Real-World Application of Oral Therapy for Infective Endocarditis: A Multicenter, Retrospective, Cohort Study

Abstract

Background

We sought to compare the outcomes of patients treated with intravenous (IV)-only vs oral transitional antimicrobial therapy for infective endocarditis (IE) after implementing a new expected practice within the Los Angeles County Department of Health Services (LAC DHS).

Methods

We conducted a multicentered, retrospective cohort study of adults with definite or possible IE treated with IV-only vs oral therapy at the 3 acute care public hospitals in the LAC DHS system between December 2018 and June 2022. The primary outcome was clinical success at 90 days, defined as being alive and without recurrence of bacteremia or treatment-emergent infectious complications.

Results

We identified 257 patients with IE treated with IV-only (n = 211) or oral transitional (n = 46) therapy who met study inclusion criteria. Study arms were similar for many demographics; however, the IV cohort was older, had more aortic valve involvement, were hemodialysis patients, and had central venous catheters present. In contrast, the oral cohort had a higher percentage of IE caused by methicillin-resistant Staphylococcus aureus. There was no significant difference between the groups in clinical success at 90 days or last follow-up. There was no difference in recurrence of bacteremia or readmission rates. However, patients treated with oral therapy had significantly fewer adverse events. Multivariable regression adjustments did not find significant associations between any selected variables and clinical success across treatment groups.

Conclusions

These results demonstrate similar outcomes of real-world use of oral vs IV-only therapy for IE, in accord with prior randomized, controlled trials and meta-analyses.

Graphical Abstract

This graphical abstract is also available at Tidbit: https://tidbitapp.io/tidbits/real-world-application-of-oral-therapy-for-infective-endocarditis-a-multicenter-retrospective-cohort-study

infective endocarditis, oral stepdown antibiotic therapy

Three randomized, controlled trials have established that oral transitional therapy for infective endocarditis (IE) is at least as effective as intravenous (IV)-only therapy [1-3]. Indeed, in the largest such trial, oral therapy resulted in significantly better clinical response, lower relapse, and higher survival rates than IV-only therapy out to 5 years of follow-up [4, 5]. Systematic reviews have demonstrated extensive supporting pharmacological and observational data for oral therapy, and meta-analyses of randomized, controlled trials have confirmed that oral therapy is at least as effective as IV-only therapy for this disease [6, 7]. Nevertheless, oral therapy remains rarely used for bacteremia in practice and less so for endocarditis [8, 9].

One potential reason for slow uptake of the use of oral therapy for IE is that data describing practical, real-world outcomes of oral therapy outside the setting of carefully conducted randomized, controlled trials are limited. We recently implemented an evidence-based expected practice [10] defining the scope of oral transitional therapy for IE in the 3 acute care, public hospitals comprising the Los Angeles County Department of Health Services (LAC DHS). We sought to compare the outcomes of patients treated with oral transitional therapy with the outcomes of patients treated with IV-only therapy in a retrospective cohort study.

METHODS

The Expected practice

Expected practices are a set of expectations set within an institution regarding how providers practice medicine to achieve more standardized, equitable care based on available clinical evidence and real-life practice conditions within the facility [10]. In contrast, clinical guidelines are created by national societies to be a resource for medical decision-making nationwide and may be difficult to apply across variably resourced care settings.

While small numbers of patients with IE were treated with oral therapy upon publication of the large Partial Oral Treatment of Endocarditis randomized, controlled trial [3] in June 2020, the LAC DHS implemented an expected practice for oral transitional therapy for IE that enabled more widespread use (Supplementary Figure 1, SupplementaryTable 1). We define oral transitional therapy as initiation on IV therapy, followed by switching to oral therapy when patients meet all of the clinical criteria specified in the expected practice, specifically, the patient was clinically stable with no immediate indication for cardiac surgical intervention; the initial course of IV therapy cleared the patient's bacteremia; there were no concerns regarding absorption of oral therapy from the gastrointestinal tract; there were no psychosocial concerns that would cause IV therapy to be preferred for compliance or level of care concerns; and an oral antibiotic regimen was available to which the etiologic organism was susceptible in vitro, and had published clinical data.

