Genomic Epidemiology of MRSA During Incarceration at a Large Inner-City Jail

Abstract

Background

Congregate settings, such as jails, may be a location where colonized detainees transmit methicillin-resistant Staphylococcus aureus (MRSA). We examined MRSA acquisition during incarceration and characterized the genomic epidemiology of MRSA entering the jail and isolated during incarceration.

Methods

Males incarcerated at the Cook County Jail were enrolled within 72 h of intake and MRSA surveillance cultures collected. Detainees in jail at Day 30 were re-cultured to determine MRSA acquisition. A survey was administered to identify acquisition predictors. Genomic sequencing of surveillance and clinical isolates was integrated with epidemiologic and jail location data to track MRSA transmission pathways.

Results

800 males were enrolled; 19% MRSA colonized at intake. Of 184 who reached Day 30 visit, 12 acquired MRSA. Heroin use before entering (OR 3.67, P = .05) and sharing personal items during incarceration (OR = 4.92, P = .01) were predictors of acquisition. Sequenced clinical USA300 isolates (n = 112) were more genetically similar than diverse intake USA300 strains (P < .001), suggesting jail transmission. Four acquired colonization isolates were within 20 single-nucleotide variant (SNVs) of other isolates; 4 were within 20 SNVs of an intake isolate, 2 for an acquisition isolate, and 1 for a clinical isolate. Individuals with genetically similar isolates were more likely to have had overlapping stays in the same buildings.

Conclusion

There was a high MRSA burden entering jail. Genomic analysis of acquisition and clinical isolates suggests potential spread of incoming strains and networks of spread during incarceration, with spread often occurring among detainees housed in similar locations. Sharing personal items during incarceration is associated with MRSA acquisition and could be a focus for intervention.

There is a high MRSA burden entering jails. Genomic analysis of acquisition, intake, and clinical MRSA isolates suggests spread of incoming and endemic strains during incarceration.

Methicillin-resistant Staphylococcus aureus (MRSA) is a significant cause of clinical infection in urban communities [1]. Congregate living (homeless shelters, military barracks, and correctional facilities), close person-to-person contact, sharing personal items, poor hygiene, environmental contamination, and compromised skin integrity promote MRSA transmission [2]. Although infection control recommendations address these factors [3], certain community settings provide special challenges.

Correctional facilities, jails and prisons, are congregate settings where outbreaks of MRSA have occurred [4–6]. In contrast to prison, jails have relatively short-term incarcerations while detainees await sentencing, with high turnover and recidivism. These features could augment MRSA spread. Models suggest that MRSA transmission occurs during incarceration; individuals colonized with MRSA are the primary source of transmission; and following discharge, in the absence of jail interventions to control MRSA, resistance spreads to the community at large [7]. However, lack of data on MRSA transmission in jails has hampered establishing these facilities as potential key points of intervention.

A complicating factor in understanding transmission dynamics of MRSA and identifying targets for interventions is that individuals entering jail may be colonized due to high-risk community exposures [8–10], illicit drug use, unstable housing, and type and location of residence [11–15]. Such individuals may then be at risk for developing MRSA infection during incarceration and have the opportunity to intermingle with other individuals, potentially increasing MRSA spread.

Prior work in urban jails demonstrated that jail-based interventions can significantly impact community disease patterns (eg, sexually transmitted diseases) [16–19]. It remains unclear if urban jails are nonhospital settings that are a controllable focus of MRSA and if a jail intervention could have downstream benefit to the community-at-large for reducing MRSA burden. Therefore, the study objectives were to (1) examine the rate of MRSA acquisition during incarceration at a large urban jail, (2) identify epidemiologic and jail-based predictors of MRSA acquisition, and (3) characterize the genomic epidemiology of colonizing and clinical MRSA strains.

