Risk of Treatment Failure in Outpatient Parenteral Antimicrobial Therapy From an Academic Medical Center in Kentucky

Abstract

Background

Outpatient parenteral antimicrobial therapy (OPAT) has many benefits, yet factors contributing to treatment failure are not well understood. This study aimed to identify risk factors for OPAT failure by comparing successful patients to those failing therapy.

Methods

Adult patients referred and discharged to receive OPAT in the home from August 2019 and May 2021 were included. Patients were excluded if they received OPAT at a nonhome location or had creatinine clearance less than 30 mL/min. Patient demographics, as well as admission-related information, were collected. The primary composite outcome was OPAT failure defined as infection nonresolution or new infection, medication modification, or hospital readmission. Bivariate followed by multivariate analysis was used to identify factors that increased a patient's risk of failing therapy. Additional analysis examined predictors of loss to follow-up.

Results

Of the 823 patients included, 588 (71.4%) successfully completed therapy with 195 (23.7%) experiencing failure. Forty (4.9%) patients were lost to follow-up. Of the failures, 109 (55.9%) experienced readmission, 63 (32.3%) experienced new or persistent infection, and 23 (11.8%) required therapy modification. Factors associated with failure included infection with vancomycin-resistant Enterococcus (odds ratio [OR]: 4.42, 95% confidence interval [CI] [1.01–19.29]), oxazolidinone treatment (OR: 4.92, 95% CI [1.90–12.82]), and increasing numbers of infusions (OR: 1.16, 95% CI [1.04–1.30]). Patients living further from the clinic were more likely to be lost to follow-up.

Conclusions

Simplifying home-based antimicrobial regimens may improve OPAT outcomes. Early identification of risk factors for OPAT treatment failure and loss to follow-up can optimize patient management and reduce adverse outcomes.

Outpatient parenteral antimicrobial therapy (OPAT) is defined as the administration of parenteral antimicrobial therapy in at least 2 doses on different days without intervening hospitalizations [1]. OPAT facilitates the transition of patients from acute care to ambulatory treatment settings, allowing institutions to optimize inpatient capacity while reducing healthcare costs and potentially improving patient satisfaction [2–6]. Outpatient parenteral antimicrobial therapy in the United States allows patients to receive necessary antimicrobial therapy at home, supported by a multidisciplinary team responsible for patient selection, education, and close follow-up.

In 2019, the University of Kentucky formalized their OPAT program, dedicating full-time employees (FTEs) to care coordination and patient selection. The program structure included a clinical nurse navigator responsible for patient evaluation and appropriateness assessment, a clinical pharmacist who reviewed antimicrobial regimens and recommended modifications, and 2 shared FTE clinic nurses who coordinated weekly laboratory information and line-related care. Patients were managed by the referring providers in the clinic setting. The clinical nurse navigator and clinical pharmacist collaborated to optimize antimicrobial regimens, with a particular emphasis on reducing the number of daily infusions. This approach aimed to enhance adherence by simplifying treatment, mirroring strategies used for oral antimicrobial and medication regimen prescriptions.

While OPAT has demonstrated efficacy in home-based treatment, data on the factors contributing to treatment failure remain limited. This study aimed to evaluate risk factors for OPAT failure in patients managed through the newly formalized OPAT program, with a focus on identifying modifiable patient- and treatment-specific factors to improve the safety and effectiveness of OPAT.

METHODS

This retrospective, single-center, case-control study compared patients who successfully completed OPAT (control) with those who experienced OPAT failure defined as infection nonresolution or new infection, medication modification secondary to toxicity or allergy, or hospital readmission within 90 days of OPAT therapy completion (case). A secondary analysis compared successfully treated patients (control) to those who were lost to follow-up (case). Patients were eligible if they were admitted to the University of Kentucky HealthCare and discharged on OPAT between August 2019 and May 2021. Patients were included if they were 18 years of age or older, enrolled in the UKHC OPAT program, and discharged home with IV antimicrobial therapy. Patients that were discharged on oral antimicrobials in addition to IV antimicrobial therapy were available for inclusion. Patients were excluded if they were less than 18 years of age, not eligible for OPAT at UKHC, or had an estimated creatinine clearance of 30 mL/min or less (per Cockcroft–Gault equation) or received their full antimicrobial course at a skilled nursing facility, inpatient swing bed, rehabilitation facility, dialysis center, or any other nonhome setting where they or their caregiver did not administer medications. Patients with intermittent stays at a rehabilitation facility before discharge home were included in the analysis.

