High Rate of Adverse Events in Patients Receiving Oral Antimicrobials Under a COpAT Monitoring Protocol

Abstract

We conducted a retrospective review to characterize adverse events in patients receiving oral linezolid, trimethoprim-sulfamethoxazole, voriconazole, and itraconazole monitored under an outpatient parenteral antimicrobial therapy program. Among 118 patients, 61% incurred at least 1 adverse event during treatment, highlighting the need for evidence-based monitoring protocols for complex outpatient antimicrobial therapy.

In a retrospective review of adverse events (AEs) in patients receiving oral linezolid, trimethoprim-sulfamethoxazole, voriconazole, and itraconazole under a complex outpatient antimicrobial therapy (COpAT) program, 61% of patients incurred at least one AE (median onset 12 days), with voriconazole courses associated with the highest AE burden. Occurrence of AE was associated with modifications in drug selection and treatment duration, highlighting the need for evidence-based monitoring protocols earlier (within 1-2 weeks) of COpAT initiation.

Outpatient parenteral antimicrobial therapy (OPAT) is a well-established care model for managing serious infections requiring prolonged antimicrobial courses outside the hospital setting. Clinical practice guidelines published by the Infectious Diseases Society of America provide recommendations on optimal program structure and laboratory monitoring to facilitate safe completion of complex intravenous (IV) antimicrobial courses [1]. A growing body of evidence supports the noninferiority of orally administered antimicrobials for select complex infections [2, 3], catalyzing the evolution of complex outpatient antimicrobial therapy (COpAT) models that leverage existing OPAT infrastructure to include oral agents for the treatment of serious infections [4, 5]. Although OPAT may provide a provisional framework for monitoring patients receiving complex oral antimicrobial regimens at home, establishing specific protocols for COpAT monitoring remains a priority [6].

Adequate laboratory surveillance is essential in this context, as the nonavailability of laboratory results has been associated with increased risk of rehospitalization during OPAT [7]. Although Infectious Diseases Society of America guidelines recommend weekly laboratory monitoring for OPAT patients, emerging literature supports a more individualized approach based on drug-specific and patient-specific risk factors [8, 9]. Comparable guidance for COpAT is lacking, leaving OPAT teams to independently balance the burden of frequent laboratory monitoring against the need for timely identification of drug-related toxicities. The objective of this study was to characterize the type, frequency, and clinical impact of AEs experienced by patients receiving complex oral antimicrobial agents under a formal COpAT monitoring program, with the goal of informing evidence-based monitoring recommendations.

METHODS

This retrospective cohort study was conducted within the OPAT program at the University of Rochester Medicine. The structure and clinical operations of this multidisciplinary OPAT program have been described previously [10]. Nonhemodialysis patients monitored and retained in care while receiving 4 most common COpAT agents in our clinic between 1 January 2019 and 31 December 2022—linezolid (LZD), trimethoprim-sulfamethoxazole (SXT), voriconazole (VRC), and itraconazole (ITC)—were included in the analysis. Patients receiving concomitant IV or multiple oral antimicrobials were excluded to control for effects of polypharmacy and accrued multiagent toxicity.

In accordance with program protocol, weekly laboratory monitoring was obtained during the first month of therapy, including a complete blood count, comprehensive metabolic panel, and therapeutic drug monitoring (TDM) levels for VRC and ITC. Monthly monitoring was recommended thereafter. All patients were reviewed during weekly multidisciplinary OPAT virtual rounds throughout the duration of therapy. VRC troughs were classified as subtherapeutic (<1.0 mg/L), therapeutic (1.0–5.5 mg/L), or supratherapeutic (>5.5 mg/L). ITC exposure was assessed as the sum of itraconazole and hydroxy-itraconazole levels against a 1.0 mg/L lower limit; the 4.0 mg/L upper limit was applied to parent drug alone, as published toxicity thresholds derive from parent concentrations [11].

An adverse event (AE) grading instrument was developed by adapting the 2017 U.S. Department of Health and Human Services Common Terminology Criteria for Adverse Events (v5.0) [12] to reflect the most clinically relevant toxicities expected with LZD, SXT, VRC, and ITC, based on prescribing information [13] and subject matter expertise. This adapted grading scale incorporated predefined ordinal severity criteria (Supplementary Table S1). AEs graded 1 or 2 were classified as mild to moderate; those graded 3 or higher were classified as severe. Events that did not conform to a predefined AE category were classified as “miscellaneous” and were not assigned a formal grade. The primary study endpoint was the occurrence of up to 3 AEs of any severity during the antimicrobial course. Secondary endpoints were patient-level and diagnosis-level risk factors, and COpAT treatment outcomes. Relationships between drug concentrations and AEs were reported descriptively.