METHODS

Study Population

The study setting was the Cook County Jail in Chicago, IL, one of the largest single-site US jails, with roughly 250 incarcerations daily and daily census of 9000–10 000 detainees. Incarcerated males were enrolled within 72 h of entering jail from January 2016–December 2017. To enroll throughout the year and given the large number of HIV-negative individuals entering the jail, we targeted enrollment at 10 HIV-negative males each week. Given our prior work demonstrating the significant impact MRSA has on HIV-infected individuals and the possible intersection with incarceration for amplifying risk [12, 20, 21], we enriched the study population for HIV-infected individuals by enrolling from the jail HIV clinic (58% of our enrolled sample was HIV-infected). The estimated prevalence of HIV-infected detainees at the Cook County Jail is ~2%. Individuals were enrolled from the jail HIV clinic within 24–48 h from jail entrance. Males were followed during incarceration and were eligible for a second study visit at Day 30 if still incarcerated.

Swab Collection and Processing

Surveillance cultures (anterior nares, throat, and bilateral inguinal) were collected at enrollment and Day 30 (if still incarcerated). Intake and Day 30 results determined colonization status. Specimens were obtained using the Copan ESwab for MRSA. Nasal swabs were collected by swabbing both anterior nares; throat swabs by swabbing the posterior pharynx; and inguinal swabs by swabbing a 10 cm² skin area bilaterally [10]. Sample sites were chosen to maximize identification of MRSA carriers [2, 20]. Swabs were inoculated into enrichment broth to increase culture sensitivity [22]. Aliquots of overnight broth cultures were inoculated on ChromID MRSA (bioMérieux, North Carolina). MRSA was confirmed by standard biochemical tests; methicillin resistance by cefoxitin disk. Confirmed MRSA isolates underwent DNA extraction.

Archived Clinical MRSA Isolates

Existing archived clinical MRSA from male detainees incarcerated during the time of the study also underwent genomic sequencing. Clinical isolates from detainees who had been in the jail for >72 h were considered “jail-onset” infections; those occurring ≤72 h into the jail stay were considered community-onset. Infections from before the study, in an individual who remained in jail during the study, also were included to put into context acquisition MRSA isolates and to better characterize the genomic epidemiology of circulating strains (See Supplemental Methods).

Whole Genome Sequencing

Whole genome sequencing (WGS) was performed on MRSA isolates from jail entry and day 30 study visits and on clinical MRSA isolates from male detainees collected during the study period (See Supplementary Methods). Details on sequenced strains are available in Supplementary Table 1 and raw sequence data are available under Bioproject PRJNA638400. Intake positive colonization isolates were previously submitted under Bioproject PRJNA530184. Representative isolates that capture all independent acquisition events were selected for each person (See Supplementary Methods, Supplementary Figures S1 and S2) Publicly available USA300 genomes used in Figure 1 were downloaded from Bioproject PRJNA374377 [23] (Supplementary Table 2). Details on variant calling and phylogenetic analysis are in the Supplementary Methods.

Figure 1.

Whole-genome phylogeny of USA300 MRSA infection and colonization isolates in the jail. Recombination-masked whole-genome alignment was used to make a maximum likelihood phylogeny of intake colonization, jail-acquired colonization, and jail-onset infection collected from individuals in the jail and publically available genomes [23]. Tree is midpoint rooted. For samples in the current study, only a single isolate per individual was included, unless genomic analysis supported multiple isolates from an individual being associated with independent acquisition events (see Methods). Overall, jail isolates span the full diversity of the USA300 phylogeny, with intermixing of intake colonization, jail-acquired colonization, and jail-onset infection isolates. However, in the background of this diversity, clustering of isolates can be observed, particularly for jail-onset infections. Publicly available isolates span the diversity of the tree, but do not interrupt clusters of jail samples. Scale bar represents substitutions per site. One isolate with a long branch was removed for visualization purposes (see full USA300 tree in Supplementary Figure S8).

