Influence of the healthcare pathway on the outcome of patients with infective endocarditis 

Abstract

Aims

To determine the prognosis of patients treated for infective endocarditis (IE) according to their healthcare pathway. To assess how the ESC guidelines are implemented concerning the performance of transoesophageal echocardiography, the use of antibiotic therapy, and the performance of valve surgery; and to compare the epidemiological profile of IE according to the type of centres in which the patients are hospitalized.

Methods and results

In a prospective multicentric study including 22 hospitals in the South-East of France, 342 patients were classified into three groups according to their healthcare pathway: 119 patients diagnosed and taken care entirely in a reference centre or hospital with cardiac surgery [Referral Center (RC) group], 111 patients diagnosed and initially taken care in a non-RC (NRC), then referred in a centre including cardiac surgery [transferred to the Referral Center (TRC) group] and 112 patients totally taken care in the NRC (NRC group). One-year mortality was 26% (88 deaths) and was not significantly different between Groups 1 and 2 (20 vs. 21%, P = 0.83). Patients in the NRC group had a higher mortality (37%) compared with patients in the RC and TRC groups (P < 0.001). ESC guidelines were not implemented similarly depending on the healthcare pathway (P = 0.04). Patients in the NRC group were significantly older (P < 0.001) and had more comorbidities (P < 0.001) than patients treated in referral centres.

Conclusion

Prognosis of patients with IE is influenced by their healthcare pathway. Patients treated exclusively in NRC have a worse prognosis than patients treated in referral or surgical centres.

Graphical Abstract

Open in new tabDownload slide

Introduction

Epidemiologic profile of infective endocarditis (IE) has changed over the past 20 years,^1–3 but mortality remains high^4–6 despite diagnostic improvement and access to surgery in up to 50% of cases.⁷ This persistent severity has justified the development of new diagnostic and therapeutic strategies to improve prognosis.^8,9 Management by a specialized multidisciplinary team within a referent medical–surgical centre seems to be a key factor in improving patients prognosis,^10,11 and is recommended by the ESC guidelines.^8,9

However, few studies have compared the prognosis and management of patients with IE according to their healthcare pathway. Indeed, three situations are usually observed: patients for whom the diagnosis and management are carried out entirely in a referral centre and/or having a cardiac surgery department, patients secondarily referred to a referral hospital or surgical centre by another care centre, patients for whom the entire care is provided within a non-referral care centre. In addition, application of European guidelines remains difficult to assess in daily clinical practice outside the major referral centres. Published epidemiological data only represent part of the IE, most often not taking into account patients hospitalized in non-referral centres and therefore introducing a significant bias in the knowledge of this pathology.¹² Finally, although the ESC guidelines suggest that patients with uncomplicated IE can be managed in a non-reference centre,⁹ if all patients with IE¹³ or only the sickest patients¹⁴ should be referred to reference centres is still a matter of debate.

The main objective of this study was to determine the prognosis of patients treated for IE according to their healthcare pathway. The secondary objectives were to assess how the ESC guidelines are implemented concerning the performance of transoesophageal echocardiography (TEE), the use of antibiotic therapy, and the performance of valve surgery, and to compare the epidemiological profile of IE according to the type of centres in which the patients are hospitalized.

Methods

Study design and patients

This was a prospective multicentric study including patients consecutively admitted for suspicion of IE in 22 hospitals in South-East of France between January 2014 and June 2017. These hospitals were both tertiary centres with cardiac surgery and second-level community hospitals. Inclusion criteria were a diagnosis of definite IE based on Duke-modified criteria¹⁵ on native, prosthetic, or pacemaker/implantable automatic defibrillator. Non-inclusion criteria were age <18 years and patients already included in the study and presenting with a new IE diagnosis. Patients were screened in the echocardiography lab of each institution and were prospectively included by local investigators.

Patient and public involvement

All patients hospitalized for suspected IE were accepted on admission to participate in this research protocol. However, they were not involved in the design, conduct, reporting, or dissemination plans of our research. The study complies with the Declaration of Helsinki. Written consent was waived by La Timone Institutional Review Board, which gave its approval for this study.