Data Collection

We conducted a retrospective chart review of Cerner-based electronic medical records (EMR) for relevant patients cared for at the 3 LAC DHS acute care safety net hospitals: Los Angeles County + University of Southern California (LAC + USC) Medical Center, Harbor–University of California Los Angeles (UCLA) Medical Center, and Olive View–UCLA Medical Center. Patients were included if they had definite or possible IE (defined below) caused by typical pathogens.

To identify patients for inclusion/exclusion, we generated an EMR query for all patients with blood cultures drawn from December 2018 to June 2022 that grew typical IE pathogens (ie, Staphylococcus spp., Streptococcus spp., Enterococcus spp., and Haemophilus spp.; and Aggregatibacter spp., Cardiobacterium hominis, Eikenella corrodens, and Kingella spp. [HACEK]). The resultant query identified 3968 episodes. We conducted 2 phases of manual chart review. Phase I focused on identifying cases of IE and phase II focused on obtaining patient-level data (ie, demographics, outcomes, length of treatment, and length of hospitalization). The manual chart review led to exclusion of additional episodes, leading to a final count of 257 eligible episodes (Figure 1).

Figure 1.

Patient selection flowchart. Abbreviations: AMA, against medical advice; ED, emergency department; EMR, electronic medical records; IE, infective endocarditis; IV, intravenous. *Included organisms: Staphylococcus spp., Streptococcus spp., Enterococcus spp., and Haemophilus spp., Aggregatibacter spp., Cardiobacterium hominis, Eikenella corrodens, and Kingella spp. [HACEK organisms].

Inclusion and Exclusion Criteria

Patients who met the definition of IE using Duke's criteria [11-13] were included in this study. Specifically, patients with definite IE met 2 major criteria or 1 major and 3 minor criteria, and patients with the diagnosis of possible endocarditis met 1 major and 1 minor criterion, 3 minor criteria, or the diagnosis was based on high clinical suspicion documented by the treating physician. Patients were excluded if they had inadequate documentation, died prior to the average time to transition to oral therapy (ie, less than 14 days), the blood culture specimen was nonclinical or from an autopsy, the blood culture was obtained from the emergency department or outpatient setting without follow-up, and if the patient was aged <18 years at time of diagnosis. Patients who were excluded for inadequate documentation were those who left against medical advice prior to diagnosis or prior to setting up therapy, transferred out of network, or were transitioned to hospice during the hospitalization. Of note, patients who were discharged home, to jail, or to a skilled nursing facility without documented specialty follow-up were included in the study (18 in the IV-only arm, 3 in the oral arm), as they were less likely to have changes in their therapy plans. Patients discharged to jail continued to be seen by providers within our health system, but were documented in their own EMR system. Of those patients, 7 were documented deceased and 14 were imputed successes.

Outcomes

The primary efficacy end point was clinical success, defined as being alive, without recurrent bacteremia, and without treatment-emergent infectious complications within 90 days. Treatment-emergent infectious complications were defined as not being present at time of diagnosis and subsequently developing while on antimicrobial therapy. Such complications included development of major arterial emboli, septic pulmonary emboli/infarcts, mycotic aneurysms, Janeway lesions, distant organ abscesses (eg, liver/spleen/renal), visceral organ infarcts (eg, splenic/renal), or distal organ hematogenous seeding (eg, osteomyelitis). Secondary outcomes included clinical success at last follow-up, treatment-related adverse events (AEs), hospital length of stay, and hospital readmission rates. AEs included drug allergy, cytopenias, acute kidney injury (AKI), IV line–related complications (ie, infection, thrombosis, dislodged/clogged lines requiring replacement, associated pain/bleeding), Clostridioides difficile colitis, gastrointestinal upset, and electrolyte/laboratory abnormalities. We also observed the average IV therapy lead-in time. Patients were assigned to the IV-only or oral cohort based on what antibiotic route they received from the hospital at discharge. Patients who were readmitted to the hospital and required new treatment for IE were included as a second encounter, whether the second episode was deemed a recurrence or new infection. All recurrences of IE occurred after the original treatment course and were counted as a failure; those patients were discharged on different regimens on the second encounter. Thus, the same patient may have been included more than once, in either arm, depending on their discharge regimen for that encounter.