Location Overlap Analysis

Transmission pairs were defined as involving one individual who acquired colonization or developed a jail-onset infection and a source. Potential sources were community-onset MRSA infections, infections that occurred outside the study period in an individual who remained in jail during the study, those already colonized at intake, other jail-onset MRSA infections, and other acquired colonization isolates (See Supplemental Methods, Supplementary Figure S3). Location sharing among genetically related transmission pairs was assessed for only USA300 isolates. Electronically available jail location data, including building and living units, were ascertained for enrolled individuals and those with jail-onset infections. For each pair within the particular single-nucleotide variant (SNV) window (eg, 0–9, 10–19), overlap in the jail or in a particular building during an epidemiologically relevant window was calculated (Supplementary Figure S4). Based on the building in which each pair overlapped, we determined if the pair shared a living unit at the same or any time during their stay. Statistical significance of overlap at difference SNV thresholds was determined using permutation tests (See Supplementary Methods).

Risk Factors and Statistical Analysis

A survey to identify predictors of MRSA colonization was administered to detainees at enrollment and included questions about drug use, sexual behaviors, housing status, and incarceration history. A survey about behaviors and activities during incarceration was administered at Day 30 to identify predictors of MRSA acquisition.

Using intake and Day 30 surveillance cultures, we determined the frequency of persistent colonization (MRSA positive at intake and Day 30), presumptive acquisition (MRSA negative at intake and positive at Day 30), loss of colonization (MRSA positive at intake and negative at Day 30), and absence of colonization (negative at both time points). Results of surveillance cultures and intake surveys were used for risk factor analysis. SAS software version 9.4 (SAS Institute, Cary, North Carolina) was used for statistical analysis. Chi-square analysis was used for categorical variables, with Fisher’s exact test for low-frequency predictors.

The study was approved by the Cook County Health institutional review board (IRB) which oversees approval for enrollment of jail detainees and the Rush University IRB; verbal consent was obtained. Approval from the Office for Human Research Protections was obtained to enroll current detainees.

RESULTS

Features of the Study Population

There were 718 unique individuals (800 incarcerations) enrolled. The prevalence of MRSA colonization at intake was 19% [10]. Strains brought into the jail by those colonized at intake were diverse (Supplementary Figures S5 and S6). Among those enrolled, 267 (33%) incarcerations lasted 30 days or longer. Of those remaining incarcerated at Day 30, 160 individuals accounting for 184 (70%) incarcerations completed the Day 30 study visit. Among those completing the Day 30 study visit, 82% were African-American and 7% Hispanic. Use of illicit drugs before incarceration was common among individuals who completed the Day 30 study visit, with 80% reporting use in the past year. Recidivism was high; 91% of individuals in the study had prior jail incarceration.

MRSA Colonization Patterns During Incarceration

Of the 184 detainees with a completed Day 30 study visit, 41 (22%) were positive and 143 (78%) were negative for MRSA at admission. Of the 143 negative at admission, 131 (91.6%) remained negative at the Day 30 study visit and 12 (8.4%) acquired MRSA (Supplementary Figure S7). Of the 12 acquisitions, 2 were the same individual who, in sequential incarcerations, separated by 3 months, was negative at intake but colonized at Day 30. For this individual, the putatively-acquired strains were >1000 SNVs apart, supporting acquisition of a new strain rather than intermittent carriage. Of the 41 incarcerations positive at admission, 17 (41%) were no longer colonized, and 24 (59%) remained colonized at the Day 30 study visit (Supplementary Figure S7). One of these 24 detainees acquired a new strain of MRSA in the nares by the Day 30 visit and maintained throat colonization with the initial strain.

For the 12 MRSA acquisitions, 9 (75%) individuals were colonized at 1 body site, 2 (17%) at 2 body sites, and 1 (8%) at 3 body sites. By body site, 2 (17%) individuals had throat colonization, 7 (58%) nares colonization, and 7 (58%) inguinal colonization. There were no differences in likelihood of acquisition of colonization by body site.

