Multidisciplinary Teams for the Management of Infective Endocarditis: A Systematic Review and Meta-analysis

Abstract Background The management of infective endocarditis (IE) is complex owing to a high burden of morbidity and mortality. Recent guidelines recommend dedicated multidisciplinary teams (MDTs) for the management of IE. The aim of this systematic review and meta-analysis was to evaluate and summarize the effect of MDT management on patient outcomes. Methods A systematic review was performed and, where feasible, results were meta-analyzed; otherwise, results were summarized narratively. Data extraction and quality assessment were performed in duplicate. Restricted maximum likelihood random effects models were used to calculate unadjusted risk ratios and 95% CIs. Results Screening of 2343 studies based on title and abstract yielded 60 full-text reviews; 18 studies were summarized narratively, of which 15 were included in a meta-analysis of short-term mortality. Meta-analysis resulted in a risk ratio of 0.61 (95% CI, .47–.78; I2 = 62%) for mortality in favor of a dedicated MDT as compared with usual care. Length of stay was variable, with 55% (10/18) of studies reporting an increased length of stay. Most studies (16/18, 88.9%) reported a decreased time to surgery and an increased rate of surgery (13/18, 73%). No studies reported on patient-reported outcomes. Conclusions This is the first systematic review and meta-analysis to assess the impact of MDT management on IE. The sum of evidence demonstrated a significant association between MDTs and improved short-term mortality. Further research is needed to evaluate benefits of virtual MDT care, cost-effectiveness, and the impact on patient-reported outcomes and long-term mortality.

Despite advances in diagnosis and management, infective endocarditis (IE) has an associated mortality rate exceeding 25% [1].Notwithstanding significant medical advancements, over the past few decades the incidence and mortality of IE have paradoxically increased in many countries [2], likely due to changing epidemiology.For instance, while rates of rheumatic heart disease in industrialized countries have declined, cases in older adults have increased because of implantable cardiovascular devices and semipermanent intravenous catheters.Furthermore, the opioid crisis has resulted in an increase in IE among people who inject drugs: a population that faces challenges with respect to timely diagnosis, appropriate treatment, adherence, and adequate follow-up-all exacerbated by the high burden of stigma [3] and ongoing risk of opiate-associated death.
In addition, the management of IE has always been complex given its propensity to cause cardiac, vascular, immunologic, neurologic, and renal complications [4].Other factors, including delays in diagnosis and/or surgical management, may further increase the risk of mortality in patients with IE [5].The subtype of IE (native vs prosthetic valve or implantable cardiac device endocarditis) may require access to specialized imaging modalities and expert interpretation for accurate diagnosis.Finally, surgical management of endocarditis and any associated surgical complications often require specialized perioperative and surgical care.
Recognizing this interdisciplinary complexity, recent guidelines have proposed that a multidisciplinary approach be employed to the diagnosis and management of IE to achieve Multidisciplinary Teams for the Management of Infective Endocarditis • OFID • 1 Open Forum Infectious Diseases R E V I E W A R T I C L E optimal outcomes.For example, the European Society of Cardiology's 2015 guidelines strongly encouraged IE management by a specialized team in reference centers (implying that patients diagnosed in community hospitals or less resourced centers might need to be transferred) based on "the weight of evidence in favor of efficacy" [6].Similarly, the 2020 American College of Cardiology/American Heart Association guidelines recommended consultation with a multispecialty heart valve team as a strong recommendation based on moderate quality evidence (nonrandomized) [7].Whether such a consultation could be virtual was not addressed.
Indeed, several published case series and observational studies have reported improved patient outcomes with the involvement of subspecialty expertise [8][9][10][11].The aim of this narrative review and meta-analysis was to evaluate and, when feasible, combine the totality of the evidence to determine the overall impact of multidisciplinary endocarditis teams with an emphasis on mortality.