Data Collection

Patient demographics, vascular access details, home health provider information, prescribed antimicrobials, and clinical outcomes were collected as part of routine pharmacists' workflow within the UKHC OPAT clinic. Data were stored in REDCap®, an electronic data capture system hosted at the University of Kentucky [7, 8]. This project was supported by the NIH National Center for Advancing Translational Sciences (grant UL1TR001998). For outcome documentation, the OPAT pharmacist classified treatment success, failure, or lost to follow-up based on provider records and assessed outcomes at the prescribed end of therapy date. Statistical analysis was performed via IBM SPSS Statistics Software Version 27 and SAS Version 9.4.

Outcomes Assessed

The primary outcome was OPAT failure, defined as either an infection nonresolution or new infection, medication modification secondary to toxicity or allergy, or hospital readmission within 90 days. Details with respect to antimicrobial duration were further classified by the persistence of initial or new infection, persistence of primary pathogen, or superinfection. Patients requiring antimicrobial duration extension solely for repeat imaging scheduling were not considered a failure unless the infection was ultimately considered unresolved and warranted further treatment. Change in antimicrobial treatment was described further via presence of toxicity, adverse effects, or allergic reaction to any of the prescribed outpatient antimicrobials. The OPAT-related hospital readmissions were defined as hospital readmissions within 90 days of OPAT index admission discharge date that also had a primary or secondary ICD-10 code that was directly related to the original infection that led the patient to OPAT. Patient-specific factors were compared to identify differences in the overall outcome of success vs failure. Secondary outcomes analysis assessed factors associated with lost to follow-up, comparing those lost to follow-up with successful patients.

Statistical Analysis

The primary analysis evaluated patient-specific factors influencing OPAT failure compared to successful patients. The secondary analysis assessed factors associated with loss to follow-up as compared to successful patients. Categorical data was expressed as counts (%) and analyzed using Pearson's χ² or Fisher's exact test, as appropriate. Continuous variables were expressed as means and standard deviations for normally distributed data or medians with interquartile ranges for nonparametric data. Independent samples t-test or Mann–Whitney U-test was used to analyze continuous variables as appropriate. Categorical variables were analyzed using Pearson's χ² or Fisher's exact test as appropriate. Significance was defined utilizing a 2-sided test with an alpha-level of 0.05. For OPAT failure, a multivariable logistic regression was performed utilizing a backwards elimination selection criterion and an alpha-level of 0.05. Due to a small sample size, no multivariable logistic regression was performed for loss to follow-up.

RESULTS

A total of 1376 patients were identified in the initial data extraction. After applying inclusion and exclusion criteria, 553 patients were excluded (Figure 1), leaving 823 patients for analysis. Of the 823 included patients who met study criteria, 783 patients completed OPAT therapy, and 40 patients were lost to follow-up. Of those who completed therapy, 588 (75.1%) were deemed as a success and 195 (24.9%) met a composite criteria for failure. Of those deemed an OPAT failure, the majority 109 (55.9%) of the patients experienced an OPAT-related hospital readmission within 90 days. In total, 63 (32.3%) patients required antimicrobial extension secondary to a new pathogen, persistence of a primary pathogen, or provider deemed nonresolution of the infection at time of the original antimicrobial end date. Twenty-three (11.8%) of patients required therapy modification secondary to toxicity, adverse drug event, or allergic reaction.

Figure 1.

Patient eligibility.