For the 24 participants who had persistent colonization, 8 (33%) were colonized at 1 body site, 6 (25%) at 2 body sites, and 10 (42%) at 3 body sites. By body site, 18 (75%) had throat colonization, 16 (67%) nares colonization, and 16 (67%) inguinal colonization. There were no differences in likelihood of persistent colonization by body site. Individuals colonized at multiple body sites were usually colonized with the same strain (Supplementary Figure S2).

Clinical MRSA Isolates

There were 142 representative clinical MRSA isolates from male detainees who underwent whole genome sequencing (WGS); 125 were jail-onset during the study period, 3 were community-onset during the study period, and 14 were isolated in the year prior. Sequenced clinical isolates were mostly from skin/skin structure infections (98.6%) and identified as USA300 (92%). Of individuals with clinical infections, 66% were African-American, 8.5% Hispanic; mean age was 37 (SD 12) years; 43% reported current illicit drug use; and 11.3% were living on the street before incarceration.

Epidemiologic Predictors of MRSA Acquisition During Incarceration

Among the 12 individuals who acquired MRSA colonization, 11 were African-American and none Hispanic. Among exposures before incarceration (Table 1), heroin use was significantly associated with acquiring MRSA colonization (P = .05). No other types of drug use were associated with acquisition. While HIV status was not associated with MRSA acquisition, taking antiretrovirals was negatively associated with MRSA acquisition (P = .08)

Table 1.

Association of Epidemiologic Factors Before and During Incarceration with Risk for Acquisition of MRSA Colonization

Epidemiologic Factor	MRSA Acquisition (n = 12)a	No MRSA Acquisition (n = 131)	OR	95% CI	P value
Exposures prior to incarceration
Race/Ethnicity
African-American (reference)	11 (92%)	106 (81%)
Hispanic	0	10 (8%)	N/A	N/A	.6
White/Other	1 (8%)	15 (11%)	0.64	0.07, 5.34	1.00
Age, mean years (SD)	39.7 (11.9)	37.6 (11.8)	1.02	0.97, 1.07	.55
Heroin use in the past year	5 (42%)	21 (16%)	3.67	1.06, 12.68	.05
Marijuana use in the past year	8 (67%)	89 (69%)	0.9	0.26, 3.16	1.00
Cocaine use in the past year	5 (42%)	48 (37%)	1.21	0.36, 4.01	.76
Ecstasy or psychedelic use in the past year	1 (8%)	31 (24%)	0.29	0.04, 2.32	.30
Other narcotics (eg, codeine, oxycontin) in the past year	2 (17%)	14 (11%)	1.64	0.33, 8.27	.63
Illicit benzodiazepine use in the past year	1 (8%)	13 (10%)	0.81	0.1, 6.8	1.00
Taking prescription drugs to get high in the past year	1 (8%)	6 (5%)	1.86	0.21, 16.9	.47
Injection drug use in past year	2 (17%)	13 (10%)	1.82	0.36, 9.2	.61
Homeless or unstable housing in the past year	7 (58%)	62 (47%)	1.56	0.47, 5.16	.46
HIV infection	9 (75%)	91 (69%)	1.32	0.34, 5.13	1.00
Taking antiretrovirals	3 (33%)	59 (66%)	0.26	0.06, 1.12	.08
Taking TMP-SMX	1 (8%)	13 (10%)	0.83	0.1, 6.91	1.00
Men who have sex with men	5 (42%)	34 (26%)	2.04	0.61, 6.85	.31
ER visit in the past year	7 (58%)	71 (54%)	1.18	0.36, 3.92	.78
Hospitalized in the past year	6 (50%)	50 (38%)	1.62	0.50, 5.30	.54
Exposures during incarceration
Participated in drug treatment	4 (33%)	22 (17%)	2.48	0.69, 8.95	.23
Sharing of personal itemsb	7 (58%)	29 (22%)	4.92	1.45, 16.67	.01
Any skin infections	1 (8%)	4 (3%)	2.89	0.30, 28.11	.36
Visit to infirmary	4 (33%)	20 (15%)	2.77	0.76, 10.09	.12
Number of times showered in the past week, mean (SD)	4.8 (1.8)	6.0 (2.7)	0.81	0.62, 1.07	.13