Data collected at the time of inclusion and during hospitalization

The following clinical data were collected: age, sex, previous cardiac and non-cardiac diseases, Charlson index score, date of first clinical signs, body temperature, presence of cardiac murmur, congestive heart failure, neurological complications, septic or cariogenic shock, or atrioventricular block on electrocardiogram at the time of admission and during hospitalization.

Biological data were collected, including haemoglobin level, platelet count, leucocytes count, C-reactive protein serum level, creatinine level, rheumatoid factor. Microbiological data included: positive or negative blood cultures, type of microorganism, and antibiogram.

Echocardiographic data collected were: presence of vegetation and its maximal length, presence of peri-annular lesion (defined as the presence of abscess and/or pseudoaneurysm and/or fistula), presence of valvular regurgitation, quantified according to current recommendations.^16,17

Type and date of the following complications were collected: congestive heart failure, peripheral embolism, neurological complications (including symptomatic and silent), mycotic aneurysm, acute kidney failure (>26 µmol/L increase of serum cretinine in 48 h or >50% increase in 7 days), glomerulonephritis, uncontrolled infection (fever and/or positive blood cultures after 7 days of appropriate antibiotic therapy and unlinked to an extracardiac cause).

Moreover, type, dose, and length of antibiotic therapy were reported. Antibiotic therapy was considered appropriate if drugs used and full duration of treatment were in accordance with ESC guidelines (taking into account patients terminating antibiotics prematurely due to death). Analysis of antibiotic treatment was conducted retrospectively by the Endocarditis team of the investigator centre (CHU La Timone, Marseille).

During hospitalization or follow-up, the indication (according to ESC guidelines) and date of surgery were also collected. Finally, surgical risk was evaluated using the Euroscore II.¹⁸

Classification of patients according to healthcare pathway

Patients were classified in three groups according to their healthcare pathway:

• Referral Center group (RC), including patients diagnosed and taken care entirely in a reference centre or hospital with cardiac surgery (four hospitals). These centres were the only surgical centres of this region of France.

• Transferred to RC group (TRC), including patients diagnosed and initially taken care in a non-referral centre, then referred in a centre including cardiac surgery.

• Non-RC group (NRC), including patients totally taken care in non-referral centres (18 centres).

Follow-up

Patients had a 1-year follow-up with a clinical evaluation or if not possible, phone contact.

Primary and secondary endpoints

Primary endpoint was death from all cause at 1-year follow-up. Secondary endpoints were:

• the evaluation of the implementation of European guidelines^8,9 in the three patients groups regarding: the use of TEE, the choice and duration of antibiotic therapy, and the performance of valvular surgery and/or extraction of pacemaker or implantable automatic defibrillator when indicated.

• comparison of demographical and microbiological data among the three groups.

Statistical analysis

Continuous variables were described by their mean, standard deviation, minimum and maximum value, 1st and 3rd quartile. Continuous variables were compared using the Mann–Whitney non-parametric U test or Student's t-test, depending on the application conditions. Categorical variables were compared using the χ² test or the Fisher test, depending on the application conditions. The event-free survival distributions were estimated with the Kaplan–Meier method. The log-rank test was used to compare survival distributions for categorical variables. The effect of continuous variables on the risk of an event occurring was estimated and tested using a Cox model. The surgery variable, which is a time dependent variable, was estimated using a Poisson approach.

Multivariate analyzes were tested using a Cox model. Multivariate analyzes were tested using a Cox model. Inclusion of variables in the fitted model was carried out according to the following criteria: result in univariate analysis with statistical significance P < 0.20, number of missing values <10% for this variable, effective presence in both categories of a bivariate variable >10%, as well as according to their clinical relevance. The final models were selected using the Akaike Information Criteria.

The comparison tests of Group 1 to Group 2, Group 1 to Group 3, Groups 1 + 2 to Group 3 were performed in bilateral situation using Bonferroni correction and were considered statistically significant for P < 0.017. For all other tests, they were performed bilaterally and were considered statistically significant for P < 0.05. Statistical analysis was performed with R software (version 3.6.2) and RStudio (v 1.2.5).