Statistical Analyses

Continuous variables were compared using the nonparametric Mann–Whitney unpaired test and dichotomous variables were compared using χ² or Fisher exact tests with α = 0.05 for significance. We conducted multivariable logistic regression for the outcomes of clinical success at 90 days and at last follow-up using SAS Enterprise Guide 7.1. The included outputs were selected a priori based on clinical significance or were included because they achieved a P value of <.05 in bivariate analysis. Adjusted odds ratios for primary outcomes were calculated with Wald 95% confidence intervals for these covariates.

RESULTS

Of 3968 patients with positive blood cultures identified by the electronic query, 275 had a diagnosis of IE, of whom 18 were excluded (16 due to loss to follow-up or transition to hospice and 2 due to ineligible gram-negative pathogens). We therefore included 257 patients with IE in the study based on manual chart review: 211 in the IV-only therapy arm and 46 in the oral transitional therapy arm (Figure 1).

Patient and Infection-related Characteristics

Demographic features were similar between groups (Table 1). However, the IV arm was significantly older and had a higher proportion of aortic valve endocarditis, diabetes, dialysis dependence, and central venous catheters at time of diagnosis. Conversely, the oral arm had significantly higher proportions of tricuspid valve endocarditis, patients with a history of injection drug use, and cases caused by methicillin-resistance Staphylococcus aureus (MRSA). Patients in the IV cohort had a longer median time to last follow-up due to progressively more patients being treated with oral therapy as more time elapsed from release of the expected practice.

Table 1.

Demographics by Treatment Arm

Demographic		Intravenous Therapy (n = 211)	Oral Therapy (n = 46)	P Value
Race				.18
Asian		13 (6.2)	2 (4.3)
Black		16 (7.6)	7 (15.2)
White		31 (14.7)	5 (10.9)
Hispanic		129 (61.1)	29 (63.0)
Othera		22 (10.4)	3 (6.5)
Gender
Female		59 (28.0)	14 (30.4)	.22
Male		152 (72.0)	32 (69.6)
Median age (IQR), y		55 (42–65)	39 (30.5–61.8)	.01
Key features
Endocarditis definite		128 (60.7)	28 (60.9)	.98
Endocarditis possible, total		83 (39.3)	18 (39.1)	.98
Clinical		44 (20.9)	12 (26.1)	.29
Duke's criteria		39 (18.5)	6 (13.0)	.29
Prosthetic valveb		31 (14.7)	4 (8.7)	.28
PPM or ICDb		20 (9.5)	1 (2.2)	.08
CVCb		26 (12.3)	1 (2.2)	.03
PPM and CVC		1 (0.5)	0 (0.0)	>.99
Valves involved
AV		47 (22.3)	3 (6.5)	.01
MV		46 (21.8)	9 (19.6)	.74
PV		4 (1.9)	0 (0.0)	.85
TV		13 (6.2)	16 (34.8)	<.0001
AV + MV		17 (8.1)	2 (4.3)	.31
AV + TV		1 (0.5)	1 (2.2)	>.99
MV + TV		4 (1.9)	0 (0.0)	.85
AV + MV + TV		2 (0.9)	0 (0.0)	>.99
Unknown/presumed		57 (26.9)	15 (32.6)	.44
ICD		12 (5.7)	0 (0.0)	.12
Otherc		8 (3.8)	0 (0.0)	.30
Any left-sided involvement		117 (55.5)	15 (32.6)	.004
Surgical management, total		51 (24.2)	14 (30.4)	.73
Early valve replacement or ICD removald		39 (18.4)	13 (28.3)
Post-Tx valve replacement or ICD removal		8 (3.8)	1 (2.2)
Early valve repair		3 (1.4)	0 (0.0)
Post-Tx valve repair		1 (0.5)	0 (0.0)
Stable housing during treatment		191 (90.5)	40 (87.0)	.47
Comorbidity
Diabetes mellitus		75 (35.5)	9 (19.6)	.04
Chronic kidney disease e		41 (19.4)	9 (19.6)	.98
End-stage renal disease dialysis-dependentb		42 (19.9)	1 (2.2)	.001
Chronic obstructive pulmonary disease		10 (4.7)	1 (2.2)	.42
Liver disease		39 (18.5)	14 (30.4)	.07
Cancer		18 (8.5)	5 (10.9)	.80
Injection drug use		38 (18.0)	17 (37.0)	.01
Median time to last follow-up (IQR), d		204 (51–495)	93.5 (26–279)	.02