^aWith inclusion of the new strain acquisition event detected with WGS for a person colonized at intake and remained colonized at DAY 30, all associations remained similar for predicting acquisitions. Heroin use before incarceration (P = .02) and sharing personal items (P = .02) both remained significant.

^bPersonal items shared by individuals who acquired MRSA included towel, toothpaste, uniform, and deodorant.

Among exposures occurring during incarceration, sharing personal items was significantly associated with MRSA acquisition (OR 4.92; 95% CI: 1.45, 16.67, P = .01). A variety of personal items were shared and no one individual item was associated with increased risk of MRSA acquisition.

Genomic Epidemiology of USA300 MRSA Intake, Clinical, and Acquisition Isolates

While USA300 intake, jail-onset clinical, and colonization acquisition isolates were overall diverse, clusters of closely related strains were identified (Figure 1, Supplementary Figure S8). Examining strains with close genetic neighbors revealed in comparison to intake isolates, jail-onset USA300 clinical and acquisition colonization isolates were more likely to have closely related genetic neighbors (Figure 2), suggesting the existence of transmission networks that included both colonized and infected individuals. Significance remained when adjusting for difference in sample sizes between the potential sources (Supplementary Figure S9).

Figure 2.

Comparison of genetic diversity of intake colonization MRSA isolates versus jail-acquired colonization and jail-onset infection. To evaluate whether jail-acquired USA300 MRSA colonization and jail-onset USA300 MRSA infections were enriched for recent transmission events, their genetic diversity was compared to that of intake USA300 MRSA colonization by creating distributions of genetic distances to closest genetic neighbors (core genome size = 2.54 Mb). Closest-pair sources for intake USA300 MRSA colonization include other intake isolates (n = 100). Closest-pair sources for jail-acquired USA300 MRSA colonization (n = 9) and jail-onset USA300 MRSA infections (n = 113) included all isolate types (n = 239 sources including intake positive colonization, jail-onset infection, jail-acquired colonization, community-onset infection (n = 3), infections that occurred in 2015 but were in jail during the study period (n = 14)). Wilcoxon rank-sum test was used to make pairwise comparisons between the 3 sets—one-sided test for A and B, two-sided test for C. Comparisons are shown for (A) jail-onset infections versus intake colonization, (B) jail-acquired colonization versus intake colonization, and (C) jail-acquired colonization versus jail-onset infections. Histograms are overlapping, not stacked, and colors are blended in overlapping parts of distributions. Significance remained when controlling for the differences in number of possible pairs for intake colonization and jail-onset infections (see Supplementary Methods, Supplementary Figures S9).

Four acquisition isolates were closely related to at least one other isolate (range 1–3) (4 were within 20 SNVs of an intake isolate, 2 to another acquisition isolate, and 1 to a clinical isolate). We observed that 50.4% of jail-onset infections were within 20 SNVs of another isolate, with 17% of clinical isolates within 20 SNVs of an intake colonization strain. 95% of jail-onset clinical MRSA isolates with a genetic neighbor within 20 SNVs were related to another clinical MRSA strain. 76% of jail-onset clinical MRSA isolates that are within 20 SNVs of another isolate are within 20 SNVs of multiple isolates (range 2–8) (Supplementary Figure S10).

For detainees with USA300 isolates, individuals with longer lengths of stay tended to have a closer genetic pair (Figure 3A, Supplementary Figure S11). Furthermore, individuals who developed MRSA infection had longer jail lengths of stay than study participants who came into the jail MRSA colonized and did not develop infection during that incarceration (Supplementary Figure S12).