Results

Patient population

Between January 2014 and June 2017, a total of 342 patients with definite IE were included. Among them, 119 were diagnosed and entirely managed in a reference centre or hospital with cardiac surgery (RC group), 111 were diagnosed and initially managed in a non-referral centre, then referred in a centre including cardiac surgery (TRC group) and 112 were totally managed in non-referral centres (NRC group). Clinical, demographical, biological, and microbiological characteristics of patients are presented in Table 1.

Overall population

In the overall population, mean age was 68 ± 15 years. Infective endocarditis were mainly community-acquired (83% of cases), Staphylococcus aureus was the most frequent bacteria involved (26% of cases). Negative blood culture IE represented 17% of cases. Left-sided IE counted for 85% of all cases and aortic (43%) and mitral (42%) localization were almost equally represented. Peri-valvular complications were observed in 16% of cases. Thirty-three percent of cases were prosthetic valve IE. More than half IE cases (58%) showed theoretical indication for valvular surgery or intracardiac device removal but only 35% had such interventions.

Characteristics according to healthcare pathway

Patients in the NRC group were significantly older and had higher Charlson index. Sources of infection did not differ between the three groups and microbiological profile of IE cases was not significantly different. Echocardiographic findings showed more severe lesions in the RC group and TRC group: longer vegetations, more peri-valvular complications, and more severe valvular regurgitation. Concerning IE complications, there was no significant difference between groups in embolic events; however, more silent embolisms were diagnosed in the RC group and TRC group. Spondylodiscitis was more frequent in the NRC group.

Patients in the RC and TRC groups had more frequent surgical indication, but surgery or intracardiac device removal was performed in only 52 patients (44%) in the RC group, 60 patients (55%) in the TRC group, and 6 patients (5%) in the NRC group, P < 0.001. In the RC group, 81 patients had an indication for intervention and 29 were not operated and in the TRC group, 91 patients had an indication for intervention and 31 were not operated. Reasons for the absence of intervention were: deaths before surgery (six patients in the RC group and two patients in the TRC group), patient’s refusal (two patients in the RC group and one patient in the TRC group) and decision of the endocarditis team due to comorbidities (21 patients in the RC group and 28 patients in the TRC group). Patients who benefited from surgery were significantly younger (69.4 ± 13 years vs. 62.4 ± 12 years, P = 0.01), had lower Charlson index (3.9 ± 1.9 vs. 2.8 ± 2.7, P = 0.007), had more frequently vegetations [86 patients (73%) vs. 95 patients (85%), P = 0.002] and severe valvular regurgitation [41 patients (35%) vs. 68 patients (61%), P < 0.001] but less renal failure [29 patients (25%) vs. 5 patients (4%), P < 0.001] and diabetes [39 patients (33%) vs. 20 patients (18%), P = 0.01].

Time from diagnosis to intervention did not significantly differ between the RC group and TRC group (mean time 30 ± 32 days vs. 22 ± 29 days, P = 0.36). In our cohort, mean time for right-sided IE was 17 ± 18 and 25 ± 28 days for left-sided IE.

Prognosis of patients according to their healthcare pathway

Among the 342 patients, 1 year mortality was 26% (88 deaths), hospital mortality was 13% (44 deaths), and 1 year recurrence rate was 4% (13 patients). Figure 1 shows survival probability according to healthcare pathway.

Hospital mortality was 11% (13 deaths) in the RC group, 8% (9 patients) in the TRC group, and 20% (22 patients) in the NRC group, P = 0.03. One-year mortality was 20% (24 deaths) in the RC group, 21% (23 deaths) in the TRC group, and 37% (41 deaths) in the NRC group, P = 0.003. One-year mortality was not significantly different between the RC and TRC groups (P = 0.83); however, it was significantly higher in the NRC group compared with the RC and TRC groups (P < 0.001).

Characteristics of patients only treated in non-referral centres

During the study period, 112 patients were managed only in a NRC. Among them, 28 patients (25%) had indication for valvular surgery or intracardiac device removal. These patients had a mean age of 76 ± 11 years and were at high surgical risk (mean Euroscore 17 ± 16%). Reasons for absence of transfer and surgery included: death before surgery for 16 patients (14%), medical choice for 14 patients (13%), patient’s refusal in three cases (3%), and high surgical risk (comorbidities) for 16 patients (14%). Six patients (5%) had intracardiac device removal (pacemaker lead or catheter) in the NRC. Among these 28 patients, 1 year mortality was 50% (14 deaths). Eighty-four patients (75%) had indication for medical treatment of IE. One-year mortality among them was 29% (24 deaths).

