified as incident. This translated into a 35% misclassification rate. In addition, without ASC data, 334 (84%) of the 396 true incident cases identified with ASCs were missed. Misclassification did not vary by unit type; the percentage of cases falsely identified as incident without ASCs was 36% in non‐ICUs and 35% in ICUs. A substantial percentage of incident cases missed without ASCs was observed in both non‐ICUs (87%) and ICUs (60%).
The number of incident cases of MRSA colonization or infection calculated with ASC data was consistently greater and the number of patient‐days at risk fewer, compared with that calculated without ASCs, in both ICUs and non‐ICUs. The combination of these 2 factors translated into higher incidences of MRSA colonization or infection with the inclusion of ASC data. However, close examination of the incident cases of MRSA colonization or infection identified hospital‐wide with each measure revealed that without ASC data, 35% of prevalent cases were falsely classified as incident and 84% of true incident cases were missed. Misclassification results were similar in the non‐ICU and ICU settings. However, when analyzed over time by quarter, the incidence measures with and without ASCs resulted in identical IRRs, indicating that both measures decreased at the same rate over time.
Understanding MRSA transmission in healthcare settings is important in evaluating MRSA control efforts within a facility; however, because true transmission is difficult to measure, proxy measures often represent the best available option to quantify the burden of MRSA transmission. Two recent position papers have recommended using the incidence of hospital‐onset MRSA colonization or infection as a proxy measure for healthcare‐associated acquisition of MRSA.7,11 Proxy measures have also been listed as an important component of MRSA prevention strategies. For settings with evidence of ongoing MRSA transmission despite adequate implementation of and adherence to basic infection control practices, the SHEA and the Infectious Diseases Society of America have recommended the use of ASCs to calculate the incidence of MRSA colonization or infection.11 However, ASCs are often costly and require substantial effort from clinical laboratories and front‐line clinical staff. In addition, recent data regarding the utility of ASCs for preventing MRSA infections have been conflicting.6,12,13 Because ASCs may not be performed in all facilities, it is important that the utility and limitations of the recommended transmission proxy measures performed with and without ASCs be understood.
Previous studies have examined the differences between the numbers of MRSA colonization or infection cases identified with and without ASC data. Several studies found that the inclusion of ASC data significantly increased the number of MRSA‐colonized patients detected.14‐16 In one study, the addition of ASCs resulted in an improved detection rate of up to 64% for mean monthly MRSA prevalence and an improved detection rate of up to 157% for mean monthly incidence in ICUs.10 Salgado et al also found that relying on clinical cultures alone identified only 15% of all patients colonized with MRSA.8
Unlike other studies, our study investigated measures of the incidence of MRSA colonization or infection in both ICU and non‐ICU settings. We found that the difference in the number of incident cases of MRSA colonization or infection detected with and without ASC data was greater in the non‐ICUs than in the ICUs. This may be attributable to the fact that clinical cultures are performed more frequently in ICUs than non‐ICUs, increasing the number of incident cases identified with clinical cultures. However, the correlation and Poisson regression analyses produced similar results in the ICUs and non‐ICUs, suggesting that the findings in this study can be applied to both hospital settings.
Although the correlation coefficients for ICUs, non‐ICUs, and the entire facility reached or approached statistical significance, the r values 0.4–0.6 reflect only a moderately linear relationship between the 2 incidence measures and are much lower than the r value of 0.87 previously observed in ICUs.10 The disparity between the 2 studies is most likely attributable to a key difference in the intervention protocol. ASCs were obtained at the time of admission to the unit in both studies, but the procedure for subsequent surveillance culture collection differed. In our study, ASCs were performed at the time of discharge or transfer from a unit, whereas in the study conducted by Huang et al,10 ASCs were performed weekly on a particular day. With the latter protocol, patients with a hospital stay of less than 7 days may not have had a second culture performed, preventing further identification of MRSA‐colonized patients.
Our analyses also found that estimates of the incidence of MRSA colonization or infection based solely on clinical culture results commonly misclassified and underestimated the number of true incident cases. This underestimation when excluding ASC data was largely attributed to the reduced number of cultures performed during a patient stay. Without ASCs, patients who do not exhibit any clinical symptoms of infection may not have cultures performed, thereby limiting the chance of identifying MRSA colonization in these patients. In this study, 35% of incident cases of MRSA colonization or infection identified in ICUs with use of clinical culture results alone were found to be misclassified when ASC results were considered, which is more than double the misclassification rate of 17% observed in a previous study by Huang et al.10 This difference in misclassification rates may be attributable to the fact that the hospital in our study implemented ASCs for a longer period of time and more broadly (ie, ASCs were performed at the time of admission, discharge, and transfer for ICU and non‐ICU settings), which increased the probability that patients with clinical culture results positive for MRSA also had an ASC performed. This may have increased the ability of ASCs to accurately identify some cases as prevalent that would otherwise have been falsely classified as incident.