Figure 3.

Individuals with closely related USA300 MRSA strains are more likely to reside in common jail locations and have longer length of stay. (A) Each square indicates the mean length of stay of unique individuals involved in a pair (y-axis) within the respective SNV distance range (x-axis). SNV distances ranges are inclusive. For A, B, and C, pairs involve a putative acquisition of USA300 MRSA (jail-acquired colonization or jail-onset infection) and a source. See distribution of length of stay in Supplementary Figure S11. (B) Each dot indicates the percentage of pairs related by the respective SNV distance that overlapped in the respective location (y-axis) in an epidemiologically relevant window. Sequential overlap indicates that 2 individuals were both in the same living unit at some point in their jail stay during an epidemiologically relevant window, but not necessarily at the same time. See Supplementary Figure 13 for results of permutation test. (C) Each dot indicates the mean time a pair of individuals related by the respective SNV distance overlapped in the jail, the same building, or the same living unit in an epidemiologically relevant window (see Supplementary Methods). See Supplementary Figure 14 to see the distribution of days overlapped in jail and in a particular building, and Supplementary Figure 15 for detailed results of permutation test. In panels B and C asterisks indicates significance by permutation test where 1 asterisk indicates significance at P < .05 and 2 asterisk indicates significance at P < .005.

Relationship of Jail Location to USA300 MRSA Colonization and Clinical Isolates

Eight buildings housed male detainees, including 4 cell-based and 4 dormitory-style buildings. Individuals incarcerated in the jail typically had opportunity for multiple movements, including between buildings (Supplementary Figure S13), to court, and to social programs (eg, church, school, drug treatment). Among individuals who acquired USA300 MRSA colonization or had a jail-onset USA300 MRSA infection, those with more closely related genetic neighbors tended to be in the jail at the same time and to overlap in particular buildings (Figure 3B, Supplementary Figure S14) and for significantly longer than did random pairs of individuals (Figure 3C, Supplementary Figures S15 and S16). Furthermore, 13 of 35 pairs of individuals who overlapped in the same building and whose MRSA isolates were within 9 SNVs overlapped at the more granular level of living unit, suggesting direct (eg, no intermediaries) transmission (Figure 3B). While more distantly related pairs (10–50 SNVs) overlapped in the same building more than random pairs (>50 SNVs), they tended to have only sequential exposures to the same living units, suggesting potential transmission with intermediaries or environmental contamination as a source (Figure 3B, Supplementary Figure S14D, E).

While all buildings were sites of overlap among individuals harboring closely related MRSA strains, certain buildings, including both dorm and cell-based buildings, had significantly more overlap than expected by chance (Figure 4, Supplementary Figure S17). We also noted that most genomic clusters of individuals whose isolates are within 20 SNVs (range 2–8 isolates) cannot be explained by overlap in a single building, indicating that transmission clusters are not necessarily confined to individual buildings (Supplementary Figure S10).

Figure 4.

Overlap in buildings among individuals with closely related MRSA strains.Colors represent pairs of USA300 isolates (including a putative acquisition of colonization or infection and a source) genetically related at different SNV thresholds (ie, ≤10, ≤20, ≤40). Random (in gray) indicates all pairs of isolates and is shown to provide the baseline location sharing of pairs of individuals for each building regardless of genetic linkage. The x-axis indicates the different buildings male detainees could stay, labeled as cell-based (subscript c) or dorm-based building (subscript d). The y-axis indicates the percent of pairs that overlap in an epidemiologically relevant window in each building (See Supplementary Methods). Asterisks indicates significance by permutation test where 1 asterisk indicates significance at P < .05 and 2 asterisks indicates significance at P < .005 (see Supplementary Methods and Supplementary Figure S17 for results of permutation test).

