Transmission of meticillin-resistant Staphylococcus aureus (MRSA) is known to occur bidirectionally between humans and animals, including in household settings, yet the role of household pets as reservoirs for human MRSA infection and factors promoting pet MRSA carriage are not fully elucidated.^1,2

Pet- and household-related characteristics previously found to be associated with pet MRSA carriage include pet species, pet age and sex, treatment with antimicrobials, exposure to veterinary clinics and close contact with a human MRSA patient (see Supporting information Methods S1 for details). Environmental contamination with MRSA^3,4 is a risk factor for MRSA carriage and reinfection in humans. Previous studies have demonstrated correlation of MRSA strains among animals, humans and their shared home environments.^5,6

This study assessed risk factors for MRSA carriage in pets living with humans previously diagnosed with MRSA skin or soft-tissue infection (SSTI). Furthermore, staphylococcal protein A (spa-)types of MRSA isolates cultured from humans, pets and household environments were compared. We hypothesised that antimicrobial use, environmental MRSA contamination and close human–pet contact may be risk factors for pet MRSA carriage and that MRSA isolates obtained from pet and environmental samples would have the same spa-type.

Human participants provided written informed consent for their participation in the study. Pet sampling was approved by the Johns Hopkins School of Public Health Animal Care and Use Committee (ACUC), and pet owners provided written informed consent. Human index patients diagnosed around one month previously with CA-MRSA SSTI, their cohabitating family members and their domestic animals were enrolled.

Between January and December 2012, in collaboration with the University of Pennsylvania School of Medicine and other institutions, 88 human index participants representing 95 households in the mid-Atlantic region were enrolled into the “Pets and Environmental Transmission of Staphylococci” (PETS) study.⁷ PETS was a sub-study of a randomised controlled trial (RCT; Epidemiology and Transmission of MRSA in the Community, NCT00966446) which evaluated the effectiveness of MRSA decontamination (one week nasal mupirocin and two chlorhexidine-based body washes) through sampling of household members for MRSA at bi-weekly home visits for six months.

The PETS study occurred in a MRSA-enriched environment (based on known MRSA patient status) and involved two household pet and home environment sampling events: one baseline visit roughly 30 days after SSTI diagnosis for initial enrollment, and one three months later for longitudinal comparison of MRSA spa-types.

At each household visit, all pets were sampled at the nares or nasal planum, the tongue, gingiva or hard palate, and inguinal and perineal regions. Sterilised electrostatic cloths (Swiffer, Proctor & Gamble; Cincinnati, OH, USA) were used as described previously^8,9 for environmental surface sampling of settled dust from nine environmental sites.

Meticillin resistance was confirmed by the presence of a mecA or mecC gene (which encode altered penicillin-binding proteins) by universal mec-testing,¹⁰ and isolate species (S. aureus) was confirmed by multiplex PCR analysis of the nuclease (nuc) gene.^11,12

Presence of detectable MRSA in this pet population is defined as pet MRSA carriage, as this study cannot distinguish between contamination, infection or transmission. Variables evaluated as potential risk factors of interest included pet characteristics (Table 1a) and household characteristics (Table 1b).

Data from the baseline visit and a three month follow-up visit were analysed using Stata (v14) software (StataCorp; College Station, TX, USA). Estimates of associations between detection of MRSA in pets and potential risk factors were obtained using survey-weighted unadjusted and adjusted logistic regression (see Methods S1 for more detail). The spa-types of MRSA isolates were compared between human, pet and environmental samples, and a longitudinal comparison of spa-types from pet isolates was conducted.

Sixty-seven (70.5%) of 95 enrolled households reported ownership of at least one pet of any species; only dogs and cats were included in analyses. Participating pets at baseline included 71 dogs and 63 cats from 57 households. Tables 1a and 1b provide pet signalment/demographics and unadjusted risk factor data for pet MRSA carriage.

Eight (14.0%) of 57 households had at least one MRSA-positive pet. The prevalence of MRSA positivity at baseline was 9.0% (12 of 134 sampled pets; Table 1a). Five (7.0%) of seventy-one sampled dogs and seven (11.1%) of 63 sampled cats carried MRSA at baseline. On unadjusted logistic regression, risk factors significantly associated with pet carriage of MRSA included antimicrobial use in the pet and pet bed (environmental) MRSA contamination. At baseline, 100% of MRSA-positive pets lived in homes where one or more human-associated environmental site (not counting the pet bedding) was MRSA-positive, and 11 pets (91.7%) with detectable MRSA carried a spa-type concordant with that detected in the home environment. At the three month visit, 86 pets were tested (38 dogs and 48 cats). In only one of seven MRSA-culture positive pets (14.3%) was the environment MRSA-negative at this time point.