Data Sources and Search Strategy
We followed the MOOSE criteria (Meta-analysis of Observational Studies in Epidemiology) [12] and PRISMA guidelines (Preferred Reporting Items for Systematic Reviews and Meta-analyses) [13].To identify potentially relevant studies for inclusion, we searched PubMed, CINAHL, and EMBASE for articles discussing multidisciplinary endocarditis management.Database search strategies are provided in the supplementary materials (Supplementary Tables 1 and 2).Results were extracted from database inception to 1 July 2022.Of 1889 initial results, there were 1465 nonduplicate records.Using the Rayyan platform [14], 2 independent blinded reviewers (H.H.-D. and A.-S. R.) screened articles for relevance by title and abstract.Conflicts between the reviewers were discussed and resolved through consensus.

Study Selection
We included studies assessing the impact of a multidisciplinary management team (primary exposure) in hospitalized adult patients diagnosed with endocarditis and reporting at least 1 of the following outcomes: short-term mortality (in hospital and/or up to 30 days postdischarge-primary outcome), longer-term mortality (1 year), morbidity (renal failure, cardiac or neurologic complications, length of hospital stay, readmission), adherence to treatment, patient satisfaction, and surgical outcomes (overall rate of surgery, time to surgery).Studies limited to pediatric patients, case reports, and publications available only as poster abstracts were excluded.

Data Extraction and Quality Assessment
Data extraction was performed in duplicate and included author, year of publication, years of data collection, location (country and type of facility), study design, population, sample size, comparator group, details of the intervention (team members, standardized protocol, and other characteristics), as well as outcomes (Table 1).Studies in the meta-analysis of shortterm mortality were assessed for quality independently and in duplicate (H.H.-D. and A.-S. R.) with the Newcastle-Ottawa Scale [15].We considered studies with 7 to 9 stars as high quality, 4 to 6 as fair, and 0 to 3 as poor.

Data Synthesis and Analysis
We used restricted maximum likelihood random effects models to calculate meta-analytic unadjusted risk ratios and 95% CIs.Heterogeneity was assessed with the I 2 statistic.A 2-sided P < .05 was considered statistically significant.All analyses were performed with the meta module in Stata version 16.1 (StataCorp LP).

Study Population
In total, 3993 episodes of IE were included: 2305 patients from preintervention/control groups, 1654 from postintervention cohorts, and 34 cases where a patient had >1 episode of IE.Sample sizes varied greatly, ranging from 6 to 645 patients.Across all studies, the mean age was 41.4 years, and 29.7% of patients were female.With respect to study type, 8 of 18 (44%) were retrospective; 5 (28%) were prospective; 4 (22%) were mixed (prospective interventional data collection with retrospective control period data); and 1 (6%) was a casecontrol study.All studies were conducted in single centers, most taking place at a tertiary care facility or reference center for IE; 15 studies (83%) explicitly mentioned this.Study settings were scattered across 12 countries, with all but 1 (94%) being high income.Most studies were conducted after 2010.
One study [33] was excluded from the meta-analysis due to the implementation of an endocarditis cardiac team during the control phase (with an interventional cardiologist, an imaging cardiologist, and a cardiothoracic surgeon), which we judged too similar to the overall intervention.
The meta-analysis revealed an overall risk ratio of 0.61 (95% CI, .47-.78) for patients treated with a dedicated multidisciplinary endocarditis team when compared with control groups (Figure 2).The I 2 value was 62%, indicating moderate Multidisciplinary Teams for the Management of Infective Endocarditis • OFID • 5 heterogeneity.Of the 15 studies assessed for quality per the Newcastle-Ottawa Scale, 12 (80%) were considered high quality and 3 (20%) fair by both reviewers (Supplementary Table 3).

Additional Analyses to Address Confounding Factors
Though not amenable to meta-analysis, several of the 15 studies that reported on in-hospital mortality data performed additional analyses that attempted to adjust for confounding (Table 3).Most studies (11/15, 73%) performed a multivariate analysis examining factors associated with mortality.Of these, 8 (73%) demonstrated that multidisciplinary team (MDT) treatment was independently associated with reduced mortality.Three studies [17,19,20] included calendar year as a term in the regression model to limit confounding by other modifications in IE care that may have occurred over time.Each observed that MDT management was still associated with reduced mortality after adjusting for the calendar year.Two studies [22,32] also performed an analysis with propensity score matching between cohorts; both revealed a reduction in mortality with MDT management.