Data are presented as number (%) unless otherwise indicated.

Abbreviations: AV, aortic valve; CVC, central venous catheter; ICD, implantable cardiac defibrillator or pacemaker; IQR, interquartile range; MV, mitral valve; PPM, permanent pacemaker; PV, pulmonary valve; TV, tricuspid valve; Tx, treatment.

Specifics not reported, not Hispanic/Latino, Middle Eastern, Armenian, Native American.

Present on admission.

Right atrial vegetation/suspected infected thrombi (4), left ventricular outflow tract vegetation (1), superior vena cava echo density (2), right ventricle mid-septum vegetation (1).

”Early” is defined by valve replacement within initial antibiotic course.

Patients with chronic kidney disease who were not dialysis-dependent.

Staphylococcus aureus was the predominant pathogen in both groups, with 52.1% vs 63.0% in the IV-only vs oral therapy arms, respectively (P = .18). Of the patients with S. aureus infection, 20.4% and 34.8% were caused by MRSA in the IV-only vs oral therapy arms, respectively (P = .04). The second most common pathogen was Streptococcus spp., causing 28.4% (IV arm) and 21.7% (oral therapy arm) of infections, followed by Enterococcus faecalis (Supplementary Table 2).

Clinical Outcomes

Clinical success rates were similar in both the IV-only arm and the oral therapy arm at 90 days of follow-up (84.4% vs 87%, P = .66) and at last follow-up (82.0% vs 76.1%, P = .36; Table 2). There were no significant differences in clinical success at either time point by any comorbidity subgroup, except that patients with liver disease at last follow-up had more success in the IV-only arm (P = .02). There were also no significant differences in the individual components of the clinical failure composite definition (death, recurrence of bacteremia, and treatment-emergent complications) or 90-day readmission rates between the 2 groups (Table 2).

Table 2.

