Article Text
Abstract
Objective To evaluate the prognostic impact of urgent cardiac surgery on the prognosis of left-sided infective endocarditis (LSIE) and its relationship to the basal risk of the patient and to the surgical indication.
Methods 605 patients with LSIE and formal surgical indication were consecutively recruited between 2000 and 2020 among three tertiary centres: 405 underwent surgery during the active phase of the disease and 200 did not despite having indication. The prognostic impact of urgent surgery was evaluated by multivariable analysis and propensity score analysis. We studied the benefit of surgery according to baseline mortality risk defined by the ENDOVAL score and according to surgical indication.
Results Surgery is an independent predictor of survival in LSIE with surgical indication both by multivariable analysis (OR 0.260, 95% CI 0.162 to 0.416) and propensity score (mortality 40% vs 66%, p<0.001). Its greatest prognostic benefit is seen in patients at highest risk (predicted mortality 80%–100%: OR 0.08, 95% CI 0.021 to 0.299). The benefit of surgery is especially remarkable for uncontrolled infection indication (OR 0.385, 95% CI 0.194 to 0.765), even in combination with heart failure (OR 0.220, 95% CI 0.077 to 0.632).
Conclusions Surgery during active LSIE seems to significantly reduce in-hospital mortality. The higher the risk, the higher the improvement in outcome.
- infective endocarditis
- cardiac surgery
- mortality
- risk score
- ENDOVAL score
Data availability statement
All data relevant to the study are included in the article or uploaded as supplemental information. Not applicable.
Statistics from Altmetric.com
Introduction
Infective endocarditis (IE) is a severe disease with a grim prognosis and high in-hospital mortality. Cardiac surgery is required in above 50% of cases with left-sided IE (LSIE).1 Formal indications are well established and discussed on the guidelines; however, timing is an important issue with little evidence and a frequent matter of disagreement in the ‘endocarditis teams’.2 3 In addition, patients with LSIE bear high surgical risk quite frequently and decision about utility or futility of cardiac surgery is controversial.4–8 ,9 Age, nosocomial origin, prosthetic IE, periannular complications, acute renal failure, cardiogenic or septic shock and virulent microorganisms such as Staphylococcus aureus or fungi are well-known risk factors and all together have demonstrated a high performance for mortality risk prediction in the ENDOVAL score.3
There are scant data on the impact of cardiac surgery on the evolution of LSIE. Observational studies agree on the prognostic benefit of cardiac surgery,1 10–12 but some others do not13–15 and there is only one randomised study, which focused on a very specific population with low surgical risk and does not represent our everyday clinical practice.16 17 Postsurgical clinical complications are strong prognostic factors in LSIE probably suggesting a dramatic change on the prognosis of IE after surgery.18
We aimed to evaluate the prognostic impact of urgent cardiac surgery on the prognosis of LSIE and its relationship to the patients’ basal risk and to the surgical indication.
Methods
Study design
We evaluated the impact of urgent cardiac surgery on in-hospital mortality of LSIE in two different ways. First, by a multivariable analysis facing cardiac surgery against the prognostic factors proposed by the European IE guidelines.2 Second, by a propensity score matching of both operated and non-operated cohorts to evaluate the impact of cardiac surgery in comparable populations. In addition, we studied the prognostic impact of cardiac surgery according to baseline mortality risk of each patient defined by the ENDOVAL score and according to surgical indication. Figure 1 summarises the study design.
Setting
We included patients with IE in an ongoing multipurpose database from January 2000 to March 2020 among three tertiary centres in Spain.
An anonymous standardised case report form was fulfilled for each episode with information about patient background, per-week clinical evolution and therapeutic and diagnostic procedures during hospitalisation. The proportion of missing data was <10% in all analysed variables and the absence of significant differences between groups were verified.
Participants
Observational study with retrospective data analysis including all patients with a final diagnosis of definite LSIE. We selected patients with definite LSIE (n=1165) and excluded those without indication for surgery (n=277) or operated after the active phase of the disease (n=283). Our study population included 605 patients with LSIE and surgical indication: 405 underwent cardiac surgery during the active phase of the disease and 200 did not undergo cardiac surgery despite having formal indications (figures 1 and 2).
Definitions
Surgical indications followed the European guidelines thoroughly for the study period and decision was considered in the setting of experienced multidisciplinary teams in each participant centre. Endocarditis team meetings are taken on a weekly basis during infective course, and urgently (<24 hours), in case of acute decompensation. We divided indications according to the guidelines into three general groups: heart failure, uncontrolled infection and prevention of embolism2 (table 1).
