Prevalence and molecular characterization of Staphylococcus aureus from human stool samples

Staphylococcus aureus is a commensal bacterium and important cause of healthcare-associated infections [1]. Nasal carriage is considered to be the most important site of S. aureus colonization [2] and is the best-studied [3]. However, other extra-nasal body sites, including the gastrointestinal tract, are known to harbor S. aureus [4‐6]. Recent studies have found S. aureus in the intestines of healthy humans [7, 8] as well as the intestines of hospitalized patients [9].

S. aureus, an in particular methicillin-resistant S. aureus (MRSA) intestinal colonization, may be more common than previously thought and has been shown to be clinically important [10, 11]. Rectal carriers have been found to be at increased risk of developing S. aureus infections [10] and gastrointestinal carriage of MRSA has been associated with nosocomial antibiotic-associated diarrhea [11]. Screening for gastrointestinal carriage has been shown to identify colonized patients who would have been otherwise missed [12]. S. aureus gastrointestinal carriage may be an overlooked reservoir, contributing to hospital infection and transmission.

The objectives of this study were to determine the prevalence of S. aureus and MRSA in human stool samples at a large university hospital, to characterize the identified isolates by molecular methods, and to assess potential risk factors for intestinal carriage. We hypothesized S. aureus and MRSA prevalence in stool will be similar to what has been reported previously in the literature [3]. Furthermore, we hypothesized having a gastrointestinal condition would increase the risk of intestinal S. aureus carriage.

Of the 781 samples retrieved from the UIHC Clinical Microbiology Laboratory, six samples were excluded due to the tubes breaking during the freezing process. An additional 148 samples were excluded due to patients providing more than one stool sample and six additional samples were excluded due to medical records that were not able to be located, leaving 621 patients included in the study (Fig. 1). Demographic and clinical characteristics of the 621 patients included in the study are described in Table 1. The average age for all patients was 51.6 years (stan. Dev.: 19 years, range: 0 to 94 years). A majority of patients were female (320, 51.5%) and Caucasian (560, 90.2%). A majority were inpatients (487, 78.4%) with 20.1% (126/625) being outpatients, and 1.3% (8/625) being housed outpatients. The average length of stay for inpatients as defined as the entire time they were admitted for the stay when their stool samples was collected was 15.8 days (stan. Dev.: 28.6 days). There were no significant differences between carriers and non-carriers for any variables listed in Table 1.

Of the 621 samples, 58 (58/621, 9.3%) were positive for S. aureus, 26 of which (26/621, 4.2%) were MRSA via the presence of the mecA gene, with an MRSA prevalence of 44.8% (26/58). One isolate was positive for PVL. Thirty unique spa types were identified, with t002 being the most prevalent at 66.7% (n = 20) followed by t012 and t1635 both at 10% (n = 3). The BURP analysis resulted in one major cluster with the hospital-associated strain t002 as the founder. The spa cc-002 grouping accounted for 24 strains (42% of all strains) and four spa types (13% of all spa types). Two spa types (three isolates) were excluded as they had less than 5 repeat sequences present in the spa gene.

Antimicrobial susceptibility testing was performed on all isolates positive for S. aureus (Fig. 2). Resistance was observed for most antibiotics tested. The highest prevalence of resistance was to erythromycin at 51.7% (n = 37) followed by oxacillin at 43.1% (n = 25) and levofloxacin at 41.4% (n = 24). Resistance to clindamycin was observed at 22.4% (n = 16). No isolates were resistant to vancomycin, daptomycin, or quinupristin/dalfopristin. Resistance to all other antimicrobials was low (Fig. 3). Twenty-six (44.8%) isolates met the definition for multi-drug resistance (MDR) with having acquired non-susceptibility to at least one agent in three or more antimicrobial categories [18]. Of the MDR isolates, one isolate was resistant to at least one agent in six antimicrobial categories and one isolate was resistant to ≥1 agent in five categories. Twelve isolates were resistant to ≥1 agent in four categories and 9 isolates were resistant to ≥1 agent in 3 categories. Three isolates met the definition of MDR by being MRSA. Twenty-two of the 58 isolates (37.9%) were susceptible to all antibiotics tested. Six isolates (10.3%) were resistant to only one antimicrobial and two were resistant to two antimicrobials (3.4%). Two isolates were phenotypically resistant to oxacillin according to the AST (both isolates had a minimum inhibitory concentration [MIC] of ≥4 μg/mL); however, both isolates were negative for the presence of the mecA and mecC genes.

