The Mortality of Infective endocarditis with and without Surgery in Elderly (MoISE) Study

Abstract

Background

Infective endocarditis (IE) is increasingly affecting older patients. However, data on their management are sparse, and the benefits of surgery in this population are unclear.

Methods

We included patients with left-sided IE (LSIE) aged ≥ 80 years enrolled in a prospective endocarditis cohort managed in Aquitaine, France, from 2013 to 2020. Geriatric data were collected retrospectively to identify factors associated with the 1-year risk of death using Cox regression.

Results

We included 163 patients with LSIE (median age, 84 years; men, 59%; rate of prosthetic LSIE, 45%). Of the 105 (64%) patients with potential surgical indications, 38 (36%) underwent valve surgery: they were younger, more likely to be men with aortic involvement, and had a lower Charlson comorbidity index. Moreover, they had better functional status at admission (ie, the ability to walk unassisted and a higher median activities of daily living [ADL] score; n = 5/6 vs 3/6, P = .01). The 1-year mortality rate in LSIE patients without surgical indications was 28%; it was lower in those who were operated on compared with those who were not despite a surgical indication (16% vs 66%, P < .001). Impaired functional status at admission was strongly associated with mortality regardless of surgical status. In patients unable to walk unassisted or with an ADL score <4, there was no significant surgical benefit for 1-year mortality.

Conclusions

Surgery improves the prognosis of older patients with LSIE and good functional status. Surgical futility should be discussed in patients with altered autonomy. The endocarditis team should include a geriatric specialist.

Infective endocarditis (IE) is a rare and severe disease with a poor prognosis at any age [1]. With the population aging, patients aged ≥ 80 years are increasingly experiencing IE; currently, they account for 10%–20% of IE cases [1, 2]. These patients have specific epidemiological features, such as a higher likelihood of healthcare-associated and prosthetic valve IE, predominance of Enterococcus spp., and a wide range of clinical presentations [1, 3, 4]. IE causes a major decline in the general status of older patients, which makes diagnostic and therapeutic management more complex given the limited functional capacity of this population [5]. In addition to medical treatment, valve surgery is often needed because of hemodynamic, infectious, or embolic complications [6], but surgical management decisions in older patients are particularly complex. According to the 2015 European Society of Cardiology (ESC) guidelines for the management of IE, “age per se is not a contraindication to surgery.” In practice, older patients with IE undergo surgery less frequently than younger ones [2, 3, 7]. However, there are equivocal data on the surgical benefits in older patients, with some studies reporting survival similar to that for younger patients [3, 7, 8] and others poorer survival [2, 9]. Some authors also report a high mortality rate for IE in nonoperated older patients (40%–45% at 1 year) and suggest that this is due to the underuse of surgery in older patients [3, 7, 10–12]. Despite the crucial implications of the surgical decision, there is still no clear evidence on who should be operated on.

Given the lack of data for this population, the Mortality of Infective endocarditis with and without Surgery in Elderly (MoISE) study sought to describe the characteristics and prognosis of left-sided IE (LSIE) in patients aged ≥ 80 years. We hypothesized that functional status at admission and surgery are major prognostic factors in this population.

METHODS

Study Population

From January 2013 to December 2020, a multicentric, prospective, observational cohort study of IE was conducted in 10 hospitals in southwest France, including the Bordeaux University Hospital, which is an expert center for endocarditis management with an endocarditis team and cardiac surgery expertise [13].

All consecutive patients aged ≥ 80 years admitted with a possible or definite diagnosis of IE according to the modified Duke criteria based on the 2015 ESC guidelines (as confirmed by the endocarditis team) and treated accordingly were eligible [6]. Cases of isolated right-sided and cardiac device–related IE were excluded. As we studied the 1-year all-cause survival and mortality rates after the IE episode, we also excluded patients with IE who relapsed within 1 year of the first episode.

Data Collection

Data on the general characteristics of LSIE were collected prospectively at diagnosis and during hospitalization and included age, sex, IE predisposition, source of infection, Charlson comorbidity index (CCI; unadjusted for age) [14], and microbiological data. Multimodal imaging examinations included transthoracic and transesophageal echocardiography (TEE), whole-body (brain, chest, and abdomen) computed tomography (CT), and fluorine-18 fluorodeoxyglucose positron emission tomography/CT (¹⁸F-FDG PET/CT). The surgical status was classified as follows: operated on, not operated on despite a surgical indication, and no surgical indications. Patients who required only implantable cardiac device extraction were not considered to have been operated on. The surgical indications according to the 2015 ESC guidelines were acute heart failure, uncontrolled infection, and embolism prevention. The decision to operate was multidisciplinary (endocarditis team, patient's choice). After discharge, all patients were followed for 1 year.