Prognostic factors of mortality

Results of univariate analysis of prognostic factors of 1 year mortality are shown in Supplementary material online, Table S1. Table 2 shows prognostic factors of mortality among the 342 patients in multivariate analysis. Belonging to the NRC group was associated with higher risk of death [heart failure (HR) 2.56; 95% CI 1.44–4.55, P = 0.001]. Heart failure (2.4; 95% CI 1.52–3.78, P < 0.001) and peri-valvular involvement (HR 3.25; 95% CI 1.86–5.68, P < 0.001) were IE complications predictive of mortality.

Evaluation of European guidelines application according to healthcare pathway

Table 3 shows the implementation of ESC guidelines in the three groups regarding the use of TEE, valvular surgery, or intracardiac device removal and antibiotic therapy. In overall population, ESC guidelines were correctly implemented in 52% of patients. Transoesophageal echocardiography was performed in most IE cases (92%); however, it was significantly less frequently used in the NRC group as compared with the two other groups (P < 0.001). Guidelines concerning surgery or intracardiac device extraction were correctly implemented in 78% of cases and there was no significant difference between the three groups. Antibiotic therapy was in accordance with guidelines in 72% of cases and there was no significant difference between patients taken care entirely or partially in RC (P = 0.66). However, treatment was less frequently in agreement with guidelines in patients in the NRC group compared with the others (P < 0.001).

Discussion

This prospective study carried out in 22 hospitals in the South-East of France including 342 IE cases divided into three groups according to their healthcare pathway shows that:

• The prognosis of patients is influenced by their healthcare pathway.

• The number of patients managed in non-referral centres is high. They are older, have more comorbidities, and less severe cardiac lesions than the others.

• Surgery or device removal is less frequently performed in those patients.

• European recommendations on the management of these patients are not followed similarly depending on their healthcare pathway.

• Both in-hospital and long-term mortalities are higher in non-referral centres.

General population

By including patients treated in NRC, our study has the advantage of providing global epidemiological information concerning the cases of IE within the south-eastern region, reflecting the management of this pathology in ‘real life’ and not only in RC. Almost a third of the patients treated for IE during the inclusion period were treated in a NRC.

Consistent with known data,^19,20 there was a predominance of IE cases in males (69%). The mean age of diagnosis of IE was 68 years in our study. This age seemed more advanced in comparison to other French (average age of 59 ± 17 years in 1999,² 62 ± 16 years in 2008²¹) or European studies (57 ± 16 years in 2001,³ 59 ± 18 years in the Euro-Endo registry²⁰). Concerning microbiological data, S. aureus was the predominant germ (26% of IE), which was in agreement with previous studies.^21,22 There was a majority of left-sided IE (85%). An embolic event was observed in 38% of cases. This rate was comparable to other studies that describe the occurrence of an embolic event in 20–50% of cases.^21,23,24

Mortality according to healthcare pathway

In our cohort, hospital mortality was 13% and 1-year mortality was 26%. Few studies have compared the prognosis of patients with IE according to their healthcare pathway. A study by Fernández-Hidalgo et al.,²⁵ compared the hospital mortality of 223 patients treated in a RC with cardiac surgery with 144 patients taken care in a NRC and secondarily referred to the RC. This study did not show any significant difference in terms of hospital mortality between the two groups. Likewise, an international multicentre study¹² carried out in RC in the management of IE did not show difference in hospital mortality between 1164 patients transferred from NRC and 1596 patients treated directly within the RC. However, these studies have some limitations: by considering only patients treated in RC they are not representative of epidemiology of patients treated in NRC. The originality of our study is to compare three groups, including a group of patients treated exclusively in NRC, these patients usually being not included in previous studies. Our main findings are a significantly higher hospital (20%) and 1 year mortality (37%) in patients treated only in NRC. Belonging to the NRC group was an independent risk factor in multivariate analysis of 1-year mortality in the total cohort of patients. This excess of mortality observed in the NRC group could be explained by various factors: epidemiological differences compared with other groups, recommendations on the management of IE that are less implemented, and a higher absence of surgery rate in patients with a theoretical surgical indication.

