Concordance of PCR and Culture from Nasal Swabs for Detection of Methicillin-Resistant Staphylococcus aureus in a Setting of Concurrent Antistaphylococcal Antibiotics

Abstract

The effect of concurrent administration of antibiotics on the detection of methicillin-resistant Staphylococcus aureus (MRSA) remains unresolved. Here, we assessed the concordance of paired nasal swabs processed using commercial PCR and culture and found high concordance in both the absence and presence of antibiotics with activity against MRSA (93.7% [95% confidence interval [CI], 88.1%, 96.8%] and 90.9% [95% CI, 84.8%, 94.7%], respectively), although PCR was more likely to be positive in the presence of antibiotics. (This study has been registered at ClinicalTrials.gov under registration no. NCT01234831.)

TEXT

MHospitals and increasingly, ambulatory settings, are endemic for methicillin-resistant Staphylococcus aureus (MRSA) (1–3). Once patients are found to have had a positive MRSA culture, they require contact precautions (CP) (4, 5). Nasal surveillance specimens are components of institutional protocols for the determination of colonization status, and in some cases, for discontinuation of CP; many protocols require such specimens to be obtained in the absence of antibiotic exposure (6). Given the widespread antibiotic use among inpatient populations, the impact of antibiotics on the ability to recover MRSA has important implications for infection control and resource utilization (7–9). To compare the effect of concurrent administration of antibiotics with activity against MRSA on the detection of MRSA from nasal surveillance swabs, we assessed the concordance of paired samples obtained in a randomized controlled trial, processed using both commercial culture and PCR assays (10). Discordance between PCR and culture results may result from the following: PCR detection of nonviable MRSA organisms, possibly affected by previous antibiotic exposure, in contrast to dependence of culture on bacterial growth; greater sensitivity of PCR than of culture techniques for detection of viable bacteria; or false-positive PCR results due to the staphylococcal cassette chromosome (SCC) elements present in the absence of the mecA gene.

(This study was presented in abstract form at ID Week, San Diego, CA, October 2012.)

We performed an analysis of subjects enrolled in an investigator-initiated randomized controlled trial (registered at ClinicalTrials.gov under registration no. NCT01234831). Eligible subjects for that trial included those patients admitted to the Massachusetts General Hospital (MGH) between 6 December 2010 and 16 September 2011 with a history of MRSA infection or colonization ≥90 days from the day of admission. The subjects were approached and attempts were made to screen them on days 1, 2, and 3 of hospitalization. Subjects who were enrolled and had at least one set of surveillance screens performed were included in this analysis. The subjects were screened simultaneously by study staff using both the BBL CultureSwab collection and transport system and inoculated onto BBL CHROMagar MRSA medium plates (CA) (Becton, Dickinson and Company) and commercial PCR using the Xpert MRSA real-time PCR assay on the GeneXpert platform (PCR) (Cepheid). The samples obtained within the 48 h after the administration of one or more doses of select antibiotics were considered to be obtained in the presence of “concurrent antibiotics.” These antibiotics included trimethoprim-sulfamethoxazole, mupirocin, ciprofloxacin, clindamycin, daptomycin, doxycycline, levofloxacin, linezolid, nitrofurantoin, quinupristin-dalfopristin, rifampin, tetracycline, tigecycline, or vancomycin (intravenous [i.v.] only). The first paired sample was analyzed due to the policy implications of using a single PCR for infection control purposes.

A total of 259 subjects were simultaneously tested using CA and PCR. The t test was used to compare age, length of hospital stay (LOS), hospitalizations in the month and year prior to admission to MGH, and MRSA history (time since last positive MRSA culture and time since original MRSA culture). The chi-square test was used to compare gender, race, admission source, admitting site, and admitting service. The concordance between CA and PCR of single paired nasal swabs in the presence and absence of antibiotics (as defined by the study) with known activity against MRSA was assessed, and the data were analyzed using the exact McNemar's test. The Wilson score method without continuity correction was used to calculate the 95% confidence interval (CI).