Bivariate analysis identified several factors associated with OPAT failure. Patients who successfully completed therapy were less likely to have been infected with vancomycin-resistant Enterococcus (VRE) (3 [0.5%] vs 5 [2.5%], P = .026), required fewer daily infusions (2 [1,2,3] vs 2 [1,2,3,4], P = .004), were less frequently treated with a beta lactam/beta lactamase inhibitor (49 [8.4%] vs 26 [13.3%], P = .040) and less likely to receive oxazolidinone therapy (7 [1.2%] vs 12 [6.2%], P < .001). Complete bivariate analysis results can be found in Table 1.

Table 1.

Patient Demographics

Characteristic	Success (n = 588)	Failure (n = 195)	P Value
Male, n (%)	352 (59.9)	115 (59.0)	.826
Age (y), mean (SD)	55.13 (14.7)	53.28 (15.2)	.133
Weight (kg), median [IQR]	88.3 [72.9–107.3]	83.3 [71.8–106]	.317
Baseline serum creatinine, median [IQR]	0.83 [0.65–1.08]	0.78 [0.61–1.03]	.088
Infection type, n (%)			.455
Bone and joint	217 (36.9)	71 (36.4)
Abscess	161 (27.4)	56 (28.7)
Bloodstream	78 (13.3)	27 (13.8)
CNS related	42 (7.1)	16 (8.2)
Infective Endocarditis	32 (5.4)	12 (6.1)
Other	49 (8.3)	13 (6.7)
Polymicrobial infection	57 (9.7)	15 (7.7)	.289
History of IVDU	25 (4.3)	4 (2.1)	.193
Charlson Comorbidity Index, median [IQR]	3 [1–5]	3 [1–5]	.461
Hemiplegia	16 (27.2)	9 (4.6)	.192
VRE infection	3 (0.5)	5 (2.5)	.026
CRE infection	4 (0.68)	1 (0.51)	1.00
ESBL infection	18 (3.1)	4 (2.1)	.619
Pseudomonas spp.	23 (3.9)	8 (4.1)	.906
Staphylococcus aureus	179 (30.4)	51 (26.2)	.255
ICU admission required	140 (23.8)	49 (25.1)	.709
Lumen count, median [IQR]	1 [1]	1 [1]	N/A
Length of stay, d, median [IQR]	8 [5–13]	9 [6–14]	.173
Infusion per day, n, median [IQR]	2 [1,2,3]	2 [1,2,3,4]	.004
Distance from clinic, miles, median [IQR]	47.2 [18.7–92.2]	47.0 [16.0–96.0]	.745
Fluoroquinolone	24 (4.1%)	14 (7.2%)	.081
Beta-lactam/beta-lactamase inhibitor	49 (8.4%)	26 (13.3%)	.040
Aminoglycoside	4 (0.68%)	3 (1.5%)	.374
Carbapenem	113 (19.2%)	49 (25.1%)	.186
Cephalosporin	263 (40.1%)	87 (44.6%)	.523
Monobactam	2 (0.34%)	3 (1.5%)	.102
Penicillin	43 (7.3%)	17 (8.7%)	.523
Clindamycin	1 (0.17%)	0 (0.0%)	N/A
Folate derivative	0 (0.0%)	5 (2.5%)	N/A
Lipoglycopeptide	2 (0.34%)	1 (0.51%)	.577
Macrolide	0 (0.0%)	1 (0.51%)	N/A
Oxazolidinone	7 (1.2%)	12 (6.2%)	<.001
Rifamycin	9 (1.5%)	2 (1.0%)	1.00
Tetracycline	1 (0.17%)	2 (1.0%)	.155
Antifungal	27 (4.6%)	12 (6.2%)	.385

Data displayed as n (%) unless specified.

Abbreviations: CRE, carbapenem-resistant Enterobacteriales; ESBL, extended-spectrum beta-lactamase; VRE, vancomycin-resistant Enterococcus.

Multivariate logistic regression analysis included VRE isolation, number of daily infusions, beta-lactam/beta-lactamase inhibitor use, and oxazolidinone therapy, based on the bivariate analysis. Additionally, age, baseline serum creatinine, and hemiplegia were included in the model due to their hypothesized impact on failure. After controlling for confounding variables, isolation of VRE (odds ratio [OR]: 4.42, 95% confidence interval [CI] [1.01–19.29], P = .048), increasing number of infusions per day (OR: 1.17, 95% CI [1.04–1.30], P = .007), and use of oxazolidinone therapy (OR: 4.92, 95% CI [1.90–12.82], P = .001) remained significantly associated with OPAT treatment failure. Complete multivariate analysis results can be found in Table 2.

