Partial oral versus full intravenous antibiotic treatment of endocarditis in people who inject drugs: A systematic review

Abstract

Background:

Prolonged intravenous (IV) antibiotic therapy may not be optimal for people who inject drugs (PWID) with infective endocarditis (IE) due to unique social and medical needs. The role of partial IV antibiotic therapy with continued oral (PO) antibiotic therapy is unclear.

Methods:

A systematic review was performed using EMBASE and MEDLINE databases. Included studies compared PO to IV antibiotic treatment for IE in PWID.

Results:

Four studies met eligibility. Observational studies included full IV treatment groups and partial IV, partial PO treatment groups for severe injection-related infections. PWID with IE comprised 41.0%–64.7% of the study populations but outcomes specific to IE were not separately reported. All-cause 90-day readmission rates were comparable between the IV treatment group (27.9%–31.5%) and partial IV, partial PO treatment group (24.8%–32.5%). Ninety-day mortality was non-significantly different between IV treatment (4.9%–10.7%) and partial IV, partial PO treatment groups (2.4%–13.0%). One small randomized clinical trial compared IV oxacillin or vancomycin with gentamicin to PO ciprofloxacin plus rifampin. The cure rates were 91% and 90%, respectively.

Conclusion:

There is limited evidence comparing IV treatment to partial IV, partial PO antibiotic treatment in PWID with IE. Observational studies suggest that PO antibiotic therapy after initial IV treatment may be equivalent to full IV treatment alone within specific parameters, but randomized trials are needed to inform recommendations. Substantial clinical and social benefits for PWID and advantages for the health care system will result if PO treatment strategies with equal efficacy can be implemented.

Introduction

Injection drug use frequently results from severe opioid and other substance use disorders and is a risk factor for the development of infective endocarditis (IE). Severe infections, such as IE, contribute to significant morbidity and mortality, especially in people who inject drugs (PWID) who are often young and otherwise relatively healthy (1).

As reflected in the evidence-based guidelines from the European Society of Cardiology and the American Heart Association, intravenous (IV) antibiotics are considered the gold standard for effectively treating IE as they achieve rapid blood and tissue therapeutic concentrations (2,3). The antimicrobial and duration of IV therapy depends on the organism, antimicrobial susceptibility, the bactericidal activity of the antibiotic, and the valve infected (2). General duration recommendations range from 2 to greater than 6 weeks from the time of negative blood cultures or surgical valve intervention (2).

Prolonged IV antibiotic therapy in PWID often necessitates a prolonged hospital stay with important social consequences, such as loss of housing, childcare, or employment (4). Furthermore, even when underlying substance use disorders are managed, prolonged hospital stays can involve symptoms of withdrawal, untreated pain, and stigmatization by the care team (4,5). Understandably, PWID may not be amenable to prolonged hospitalization for antibiotic treatments, which can lead to early patient-directed discharge and premature cessation of antibiotic therapy (4,5). Therefore, there is a substantial need to develop effective outpatient care practices that are acceptable to PWID with IE. Outpatient parenteral antibiotic treatment (OPAT) programs provide a potential solution, but PWID are often considered ineligible due to social factors such as insecure transportation and housing (5). Furthermore, prolonged venous access is not without potential complications such as phlebitis and thrombosis (5).

Given the challenges in completing IV antibiotic treatment in PWID with IE, outpatient oral (PO) antibiotic therapy is an attractive alternative. There is a growing body of evidence to suggest that PO antibiotic therapy following initial IV therapy may be safe and effective for some cases of IE, however, PWID are a unique population and are often excluded from research (4–6). This systematic review examines the current literature for PO versus IV antibiotic treatment of IE in PWID.

Methods

Our systematic review protocol adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Figure 1).

Figure 1:

PRISMA study flow diagram

Search strategy

A systematic search strategy was developed to capture all articles of PO versus IV antibiotic treatment of IE in PWID, limited to the English language. The EMBASE and MEDLINE databases were searched in November 2022. Recent and relevant reviews were searched for additional potential studies. The search strategy is presented in Figure 2.