Antibiotic exposure was identified in 51% (132/259) of the screened patients, while 49% (127/259) were sampled in the absence of antibiotics, forming the two comparison groups. There were no observed statistically significant differences in the antibiotic receipt or nonreceipt groups in age, gender, admission source, admitting site, admitting service, and number of hospitalizations at our institution in the prior month or year. Compared to the off-antibiotics group, the group with concurrent antibiotic exposure had a longer mean length of stay (8 days versus 6 days, P = 0.03) and a greater median number of inpatient hospitalizations at MGH in the prior year (1 versus 0, P = 0.03). The group with concurrent antibiotic exposure had a more recent history of MRSA infection or colonization than the nonreceipt group, with a significantly lower mean number of days since the last positive MRSA isolate was detected (586 days versus 872 days, P < 0.01) and mean number of days since their original MRSA isolate was detected(951 days versus 1,225 days, P = 0.02) (see Table SA in the supplemental material).

There was high concordance between the culture and PCR assays across all subjects tested (Table 1). Among 127 paired samples that were obtained in the absence of antibiotics, 119 (93.7%; 95% CI, 88.1%, 96.8%) were concordant. For those without exposure to MRSA-active antibiotics, the exact McNemar's test revealed no significant difference in the proportion of positive test results between CA and PCR (33.1% [42/127] versus 36.2% [46/127]), respectively; P = 0.29). Of the 132 paired samples obtained in the presence of antibiotics, 120 (90.9%; 95% CI, 84.8%, 94.7%) were concordant. For those with exposure to MRSA-active antibiotics, however, there was a greater tendency for PCR to give positive results (34.1% [45/132] positive for CA and 41.7% [55/132] positive for PCR; P < 0.01). Of the 12 patients with discordant results, the single patient with a culture-positive/PCR-negative discordance had received a single dose each of vancomycin and a fluoroquinolone; in the 11 patients with a culture-negative/PCR-positive discordance in the presence of MRSA-active antibiotics, eight had received at least one dose of vancomycin during the 48-h time window and three had received at least one dose of a fluoroquinolone.

TABLE 1.

Concordance of paired nasal swabs^a

^aShaded cells represent concordant paired results.

^bCA, chromogenic agar.

^cAntibiotics are those defined in the study protocol.

Regardless of the assay employed, efficient strategies for discontinuation of CP are needed in order to improve patient care and resource utilization. This study demonstrates high concordance in both the absence and presence of concurrent antibiotic use; however, PCR was significantly more likely to give positive results in the presence of concurrent antibiotic administration. The latter may reflect the higher sensitivity of PCR than culture for the detection of viable organisms, the detection of DNA from nonviable organisms, or false positives resulting from staphylococcal cassette chromosome (SCC) elements in the absence of the mecA gene. This study has limitations, most notably that the current MRSA status of subjects with a history of MRSA was not known with certainty. As such, it was not possible to evaluate standard test characteristics, such as sensitivity and specificity. Broth enrichment techniques would attenuate this limitation. In addition, it was not possible to randomize subjects in the study to receipt or nonreceipt of antibiotics, and there were differences observed between the groups, notably in the length of stay and timing of the most recent MRSA isolate.

In conclusion, PCR and culture surveillance swabs for MRSA had a high concordance rate, although PCR was more likely to be positive in the setting of concurrent antibiotic administration. Others (11, 12) have compared molecular methods with culture methods and found discordance between the samples, although not in the context of assessing for clearance of colonization and discontinuation of CP. Given our findings, discontinuation policies should carefully consider the interpretation of assays obtained in the presence of antibiotics with activity against MRSA. A conservative interpretation would use any positive PCR result as evidence of continued colonization with viable MRSA. Infection control programs will need to make individual judgments as to whether a negative PCR result is sufficient to proceed with discontinuing precautions in patients with negative results obtained while on antibiotics.