Discussion

In this study evaluating the safety of prolonged courses of oral antibiotics, approximately one in four patients experienced an adverse event. Gastrointestinal adverse events were most common, but other adverse events were also observed. Combination regimens had more ADEs than monotherapy regimens although this was not statistically significant. Among combination regimens, linezolid, TMP/SMX, and fluoroquinolones had the highest ADE rate while combinations with tetracyclines and beta-lactams had lower ADE rates. Among monotherapy regimens, linezolid and TMP/SMX had the highest ADE rates, while fluoroquinolones and beta-lactams had lower ADE rates. In our analysis looking at both monotherapy and combination regimens, linezolid and TMP-SMX had the highest rates of adverse events while amoxicillin, amoxicillin-clavulanate, and tetracyclines had the lowest.

Overall, there are very few studies that have evaluated the safety profile of oral antibiotics used as COpAT. Sawey et al. published a study comparing ADE of patients receiving oral linezolid, TMP/SMX, voriconazole, and itraconazole under a COpAT monitoring protocol. Amongst patients who received linezolid, they also reported GI symptoms as the most common ADE. Regarding TMP/SMX, their study reported electrolyte and renal injury as the most common ADE followed by GI symptoms. ^{Reference Sawey, Burgoyne and Winbush13}

In our study, 46 patients (26%) experienced 73 ADEs. This is overall higher compared to previous studies of ADE rates in OPAT patients. Keller et al. reported 18% of patients experienced an ADE in a prospective cohort study describing ADE rates in OPAT patients. ^{Reference Keller, Williams and Gavgani14} However, OPAT in general introduces the risk for vascular access complications which is inherently avoided with oral therapy in COpAT patients. In an observational cohort study, Underwood et al. concluded catheter-related adverse events occur more frequently than drug-related adverse events. ^{Reference Underwood, Marks, Collins, Logan and Pollara15} Therefore, while COpAT does not eliminate the risk for ADE, it still my offer an advantage of removing vascular access complication risk.

Our findings highlight several important considerations for clinicians when prescribing prolonged courses of oral antibiotics. Our rate of ADE reinforces the need for continued careful monitoring for COpAT patients. Although GI symptoms were the most frequent ADE across antibiotic classes, the type of ADEs varied substantially by antibiotic class suggesting that regimen choice should be individualized based on patients’ preexisting risk factors, medical problems, and even social factors that may affect laboratory monitoring and follow-up. For example, with linezolid, the occurrence of bone marrow suppression within a median of about two weeks suggests the need for early laboratory monitoring and potentially limiting to shorter treatment courses when alternatives exist. In regards to TMP/SMX, the rates of nephrotoxicity and hyperkalemia, particularly in patients with diabetes, suggests the need for closer laboratory monitoring and caution when used in patients with preexisting risk factors for renal toxicity. Most patients who developed an ADE with TMP/SMX were discharged on higher doses (at least 4 DS tablets daily). Therapeutic drug monitoring may also have a role in preventing ADEs for antibiotics with established exposure-toxicity relationships, such as linezolid. ^{Reference Cheng, Wu and Kuo16}

In contrast, oral beta-lactams and tetracyclines had lower ADE rates suggesting they are better tolerated for prolonged courses. However, this may be challenging in certain infections such as bone and joint infections as oral beta-lactams may be considered less effective given their poor bone penetration and bioavailability compared to other classes. ^{Reference Landersdorfer, Bulitta, Kinzig, Holzgrabe and Sörgel17,Reference Preiss, Kriechling and Montrasio18} Similarly, doxycycline typically achieves low serum concentrations compared to other classes such as linezolid or TMP/SMX, so is typically not used for patients with endovascular infections or bacteremia. ^{Reference Dagher, Fowler, Wright and Staub19} Again, the need for an individualized approach for each patient is highlighted.

Fluoroquinolones were the third most prescribed antibiotic in our study. Gastrointestinal adverse events were the most common ADE in our study and serious complications were rare. Gastrointestinal symptoms are known to be a common side effect associated with fluoroquinolones; however, concerns about tendinopathy and cardiovascular risk remain prominent in the literature. ^{Reference Baggio and Ananda-Rajah20} There was only one report of joint pain in our study and no reports of cardiovascular complications from fluoroquinolone use, which may suggest more serious complications are less frequent than anticipated.

There were several limitations in our study, primarily its retrospective design and inability to attribute causality. In an effort to report clinically meaningful adverse events, we relied on clinician documentation of ADEs, which may be imprecise. If laboratory ADEs occurred but were not documented by the clinician, then these were not counted towards the incidence of ADEs. Similarly, we relied on documentation to determine when an acute kidney injury occurred as opposed to an expected rise in serum creatinine with TMP/SMX. Many patients were discharged on combinations of antibiotics making it difficult at times to ascertain the true ADE risk for each agent. We attempted to address this by providing analyses of ADEs for patients on monotherapy regimens. The number of ADEs observed was rather low and consequently our effect estimates have large CIs. Furthermore, although all patients had follow-up appointments scheduled in our infectious disease clinic, some patients did not attend their scheduled visit and were lost to follow-up. We still included these patients as we felt this is common in real-world clinical practice and frequently encountered. Regardless, this may lead to an underestimate of the incidence of ADEs. Additionally, we assumed a constant hazard throughout each day of antibiotic exposure, which may not be correct. This single center experience may also not apply to other centers with different patient populations or prescribing practices. Based on our patient cohort demographics, our results may not fully represent all patients discharged on oral therapy, especially in regard to an older patient population or patients with more comorbidities. Lastly, we only assessed one side of the risk-benefit equation and there may be many circumstances where more hazardous therapies are more likely to be effective.

In summary, most patients were able to tolerate prolonged courses of oral antibiotics with GI side effects as the most common ADE. Linezolid and TMP/SMX were found to have higher ADE rates, with GI symptoms, bone marrow suppression, nephrotoxicity, and hyperkalemia more commonly observed. Our findings suggest a few implications: first, clinicians should be judicious in selecting linezolid and TMP/SMX for prolonged courses and ensure adequate lab monitoring is in place. TMP/SMX may not be a preferred choice in patients with diabetes if other options are available. Second, the relatively favorable safety profile of oral beta-lactams and tetracyclines should be considered alongside their efficacy relative to other antibiotic classes depending on the infectious syndrome. Third, structured protocols for laboratory monitoring and therapeutic drug monitoring especially for higher risk patients such as those with diabetes or baseline renal impairment may be worth incorporating to mitigate risks associated with higher-toxicity regimens. Lastly, although the use of COpAT is rising to help mitigate potential harms associated with IV antibiotic use, the risk of ADEs associated with COpAT is still present; therefore, Antimicrobial Stewardship remains essential to guide antibiotic selection and duration. Our results emphasize the importance of weighing patient-specific risk factors and drug tolerability when planning prolonged oral antibiotic courses. However, more evidence-based studies that compare the efficacy and the safety profile of various oral antibiotics are needed to assist in the development of protocols for patients receiving COpAT.