A minimum of 2 reviewers (C.S. and A.W.) independently assessed each patient chart for the first 3 chronologically sequential AEs occurring during the treatment course, demographic characteristics, and baseline comorbidities with confirmatory review by ID-trained clinicians (C.B., S.M., A.Y.). Statistical analyses were performed using chi-square and Fisher exact tests where appropriate, with analysis of variance used for continuous variables; analyses were conducted in GraphPad and RStudio. Claude Opus 5 (Anthropic PBC) was used to generate descriptive statistics for TDM portion of the analysis. The study was deemed a quality improvement initiative and granted exemption from formal review by the Institutional Review Board.

RESULTS

A total of 118 unique COpAT patients met eligibility criteria following exclusion of 6 patients due to loss to follow-up. Patient demographic characteristics and baseline comorbidities are summarized in Table 1. Comorbidity burden was not different between the AE and non-AE groups. The median age of patients who experienced an AE was modestly higher than those who did not (62.5 vs 56.5 years; P = .542). Pulmonary diagnoses were more common in the AE (P = .048) group, whereas skin and soft-tissue and central nervous system infections less common (P = .031, 0.028, respectively).

Table 1.

Patient Demographics

Characteristic	All Patients (n = 118)	With AE (n = 72)	Without AE (n = 46)	P Value
Age, median (y)	60	62.5	56.5	.542
Male sex, n (%)	75 (63.6)	43 (59.7)	32 (69.6)	.279
Comorbid conditions
Chronic kidney disease	9 (7.6)	8 (11.1)	1 (2.2)	.075
Essential hypertension	54 (45.8)	36 (50.0)	18 (39.1)	.248
Diabetes mellitus	34 (28.8)	20 (27.8)	14 (30.4)	.756
Liver disease	6 (5.1)	5 (6.9)	1 (2.2)	.250
Active malignancy	12 (10.2)	7 (9.7)	5 (10.9)	.841
Concurrent SSRI/SNRI use	18 (15.3)	12 (16.7)	6 (13.0)	.593
Concurrent ACEi or ARB use	28 (23.7)	16 (22.2)	12 (26.1)	.630
Anticoagulation with warfarin	5 (4.2)	3 (4.2)	2 (4.3)	.962
Concurrent other antimicrobial use	9 (7.6)	6 (8.3)	3 (6.5)	.718
Infectious disease diagnosis—n (%)
Pulmonary	41 (34.7)	30 (41.7)	11 (23.9)	.048a
Bacteremia/endocarditis	12 (10.2)	6 (8.3)	6 (13.0)	.409
Skin and soft tissue	8 (6.8)	2 (2.8)	6 (13.0)	.031a
Bone and joint	39 (33.1)	28 (38.9)	11 (23.9)	.092
Intra-abdominal	4 (3.4)	2 (2.8)	2 (4.3)	.646
Central nervous system	3 (2.5)	0 (.0)	3 (6.5)	.028a
Urinary tract	4 (3.4)	1 (1.4)	3 (6.5)	.133
Other	7 (5.9)	3 (4.2)	4 (8.7)	.310
Oral antimicrobial agent—n (%)
TMP-SMX	34 (28.8)	19 (26.4)	15 (32.6)	.467
Itraconazole	13 (11.0)	7 (9.7)	6 (13.0)	.574
Linezolid	42 (35.6)	24 (33.3)	18 (39.1)	.521
Voriconazole	29 (24.6)	22 (30.6)	7 (15.2)	.059

^a P < .05 by Chi-square or Fisher's exact test.

Abbreviations: AE, adverse event; ACEi, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; SNRI, serotonin-norepinephrine reuptake inhibitor; SSRI, selective serotonin reuptake inhibitor; TMP-SMX, trimethoprim-sulfamethoxazole.

Overall, 72 of 118 patients (61%) incurred at least 1 of the 136 total AEs documented during the study period. The majority of AEs (n = 98, 72.1%) were mild-to-moderate in severity (grade ≤2). Severe AEs (grade ≥3) were identified in 13 instances, affecting 11% of patients, with rates not differing between agents. The median time from treatment initiation to first AE was 12 days, with no statistically significant difference between agents (Table 2). More than 50% of patients receiving each of the 4 agents incurred at least 1 AE during their course of therapy. AE occurrence of any severity was significantly associated with modification of the drug regimen (P = .001) and shortened or altered treatment duration (P = .0192) (Supplementary Figure S1)

Table 2.