Urgent cardiac surgery is considered surgery during active phase of the disease, which means under antibiotic treatment.19 For a better understanding of surgical urgency, we include the variable time between diagnosis and surgery and the time between heart team discussion and surgery.
ENDOVAL score is an in-hospital mortality risk estimator for LSIE based on European IE guidelines prognostic factors and irrespective of whether patients undergo cardiac surgery or not. It is accessible as an informatic application via Google Play Store (ENDOVAL score).3
Risk-E Score is a mortality risk score for patients with LSIE with surgical indication, which includes LSIE-specific factors (microorganisms, periannular complications and sepsis) besides classical variables in heart surgery (age, haemodynamic condition and renal failure), with better performance than other surgical scores.20 It is accessible at MDCal Medical Calculator (Google Play Store and App store) and at Risk-E Score (Google Play Store).
Prohibitive surgical risk was considered by endocarditis team taking into account the surgical risk, comorbidities, clinical status and possibilities of recovery. Until development of Risk-E Score, other scores were used to evaluate the mortality risk related to cardiac surgery, however no absolute value was considered prohibitive for any score.
We have analysed possible predisposing events as situations with theoretical risk of bacteraemia occurring within the previous 2 months of the beginning of the disease. Comorbidity was considered when at least one of the following conditions was present chronic kidney disease (creatinine clearance <60 mL/min), chronic pulmonary obstructive disease or immunosuppression.3
Patients were referred in case of the diagnosis was performed in another hospital, and were then transferred for specific management at one of our institutions.
Acute onset included all episodes with a time between symptom onset and diagnosis below 15 days.
Statistical methods
Categorical variables are reported as frequency (n) and percentages and continuous variables as mean value and SD. Normal distribution of quantitative variables was verified with the Kolmogorov-Smirnov test and visually through Q-Q plot graphics. Qualitative variables were compared with the χ2 test and Fisher’s exact test. Continuous variables were compared with Student’s t-test.
Multivariable analysis was performed by a logistic regression model with the maximum likelihood method using forward stepwise selection to prove the independent relationship of cardiac surgery and in-hospital mortality. The analysis includes cardiac surgery and the prognostic factors proposed by European Guidelines (age, nosocomial origin, prosthesis, comorbidities, S. aureus, Gram-negative bacilli, fungi, vegetation, periannular complications, left ventricle systolic dysfunction, heart failure, renal failure and septic shock). The ratio variable/event was controlled to avoid overfitting. ORs adjusted for each of the variables included along with their 95% CIs were calculated. Non-collinearity was checked among the variables. The area under the receiver operating characteristic curve (ROC curve) was used to measure the discriminatory capacity. Calibration was evaluated with the Hosmer-Lemeshow test and with plots. The errors and CIs were calculated using bootstrap samples.
The propensity matching used a logistic regression in which the dependent variable was an indicator of whether a patient had been or had not been operated. The independent variables were either those clinically relevant or those which showed significant differences in the comparative analysis (age, comorbidities, Streptococcus gallolyticus, viridans group streptococci, S. aureus, periannular complications, severe valvular dysfunction, renal failure, septic shock, ischaemic stroke and brain haemorrhage). Pairs of patients were derived using greedy nearest neighbour method 1:1 with 1/5 of the SD of the logit of the propensity score as calliper. The MatchIt package (Ho et al 21) was used. Qualitative variables were compared with the McNemar test and continuous variables were compared with paired sample t-test.
The statistical analyses were performed with the use of R software, V.3.6.1 (R Project for Statistical Computing) and IBM SPSS Statistics, V.26.0 (IBM, Armonk, New York, USA). Differences were considered statistically significant when p value was <0.05.
Patient and public involvement
Patients and/or the public were not involved in the design, conduct, reporting or dissemination plans of this research. The data of enrolled patients were extracted from available case files after approval from the institutional ethics committee.
Results
Baseline features of patients
The description of the main features of LSIE cohort and each group according to ENDOVAL score mortality estimation is presented in table 1 of online supplemental file 1. Differences of operated and non-operated patients with LSIE who had a formal indication for surgery is depicted in table 2. Non-operated patients with surgical indication are older and present more comorbidities. In these patients, IE is caused mostly by S. aureus and clinical complications such us heart failure, renal failure, septic shock and neurological events are also more frequent. Among surgical indications, non-operated patients presented more uncontrolled infection and less heart failure. In-hospital mortality is significantly higher in non-operated patients with LSIE. The main reason for not performing surgery is prohibitive surgical risk (73%) as depicted in figure 3.
Supplemental material
Prognostic impact of cardiac surgery
The independent predictive value of cardiac surgery is evaluated in two different ways.