Descriptive Outcomes
With regard to the composition of teams, all IE teams included cardiac surgery.All but 1 (94%) [28] included cardiology and infectious diseases.Other common specialties were neurology, echocardiography/radiology, and microbiology.Most teams (10/18, 56%) met on a regular basis, which varied from weekly to monthly.Two teams (11%) met on an ad hoc basis.Six studies (33%) reported that patients received a multidisciplinary assessment but did not explicitly define any meeting frequency.

Other Outcomes Postimplementation
Most other outcomes were reported inconsistently among studies (not amenable to meta-analysis) and so are described narratively (Table 2).No study adequately adjusted for the competing risk of death in evaluating any of the durations.Regarding length of stay, results were mixed, as was reporting (means vs medians).Eleven studies (61%) cited lengths of stay that varied from 13.5 to 42.1 days.Of the 11 studies that reported this outcome, 6 (55%) indicated an increased length of stay and 5 (45%) identified a decrease, but only 4 (36%) were statistically significant.As for surgical outcomes, time to surgery was noted in 8 studies.A majority (7/8, 88%) revealed a decrease in the time to surgery.Rate of surgery was reported in 11 studies, of which 8 (73%) noted an increased rate.Only 3 studies reported on adherence to treatment, all of which (100%) cited increased adherence to appropriate antimicrobial therapy.
No studies reported on quality of life or patient satisfaction.