For the MoISE study, we retrospectively collected more specific data on geriatric characteristics at admission from the electronic medical, nursing, and social records. These included daily nurse support before admission, number of usual medications, body mass index, and functional status (activities of daily living [ADL] score). The ADL score ranges from 0 to 6 (1 point each for independent bathing, dressing, toileting, transferring, continence, and feeding), with lower scores indicating poorer functional status [14, 15]. Data on antimicrobial therapy were collected, including the route of administration, adverse events, and use of suppressive treatment. Data on complications that occurred during the hospital stay were extracted, including surgical revision, acute kidney injury (AKI; serum creatinine increased by ≥26.5 µmol/L or ≥1.5-fold relative to baseline), imaging-confirmed stroke, delirium, falls, and pressure ulcers. Data on the ability to walk unassisted and daily nurse support at discharge were also collected.

Statistical Analyses

We analyzed the baseline general and geriatric characteristics of the enrolled patients, management of IE, and prognosis according to surgical indications and functional status.

Categorical variables are presented as numbers and percentages, and continuous variables are presented as median with interquartile range. Bivariate comparisons were made using the χ² test, Fisher test (categorical variables), or Wilcoxon test (continuous variables). One-year survival was defined as the time between hospital admission and all-cause death during the year following IE. Patients alive after 1 year were censored. The Kaplan–Meier method was used to plot survival curves.

An initial analysis sought to identify general and geriatric characteristics associated with mortality. Univariate Cox regression analyses of the entire population were used to estimate the hazard ratios (HRs) of all-cause death and the 95% confidence intervals. Multivariate analysis was used to estimate HRs for geriatric factors, with adjustment for surgical status and general factors significantly associated with mortality in univariate regression analyses. Univariate regression was also conducted for subgroups based on surgical status.

A second analysis investigated the interaction between functional status (ADL score) at admission and the outcome of surgery when indicated. We estimated the unadjusted and adjusted HRs for all-cause death of the performance of surgery in subgroups defined according to the ADL score (0–6) and the ability to walk unassisted. To account for confounding factors, we adjusted the Cox model for general characteristics significantly associated with mortality in the first analysis. Sensitivity analyses were conducted using a propensity score that included predefined variables that influence both surgical decision and survival: age, sex, CCI, healthcare-associated, valvular localization, prosthetic valve, causative pathogen, presence of vegetation, and cerebral embolism. Cox models with inverse probability of treatment weighting and adjusted on quartile of propensity score were used.

A 2-sided P value < .05 was considered statistically significant for the Wald test. Schœnfeld residual tests were used to verify that the proportional hazards assumption was met, and curve analysis was performed for variables of interest. The linearity of continuous variables was verified using cubic splines. No imputation method was applied for missing data. All analyses were conducted using R Studio (ver. 1.1.463; R Development Core Team, Vienna, Austria) and the survival, survey, and cobalt R packages.

Ethics Approval and Consent

Patients were informed about the study and the anonymous use of their medical data for clinical research but did not have to provide individual consent in accordance with French legal standards. The French Data Protection Authority approved the study, and the study was conducted in accordance with the ethical standards of the Declaration of Helsinki.

RESULTS

Of 172 patients with IE aged >80 years, the MoISE study included 163 (Figure 1).

Baseline Characteristics

Table 1 summarizes the patient characteristics according to surgical indications and performance. Nonoperated patients were older, more likely to be women, and had more comorbidities than those who underwent surgery. There was less mitral involvement in the operated patients, but no significant difference in the rate of prosthetic valve infection or cerebral embolism or in the causative pathogen (Table 1). Table 2 describes the geriatric characteristics according to surgical status. Operated patients were less likely to be visited by a nurse at home before admission (2.6% vs 17%, P = .02), were twice as likely to be able to walk unassisted at admission (74% vs 35%, P < .001), and had a higher median ADL score (5/6 vs 3/6, P = .001) than those not operated on.

Medical and Surgical Management

Supplementary Table 1 describes the medical and surgical management procedures. Most patients were operated on electively (>7 days; n = 31 of 38, 81.2%). Moreover, 67 patients did not undergo surgery despite theoretical indications (63.8%), mostly because their general condition was deemed incompatible with postoperative resuscitation (n = 57 of 67, 85.1%). Other reasons included patient refusal (n = 6), neurological contraindications (n = 1), death before surgery (n = 1), and nonrecognition of a contraindication to embolism (n = 2). AKI occurred in 35.0% (n = 57) of patients and was more frequent in those with surgical indications (Supplementary Table 1). One in 5 patients had delirium while in the hospital (Supplementary Table 1).