Figure 2:

Search terms and strategy

Eligibility criteria

Included observational or interventional studies were required to compare PO to IV antibiotic treatment of any duration for the treatment of IE in PWID. Studies that did not have 100% PWID were excluded. No microbiologic organism of IE was prioritized, but studies that focused on culture-negative endocarditis were excluded. Case series and articles without original data were excluded.

Study selection

All title and abstracts identified by the literature search were independently screened by two investigators (AB and HLJ). Based on screening, full text articles were subsequently reviewed by the same two reviewers. Discrepancies were resolved by consensus.

Data extraction

One reviewer (AB) conducted the data extraction by collecting comparable and relevant outcomes. Data extraction was verified by additional reviewers (HLJ, CA).

Risk of bias of randomized studies and quality assessment of non-randomized studies

Quality assessment and risk of bias of non-randomized studies were conducted using the Newcastle-Ottawa Scale (7). The Newcastle-Ottawa scale includes criteria for three sections: selection of study groups, comparability of groups, and ascertainment of outcome with scoring based on a designated number of stars for each section. Within the selection of study groups category, four possible stars can be awarded based on representativeness of the exposed cohort, selection of non-exposed cohort, ascertainment of the exposure, and demonstration that the outcome of interest was not present at the start of the study. The comparability of the study assesses if the study controls for the most important factor and any additional factors, with up to two stars awarded. Assessment of the outcome ascertainment is determined based on the method of ascertainment, if the follow-up was long enough for the outcome to occur, and the adequacy of follow-up of cohorts, with a total of three possible stars.

Risk of bias and quality assessment for randomized trials were performed using the Cochrane Collaboration Modified Tool that assesses features of the study to judge for selection bias, performance bias, detection bias, attrition bias, reporting bias, and other biases (8).

Results

The initial literature search retrieved 665 studies and after removing duplicates, 609 were screened. Title and abstract screening selected 15 studies for full-text review. Four studies, including three observational and one randomized trial, remained for final analysis. Tables 1–3 summarize the characteristics of the selected studies.

Table 1:

Observational studies of PO versus IV antibiotic treatment in PWID with serious injection-related infections

Reference	Design	Study period	Patients, no. (%)			Microbiology	90-day readmission rate				90-day mortality
			Full IV	Partial IV no PO	Partial IV partial PO		Full IV	Partial IV no PO	Partial IV partial PO	p-Value	Full IV	Partial IV no PO	Partial IV partial PO	p-Value
Lewis et al., 2022 (9)	Prospective cohort	2019–2021	61 (36.7)	NA*	105 (63.3)	65.6% S. aureus	27.9%	NA*	24.8%	0.739	4.9%	NR†	2.9%	0.625
Marks et al., 2020 (4)	Retrospective cohort	2016–2019	143 (48.8)	67 (22.9)	83 (28.3)	55.2% S. aureus	31.5%	68.7%	32.5%	<0.001	4.9%	4.4%	2.4%	0.489
Wildenthal et al., 2022 (6)	Retrospective cohort	2016–2021	122 (53.7)	36 (15.9)	69 (30.4)	100% S. aureus	31.1%	53.8%	26.1%	0.02	10.7%‡	44.4%‡	13.0%‡	<0.001

^* Not applicable as not an included treatment group in study

^† Not reported

^‡ Composite of microbiologic failure or death

IV = Intravenous; PO = Oral; PWID = People who inject drugs; S. aureus = Staphylococcus aureus

Table 3:

Clinical trial of oral versus intravenous antibiotic treatment in people who inject drugs with infective endocarditis (IE)

Reference	Patients, no. (%)		IE patients no. (%)	Microbiology	Mortality			Cure†			Toxicity‡
	IV	PO			IV	PO	p-value	IV	PO	p-value	IV	PO	p-value
Heldman et al, 1996 (10)	45 (52.9)	40 (47.1)	85 (100)	93.5% S. aureus	2.1%	0%	NR*	91%	90%	0.9	61.5%	2.8%	<0.001

^* Not reported

^† Defined as sterile blood cultures days 6 and 7, and 34 and 35 after cessation of antibiotics

