Discussion

In our analysis of S. aureus wound isolates in PWUF in Philadelphia, we found significantly greater nonsusceptibility to oxacillin, clindamycin, and tetracycline compared to isolates from our health system’s general outpatient population. Of the three oral agents routinely prescribed for MRSA infections, doxycycline was the most commonly prescribed discharge antibiotic in our study. Doxycycline demonstrates high in vitro activity against S. aureus among general outpatients in our health system’s antibiogram, making it an often-preferred antibiotic for empiric and targeted S. aureus therapy, in addition to its standardized dosing and high tolerability compared to clindamycin and trimethoprim-sulfamethoxazole. However, one-quarter of PWUF discharged with doxycycline were presumably discharged on inactive therapy. Lloyd-Smith and colleagues analyzed susceptibility patterns among 37 S. aureus isolates from the wounds of patients presenting to a Vancouver supervised injection facility in 2008. Roughly half of S. aureus isolates were MRSA and 92% of all isolates demonstrated susceptibility to tetracycline. ^{Reference Lloyd-Smith, Hull and Tyndall1} Differences in resistance profiles between that study and ours may, in part, be attributed to the changing epidemiology of S. aureus over time. Tetracycline resistance has significantly increased over the past decade among outpatients with MRSA infections. ^{Reference Carrel, Smith and Shi20} The USA300 clone has become the predominant clone implicated in SSTIs within the United States. ^{Reference Chua, Moore and Perri21,Reference Talan, Krishnadasan and Gorwitz22} USA300 has acquired methicillin, clindamycin, and tetracycline resistance and has been theorized to move westward across North America within the past 25 years, possibly accounting for the difference in susceptibilities from Lloyd-Smith and colleagues. ^{Reference Tenover and Goering23–Reference Challagundla, Luo and Tickler25} Non-USA300 clones, including sequence types 5 and 9, may also contribute to the high-degree of drug resistance, including tetracyclines. ^{Reference Chen, Sun and Hong26,Reference Hofstetter, Jacko and Shumaker27} Additionally, S. aureus transmission has been demonstrated to be dependent on drug-use networks. ^{Reference Quagliarello, Cespedes and Miller28,Reference Leung, Padgett, Robinson and Brown29} Infections among social networks of PWUF within Philadelphia may be enriched with tetracycline-resistant S. aureus compared to the general population or PWUF in other geographic regions.

Roughly half of patients were discharged prior to the availability of S. aureus susceptibilities. Hazen and colleagues estimated that 15% of patients with SSTIs secondary to injection drug use directed their own discharge from the hospital, placing these patients at increased risk for 90-day readmission. ^{Reference Hazen, Pizzicato, Hom, Johnson and Viner30} Managing withdrawal by administering medications for opioid use disorder to patients with serious infections to prevent premature discharge have been met with mixed success. ^{Reference Nolan, Marks, Liang and Durkin10,Reference Suzuki, Robinson and Mosquera31} Within our study, patients discharged prior to susceptibility availability were more likely to be discharged BMA. These patients were also less frequently administered methadone or buprenorphine compared to patients discharged after susceptibilities were available. Methods of encouraging patients to stay hospitalized until more microbiologic data become available are crucial to ensure active oral therapy upon discharge. For those that do elect to leave the hospital, advances in rapid molecular diagnostics for S. aureus susceptibilities as well as multidisciplinary follow-up can limit discharges on inactive therapy. ^{Reference Lewis, Liang and Schwarz32} This highlights the need for joint infectious diseases and addiction medicine management of patients who use drugs presenting with acute SSTI. Opioid withdrawal is often difficult to control as the potency of illicit opioids has increased and opioids are frequently adulterated with other substances, such as xylazine, that may complicate withdrawal. ^{Reference O’Neil and Kovach33,Reference Wildenthal, Schwartz and Nolan34} In the face of an evolving antimicrobial resistance and drug-use landscape, further research across disciplines is needed to better provide antimicrobial therapy for patients at risk for opioid withdrawal.