Functional Prognosis and Mortality

Mortality was significantly higher at 1, 3, 6, and 12 months in nonoperated patients with surgical indications compared with those operated on or those without surgical indications. The median overall survival was 50 days. The mortality rate at 12 months was 28% in patients without surgical indications, 16% in operated LSIE patients, and 66% in nonoperated patients with indications (Table 2).

Table 2 summarizes the discharge characteristics of living patients. More at-home nurse support than before admission was needed for only 3 of 34 (8.8%) patients who were operated on vs 17 of 28 (60.7%) patients who were not operated on despite surgical indications and 19 of 45 (42.2%) patients without surgical indications.

Factors Associated With Mortality

Figure 2 presents the patient survival data according to surgical and functional status. The general characteristics associated with all-cause mortality are shown in Supplementary Table 2. Aging, female sex, high comorbidity burden, and cerebral embolism were associated with a higher risk of death.

Table 3 presents the geriatric characteristics associated with all-cause mortality. Impaired functional status at admission was strongly associated with an increased risk of death, regardless of age, sex, CCI, cerebral embolism, or surgical status. The occurrence of AKI, delirium, and pressure ulcers during hospitalization was significantly associated with a higher risk of death, but only delirium remained significant in the multivariate analysis. Supplementary Table 3 presents the results of subgroup analysis according to surgical status.

In patients with surgical indications, there were significant interactions between performing surgery and the ADL score at admission (P = .01) and the ability to walk unassisted (P = .04). Figure 3 and Supplementary Figure 1 show the Kaplan–Meier 1-year survival curves according to functional status and whether or not surgery was performed. For patients unable to walk unassisted or with an ADL score <4, surgery did not significantly reduce the 1-year mortality rate in crude and adjusted Cox models and in sensitivity analysis (Supplementary Table 4).

DISCUSSION

In a heterogeneous population of older patients with LSIE, we confirmed our hypothesis that functional status at admission and cardiac surgery were major interacting prognostic factors. Indeed, operated patients had better functional status at diagnosis. The 1-year prognosis of the selected older patients with LSIE who underwent surgery was good (mortality, 16%), while it was extremely poor in nonoperated patients, especially when surgery was theoretically indicated (mortality 66%). The surgery significantly improved survival in patients with good functional status defined as an ADL score ≥4. In those unable to walk or with ADL score <4, there was no significant survival improvement.

In the MoISE study, the surgical rate (23%) in older patients is consistent with some hospital-based studies but is more than twice as high as in recent Scandinavian and Japanese national registry studies (<10%) [2, 16, 17]. A high comorbidity burden could partly explain the low rate of surgery in older patients. Indeed, almost 60% of older patients in our study did not undergo surgery despite theoretical indications. Moreover, some indications may be missed in older patients because of difficulty in performing TEE, although essential to establish the surgical indications, because of confusion or contraindication to general anesthesia. The TEE rate is not well reported in the literature but is surely lower than in younger patients [3, 7, 10, 12]. In the MoISE study, we found a low rate of TEE (77%); among 38 patients who did not undergo TEE, 36 (95%) were not operated on.

According to observational studies, good functional status before admission for LSIE is important with respect to the decision to perform surgery. In the French Elderl-IE study, which prospectively enrolled 120 IE patients aged >75 years, the estimated ADL score before admission was better in operated patients. Our results showed that no nurse intervention at home was a proxy for good functional status before admission. Indeed, cardiac surgery is infeasible in older patients who are bedridden or very debilitated. Furthermore, Forestier et al and Nagai et al showed that the ADL score before IE developed had significantly decreased by admission, even in previously independent older patients [10, 18]. Early recognition and diagnosis of LSIE are needed in older patients to avoid functional decline due to IE and its complications.

One of our key findings concerns the major impact of functional status at admission on LSIE management and prognosis. In the Elderl-IE study, the ADL score at admission for IE was significantly associated with mortality [10], as was the case in the MoISE study. Functional status at the time of admission might be better for estimating prognosis in older patients and therefore for assessing the expected benefits of surgery. Interestingly, our study provides the first data regarding which patients are most likely to benefit from surgery. Surgery for LSIE should probably not be performed in patients unable to walk unassisted at admission and in patients with an ADL score <4, as we did not observe significant benefits in these cases despite immortal time bias. However, the power of our study may have been insufficient to reveal the benefits. Conversely, older patients able to walk unassisted or with an ADL score ≥4 should undergo surgery as their prognosis improves markedly. Obviously, the surgical decision requires a comprehensive geriatric assessment of each patient and cannot be based only on the functional status at admission. The Elderl-IE study emphasized that malnutrition at admission also predicted a poor prognosis. Thus, a geriatrician is needed on the endocarditis team to help with surgical decision-making.