Adverse Events by Antimicrobial Agent

	All Agents	SXT (n = 34)	ITC (n = 13)	LZD (n = 42)	VRC (n = 29)	P Value
Unique patients (n)
	118	34	13	42	29
Treatment duration median days (IQR)b	30 (16, 94)	34.5 (14.25, 79.25)	102 (75.75, 225.25)	21 (14, 28.75)	97 (34.5, 168.5)	<.001a
Unique patients with AE—n (%)	72 (61%)	19 (56%)	7 (54%)	24 (57%)	22 (76%)	.022a
Patients with ≥ 2 adverse events—n (%)	45 (38%)	11 (32.4%)	5 (39%)	15 (36%)	14 (48%)	.014a
Median days to AE (IQR)	12 (5, 25)	9 (5, 13.5)	11 (7, 28)	9.5 (5, 21)	17 (6, 48)	ns
Unique adverse events (n)
	136	36	16	44	40
Mean adverse events per patient (range 0–3)	1.85	1.89	2.29	1.83	1.28	.005a
Mild-mod adverse events (grades 1–2)—n (%)	98 (73%)	30 (83%)	11 (69%)	32 (73%)	25 (63%)	.012a
Severe AE (grades 3–4)—n (%)	13 (10%)	3 (8.5%)	2 (12%)	5 (11%)	3 (7.5%)	ns
Ungraded AE—n (%)	22 (18%)	3 (8.5%)	3 (19%)	7 (16%)	12 (30%)	ns
Patient outcome
Completed therapy with another oral agent—n (%)	8 (7%)	4 (11.7%)	0	2 (5%)	2 (7%)	ns
Completed therapy with same drug—n (%)	70 (59%)	22 (65%)	6 (46%)	29 (69%)	13 (45%)	<.001a
Duration of therapy modified—n (%)	36 (31%)	8 (23.3%)	6 (46%)	11 (26%)	11 (38%)	ns
Unknown—n (%)	4 (3%)	0 (0%)	1 (8%)	0 (0%)	3 (10%)	ns

^a P < .05.

^bTreatment duration reflects full course length.

AE, adverse event; IQR, interquartile range; ITC, itraconazole; LZD, linezolid; SXT, trimethoprim-sulfamethoxazole; VRC, voriconazole; ns, not significant.

Bolded numbers are totals corresponding to the bolded row name.

Gastrointestinal (GI) AEs were the most common in the total patient sample, accounting for 31.2% of AEs (n = 43), with a predominance of upper GI symptoms (Supplementary Table S2). For LZD, GI AEs accounted for 50% of all events (n = 22), including 2 severe events. Among the 36 AEs documented in SXT-treated patients, electrolyte/renal abnormalities (n = 17, 47.2%) and GI events (n = 11, 30.6%) were most frequent; 3 events were severe, including 1 grade 4 hyperkalemia. For patients receiving ITC, the 16 total AEs were primarily GI (37.5%) (Supplementary Figure S2). ITC courses were notably longer in duration, were associated with a higher mean number of AEs per patient, a higher proportion of mild-to-moderate events, and the greatest likelihood of requiring therapy modification (Table 2).

For VRC-treated patients, 40 AEs were documented; neurological events accounted for 25% (n = 10), of which 60% were vision toxicity. All vision toxicity events in this cohort were mild to moderate in severity, consistent with previously published data [11]. Two patients experienced severe elevations in alkaline phosphatase while receiving VRC; 1 additional severe AE was classified as financial burden. VRC courses were associated with the highest prevalence (76%) and the greatest frequency of ≥2 AEs per treatment course (48.3%) (Table 2).

Among VRC-treated patients, 26 of 29 (90%) had ≥ 1 level obtained, with median time to initial measurement of 7 days (IQR 5–12). Among 143 VRC troughs, 1 was uninterpretable because of multiply-timed levels per single day, 93 (65%) were therapeutic, 37 (26%) subtherapeutic, and 12 (8%) supratherapeutic, with overall median trough 1.6 mg/L (IQR 0.9–2.5). Sixteen of 26 monitored patients (62%) had at least 1 subtherapeutic level. Among 8 of 26 (31%) VRC patients with ≥1 supratherapeutic level, all 8 (100%) experienced an AE. The supratherapeutic level was the first one collected in 6/8 (75%) patients, occurring at median 8 days after initiation (IQR 5–18). The supratherapeutic level preceded the first AE in 3 patients, detected concurrently in 3, and occurred after the event in 2. All 3 VRC patients with a hepatobiliary AE had a supratherapeutic level. VRC courses without supratherapeutic levels were affected by AE in 11 of 18 (61%) patients.