Implementation of European guidelines

In our study, we observed a more frequent use of TEE (92% of patients) than in other multicentre studies (60–75%^3,20). However, within the NRC group, TEE was significantly less frequently performed compared with the other groups. Transoesophageal echocardiography has a key role in the diagnostic process of IE and in particular in the detection of peri-annular complications and should be performed systematically, in accordance with current recommendations.^8,9,26 Thus, it is possible that these complications, often requiring surgery, have not been diagnosed in some patients, especially in the NRC group.

Regarding valve surgery or the extraction of intracardiac material, there was no significant difference in application of recommendations between the three groups. However, we noted within each group patients with a theoretical indication for surgery for who surgery was finally not performed. Within the NRC group, 25% of patients presented a theoretical indication and only 5% received interventional management. Conversely, in the RC and TRC groups, 65% of patients with theoretical surgical indication benefited from intervention. Different studies have already shown the poor prognosis of patients with a theoretical indication for surgery and not operated,^20,27 which may in part explain the higher mortality of patients in the NRC group. In addition, the main multicentre studies carried out over the last few years highlight a surgery rate close to 50% (52% in Euro Heart Survey,³ 51% in the Euro-Endo register,²⁰ 48% in the ICE register⁴). Among the patients treated exclusively in RC, the rate of surgery or extraction of intracardiac material was 44%, close to previous published data. On the other hand, if we considered all the patients for whom the diagnosis of IE was made in a non-referral centre (TRC and NRC groups), the overall rate of valve surgery or extraction of intracardiac material was only 30%. Several factors could explain this discrepancy with the pre-existing studies. Indeed, most studies on IE focused on patients treated in RC with a cardiac surgery department. It is possible that a referral bias may cause an overestimation of the rate of cardiac surgery in this pathology, by not considering many IE of ‘medical’ management and treated exclusively in NRC. On the other hand, it is also possible that among the patients diagnosed in a NRC, surgical indications were sometimes not identified, especially in the NRC group in which, unlike the TRC group, it was not possible to re-evaluate the surgical indications by a multidisciplinary team in a reference centre. In a same way, a Spanish study by López-Dupla et al.²⁸ in a university centre without a cardiac surgery department showed that the establishment of a multidisciplinary team including a cardiac surgeon was associated with an increase in the rate of surgical indication identified (14.5 vs. 34.5%). Finally, the European guidelines concerning the antibiotic treatment were significantly less applied among patients in Group 3 with 35% of inappropriate antibiotic therapy. Here again, few studies concerning the antibiotic treatment used in non-referral centres are available. Fernández-Hidalgo et al.²⁵ showed that among the patients diagnosed in a NRC and referred secondarily to a surgical centre, the antibiotic treatment initiated in the non-referral centre was inadequate in 54% of cases. In this study, inadequate treatment was associated with an increased risk of in-hospital mortality in multivariate analysis. The goal of our study was to evaluate the implementation of ESC guidelines in ‘real life’. It is however important to highlight that in some cases, there may be logical reasons why patients do not undergo an indicated procedure such as TEE, for instance when a patient is not operable, or surgery.

Epidemiological characteristics of infective endocarditis according to healthcare pathway

The epidemiological profile of patients in the RC and TRC groups was comparable. These were patients with more severe IE than in the NRC group with more haemodynamic complications, a higher rate of vegetations, more peri-annular complications. These elements partly explain the higher surgical indication rate in these two groups. The patients in the NRC group were significantly older, presented more comorbidities, and a greater operative risk. Some studies focusing on IE cases in the elderly have shown, including during treatment in a RC, limited access to surgery and significant mortality in these patients when surgery, although indicated, was not performed.^29,30

Limitations

Some data could not be recorded, such as the reason for the transfer of patients to a referral centre, the time between diagnosis and the transfer of patients, and some factors who could have influenced transfer decision (such as cirrhosis or poor nutritional status). Due to the large number of participating non-referral centres, the reasons for transferring patients may have differed from one centre to another. Moreover, modified Duke criteria were used for diagnosis of IE and have limited sensitivity. This study reflects the epidemiology and practices used in the south-eastern region of France and cannot be generalized to all the other centres. Finally, the study design was meant to compare prognosis of patients according to the healthcare pathway but reasons explaining the different prognosis observed could only be hypothesized.