Medical record abstractions were completed on 189 patients (58 cases and 131 controls). A majority of cases were Caucasian (n = 168, 88.9%) and slightly over half were male (n = 98, 51.8%). In the univariate analysis, most covariates were not significantly associated with S. aureus intestinal colonization. We observed a borderline significant association with sex with females having increased odds of intestinal colonization, though the confidence interval included 1.0 (OR: 1.84, 95% CI: 0.98–3.44). Having a history of a gastrointestinal condition, the primary predictor, was significant in the unadjusted, univariate model with those having any disorder of the gastrointestinal tract having increased odds of intestinal colonization (OR: 1.96, 95% CI: 1.04–3.7). The most prevalent disorder of the GI tract was gastroesophageal reflux disease (GERD) at 31 of the 189 patients (16.4%) of the total population analyzed for the nested study. The second most prevalent condition was lower GI bleed (n = 16, 7.4%) followed by Crohn’s disease (n = 13, 6.9%) and any inflammatory disorder of the intestines including inflammatory bowel disorder (n = 13, 6.9%). No other covariates were significantly associated with intestinal colonization in the univariate analyses (Table 2).

The final adjusted model included having a disorder of the gastrointestinal tract, age, sex, laxative usage, anti-motility agent usage, having a S. aureus infection, whether the patient died, ICU length of stay, and the reason for specimen collection. An interaction between having a gastrointestinal disorder and age was forced into the model as the interaction term was verging on significance (p = 0.058) and did not increase the AIC by more than 3. In the final model, having a disorder of the GI tract was significantly associated with increased odds of being intestinally colonized with S. aureus, though the confidence interval is very wide (OR: 13.9, 95% CI: 1.67–115.7). No other covariates included in the final model were significantly associated with intestinal S. aureus colonization (Table 2).

We identified S. aureus in the intestines of patients at UIHC and report an intestinal colonization rate of 9.3%. During the 1950’s, S. aureus intestinal colonization was studied in greater detail than seen in the current literature [19]. Studies from this time report carriage prevalence ranges between 8% and 30% [3, 20, 21]. While researchers have known of the relationship between intestinal S. aureus carriage and antibiotic-associated diarrhea, as well as the increased risk of other S. aureus infections, intestinal carriage has been studied far less than carriage at other anatomical sites, particularly the nares. Recent studies of S. aureus intestinal carriage have detected intestinal carriage rates ranging from 10% [22] to as high as 37% [10]. It has been estimated the average rate of intestinal colonization in hospitalized patients is roughly 20% [3], though the number of studies included in reaching this estimate is small and the populations vary greatly. Our observed prevalence of 9.3% is lower than the estimated average and lower than most studies have reported to date. This may be due to the fact the population screened for this study included patients from many departments across a large teaching hospital as well as outpatients seen in the emergency department. Use of frozen instead of fresh stools may also have reduced our recovery rate.

Many of the studies in recent years have focused on MRSA carriage and have reported MRSA intestinal carriage rates frequently ranging from 5% [22, 23] to 10% [24] with some studies finding a MRSA prevalence as high as 22% in high-risk populations [25]. The observed prevalence of MRSA reported here is 4.2%, which is lower than others have reported. Of the S. aureus isolates identified, 44.8% met the definition for MDR-SA. Two of these isolates were phenotypically resistant to oxacillin; however, they were not positive for either resistance gene – mecA or mecC –tested for.

We found one isolate to be positive for the PVL gene [26]. As PVL genes are mainly associated with community- associated S. aureus, it is not surprising we found such a low prevalence in clinical cohort. Spa type t002 was the most frequently identified spa type accounting for 66.7% of all identified spa types. This spa type is most frequently associated with hospital strains and typically belongs to the pulsed-field type USA100 [27]. Spa type t012, the second most prevalent spa type observed in this study (10% of all spa types) is frequently found in community settings and may be associated with younger age [28].