DISCUSSION

There is a high burden of MRSA entering Cook County Jail; 19% of males arrive colonized [10]. Beyond that, we detected presumptive acquisition of MRSA colonization by day 30 of incarceration in an additional 8.4%, with sharing personal items a major risk factor. Genomic analyses, especially the small SNV differences among acquired and clinical MRSA isolates, suggests potential spread of incoming as well as of prevalent MRSA strains, with transmission potentially occurring among detainees housed together.

The ~8% acquisition rate is higher than that reported among individuals in other congregate settings [24] and more in line with rates in intensive care units [25, 26] where MRSA, at least for some units, may be viewed as endemic. Interestingly, we observed that heroin use before entering the jail was significantly associated with acquisition. It is unclear if individuals who use heroin tend to congregate with certain populations, are housed in similar locations, or are characterized by factors or behaviors occurring during incarceration that were unmeasured. Nevertheless, a unique feature and likely a major challenge of jails is the exceedingly high incoming MRSA prevalence [10].

While it did not attain statistical significance, we observed that more frequent showering was negatively associated with acquisition of MRSA colonization. A prior case-control study in the LA County Jail observed that sharing soap with other inmates and less frequent showering were 2 factors associated with developing a MRSA infection during incarceration [27]. Individuals who developed MRSA infection tended to have longer lengths of detention, which may reflect the longer period of observation but also more opportunities for interactions with others. In the community, recommendations for hygiene and against sharing personal items remain critical components of education patients receive to prevent MRSA [3]. Education regarding sharing of personal items and hygiene is a key, although difficult to enforce, MRSA control intervention for congregate living settings such as jails.

Using WGS, we observed that clinical USA300 infections occurring during incarceration have greater genomic similarity to each other in comparison to the diverse intake USA300 colonization strains. This finding suggests that infections could have originated from transmission within the jail. We observed that individuals with genetically similar MRSA strains were more likely to overlap in jail, suggesting spread among detainees housed in similar locations. As support for possible transmission, 35 of the 61 pairs within 9 SNVs overlapped in the same building, with 20 pairs having directly (n = 13) or sequentially overlapped in the same living units. While pairs separated by moderate SNV thresholds (20–50) still had significant overlap in a particular building, they often did not directly overlap in a living unit within that building. However, 10%–16% of pairs at moderate SNV thresholds had sequential occupancy of the same living unit, suggesting persistent environmental contamination or exposure to a more persistent MRSA strain in the jail.

Our observation that clusters of genomically-similar infection isolates existed even among people not sharing the same building or living unit suggests there could be virulent sublineages of MRSA that are more likely to cause infections or certain infections (eg, draining wounds) are more likely to be involved in MRSA spread. These hypotheses warrant additional investigation as they could provide possible targets for interventions.

The continual influx of MRSA-colonized individuals into a setting characterized by close person-to-person contact, compounded by reduced opportunities for infection control, shows how challenging infection control can be in congregate settings. Prior mathematical models of the LA County Jail MRSA outbreak predicted that MRSA spread becomes more problematic as there are increased numbers entering the system with MRSA [7]; our results support this prediction. This model also noted that as more infections occurred during incarceration, increased spread to the surrounding community could occur [7]. Further understanding both the downstream impact in the community of high MRSA burden in the jail as well as delineating the role recidivism contributes to overall MRSA burden is warranted.

Our study has limitations. First, we performed surveillance for MRSA colonization acquisition at Day 30 and thus may have missed acquisitions at shorter jail stays. Second, we cannot state definitively whether Day 30 positive swabs are acquisitions or intermittent colonization. However, the epidemiologic risk factors (ie, sharing personal items) and genomic data lend support to these instances being acquisitions. Although given the small number of acquisitions, though statistically significant, the effects for heroin use and sharing personal items should be interpreted with due caution. Third, we did not ascertain infection control behaviors (eg, showering frequency, sharing personal items) among individuals who developed an MRSA infection and therefore cannot comment on the influence of such behavior on developing an infection. Fourth, we screened only a small percent of incoming detainees and likely missed some intake colonization that put detainees at risk for developing endogenous infections and serving as potential sources of MRSA transmission to others. Even with limited sampling, we observed that 17% of jail-onset clinical USA300 MRSA isolates were within 20 SNVs of an intake MRSA strain.