^‡ Defined as hepatotoxicity (rise in serum transaminases from normal to <150 IU L⁻¹ or from an abnormal baseline to a threefold increase associated with at least 7 days of study drug treatment) or nephrotoxicity (50% reduction in calculated creatinine clearance using Cockcroft and Gault formula)

Observational studies

The three observational studies included one prospective and two retrospective study designs. All the observational studies included hospitalized-only patients with evidence of active or recent injection drug use and serious injection-related infection. All studies defined serious infections as: Staphylococcus bacteremia, infective endocarditis, septic arthritis, epidural abscess, or osteomyelitis as diagnosed by an infectious diseases specialist. No study described the criteria for a diagnosis of IE. The range of included patients with IE was 41.0%–64.7%, but outcomes specific to IE were not separately reported. Table 2 shows the distribution of injection-related infections included in the studies. All studies included an IV treatment group according to usual care, as well as a treatment group consisting of partial IV and partial PO therapy. Both retrospective studies also included a partial IV treatment group where patients did not complete the recommended antibiotic treatment and did not receive any other antibiotic treatment. The prospective study provided a multidisciplinary Bridge Program to patients who remained in hospital or were discharged with oral antibiotics aimed to improve health education, medication adherence, and navigation through post-discharge appointments (9).

Table 2:

Differential rates of serious injection-related infection in observational studies

Reference	IE, no. (%)			Osteomyelitis n (%)			Septic arthritis, no. (%)			Epidural abscess n (%)
	Full IV	Partial IV no PO	Partial IV partial PO	Full IV	Partial IV no PO	Partial IV partial PO	Full IV	Partial IV no PO	Partial IV partial PO	Full IV	Partial IV no PO	Partial IV partial PO
Lewis et al., 2022 (9)	32 (52.5)	NA*	36 (34.3)	17 (27.9)	NA*	39 (37.1)	7 (11.5)	NA*	25 (23.8)	6 (9.8)	NA*	9 (8.6)
Marks et al., 2020 (4)	97 (67.8)	38 (56.7)	31 (37.3)	32 (22.4)	23 (34.3)	41 (49.4)	14 (9.8)	16 (23.9)	16 (19.3)	15 (10.5)	11 (16.4)	8 (9.6)
Wildenthal et al., 2022 (6)	90 (73.8)	22 (61.1)	36 (52.1)	18 (14.8)	9 (25.0)	16 (23.2)	25 (20.5)	10 (27.8)	20 (29.0)	16 (13.1)	5 (11.6)	14 (20.3)

^* Not applicable as not an included treatment group in study

IV = Intravenous; PO = Oral

All-cause 90-day readmission rates were comparable, reported as 27.9%–31.5% in the full IV treatment groups, 53.8%–68.7% in the partial IV, no PO treatment group, and 24.8%–32.5% in the partial IV, partial PO treatment group. All studies found the difference between all-cause 90-day readmission rates to be non-significantly different between the full IV treatment group and partial IV, partial PO treatment group (4,6,9).

The 90-day mortality, reported singularly or as part of a composite outcome, was not found to be significantly different between full IV treatment and partial IV, partial PO treatment groups for any of the observational studies (4,6,9).

The three observational studies had similar risk of bias as shown in Table 4. All studies scored appropriately on cohort selection but within comparability domains, showed a potential higher risk of bias. The included observational studies did not all describe important confounders such as the antibiotic regimes. Additionally, the distribution of important confounders, such as the proportion of PWID with IE, methods of source control, and addiction services consulted, were varied. All studies lost stars in the outcome domain due to a substantial amount of participants lost to follow-up or no description of those lost.

Table 4:

Risk of bias assessment using Newcastle–Ottawa scale for non-randomized studies

Reference	Design	Selection	Comparability	Outcome	Total
Lewis et al., 2022 (9)	Prospective cohort	****	*	**	7/9
Marks et al., 2020 (4)	Retrospective cohort	****	*	**	7/9
Wildenthal et al., 2022 (6)	Retrospective cohort	****	*	**	7/9

The category of selection of study groups is out of 4 possible stars, comparability of groups is out of 2 possible stars, and ascertainment of outcome is out of 3 possible stars.