Prevalence of intestinal carriage is dependent on the method of specimen collection employed. Varying rates of colonization are seen by method and anatomical site used to define “gastrointestinal.” The two main methods used to collect specimens are swabs and stool samples. Swabs collected from the rectum, anus, perineum, and the groin or inguinal region are typically accepted as representing intestinal carriage of S. aureus in addition to stool specimens [3]. Several studies have assessed which method and region provide the most consistent rate of colonization, but these studies are few and do not provide a consensus. Rectal swabbing has been shown to yield more S. aureus than stool cultures [29]. This may explain why our study reports a lower prevalence compared to many of the studies cited above which used rectal or perineal swabbing.

Intestinal colonization may be an important reservoir for the dissemination of S. aureus in the health care setting. Intestinal colonization is known to lead to increased risk of infection [30‐32], though the mechanism is not clear. It has been hypothesized intestinal colonization increases colonization or contamination of the skin which in turn increases contamination of the patient’s environment. Environmental contamination then increases both the risk of infection as well as the potential for nosocomial transmission [31]. While we were unable to investigate colonization of other body sites in this study, S. aureus intestinal colonization has been associated with an increased risk of skin colonization [10]. Nasal and intestinal carriage are also frequently observed in the same patient; however, intestinal colonization alone does increase the detection sensitivity [3, 22, 23, 33‐35]. In a meta-analysis by McKinnell et al., it was found rectal screening increased yield by 20%, with rectal screening having the greatest impact in hospitals with a low MRSA prevalence (23% increase) [35].

To assess potential risk factors for intestinal colonization, we performed a nested case-control study. We hypothesized having a disease or condition of the gastrointestinal tract would put patients at increased risk of S. aureus intestinal colonization. In the univariate analysis we found gastrointestinal conditions did significantly increase the odds of being colonized with S. aureus (OR: 1.96, 95% CI: 1.04–3.7) and this association remained significant after adjusting for the other variables included in the final model (OR: 13.9, 95% CI: 1.67–115.7), though the 95% confidence interval was very wide due to a small sample size. No other variables abstracted from the medical record were significantly associated with an increased risk of intestinal colonization in either the univariate or adjusted models. Other studies have reported several potential risk factors for intestinal S. aureus colonization. It has been shown that a stay in an extended care facility or nursing home can significantly increase the risk of intestinal colonization [36]; however, we were unable to assess nursing home stays in the present study. Studies have also reported length of stay [37], a recent history of antimicrobial usage [37‐39], a history of MRSA infection [38], and dependence on healthcare workers to perform activities of daily living [37] all significantly increasing the risk of intestinal S. aureus carriage. While we were not able to assess activities of daily living in the medical record, we did not detect a significant association with LOS, antibiotic usage, or the history of MRSA infections and colonization in our population. This may be due to the inclusion of a single hospital in this study, smaller sample size, and the reliance on medical record data.

Our study has several limitations. The first is the specimens analyzed were not collected for research purposes, but for routine medical care and as such we were only able to address intestinal colonization in a cross-sectional study and were unable to determine the duration of colonization in these patients. Future studies are needed to determine colonization duration, as well as the rates of persistent versus intermittent carriage. Furthermore, we were unable to determine if the S. aureus strain the patient was colonized with was the same strain causing an S. aureus infection, or otherwise present in other locations of the patient’s body or environment. This study was conducted in a single hospital in Iowa with a predominantly white population which may limit the generalizability of the study findings. However, UIHC is a large teaching hospital with a catchment area including several surrounding mid-western states. S aureus was identified from the stool of patients at the UIHC. Lastly, the samples used for this study were collected as part of routine clinical care and as such, all participants had some degree of gastrointestinal symptoms. A history of gastrointestinal conditions increased the risk of intestinal S. aureus carriage. Many of those isolates are resistant to antibiotics and may serve as a reservoir for subsequent infections and transmission events.