Our study demonstrates that not only is there a high level of MRSA colonization at jail entrance, acquisition of colonization and infection may occur during incarceration. Genomic analyses support this contention and suggest that spread occurred more frequently in certain jail locations. Sharing personal items was associated with acquisition of MRSA; an education campaign aimed at this practice could be a strategy to help curb spread. While our study examined MRSA, such an education campaign could even be extended to COVID-19, a pathogen that significantly impacted jails. Future study with more detailed epidemiologic analysis and environmental sampling within the jail complex might further inform and target interventions. Finally, the utility of an intervention at intake and/or discharge may be another focus of research as the jail remains a critical component of MRSA epidemiology in urban areas.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Glossary

Nonstandard Abbreviation

IRB

institutional review board

MRSA

methicillin-resistant Staphylococcus aureus

SNV

single-nucleotide variant

WGS

whole genome sequencing

Notes

Author contributions. Kyle Popovich: Principle Investigator for NIH R01 that funded project, led project development, coordinated communication and data flow among team members, supervised enrollment, lab processing, and biostatistical analysis, directed interpretation of results, and drafted manuscript. Stephanie Thiede: Performed genomic, phylogenetic, and location overlap analyses, created figures and drafted supplemental materials, assisted with interpretation of results, and assisted with drafting the manuscript. Chad Zawitz: Assisted with enrollment at the jail, serves as the content expert for MRSA in correctional facilities and detainee health, assisted with interpretation of results, and reviewed the manuscript. Alla Aroutcheva: Supervised laboratory processing of specimens, assisted with interpretation of results, and reviewed the manuscript. Darjai Payne: Project coordinator, led enrollment and data collection for the study, performed lab processing of specimens, and reviewed the manuscript. William Janda: Assisted with laboratory work flow and reviewed the manuscript. Michael Schoeny: Assisted with project development, biostatistician for project, assisted with interpretation of results, and reviewed the manuscript. Stefan Green: Assisted with project development, performed whole genome sequencing on isolates, and reviewed the manuscript. Evan Snitkin: Assisted with project development, supervised genomic, phylogenetic, and location overlap analyses, assisted with interpretation of results, and assisted with drafting the manuscript. Robert A. Weinstein: Assisted with project development, provided content expertise on study enrollment, detainee health, MRSA epidemiology through duration of the project, assisted with interpretation of results, and reviewed the manuscript.

Acknowledgments. We thank the Health Research and Solutions Center at Cook County Health for their assistance with data collection, Jon Zelner at the University of Michigan on guidance on statistical analyses, and Ali Pirani at the University of Michigan for bioinformatics support. We thank Bala Hota and Mary Hayden at Rush University Medical Center with their assistance with the early development of this study. We thank Connie Mennella, chair of correctional health at Cermak Health Services, with her assistance in the early planning of the project. We thank the individuals who participated in this study.

Financial support. The project described was supported by Grant Number R01AI114688 (PI: KJP) from the National Institute of Allergy and Infectious Diseases. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Allergy and Infectious Diseases or the National Institutes of Health. SNT was supported by the Molecular Mechanisms of Microbial Pathogenesis training grant (NIH T32 AI007528).

Potential conflicts of interest. R.W. reports participation in clinical studies where participating hospitals or nursing homes received contributed product from Sage Products Inc., Molnlycke, Clorox, Medline, or Bio-K+. Neither R.W. nor his hospital received product, funding, or payments. All other authors have no conflicts of interest to disclose. The author has submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.