JSTOR

Methods

Ten standardized environmental sites were sampled in 15 hospital rooms: the bed rail, table over the bed, dresser, door handle, television remote control, nurse call button, floor, blood‐pressure cuff, bathroom handrail, and toilet seat or commode. Sampling was conducted before standard daily cleaning. Ten rooms housing patients with a high perceived risk of contaminating the environment with MRSA were selected; 7 of these patients had MRSA recovered from multiple sites, 2 had it recovered from stool, and 1 from urine. Five additional rooms were chosen that had a low perceived risk of MRSA environmental contamination; these rooms had housed MRSA‐negative patients for at least 1 month prior to sampling. Sterile cotton swabs were moistened in brain‐heart infusion broth (BD Diagnostics) and used to sample a surface area of approximately 25 cm². Each swab sample was plated onto a chromogenic MRSA‐selective medium (BBL Chromagar MRSA medium; BD Diagnostics) and then processed by RT‐PCR, as described elsewhere.³ After the RT‐PCR procedure, 500 μL of brain‐heart infusion broth was added to the original swab sample, and 20 μL was plated onto MRSA‐selective medium after 24 and 48 hours incubation. The MRSA‐selective medium was incubated at 36°C and checked at 24 and 48 hours. Mauve‐colored colonies growing on MRSA‐selective medium were confirmed to be MRSA by Gram staining, coagulase testing (Staphaurex; Remel), and growth on oxacillin screening agar (BD Diagnostics).

Results

MRSA was cultured from 16.7% of the 150 surface samples cultured by direct plating and from 26.7% of those cultured by direct plating plus enrichment broth, compared with 60.0% of the surfaces tested with RT‐PCR (26.7% vs 60.0% [ , by the Fisher exact test]). Compared with culture including enrichment broth, the sensitivity of RT‐PCR was 92.5%, the specificity was 51.4%, the positive predictive value was 41.1%, and the negative predictive value was 94.9% (Table). The inclusion of the enrichment step in the culture method increased the positive predictive value of culture from 26.7% to 41.1% but reduced its negative predictive value from 98.3% to 94.9% (Table). Three swab samples were negative for MRSA by RT‐PCR but were positive by culture (Table). One swab sample was unresolved by the RT‐PCR system, even after the manufacturer’s recommended freeze‐thaw procedure. Both culture and RT‐PCR identified MRSA contamination on significantly more surfaces in rooms occupied by MRSA‐positive patients, compared with rooms occupied by MRSA‐negative patients (32.0% vs 16.0% for culture [ ]; 69.0% vs 42.0% for RT‐PCR [ , by the Fisher exact test]), and there was a positive correlation ( ) between the proportion of sites identified as contaminated by RT‐PCR and those identified by standard culture. MRSA was cultured from 16.0% of 50 surface samples in the 5 rooms that had not housed a patient known to be colonized or infected with MRSA for over a month (data not shown).

Table.  Comparison of Real‐Time Polymerase Chain Reaction (RT‐PCR) Compared With Culture for the Detection of Methicillin‐Resistant Staphylococcus aureus on Environmental Surfaces

Open New Window

Discussion

The sensitivity and negative predictive value of the RT‐PCR system were over 90% and comparable with previous assessments of the RT‐PCR system.^3,4 The 3 swab samples that initially had false‐negative results later yielded isolates that were positive by RT‐PCR when tested from pure culture, suggesting that the amount of MRSA on these swabs was below the limit of detection for the RT‐PCR system. This is supported by the fact that samples from 2 of the 3 sites gave positive results only in culture that included enrichment broth, and the sample from the remaining site grew only 2 MRSA colonies in culture (data not shown). However, the specificity and positive predictive value of the RT‐PCR system were poor. This could be explained by the detection of DNA from dead or nonculturable MRSA, which are unlikely to play a role in transmission, or the reaction of one of the primers with an SCC fragment not containing functional mecA in methicillin‐susceptible S. aureus or coagulase‐negative staphylococci. PCR detection of DNA from dead or nonculturable bacteria in environmental samples has been described elsewhere,⁸ and Desjardins et al.⁹ reported that false‐positive RT‐PCR results were most likely caused by methicillin‐susceptible S. aureus. Because we used selective media, we were not able to determine whether the poor specificity was due to the detection of methicillin‐susceptible S. aureus, coagulase‐negative staphylococci, or DNA from dead or nonculturable MRSA. It is also possible that the poor specificity and poor positive predictive value of the RT‐PCR system could be explained by the low sensitivity of the MRSA‐selective medium, which has not been validated for the detection of MRSA from environmental surfaces. However, MRSA‐selective medium has proven more sensitive than nonselective culture media for the detection of MRSA from nasal swab samples,¹⁰ and we used nonselective enrichment broth to increase the sensitivity of the culture method. Furthermore, the frequency of environmental contamination identified by standard culture was comparable to that observed in other studies.⁵

The correlation between the frequency of contamination identified by RT‐PCR and that identified by culture was not strong ( ), and in 4 rooms where no MRSA was detected by culture, RT‐PCR results indicated contamination on 30% of the sites sampled in the first room, 50% of those in the second room, 60% of those in the third room, and 80% of those in the fourth room. Therefore, RT‐PCR was not an accurate measure of the frequency of contamination. The recovery of MRSA from culture of samples from 16% of the surfaces in rooms that had housed MRSA‐negative patients for at least a month was a surprising finding, which could be explained by MRSA environmental contamination persisting for long periods despite cleaning, by undetected patient colonization, or by the importation of environmental MRSA by hospital workers or visitors.

In summary, the poor specificity, poor positive predictive value, and high cost of this RT‐PCR method, compared with standard culture, mean that the method is not suitable for the detection of MRSA on environmental surfaces.