Despite the low correlation coefficients and substantial rates of misclassification and underestimation of true incident cases of MRSA colonization or infection, similar incidence trends were found between measurements of the incidence of MRSA colonization or infection calculated with and without ASC data. This suggests that in spite of the month‐to‐month differences in incidence, at this one facility trends over time are consistent between the 2 incidence measures. Thus, incidence measures based on clinical cultures alone may be useful in monitoring MRSA transmission trends over time.
This study is primarily limited by the fact that it was conducted using data from only one facility, which reduces the generalizability of its findings. Previous studies have combined data from multiple ICUs and may be more representative; however, the evaluation of data from both ICU and non‐ICU settings makes our current study novel. In addition, our study was designed to compare 2 measures of incidence; although these measures are often recommended and used to measure transmission, our study does not allow us to further assess how accurately these 2 measures reflect true MRSA transmission.
In our review of data from one Veterans Affairs medical center, measurements of the incidence of MRSA colonization or infection based solely on clinical cultures appeared to underestimate the true magnitude of MRSA cases when ASCs were included; however, temporal changes identified using both measures were similar. This suggests that the incidence of MRSA colonization or infection calculated without ASC data may be appropriate for assessing trends in MRSA transmission over time, a crucial element for evaluating the progress and success of activities designed to prevent MRSA transmission.
Potential conflicts of interest. All authors report no conflicts of interest relevant to this article.
- 1.Cosgrove SE, Sakoulas G, Perencevich EN, Schwaber MJ, Karchmer AW, Carmeli Y. Comparison of mortality associated with methicillin‐resistant and methicillin‐susceptible Staphyloccocus aureus bacteremia: a meta‐analysis. Clin Infect Dis 2003;36:53–59.
- 2.Engemann JJ, Carmeli Y, Cosgrove SE, et al. Adverse clinical and economic outcomes attributable to methicillin resistance among patients with Staphylococcus aureus surgical site infection. Clin Infect Dis 2003;36:592–598.
- 3.Cosgrove SE, Qi Y, Kaye KS, Harbarth S, Karchmer AW, Carmeli Y. The impact of methicillin resistance in Staphylococcus aureus bacteremia on patient outcomes: mortality, length of stay, and hospital charges. Infect Control Hosp Epidemiol 2005;26:166–174.
- 4.Centers for Medicare and Medicaid Services. 9th Statement of Work. Washington, DC: US Department of Health and Human Services, 2008. http://www.cms.gov/QualityImprovementOrgs/downloads/9thSOWBaseContract_C_08‐01‐2008_2_.pdf. Accessed December 14, 2009.
- 5.US Department of Health and Human Services. HHS action plan to prevent healthcare‐associated infections. Washington DC: US Department of Health and Human Services, 2009. http://www.hhs.gov/ash/initiatives/hai/actionplan/. Accessed December 14, 2009.
- 6.Huang SS. Health care‐associated infection: assessing the value and validity of our measures. Clin Infect Dis 2009;48:1116–1122.
- 7.Cohen AL, Calfee D, Fridkin SK, et al. Recommendations for metrics for multidrug‐resistant organisms in healthcare settings: SHEA/HICPAC position paper. Infect Control Hosp Epidemiol 2008;29:901–913.
- 8.Salgado CD, Farr BM. What proportion of hospital patients colonized with methicillin‐resistant Staphylococcus aureus are identified by clinical microbiological cultures? Infect Control Hosp Epidemiol 2006;27:116–121.
- 9.Eveillard M, Lancien E, Barnaud G, et al. Impact of screening for MRSA carriers at hospital admission on risk‐adjusted indicators according to the imported MRSA colonization pressure. J Hosp Infect 2005;59:254–258.
- 10.Huang SS, Rifas‐Shiman SL, Warren DK, et al. Improving methicillin‐resistant Staphylococcus aureus surveillance and reporting in intensive care units. J Infect Dis 2007;195:330–338.
- 11.Calfee DP, Salgado CD, Classen D, et al. Strategies to prevent transmission of methicillin‐resistant Staphylococcus aureus in acute care hospitals. Infect Control Hosp Epidemiol 2008;29:S62‐S80.
- 12.Robicsek A, Beaumont JL, Paule SM, et al. Universal surveillance for methicillin‐resistant Staphylococcus aureus in 3 affiliated hospitals. Ann Intern Med 2008;148:409–418.
- 13.Harbarth S, Fankhauser C, Schrenzel J, et al. Universal screening for methicillin‐resistant Staphyloccocus aureus at hospital admission and nosocomial infection in surgical patients. JAMA 2008;299:1149–1157.
- 14.Hindler JF, Stelling J. Analysis and presentation of cumulative antibiograms: a new consensus guideline from the Clinical and Laboratory Standards Institute. Clin Infect Dis 2007;44:867–873.
- 15.White RL, Friedrich LV, Burgess DS, Brown EW, Scott LE. Effect of removal of duplicate isolates on cumulative susceptibility reports. Diagn Microbiol Infect Dis 2001;39:251–256.
- 16.Lucet JC, Chevret S, Durand‐Zaleski I, Chastang C, Regnier B. Prevalence and risk factors for carriage of methicillin‐resistant Staphylococcus aureus at admission to the intensive care unit: results of a multi‐center study. Arch Intern Med 2003;163:181–188.