Clinical Outcomes

Outcome	Intravenous Arm (n = 211)	Oral Arm (n = 46)	P Value
Clinical success at 90 days	178 (84.4)	40 (87.0)	.66
Alive	193 (91.5)	41 (89.1)	.61
Lack of recurrence of bacteremia	204 (96.7)	45 (97.8)	.69
Absence of treatment-emergent  complications	185 (87.7)	44 (95.7)	.12
Clinical success at 90 days by comorbiditya
Diabetes mellitus	58 (77.3)	9 (100)	.11
CKD	37 (90.2)	6 (66.7)	.06
Dialysis-dependent	33 (78.6)	0 (0.0)	.26
COPD	9 (90.0)	0 (0.0)	.19
Liver disease	35 (89.7)	11 (78.6)	.29
Cancer	15 (83.3)	3 (60.0)	>.99
Injection drug use	32 (84.2)	14 (82.4)	.86
Stable housing during therapy	162 (84.8)	34 (85.0)	.98
Clinical success at last follow-up	173 (82.0)	35 (76.1)	.36
Alive	182 (86.3)	39 (84.8)	.79
Lack of recurrence of bacteremia	203 (96.2)	43 (93.5)	.41
Absence of treatment-emergent complications	192 (91.0)	43 (93.5)	.59
Clinical success at last follow-up by comorbiditya
Diabetes mellitus	59 (78.7)	9 (100)	.12
CKD	35 (85.4)	6 (66.7)	.19
Dialysis-dependent	31 (73)	0 (0.0)	.33
COPD	10 (100)	0 (0.0)	.09
Liver disease	34 (87.2)	8 (57.1)	.02
Cancer	13 (72.2)	1 (20.0)	.06
Injection drug use	32 (84.2)	14 (82.4)	.86
Stable housing	156 (81.7)	30 (75.0)	.33
Readmissions within 90 days	72 (34.1)	12 (26.1)	.29

Data are presented as number (%) unless otherwise indicated.

Abbreviations: CKD, chronic kidney disease, COPD, chronic obstructive pulmonary disease.

A patient may have more than 1 comorbidity and more than 1 adverse event.

We also conducted a sensitivity analysis to evaluate outcomes in patients who had definite IE (excluding possible IE patients). Also, there were no significant differences in primary outcomes at 90 days between the IV-only and oral cohorts (clinical success: 82.0% vs 89.3%, P = .27; Table 3). Nor were there any differences between the individual components of the composite success end point.

Table 3.

Primary Clinical Outcomes in Patients With Definite Infective Endocarditis

Outcome	Intravenous Arm (n = 128)	Oral Arm (n = 28)	P Value
Clinical success at 90 days	105 (82.0)	25 (89.3)	.27
Alive	120 (91.5)	27 (96.4)	.56
Lack of recurrence of bacteremia	124 (96.7)	27 (96.4)	>.99
Absence of treatment-emergent complications	111 (87.7)	27 (96.4)	.13

Data are presented as number (%) unless otherwise indicated.

Of note, a similar rate of patients in the IV-only cohort vs oral cohort failed to complete their planned duration of therapy. Specifically, 7.1% (n = 15) of patients receiving IV-only therapy failed to complete therapy, 3 due to leaving against medical advice and 12 due to death. Only 6.5% (n = 3) of patients in the oral therapy arm did not complete therapy, 1 due to reported stolen medications, 1 due to miscommunication of duration of therapy, and 1 due to decision to transition to hospice therapy after discharge but prior to completion of therapy. Of the patients in the IV-only arm, 19.4% (n = 41) had prolonged hospitalizations to enable them to complete their therapy.

Adverse Events

There was a statistically significant increased number of AEs in the IV-only group compared with the transitional therapy group. Specifically, 27.5% of patients in the IV-only group developed AEs compared with 8.7% in the oral group (P = .004; Table 4). The most notable differences were patients who experienced AKIs (P = .048) and IV line–related AEs (P = .04). Of note, 6 patients developed AEs while on IV therapy prior to transitioning to oral therapy: 1 drug rash, 2 AKIs, and 3 IV line–related problems. These were not accounted for in the total number of AEs in either arm. We analyzed AEs by comorbidities, and there were no significant correlations. Patients with stable housing in the IV arm still had significantly more AEs.

Table 4.