Table 2.

Success vs Failure Multivariate Analysis

Characteristic	P Value	OR	95% CI
Baseline serum creatinine	.484
Hemiplegia	.421
Age	.145
VRE and oxazolidinone	.986
BL/BLI	.073
VRE	.048	4.42	1.01–19.29
Infusions per day	.007	1.17	1.04–1.30
Oxazolidinone	.001	4.92	1.89–12.82

Abbreviation: VRE, vancomycin-resistant Enterococcus.

Among the 40 patients lost to follow-up, bivariate analysis showed they were less likely to have hemiplegia (4 [10.0%] vs 16 [27.2%], P = .033], but were younger (50.1 years [±15.8] vs 55.1 [±14.7], P = .037), had a lower body weight (72.1 kg [61.5–95.6] vs 88.3 [72.9–107.3] P = .005), and lived further from the ID OPAT clinic (80.0 miles [30.3–130.8] vs 47.2 miles [18.7–92.2], P = .030). Given the small sample size, a multivariable logistics regression model could not be performed for loss to follow-up analysis. Additional results can be found in Table 3.

Table 3.

Success vs Loss to Follow-Up Bivariate Analysis

Characteristic	Success (n = 588)	Loss to Follow-Up (n = 40)	P Value
Male, n (%)	352 (59.8%)	27 (67.5%)	.339
Age (y), mean (SD)	55.13 (14.7)	50.08 (15.8)	.037
Weight (kg), median [IQR]	88.3 [72.9–107.3]	72.1 [61.5–95.6]	.005
Baseline serum creatinine, median [IQR]	0.83 [0.65–1.08]	0.73 [0.58–1.05]	.104
Polymicrobial infection	57 (9.7%)	3 (7.5%)	1.000
History of IVDU	25 (4.3%)	3 (7.5%)	.413
Charlson Comorbidity Index	3 [1,2,3,4,5]	3 [1,2,3,4]	.160
Hemiplegia	16 (27.2%)	4 (10.0%)	.033
VRE infection	3 (0.5%)	0 (0.0%)	N/A
CRE infection	4 (0.68%)	0 (0.0%)	N/A
ESBL infection	18 (3.1%)	1 (2.5%)	1.000
Pseudomonas spp.	23 (3.9%)	1 (2.5%)	1.000
Staphylococcus aureus	179 (30.4%)	14 (35%)	.545
ICU admission required	140 (23.8%)	13 (32.5%)	.215
Lumen count, median [IQR]	1 [1]	1 [1]	N/A
Length of stay, d, median [IQR]	8 [5–13]	9.5 [6–20.5]	.068
Infusions per day	2 [1–3]	2 [1–3]	.503
Distance from clinic, miles, median [IQR]	47.2 [18.7–92.2]	80.0 [30.3–130.8]	.030

Data displayed as n (%) unless specified.

Abbreviations: CRE, carbapenem-resistant Enterobacteriales; ESBL, extended-spectrum beta-lactamase; VRE, vancomycin-resistant Enterococcus.

DISCUSSION

This study attempted to identify institutional specific risk factors for OPAT treatment failure, focusing on both modifiable and nonmodifiable factors to enhance patient selection and optimize transitions of care. Our findings suggest that antimicrobial selection and regimen complexity play a role in OPAT treatment success. Notably, we observed that an increased number of daily infusions was associated with higher risk of treatment failure (OR: 1.167, 95% CI 1.044–1.304), reinforcing prior research that linked simplified regimens to improved adherence (9). This investigation did limit inclusion of patients with reduced kidney function (defined as ≤30 mL/min) in hopes of limiting any potential bias toward reduced number of daily infusions within the study cohort. To our knowledge, this would be the first report demonstrating a direct association between increased infusion frequency and treatment failure, highlighting the importance of regimen simplification during transitions of care. Although antimicrobial stewardship often prioritizes inpatient efficacy, these findings support incorporating home-based regimen simplification into OPAT decision-making [9, 10].