Clinical trial

One randomized clinical trial was identified, published in 1996. Heldman et al. (10) investigated 4-week treatment regimens comparing PO versus IV for PWID with right-sided staphylococcal endocarditis, without metastatic infection apart from septic pulmonary emboli. The IV treatment group received either oxacillin 2 g every 4 hours or vancomycin 1 g every 12 hours (each with gentamicin 2 mg/kg every 8 hours for the first 5 days), while the PO treatment group received ciprofloxacin 750 mg twice daily plus rifampin 300 mg t Treatment group randomization occurred prior to any antibiotic treatment start; therefore, the PO groups did not receive an initial IV regimen. Patients completed the 28-day treatment as inpatients only.

Over the 1-month follow-up period, there were no reported deaths in the PO treatment group and one death in the IV treatment group related to an acute cerebral hemorrhage. Cure rates were not significantly different with 91% in the IV group and 90% in the PO group. Hepatotoxicity or renal toxicity were significantly higher in the IV treatment group (Table 3).

The risk of bias for the single clinical trial is shown in Table 5. There was low risk of bias based on random sequence generation, outcome data, and other biases. High risk of bias was found due to the lack of allocation concealment and any participant or personnel blinding.

Table 5:

Risk of bias assessment using Cochrane Collaboration modified tool for randomized controlled trials

Reference	Random sequence generation	Allocation concealment	Selective reporting	Blinding of participants and personnel	Blinding of outcome assessment	Incomplete outcome data	Other bias
Heldman et al., 1996 (10)	+	-	?	-	-	+	+

+ indicates low risk of bias, - indicates high risk, ? indicates unclear risk

Discussion

Observational evidence suggests that there is no difference between an exclusive full IV treatment strategy and a combined strategy of partial IV, partial PO treatment with respect to 90-day readmission and 90-day mortality. The limited evidence may also suggest that PWID discharged before clearance of bacteremia or those who received partial IV and no PO therapy have worse outcomes. The observational studies were small (n = 166–293) and mortality was only reported over 90 days of follow-up. Similarly, the clinical trial evidence showed no difference between death and cure rates between IV and PO treatment. The clinical trial included only right-sided IE, was also small (n = 85), and mortality was only reported over 30 days of follow-up.

A significant limitation of all of the observational studies is the inclusion of patients with a variety of infectious syndromes, with no analysis of outcomes for IE specifically. With inconsistent rates of IE within the included studies and lack of results specific to this infection, a bias toward the null may arise; therefore, only limited conclusions can be made. Additionally, the research question is highly specific and resulted in a small study yield. Due to the lack of available comparable studies, a quantitative meta-analysis could not be performed. Our systematic review cannot make treatment recommendations based on the existing evidence.

A recent randomized controlled trial comparing partial PO versus full IV antibiotic treatment (POET) of IE found that transitioning to PO treatment after a minimum of 10 days of IV treatment was non-inferior to continued IV treatment after 6-month follow-up (11). Of note, PWID comprised only 1.3% of the IE population studied in the POET trial. Similarly, previous reviews of the efficacy of partial PO antibiotic treatment in the IE population have found limited yet supportive evidence when used in combination with close multidisciplinary outpatient monitoring and follow-up (3).

It is important to distinguish outcomes for PWID specifically, as this population differs from the general population. PWID are often younger with fewer chronic medical comorbidities, yet have more mental health disorders and chronic viral infections (5,12). Additionally, PWID are at risk for bacteremia and disproportionally report higher rates of poverty and homelessness (12). These differences result in unique medical and social needs that require attention by the care team (5,12–14). Therefore, outcomes specific to this population need to be independently studied.

IE is of primary concern as a serious injection-related infection specifically among PWID. Due to significant morbidity and mortality and the unique issues that arise among PWID, specialized interdisciplinary care teams are recommended to address the complex issues inherent to PWID with IE. It is well documented that the rate of injection drug use is increasing, as is the incidence of IE (15,16). The population of PWID, are at a 20-fold higher risk of bacterial endocarditis and the long-term survival is poor (17,18). Hospital stays for PWID with IE may include intensive care and surgical intervention, and regardless of intervention, often require prolonged in-patient stays for multidisciplinary team management. A simulated multidisciplinary model of antibiotic therapy for PWID with IE found that partial PO antibiotic therapy was associated with higher rates of treatment completion, less death, and was less expensive than standard of care IV antibiotic therapy (18). At a minimum, multidisciplinary care teams should include access to Infectious Diseases specialists, Internal Medicine consultants, Addiction Medicine specialists, Pharmacists, Social Workers, Cardiac Surgeons, and Intensivists (18). This multidisciplinary care should extend to the outpatient setting, with health care visits two to three times per week, as in the POET trial (11). Extensive outpatient support, such as the Bridge Program implemented by Lewis et al., may be the key to PO antibiotic therapy success (9).