The prognosis of LSIE in the MoISE study when surgery was performed is consistent with the 1-year mortality rate in operated older patients, which ranges from 6% to 39% [3, 11]. Of note, none of the 7 patients who were operated on for mitral endocarditis died during the 1-year follow-up, although this condition was associated with higher mortality in LSIE patients in recent Spanish and Danish studies [11, 17]. There was no association between the surgical indication and survival in operated older patients, suggesting that there is no prohibitive surgical indication in this population (Supplementary Table 2).

Nonperformance of surgery despite an indication led to an extremely poor prognosis, with a high initial mortality mostly due to refractory heart failure related to endocarditis and a 1-year mortality rate of 66%. Given the severely impaired functional status of those patients, it is unlikely that surgery would improve their prognosis.

In our study, the older patients without surgical indications had a slightly poorer baseline general condition than operated patients, leading to an intermediate prognosis (1-year mortality rate of 28%). We cannot exclude that we missed surgical indications because of the low TEE rate in these patients (74%). Three published studies reported mortality rates of 15%–46% in older patients without surgical indications, which is consistent with our results [3, 7, 10]. These LSIE cases that were not operated on are a less aggressive form of endocarditis that is not associated with major cardiac damage or large vegetations and are more easily managed. PET/CT was more often used in these patients (64%) without typical echocardiographic abnormalities in order to establish the diagnosis of IE. Surgery is not expected to improve the prognosis a priori, and management of functional decline is a major prognostic factor in these patients.

Functional status in older patients is severely impaired after discharge for LSIE [10]. Indeed, among patients who were not operated on and did not die during their hospital stay for LSIE, 12% were institutionalized at discharge and 37% returned home but needed more nursing support. Conversely, 91% of the selected older patients who underwent surgery returned home without more nursing support. In previously independent older patients, LSIE can lead to dependency due to cardiac or infectious complications, as well as malnutrition, delirium, falls, pressure ulcers, and other common complications of a long hospital stay. We demonstrated that these complications are associated with higher mortality. Thus, these complications need to be prevented in order to improve the prognosis of IE in older patients, which would be aided by the integration of geriatricians into the endocarditis team [19].

The MoISE study includes 163 patients aged ≥80 years; it is one of the largest studies of older patients with LSIE and the largest French study of this group [3, 10]. However, only 38 patients were operated on, which might have led to low power of the analysis. We attempted to comprehensively analyze a selection bias for the performance of surgery due to the major imbalance in the baseline characteristics between the patients who underwent surgery and the patients who did not. Because the surgical indications did not change during the study, the time-dependent indication bias was low. Although most of the data for the LSIE patients were collected prospectively, we acknowledge the inherent limitations of retrospective data collection (eg, ADL score). However, the medical and nursing records were comprehensive and the missing data rate was low, making our results consistent with those of the prospective study of Forestier et al [10].

CONCLUSIONS

Cardiac surgery can be performed safely in patients aged >80 years with good functional status (as determined by an ADL score ≥4 on admission) considering other prognostic factors, including age, comorbidities, and cerebral embolism. LSIE leads to dependency after hospital discharge in half of all older patients, which should be considered when evaluating them. The integration of geriatricians into the endocarditis team would facilitate the complex management of these patients and the decision to perform surgery and, ultimately, would improve the prognosis of older patients with LSIE.

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.

Notes

Author Contributions. F. C. and N. I. coordinated the cohort and the prospective data collection. V. H., F. C., C. R., and N. I. initiated the study. V. H. collected retrospective data and conducted statistical analyses. V. H., F. C., C. R., and N. I. drafted the manuscript. All authors reviewed and edited the drafted manuscript and approved the final manuscript.

Acknowledgments. We thank each member of the Mortality of Infective endocarditis with and without Surgery in Elderly study group that included patients in this study from the Bordeaux Endocarditis Team: G. Tlili, Angouleme hospital; C. Ngo Bell, Arcachon hospital; A. Barret and M. Videcoq, Bayonne hospital; C. Alexandrino, M.O. Vareil, and H. Wille, Dax hospital; K. André and E. Nyamankolly, Langon hospital; D. Girard and M. Ducours, Libourne hospital; O. Caubet and H. Ferrand, Mont de Marsan hospital; F. Lacassin, Pau hospital; W. Picard and C. Pavin, Perigueux hospital; and B. Castan, C. Aguilar.

Disclaimer. The English in this document has been checked by at least 2 professional editors, both native speakers of English. For a certificate, please see:

http://www.textcheck.com/certificate/pYn3I8.

Data availability. The datasets used during the study are available from the corresponding author on reasonable request.

References

Author notes