Seventy four pets were tested both at baseline and three months, and of these 13 (17.6%) were MRSA-positive at either or both time points. Of the nine longitudinally-sampled pets that were MRSA-positive at the first visit, five (55.5%) pets cleared MRSA carriage during this time and four (44.4%) pets had persistent carriage. Four (5.4%) pets had incident MRSA carriage (i.e. MRSA not present at baseline and then isolated at three month sampling). All pets with persistent carriage carried the same MRSA spa-type on both visits. Longitudinal comparisons of spa-types recovered from animal isolates at baseline and three months are provided in Table S5. See Results S1 and Table S4 for final risk factor model variable analysis at three months.

A priori, pet signalment variables (pet species, age and sex) were fixed in the multivariate model. On unadjusted analyses, pet neuter status, antimicrobial use, kennel exposure, fleas on pet exam and MRSA isolation from pet bedding were all associated with MRSA carriage at a p-value <0.20 and therefore evaluated in adjusted analyses.

Because no pets carrying MRSA were unexposed to household environmental MRSA, this variable could not be included in multivariate models. A number of factors were of marginal significance (p > 0.05 and p < 0.20) in unadjusted analysis at baseline and were considered for multivariate evaluation. These included flea infestation at household examination, neutered status and exposure to a boarding kennel within the past year. False discovery rate analysis suggested that only pet bed MRSA contamination would remain significant after consideration of multiple comparisons. After adjusting for all other variables, pet neuter status and kennel exposure were no longer significant (p = 0.729 and p = 0.354, respectively).

On sensitivity analysis for an adjusted model including veterinary clinic visits instead of pet antibiotic use, the associations between MRSA isolation from pet bedding and fleas on pet exam were unchanged, and an association between vet clinic visit association and pet MRSA carriage was not significant (p = 0.118).

Stepwise forward and backward logistic regression selection resulted in retention of pet antimicrobial use, presence of fleas on the pet and pet bed isolation of MRSA in a multiple logistic regression model. Inclusion of the human–animal contact variable did not improve model fit.

The final multivariate logistic regression model at baseline included pet species, age, sex, antimicrobial use, fleas on pet exam and MRSA isolation from pet bedding (Table 2).

All pets in this study had documented household exposures to a person with a MRSA SSTI, putting all of them at risk for MRSA acquisition. Pet bed (environmental) MRSA positivity was associated with higher odds of pet MRSA carriage in this population, and all positive pets resided in a house with at least one human-associated (nonpet-bed) environmental surface positive at baseline. This provides evidence that contaminated home environments, whether contaminated by a person or a pet, probably play some role in MRSA carriage, transmission or colonisation in pets exposed to MRSA. If the index human had a draining wound this could contribute to environmental MRSA contamination; however, this detail was not available from extracted data. Other risk factors identified here included antimicrobial use in the pet, and pet infestation with fleas.

The results reported in Table 3 strongly support the hypothesis that MRSA strain type can be shared amongst people, pets and their home environments. Most homes in our study had the same MRSA spa-type in all three reservoirs. A recent study in households of children with CA-MRSA⁶ also found that MRSA strains were shared among these three reservoirs (humans, pets and environment). The majority of pet isolates had spa-type t008, associated with the USA300 clone, which is the most common CA-MRSA strain in humans in the USA.

Human index participants with MRSA infection initially were treated by a physician, typically with antimicrobial drug therapy, which limits the ability to generalise these findings to pet populations in homes where antibiotics were not prescribed for human use. Additionally, results from this mid-Atlantic population may not reflect those found elsewhere.

Currently available guidelines¹⁴ state that while human-to-pet transmission of MRSA is possible, re-infection of the human patient from the pet is less likely, pet carriage of MRSA is usually temporary, and pet screening is recommended only within a broader household approach to prevent recurrent MRSA infection. We agree that early removal of pets from the homes of humans infected with MRSA may not be necessary if clinical management includes targeted testing. Testing the pet may not be beneficial without consideration of planning for concurrent environmental and pet MRSA decontamination.

In conclusion, this study – which focused on potential risk factors for pet MRSA carriage in order to guide best practices for risk-based MRSA testing – concluded that environmental contamination and pet antimicrobial use are risk factors for pet MRSA carriage. Longitudinal and interventional studies are needed to build better causal understanding of transmission pathways and dynamics. Next steps should identify whether interventions targeting the home environmental reservoir (including the pet’s bedding) can reduce risk of MRSA carriage in exposed pets.