Transitions of care for PWUF with a S. aureus infection are difficult. With the high incidence of resistance to commonly used oral S. aureus agents, prescription of active empiric therapy on discharge is challenging. A study of patients who inject drugs that were prescribed oral antibiotic therapy after discharging BMA estimated that nearly one-quarter of patients were lost to follow-up after discharge. ^{Reference Lewis, Liang and Schwarz32} Thus, these patients may be difficult to contact after discharge to inform of the need to switch antibiotic therapy to an active in vitro agent. Despite recent evidence, PWUF are often not considered candidates for outpatient parenteral antimicrobial therapy. ^{Reference D’Couto, Robbins, Ard, Wakeman, Alves and Nelson35} Prescription of oral linezolid on discharge carries a risk of serotonin syndrome when combined with fentanyl use. While the risk of serotonin syndrome with medically prescribed doses of fentanyl is low, the risk with substantially higher illicit doses requires further investigation. ^{Reference Gatti, Raschi and De Ponti36} Highly active oral anti-MRSA agents, such as delafloxacin, omadacycline, and tedizolid, are cost-prohibitive for many PWUF without commercial insurance. For those with medical coverage, these agents often require prior authorization. Patients may be unwilling to stay hospitalized during benefits investigation, particularly when experiencing opioid withdrawal symptoms. Half of patients discharged BMA were discharged without an antibiotic. Patients may elect to leave the hospital quickly, leaving little time to prescribe antibiotics to an outpatient pharmacy. The inpatient use of long-acting lipoglycopeptides, such as dalbavancin and oritavancin, are attractive options to provide adequate therapy for SSTIs administered as a single dose. While the high cost of lipoglycopeptides has come under scrutiny by health systems, prevention of treatment failure and subsequent readmission may offset the cost. ^{Reference Lueking, Wei, Mang, Ortwine and Meisner37}

Our study has several limitations. Primarily, isolation of S. aureus could represent superficial colonization as opposed to active infection. We restricted inclusion to patients who were determined to be acutely infected and required antibiotic therapy per the treating provider. However, S. aureus may not have been an infecting pathogen among some PWUF given the polymicrobial nature of many cultures and nonsterile collection. Second, in vitro susceptibility testing for tetracycline at our institution is used as a surrogate for doxycycline. Isolates may have maintained susceptibility to doxycycline, inflating estimates of doxycycline nonsusceptibility and the proportion of patients discharged on inactive doxycycline therapy. Third, we were unable to obtain antibiotic administration data outside of our health system. This population is highly transient with frequent healthcare presentations; therefore, measures of previous antibiotic exposure are underestimated. ^{Reference Soria, Johnson and Collins38} Associations between S. aureus susceptibility and prior antibiotic exposure were likely underpowered. Fourth, pathogens other than S. aureus present in wound cultures were not collected from the EHR. We were unable to estimate the burden of polymicrobial infections, and antibiotics prescribed may have had indications other than S. aureus infection. Lastly, mode of fentanyl use was not collected. As such, S. aureus risk and susceptibility as a factor of route of drug use was not assessed. However, it is assumed that most fentanyl use in Philadelphia is via injection.

To our knowledge, this is the largest analysis of S. aureus susceptibility among PWUF within a large, multisite hospital network. Our results suggest that clindamycin and doxycycline should not be prescribed empirically in this population for acute S. aureus SSTI, particularly within Philadelphia, given high rates of in vitro nonsusceptibility. PWUF are medically complex. Highly resistant, often polymicrobial infections, and complex substance withdrawal necessitate collaboration with experts in infectious diseases, emergency medicine, psychiatry, and addiction medicine to provide optimal care. Institutional protocols that provide joint guidance on controlling opioid withdrawal and optimal antibiotic therapy based upon local susceptibility patterns warrant development.