Table 2 presents the result of propensity score matching of non-operated and operated patients with LSIE. Cardiac surgery maintains its protective factor after eliminating differences between groups (in-hospital mortality 40% vs 66% in operated and non-operated patients with LSIE, respectively, p<0.001).
Table 3 shows the multivariable analysis of in-hospital mortality in the global cohort. Cardiac surgery is an independent predictor of survival (OR 0.260, 95% CI 0.162 to 0.416) considering the most important predictive variables proposed by the European guidelines. Area under ROC curve of 0.841; 95% CI 0.809 to 0.873. P value of Hosmer-Lemeshow test of 0.631 (good fit of multivariable model).
The distribution of timings to surgery is depicted in online supplemental table 2. The median of time between diagnosis and surgery is 6 days and between endocarditis team discussion and surgery is 1 day. Patients who died were operated more urgently (median time between endocarditis team discussion and surgery 1 (0–2) vs 2 (1–4) days, p<0.001).
Prognostic impact of cardiac surgery according to baseline mortality risk
The association between cardiac surgery and in-hospital mortality is different depending on the theoretical risk of mortality of each patient estimated by the ENDOVAL score (figure 4). The greatest prognostic benefit of cardiac surgery is seen in patients at highest risk (predicted mortality of 80%–100%, OR 0.08, 95% CI 0.021 to 0.299).
Prognostic impact of cardiac surgery according to surgical indication
The prognostic benefit of cardiac surgery is similar along all surgical indications, however after propensity matching it only reaches statistical significance among uncontrolled infection indication, alone or in combination with heart failure (figure 5).
Discussion
We have addressed the role of cardiac surgery in the short-term prognosis of LSIE and specifically the impact of mortality risk assessed by the ENDOVAL score and surgical indication on surgical results. The main findings of our work are: (1) there is a high proportion of LSIE cases with surgical indication who do not undergo cardiac surgery. They present worse clinical profile and prognosis; (2) cardiac surgery is an independent protective factor of in-hospital mortality for active LSIE with surgical indication; (3) the higher the mortality risk estimated by the ENDOVAL score, the higher the protective effect of cardiac surgery; (4) uncontrolled infection is the indication with greatest benefit from cardiac surgery.
In our series, we found that one-third of patients with LSIE and a surgical indication did not undergo cardiac surgery during the active phase of the disease. Chu et al reported a 24% rate and found that host factors related to poor prognosis, namely sepsis, S. aureus aetiology and neurological complications were strongly related to not performing surgery.22 Our cohort is a riskier population; older, with more comorbidities and a worse clinical profile with higher incidence of S. aureus and septic shock among other risky markers, which justifies our higher proportion of non-operated patients with LSIE. This group reported a list of reasons for not surgical treatment: poor prognosis (33.7%), haemodynamic instability (19.8%), stroke (22.7%), sepsis (21%), which in the end represent our concept of prohibitive surgical risk (73%).
Non-operated patients have worse clinical profile and higher mortality. These patients are older and present more comorbidities which worsen their operative risk as these are important risk predictors of surgical mortality scores.20 23 24 In addition, similar to previously published reports, the higher prevalence of S. aureus may justify higher incidence of septic shock, renal failure and neurological complications. On the contrary, severe valvular dysfunction was more frequent in operated patients, which represents the most frequent cause of heart failure.1 2 Although it has been seen that uncontrolled infection is a stronger predictor of poor outcome after surgery than heart failure, our results strongly suggest that outcome improves with surgery especially for even inpatients with uncontrolled infection.19
Cardiac surgery is an independent protective factor of in-hospital mortality for active LSIE with surgical indication both by multivariable analysis and propensity score matching. This protective role is independent of prognostic factors proposed by the European guidelines on IE and possible confounders like age, comorbidities, S. aureus, S. gallolyticus, viridans group streptococci, renal failure, septic shock, neurological complications, periannular complications and severe valvular dysfunction. Observational studies suggest a positive impact of cardiac surgery in LSIE evolution,1 10–12 however, there are some which failed to demonstrate so,13 14 and even more, Tleivie et al reported a higher mortality among operated patients.15 It would be possible that operative mortality partially offsets the clinical benefit of cardiac surgery in the short-term prognosis,25 however, still consistent randomised data are lacking.13
For the first time, we have evaluated the prognostic benefit of cardiac surgery according to the baseline risk of the patient with LSIE. For this analysis we have used the ENDOVAL score, an easy tool for mortality prediction derived from the prognostic variables proposed in the European guidelines.3 Our results show that cardiac surgery is of benefit for all patients with LSIE and a surgical indication, but offers its higher impact among patients at higher risk. The higher the risk, the higher the benefit of surgery. Patients with LSIE with very high risk and a surgical indication who do not undergo cardiac surgery present a mortality of 94%, and surgery dramatically improves outcome, with an OR of 0.08 when the estimated mortality ranges between 80% and 100% (figure 4). As we can see from our work and others, there is reluctance to proceed to surgery in patients with indication when the risk is too high.22 Our work supports sending those patients to surgery, which might help reduce mortality. Therefore, in patients with LSIE, high risk related to their infective process should not refrain from sending a patient to surgery but reinforce the indication.