Adverse Events

AE	Intravenous Arm (n = 211)	Oral Arm (n = 46)	P Value
Total patients with AEsa	58 (27.5)	4 (8.7)	.004
Drug allergy (eg, rash)	6 (2.8)	0 (0.0)	.49
Cytopeniab	10 (4.7)	2 (4.3)	.66
Acute kidney injury	23 (10.9)	1 (2.2)	.048
IV-line infection	4 (1.9)	0 (0.0)	.85
IV line–related thrombosis	4 (1.9)	0 (0.0)	.85
Other IV line–related complicationsc	9 (4.3)	0 (0.0)	.23
All IV line–related AEs	17 (8.1)	0 (0.0)	.04
Clostridioides difficile colitis	2 (0.9)	0 (0.0)	>.99
Gastrointestinal upset	4 (1.9)	1 (2.2)	>.99
Otherd	5 (2.4)	3 (6.5)	>.99
Total AEs by comorbiditya
Diabetes mellitus	22 (30.7)	0 (0.0)	.06
Chronic kidney disease	13 (31.7)	1 (11.1)	.21
Dialysis-dependent	15 (35.7)	0 (0.0)	>.99
Chronic obstructive pulmonary disease	2 (20.0)	1 (100)	.30
Liver disease	12 (30.8)	2 (14.3)	.20
Cancer	3 (16.7)	1 (20.0)	.97
Injection drug use	12 (31.6)	2 (11.8)	.11
Stable housing	53 (27.7)	3 (7.5)	.003

Data are presented as number (%) unless otherwise indicated.

Abbreviations: AE, adverse event; IV, intravenous.

A patient may have more than 1 comorbidity and more than 1 AE.

Cytopenia includes leukopenia, anemia, and thrombocytopenia.

Dislodged/clogged IV lines requiring replacement, pain associated with IV line, bleeding from IV line.

Other AEs in IV arm: transaminitis, elevated creatine kinase (CK) in association with daptomycin. Other AEs in oral arm: hyperkalemia, concern for serotonin syndrome.

Multivariable Adjustment

Two multivariable logistic regression models were constructed to assess likelihood of clinical success at 90 days follow-up and last follow-up across oral and IV groups with selected variables included to adjust for imbalances. The variables included all demonstrated statistically significant imbalances between the 2 cohorts: age, dialysis dependence, aortic valve involvement, and MRSA infection. Despite also demonstrating statistical imbalances between the oral and IV cohorts, central venous catheters were not included in the multivariable model because they were significantly collinear (P <.001) with hemodialysis, and IV drug use was not included because it was significantly collinear (P = .0024) with MRSA infection. None of the included variables were significantly correlated to each other, indicating that there were no other collinear or confounding effects. The overall multivariable models failed to demonstrate significant impact of any of the combined effect of variables on clinical success at 90 days follow-up or at last follow-up (Table 5).

Table 5.

Multivariable Adjustment for Outcomes

Variables Analyzed	Beta Coefficient	Odds Ratio	95% Confidence Interval	P Value
Clinical success at 90 days
Oral therapy (vs IV-only as referent)	−0.05	0.95	0.35–2.58	.92
Age	−0.02	0.98	0.96–1.00	.07
Dialysis-dependent	−0.59	0.56	0.24–1.31	.18
Aortic valve involvement	−0.49	0.61	0.28–1.35	.22
Staphylococcus aureus, MRSA	−0.76	0.47	0.21–1.03	.06
Clinical success at last follow-up	…	…	…	…
Oral therapy (vs IV-only as referent)	−0.65	0.52	0.23–1.19	.12
Age	−0.02	0.98	0.96–1.00	.05
Dialysis-dependent	−0.69	0.50	0.23–1.12	.09
Aortic valve involvement	−0.19	0.82	0.40–1.72	.61
S. aureus, MRSA	−0.23	0.80	0.38–1.69	.55

Abbreviations: IV, intravenous; MRSA, methicillin-resistant Staphylococcus aureus.

Therapy Prior to Oral Transitional Therapy

There was heterogeneity in duration of IV lead-in therapy prior to transition to oral therapy. Most cases transitioned at time of discharge. We did not observe an association between duration of IV therapy prior to oral transition and clinical success (Table 6). The IV lead-in duration ranged from less than 7 days to more than 21 days, with a median of 15.5 days, which correlated with the median length of stay as noted below.

Table 6.