A total of 59 ITC levels was distributed among all 13 (100%) ITC patients, with median time to first level of 12 days (IQR 6–16) after initiation, and 2 measurements uninterpretable because of missing metabolite information. Median ITC plus hydroxy-itraconazole trough was 2.2 mg/L (IQR 1.5–4.6), with 6 of 57 paired levels (11%) subtherapeutic without notable hypothesis generating patterns. (Supplementary Table S3).

DISCUSSION

This study demonstrates a high burden of AEs among patients receiving LZD, SXT, VRC, and ITC under a formal COpAT monitoring program, with more than 60% of patients incurring at least 1 AE and nearly two-thirds of affected patients experiencing 2 or more events per course. These findings are clinically significant given the ongoing shift toward oral antimicrobial agents for infections historically managed with IV therapy and underscore the necessity of structured monitoring.

The finding that a substantial proportion of patients experienced their first AE within the first 2 weeks of therapy highlights this as a particularly vulnerable time period and supports recommendations for close clinical follow-up early in the treatment course. Previous research has described higher levels of variability in follow-up appointment schedules for COpAT patients compared to patients on traditional OPAT [4] Focusing efforts on ensuring follow-up and monitoring early in the treatment course for COpAT patients could help to maximize the impact of these appointments.

Laboratory abnormalities were common and clinically relevant, but not always consistent with expected side effect profiles of the medications [13]. For example, although electrolyte and renal AEs were prevalent among SXT recipients, hematological AEs (including thrombocytopenia and leukopenia) was observed in 14% of LZD-treated patients [14], compared to a preponderance (50%) for GI toxicity (Supplementary Figure S2). Conclusions about these findings are limited, however, by the inability to capture dosing, longitudinal trends in renal function, and effect on linezolid drug levels. Alternatively, the high prevalence of neurological AEs with VRC, with predominance of vision toxicity, is consistent with the known neurotoxic profile of this azole [11] and supports the need for structured ophthalmologic and neurological symptom assessment during VRC therapy.

Supratherapeutic VRC exposure was uncommon; however, every patient with a supratherapeutic level experienced an AE, including all 3 hepatobiliary events observed in the VRC group. Supratherapeutic levels occurred early, mirroring the timing of first AEs; however, the temporal sequence is less clear and merits further study. Although these findings are hypothesis generating and consistent with the concentration-dependent hepatotoxicity described for VRC [11], study sample size supports descriptive reporting rather than comparative analytics. Subtherapeutic exposure was more common, affecting 62% of monitored VRC courses. Both patterns support obtaining drug levels for TDM within the first 2 weeks of therapy or sooner.

Despite the absence of statistically significant differences in severe AE rates between agents, the clinical impact of AEs was substantial: their occurrence was significantly associated with regimen modification and changes in treatment duration. These data corroborate the recommendations for multidisciplinary team oversight of patients receiving oral antimicrobials for prolonged durations [4]. Our analysis adds to the evidence base by providing novel and concrete data points to support a recommendation for weekly complete blood count and comprehensive metabolic panel and TDM monitoring for a minimum of the first 4 weeks of COpAT—a recommendation that, to date, has been less restrictive in frequency (ie, every 2–4 weeks) and largely based on expert opinion [15]. In addition, clinicians initiating COpAT should maintain heightened awareness of GI AEs across all agents and be particularly vigilant for the elevated AE burden associated with azole antimicrobials. Patients on extended courses should continue to undergo regular laboratory monitoring at intervals to be defined in larger prospective studies, which are currently lacking. Further development of evidence-based, agent-specific monitoring protocols is a research priority and will be essential to optimize safety and outcomes in this growing patient population.

Limitations of this study include its single-center retrospective design and modest sample size which may limit generalizability. The study design offers limited ability to systematically account for drug–drug interactions, polypharmacy or concurrent illness effects, or inter-patient variability in monitoring frequency. Therapeutic drug concentrations were only analyzed for ITC and VRC, and dosage information was not comprehensively captured, which is particularly relevant given the dose-dependent toxicity of agents such as LZD [14]. Future prospective studies should address these gaps.

Supplementary Data

Supplementary materials are available at Open Forum 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.