In addition, the positive prognostic impact of surgery is similar across the different surgical indications (figure 5), however it only reaches statistical significance among uncontrolled infection indication alone or in combination with heart failure. It is probable that due to the loss of patients in the propensity matching analysis we could not have reached statistical significance for heart failure and prevention of embolism indications. Previous works have suggested different impact of surgery according to the surgical indication, with none of them achieving significant results for prevention of embolism.19 26 27
From a practical point of view, our results support cardiac surgery in any context of LSIE provided an accepted indication is present, and irrespective of whether baseline risk is high. Of course, given the many confounders that can have influence in the course of this kaleidoscopic disease, which cannot be adequately reflected in the available scores, the ‘endocarditis team’ discussions are mandatory, particularly in these patients in whom futility and utility are separated by just a thin line.28 29
We are aware of some limitations, main of them due to the retrospective design of the article. By using the ENDOVAL score for patient risk stratification we have to assume the limitations of a multivariable model despite its good performance.3 Propensity score matching reduces the impact of treatment-selection bias when using observational data to estimate causal treatment effects; nonetheless, unmeasured characteristics and confounders are not balanced with this method, and some other methodological biases are not fixed.30 In this regard, the ‘endocarditis team’ decision is an intangible variable which could reflect non-measurable differences among populations. Importantly, a difficulty in the analysis of the impact of cardiac surgery for acute infective endocarditis is that the in-hospital clinical benefit may be offset by operative mortality and requires follow-up to be better ascertained, which was not available in our study.25
In conclusion, cardiac surgery during active LSIE, when indicated, seems to significantly reduce in-hospital mortality. The higher the risk, the higher the improvement in outcome if surgery is undertaken.
Key messages
What is already known on this subject?
Left-sided infective endocarditis (LSIE) is a severe disease with high in-hospital mortality.
Cardiac surgery is required in >50% of cases and, although formal indications are well established on the guidelines, patients with LSIE bear high surgical risk quite frequently and decision about utility or futility of cardiac surgery is controversial.
What might this study add?
The main findings of our work are: (1) there is a high proportion of LSIE cases with surgical indication who do not undergo cardiac surgery. They present worse clinical profile and prognosis; (2) cardiac surgery is an independent protective factor of in-hospital mortality for active LSIE with surgical indication; (3) the higher the mortality risk estimated by the ENDOVAL score, the higher the protective effect of cardiac surgery; (4) uncontrolled infection is the indication with greatest benefit from cardiac surgery.
How might this impact on clinical practice?
From a practical point of view, our results support cardiac surgery in any context of LSIE provided an accepted indication is present, and irrespective of whether baseline risk is high.
Of course, given the many confounders that can have influence in the course of this kaleidoscopic disease, which cannot be adequately reflected in the available scores, the ‘endocarditis team’ discussions are mandatory, particularly in these patients in whom futility and utility are separated by just a thin line.
Data availability statement
All data relevant to the study are included in the article or uploaded as supplemental information. Not applicable.
Ethics statements
Patient consent for publication
Ethics approval
The protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the local ethical committees.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Twitter @manuel_carneri
Correction notice This article has been corrected since it was first published. Author 'Alberto San Román' has been updated to 'J Alberto San Román' and their affiliations have been amended.
Contributors PEGG, JL, IV, CS, CO, MC and ASR were involved in the conceptualisation of the study. PEGG, CS, GC, CO, AJ, JBP, SdS, LM, MC, EM, PP and MdM were involved in the acquisition of the data and conduct of the study. IG and MC performed the statistical analysis. PEGG, JL, IV, CS, CO, SdS, EM, MC and ASR were involved in the interpretation of the data. All authors contributed to drafting this manuscript, with PEGG taking a lead role; he is also the guarantor of the manuscript. All authors gave intellectual input to improve the manuscript, and have read and approved the final version.
Funding This work was supported by 'Gerencia Regional de Salud de la Junta de Castilla y León' (GRS 1523/A/17).
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.