Clinical Outcomes Based on Intravenous Lead-In Duration

Duration of Intravenous Therapy Lead-In	≤7 Days (n = 9)	8–14 Days (n = 12)	15–21 Days (n = 14)	>21 Days (n = 11)
Clinical success at 90 days	8 (88.9)	9 (75)	14 (100)	9 (81.8)
Clinical success at last follow-up	6 (66.7)	6 (50)	13 (92.9)	9 (81.8)
Readmitted within 90 days	3 (33.3)	6 (50)	0 (0.0)	3 (27.3)

Data are presented as number (%) unless otherwise indicated.

Durations of Therapy and Hospitalization

There were no significant differences between median length of treatment and hospitalization between the 2 groups, except, as expected, the duration of IV therapy was longer in the IV-only therapy arm. The median duration of hospitalization for the IV-only arm was 16 days, whereas it was 14.5 days for the oral transitional arm (Table 7). The median total duration of therapy was the same for both groups.

Table 7.

Length of Therapy and Length of Hospitalization

Length of Therapy and Hospitalization	Intravenous Arm (n = 211)	Oral Arm (n = 46)	P Value
Median length (interquartile range), d
Length of inpatient stay	16 (10–31)	14.5 (8.8–23.3)	.20
Duration of inpatient IV therapy	16 (10–31)	15.5 (8.8–21.3)	.07
Duration of therapy post-discharge	23 (5–33)	27.5 (14–35)	.10
Total duration of IV therapy	42 (42–47)	15.5 (8.8–21.3)	<.0001
Total duration of antibiotic therapy	42 (42–47)	42 (42–46)	.59

Durations reported as median length of stay in days, with interquartile range.

Abbreviation: IV, intravenous.

Oral Transitional Therapy

An assortment of oral regimens was administered at the time of transition, all of which resulted in similar success rates at last follow-up (Supplementary Table 3). The most commonly used therapy was oral linezolid 600 mg twice a day with or without rifampin. Oral linezolid was administered to 30 (65.2%) patients, of whom 26 (86.7%) and 4 (13%) were treated without vs with rifampin, respectively. The clinical outcomes did not differ between the linezolid group with or without rifampin (Table 8). The next most common oral therapy used was a high-dose penicillin (ie, ampicillin, amoxicillin, dicloxacillin, or penicillin V) with or without combination therapy (8 of 46, 17.4%). Other oral therapies used included a fluroquinolone with rifampin or trimethoprim-sulfamethoxazole with or without rifampin.

Table 8.

Clinical Outcomes of Linezolid Therapy

Outcome	Linezolid 600 mg BID (n = 26)	Linezolid 600 mg BID + Rifampina (n = 4)	P Value
Oral antibiotic discharge regimen
Clinical success at 90 days	24 (92.3)	4 (100)	>.99
Clinical success at last follow-up	20 (76.9)	4 (100)	.93
Recurrence of bacteremia at 90 days	0	0	>.99
Recurrence of bacteremia at last follow-up	2 (7.7)	0	>.99
Readmission at 90 days	8 (30.8)	0	.37

Data are presented as number (%) unless otherwise indicated.

Abbreviation: BID, twice a day.

Rifampin dose was either 600 mg every day (n = 2) or 300 mg 3 times per day (n = 2).

DISCUSSION

A long-standing dogma exists, based on little evidence, that deep-seated infections such as IE require prolonged IV antibiotic therapy to achieve favorable clinical outcomes. However, recent randomized, controlled trials have demonstrated equivalent outcomes and fewer adverse side effects when oral transitional therapy is used in the treatment of IE [1-3]. Currently, despite these studies, providers may hesitate to use oral transitional therapy due to lack of clinically applicable regimens with supportive real-world data. Our results are reassuringly aligned with these randomized studies, demonstrating similar clinical outcomes of oral transitional vs IV-only therapy in real-world settings outside of carefully conducted clinical trials.