Interestingly, infectious diagnosis was not a significant predictor of OPAT failure (P = .455). Most patients were discharged with infections related to bone or joint (n = 288), skin or soft tissue (n = 217), or bacteremia (n = 105), with fewer cases involving infections of the central nervous system (n = 58), gastrointestinal or genitourinary system (n = 21), endocarditis (n = 44), cardiac devices (n = 15), respiratory (n = 8), or other causes (n = 27). Additionally, Charlson Comorbidity Index (CCI) did not differ between the groups (3 [1,2,3,4,5] vs 3 [1,2,3,4,5], P = .461), which contrast with previous studies that inconsistently linked CCI to hospital readmissions [11–15]. These findings likely reflect the predischarge screening process, where high-risk patients may have been transitioned to alternative care settings rather than home-based OPAT.

In our study, patients with VRE infections were more than four times as likely to experience OPAT failure (OR: 4.421, 95% CI [1.01–19.29]), consistent with prior studies demonstrating increased failure risk in OPAT patients treated for resistant organisms (OR, 1.57; 95% CI, 1.03–2.36) [11 ]. For this reason, closer monitoring of OPAT patients with VRE infections may be warranted. Additional studies should further evaluate whether treatment failure is linked to resistance mechanisms of the pathogen, or the specific antimicrobial agents employed in treatment.

Additionally, oxazolidinone therapy was independently associated with increased failure risk (OR: 4.922, 95% CI [1.889–12.823]). Although these agents are commonly used for VRE infections, we confirmed that oxazolidinone use and VRE infection were independent risk factors for failure (P = .986). Additional statistical analysis was done to rule out significant interaction between these study variables, remaining significant in the multivariate model. The exact mechanism underlying this association remains unclear and warrants further investigation. Clinicians opting to treat patients with oxazolidinones in OPAT therapy should consider enhanced monitoring and follow-up strategies to mitigate failure risk.

Patients lost to follow-up were less likely to have hemiplegia, but were younger, had a lower body weight/BMI, and lived farther from the OPAT clinic. The median travel distance, as defined by an estimated driving mileage, was significantly greater in those lost to follow-up (80.0 miles [30.3–130.8] vs 47.2 [18.7–92.2], P = .030), with a mean difference of 33 miles between groups. These findings highlight the geographic barriers faced by rural patients, which could contribute to missed follow-up visits and reduced treatment adherence. While telehealth services were not assessed in this study, their potential role in mitigating geographic disparities in OPAT monitoring warrants further exploration.

This study provides new insights into previously unrecognized risk factors for OPAT failure, particularly the impact of regimen complexity and antimicrobial selection. The large sample size enabled robust statistical analyses, including multivariate logistics regression, which helped control for confounding variables. Additionally, the structured OPAT referral process at our institution ensured that all patients received an ID consult, diagnostic workups, and a multidisciplinary assessment before OPAT initiation, likely contributing to the high success rate observed in our cohort.

However, several limitations exist. As a retrospective, single-center study, findings may not be generalizable to other facilities with different patient populations and OPAT structures. Additionally, the small sample size of the lost to follow-up group prevented multivariable logistics regression for this endpoint, limiting our ability to fully account for potential confounders. Lastly, as data collection relied on electronic medical records and REDCap documentation, missing or incomplete records may have introduced bias.

CONCLUSIONS

Overall, this study identified modifiable risk factors for OPAT-related treatment failure, reinforcing the importance of individualized patient assessment when selecting OPAT candidates. Notably, we provide the first evidence linking increased daily infusions to a higher risk of failure, a factor previously associated with only reduced adherence. Clinicians should prioritize regimen simplification, close monitoring for VRE infections, and enhanced follow-up for patients receiving oxazolidinones. Additionally, geographic distance should be considered when discharging patients on OPAT, particularly for rural populations. These findings emphasize the need for a patient-centered approach in OPAT management, balancing the benefits of home-based antimicrobial therapy with individualized risk assessment.