Conclusion

Prognosis of patients with IE is influenced by their healthcare pathway. Patients treated exclusively in NRC have a worse prognosis than patients treated in RC. These patients are often fragile and their management difficult. Given the difference in mortality between patients managed in referral vs. non-referral centres, the current recommendations of treating patients with non-complicated IE in non-referral centres should be questioned. Infective endocarditis is a deadly disease with a better prognosis when managed in RC.

Supplementary material

Supplementary material is available at European Heart Journal: Acute Cardiovascular Care online.

Acknowledgement

We thank all the French centers who have participated to this work.

Funding

This study was supported by a grant from: the PHRCI (Projet Hospitalier de Recherche Clinique Inter-régional 2012).

Data Availability

The data underlying this article will be shared on reasonable request to the corresponding author.

Investigators and sites

La Timone Hospital Marseille: Florent Arregle, MD; Mary Philip, MD; Sandrine Hubert, MD; Frédérique Gouriet, MD PhD; Jean-Paul Casalta, MD; Frédéric Collart, MD; Alberto Riberi, MD; Hélène Martel, MD; Sébastien Renard, MD; Laurence Camoin, MD, PhD; Anne Claire Casalta, MD; Hubert Lepidi, MD, PhD; Michel Drancourt, MD, PhD; Didier Raoult, MD, PhD; Gilbert Habib, MD, PhD. CHU Nord Marseille: Franck Thuny, MD; Yves Frances, MD; Karim Aissi, MD; Philippe Brouqui, MD; Matthieu Million, MD; CHU de Montpellier Catherine Sportouch, MD; Stéphane Cade, MD; Frédéric Cransac, MD. CHU de Nîmes: Patrick Messner-Pellenc, MD; Camille Soulier, MD; Jean-etienne Ricci, MD; Hôpital d’Instruction des Armées Laveran, Marseille: Laurent Fourcade, MD; Pierre-Laurent Massoure, MD; Jean-Marie Gil, MD; Hôpital Ambroise Paré-Desbief, Marseille: Emmanuel Philip, MD; Fondation Saint Joseph, Marseille: Patrick Khanoyan, MD; Roger Rosario, MD; Nicolas Michel, MD; CH de Martigues: Serge Yvorra, MD; Virginie Chellini, MD; CH d’Avignon: Stéphanie Branger, MD; Stéphane Andrieux, MD; Saida Cheggour, MD; Salah Aboukoudir, MD. Clinique Rhône-Durance, Avignon. Laurent Meille, MD; Catherine Meulemans, MD; CH Toulon: Jean-Michel Tartière, MD; Danielle Raufast, MD; Isabelle LeCardonnel, MD; Hôpital d’Instruction des Armées Saint-Anne, Toulon: Christophe Jego, MD; Gilles Cellarier, MD; Raphael Poyet, MD; Polyclinique Les Fleurs, Ollioules: Philippe Pietri, MD; Philippe Villain, MD; Jean-Paul Giorgi, MD; Chirine Parsai, MD; Olivier Brissy, MD; CH Aix-en-Provence: Bernard Jouve, MD; Tewfik Benchaa, MD; CH d’Arles: Salah Si Ahmed, MD; François Saint-Pierre, MD; CH de Draguignan: André-François Chaix. Pierrick Boyer, MD, Clinique La Casamance, Aubagne Véronique Haddad, MD. CH des Alpes du Sud, Gap-Sisteron: Christian Tafani, MD; Amar Hidoud, MD; Audrey Boudes, MD; Hakim Berguigua, MD; Fabien Devemy, MD; Bernard d’Hautefeuille, MD; CH de Manosque Alain Rapuzzi, MD; Clinique des Franciscaines, Nîmes: Vlad Ciobataru, MD; Hôpital de Fréjus: Xavier Lamit, MD; Hôpital de Bastia: Aurelia Tho, MD; Thierry James, MD; Abdelkader Bensalah, MD.

References

Author notes