We used a variety of oral antibiotics (Supplementary Table 3) based on the options provided by our institution's expected practice (Supplementary Figure 1, SupplementaryTable 1), extrapolated from previous studies [1-4, 14-19]. Even with this variety of antibiotic choice, the clinical outcomes remained similar in the 2 groups. As mentioned, linezolid was the most frequently used antibiotic, despite the fact that it is “bacteriostatic.” The excellent outcomes with linezolid therapy underscore the lack of clinical relevance of the “static vs cidal” paradigm, a myth debunked in previous randomized, controlled trials and a 2018 systematic review [20].

We found some differences in demographics, but in a manner that could drive poor outcomes in both arms. Specifically, IV only–treated patients were older, with higher rates of diabetes and end-stage renal disease, and had more left-sided disease, which may portend worse outcomes. However, they had lower rates of substance use disorder and lower proportion of infections caused by MRSA, which portends better outcomes [21-24]. Not surprisingly, clinicians favored discharge without an IV line among patients with substance use disorder due to concern of line misuse [25, 26]. We also noted higher percentages of patients with central venous catheters and dialysis-dependent patients in the IV-only arm, which was anticipated as these patients already had long-term IV access and could more easily obtain IV therapy in the outpatient setting. We adjusted for these demographic differences using multivariate analyses and found that none of the conditions that we had selected for were significantly associated with the patients’ clinical outcomes.

While clinical success rates were similar between the cohorts, the AE rate was significantly higher in the IV-only arm, reaffirming the findings from other studies that found long-term IV therapy and IV lines were associated with high rates of AEs in treating IE [6, 27-29]. There were no specific comorbidities associated with higher rates of AEs. Of note, there were more patients who had stable housing in the AEs IV cohort compared with the oral cohort, but that did not positively impact their clinical outcomes or decrease their rates of AEs.

We were surprised to note that the length of hospital stay did not significantly differ between the 2 cohorts due to extensive use of outpatient IV therapy in the IV-only arm and apparent hesitancy to transition to oral therapy early in the oral arm. However, the data demonstrate that longer IV lead-in periods did not improve outcome in the oral therapy arm, providing reassurance that prolonging IV therapy prior to transitioning to oral therapy is unnecessary and may be harmful given higher AE rates. We hope that with time and experience, clinicians will transition to oral therapy more rapidly when clinical criteria specified in the expected practice are met.

The primary limitation of our study is its retrospective nature. We cannot exclude the possibility of missing follow-up outside of our public hospital network. However, in a safety net healthcare system like ours, patients are less likely to receive care at outside institutions given limited insurance. Fortunately, we have 1 EMR throughout our DHS system that captures all blood cultures, subsequent hospitalizations, emergency department visits, and clinic visits, making it less likely that we missed future recurrences or treatment-emergent complications during the study period.

In summary, the results from our 3 public hospitals demonstrate that it is possible to select patients with IE based on rational clinical criteria who can be safely treated with oral therapy, including patients infected with MRSA. Treating patients with oral therapy resulted in treatment success rates similar to IV-only therapy but with significantly less harm. Our findings provide real-world effectiveness confirmation of the efficacy data demonstrated in 3 concordant randomized, controlled trials. Additional trials are ongoing, and results will be valuable additions to the available literature when published [30].

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.

Note

Patient consent. Given its retrospective nature, this work did not in­clude factors that necessitated patient consent. The study was approved with waiver of informed consent by the University of Southern California Biomedical Research Institutional Review Board.

Author notes

Potential conflicts of interests. N. W. D. reports receipt of a speaker’s fee for IDWeek 2022, funded by a collaboration of societies that help fund IDWeek, including Infectious Diseases Society of America, The Society of Healthcare Epidemiology of America, HIV Medicine Association, Pediatric Infectious Diseases Society, and Society of Infectious Diseases Pharmacists. All remaining authors: No reported conflicts of interest. All authors have 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.