Antibiotic treatment strategies for IE must be individualized, with clinical, microbiological, and pharmacological factors balanced with the patient's medical and social needs. Regardless of the antibiotic delivery method, the treatment must be directed at a susceptible pathogen, allow sufficient bioavailability and distribution, and be given for the appropriate duration (3,19). PO antibiotic agents including select beta-lactams, fluoroquinolones, and tetracyclines have favorable pharmacokinetic profiles with promising clinical results (3,19). However, many questions remain, including when to step down to PO treatment, PO treatment dosages, single or combination PO treatment, the impact of minimum inhibitory concentration on PO treatment efficacy, the role of PO treatment in prosthetic valve IE, and the timing of step down to PO treatment in relation to source control.

One treatment domain of the recruiting Staphylococcus aureus Network Adaptive Platform (SNAP) Trial will investigate the potential non-inferiority of an early switch to PO treatment in comparison to full IV treatment for S. aureus bacteremia including IE with PWID anticipated to be 30% of the study population (ClinicalTrials.gov Identifier: NCT05137119) (20). Randomization to early PO stepdown will be based on clinical stability at day 14 of treatment for patients with IE who have achieved source control and clearance of bacteremia by day 5. The primary endpoint is all-cause mortality at 90 days, and specific outcomes among PWID are premature discharge and 90-day hospital readmission. It is hoped that this trial will help to address remaining clinical questions in this treatment context.

Future studies investigating IE treatments need to increase inclusion of PWID or focus on outcomes in PWID exclusively. Ideally, PWID with IE would be randomized to partial IV, partial PO treatment or full IV treatment, similar to the POET trial (11). Outcomes of interest should include mortality, reinfection rates, treatment completion, and surgical intervention. Beyond the usual challenges of operating clinical trials, there are additional ethical considerations when conducting research with PWID, specifically regarding stigmatization and views that PWID cannot provide informed consent (21). We and others argue that paternalistic exclusion of PWID from research further perpetuates their vulnerability and prohibits access to evidence-based treatment (21).

Treatment of IE in PWID is becoming increasingly common and complicated. Evidence comparing the outcomes of partial PO to full IV antibiotic treatment for these patients is scarce, mostly originating from observational studies. Observational studies suggest that PO antibiotic therapy after initial IV treatment may have equivalent outcomes to full IV treatment alone. There is a significant need for stronger evidence, especially from well-designed clinical trials, in order to definitively provide treatment recommendations. If an equally effective PO treatment strategy can be implemented, there may be substantial clinical and social benefits for these patients, in addition to advantages for the health care system. Treatment strategy decisions in this complex population ultimately require a multidisciplinary team approach with a thoughtful consideration of the patient's social context along with outpatient support programs to optimize follow-up and address barriers to patient care.

Contributors:

Conceptualization, A Brown, D Webster, C Adams; Methodology, A Brown, HL Jefferson, C Adams; Validation, P Daley; Formal Analysis, A Brown, HL Jefferson, C Adams; Data Curation, A Brown, HL Jefferson; Writing – Original Draft, A Brown; Writing – Review & Editing, A Brown, HL Jefferson, P Daley, WDT Kent, D Webster, C Adams; Visualization, A Brown, D Webster, C Adams; Supervision, C Adams

Ethics Approval:

Ethics approval was not required for this article.

Informed Consent:

N/A

Registry and the Registration No. of the Study/Trial:

N/A

Funding:

No funding was received for this work.

Disclosures:

The authors have nothing to disclose.

Peer Review:

This manuscript has been peer reviewed.

Animal Studies:

N/A