DISCUSSION
To our knowledge, this is the first systematic review and metaanalysis to assess the impact of multidisciplinary endocarditis team management.While limited to single-center observational studies, the data suggest that IE team implementation is associated with decreased short-term mortality, which, if correct, is clinically important.We agree with authors who have previously hypothesized that the mortality reduction achieved by MDTs is likely multifactorial.Potential mechanisms were as follows: earlier detection, diagnosis (including reduced time to echocardiography), and initiation of treatment; decreased time to surgery; earlier and more frequent assessments by infectious diseases consultants; improved adherence; and more appropriate treatments (antimicrobial agent, dose, duration).We noted variability in the length of stay across studies preand postintervention.This is consistent with Bikdeli et al [34], who examined trends in length of stay in a large cohort of hospitalized patients with IE and found it to be highly variable.When length of stay is evaluated, adjustment for the competing risk of death will be very important.Multidisciplinary teams could increase length of stay for various reasons, such as delayed discharge due to more investigations, a higher rate of surgery, or, paradoxically, higher survival.Yet, a decreased length of stay might be possible if the MDT model renders the process of hospital discharge more efficient.Consequently, length of stay may not be a suitable outcome to demonstrate the success of IE teams.
Increased rates of surgery and decreased time to surgery were noted in many studies reporting these outcomes.This is in line with Regunath et al [35], where they deployed a series of quality improvement tools to improve multidisciplinary IE care and identified that the decision to operate was a leverage point to improve care.Strategies include early identification of patients at high risk to quickly escalate the evaluation for surgical management.As "any solution for this leverage point will mandate collaboration of all of the essential subspecialty services" [35], one can hypothesize that reduced time to surgery could be achieved with earlier identification of urgent cases and a facilitated, prompt, direct communication among involved experts.Nonetheless, since not all studies reported on this outcome, it is subject to a reporting bias.
Our review had strengths, such as its systematic approach and large number of identified patients.It also aligns with and provides stronger supporting data for the specialty guidelines [6], which suggest that patients with IE should be managed by specialized endocarditis teams that meet on a regular basis and select therapies based on the best available evidence.
There are limitations to this review worth discussing in some detail, many of which are inherent to the design and outcome measurements in the studies.For instance, most studies were designed as pre-and postintervention groups, which are not ideal to meta-analyze because they are not contemporaneous and thus are subject to bias due to temporal changes in outcomes that confound the relationship between outcome and intervention.Mortality reductions might therefore reflect general improvements in medical knowledge and care arising during the elapsed time between pre-and postintervention measurements rather than the advantage of the MDT management itself.Along these lines, studies performed prior to 2014 contributed substantially to the observed overall reduction in mortality.However, it should be noted that mortality from IE has been increasing over time [2].As such, improvements seen postintervention would go against the general time trend of an observed increase in mortality.As mentioned, while several studies did perform multivariate analyses to adjust for potential confounders, only 3 [17,19,20] included calendar year as a term in their models to limit this form of bias.In these 3 studies, MDT management remained beneficial.
There are other limitations worth mentioning.First, the observational nature of the studies risks residual confounding, confounding by indication, immortal time, temporal trends, selection bias, reporting bias, and other biases.As many of these centers were referral centers, it is also impossible to know whether there were systemic biases in who was even evaluated and accepted for transfer.Second, even the primary mortality outcome was subject to moderate heterogeneity.This could be due to several factors, such as the lack of a standardized intervention design and the diversity across the studies in context, location, and sample size.Third, meta-analysis of other outcomes was not feasible given the lack of consistent  reporting.Whereas mortality from IE is generally affected by many factors-an important one being the timing of surgery among patients with surgical indications-most studies did not report this outcome.Among the few studies that reported time to surgery, definitions were variable, making it impossible to meta-analyze the results.Fourth, we included all types of IE cases, but people who inject drugs and patients with cardiac-implanted devices have several specific diagnosis and management challenges [36].Fifth, studies were mostly conducted in high-income countries; therefore, the findings may not be generalizable to less-resourced settings.Studies did not evaluate cost-effectiveness; many were conducted in cardiac referral centers; and there was little discussion surrounding how to operationalize patient transfers or how to ensure equitable access to the required expertise in remote or underserved populations.Sixth, the overall proportion of female patients in our review was lower than the prevalence of IE among females in the literature, so this population was underrepresented in the studies [37].Last, apart from adherence to treatment, none of the studies examined the impact on any patient-reported outcomes.
While MDTs seem reasonable on face value and are supportable by this analysis, the answer to many of the methodological limitations of the observational evidence will be to perform randomized controlled trials.MDTs represent a substantial investment of health care resources and may require an increase in patient transfers; therefore, it is reasonable to conclusively demonstrate value to justify those investments.Additionally, publishing guidance on a standardized framework for MDT management could be helpful, especially if such guidance were to include meaningful outcome definitions to be used in future observational, quasi-experimental, and randomized studies.

CONCLUSION
This systematic review and meta-analysis demonstrated a significant association between team-based management and reduced short-term mortality.While this supports the current guideline recommendations, the resources required for universal deployment are substantial, and deployment will not be immediate or simultaneous.As such, a cluster randomized or stepped-wedge controlled trial could serve as a means of scalable deployment and an evidence-based approach to measuring the impact on important outcomes.

Table 1 . Characteristics of Studies
Multidisciplinary Teams for the Management of Infective Endocarditis • OFID • 3

Table 2 . Length of Stay and Time to Surgery for Patients With Infective Endocarditis
a Median (range).b Mean (standard deviation).

Table 3 . Additional Analyses Performed in Studies Examining the Effect of Multidisciplinary Team Implementation on Mortality
Multidisciplinary Teams for the Management of Infective Endocarditis • OFID • 9

Table 3 . Continued
Unadjusted and adjusted mortality after propensity score matching showed no difference in in-hospital mortality, but mortality at 6 mo and 1 y tended to be lower in patients managed by the MDT.Mortality at 6 mo: 23.5% in MDT cohort vs 36.8% in controls, P = .0926.At 1 y: 26.5% in MDT cohort vs 41.2% in controls, P = .0699… Abbreviations: IE, infective endocarditis; HR, hazard ratio; MDT, multidisciplinary team; OR, odds ratio.
a Data in parentheses indicated 95% CI.