We attempted to identify all relevant studies regardless of publication status (published, unpublished, in press, ongoing) written in English or Portuguese. We identified relevant studies from several databases (Table 1). In particular, we searched the conventional databases Scopus, Medline and Pubmed for articles published in any journal. In addition, we performed specific searches in targeted journals, as described in Table 1.

The search strategy used a combination of subject headings and free-text keyword searching to capture the main search headings of “endocarditis” or “valve infection” and “Portugal” and “Portuguese”. Each database search strategy was adapted to match the appropriate subject headings used by that database.

Citations were screened based on the title/abstract and full texts were manually obtained for all potentially relevant articles. Additionally, we signed up to PubMed to receive automated electronic notification of any new articles containing the above search terms.

In addition to the search methodology described above, we also manually searched papers from the Portuguese Congress of Cardiology editions from 2010 to 2019, as well as several Portuguese medical journals (Table 1). We also conducted a separate citation search on Google Scholar, SIGLE and RCCAP, which are repositories of Master and PhD thesis in Portugal.

Various authors were contacted via telephone and/or email to confirm/complete information from published series and to identify ongoing or unpublished studies relevant to the review.

All abstracts and articles generated were uploaded to Mendeley™. They were screened to exclude items that did not fulfill the inclusion criteria and duplicates. We independently reviewed all texts. All authors were involved in the discussion and all agreed on the final inclusion.

Any study that reported an IE outcome was eligible for inclusion. We included studies that fulfilled the following conditions: 1) study included a population of adult patients (>17 years old) diagnosed with IE; 2) patients hospitalized in Portuguese hospitals, regardless of the department; 3) studies should report demographics, study design and setting, clinical, imaging and microbiological aspects. With respect to in-hospital complications, we also analyzed the pooled data for heart failure, embolization, uncontrolled infection, acute kidney failure and cardiac surgery, when available; 4) studies should report outcomes. In-hospital mortality and mortality at 30 days post-discharge were classified as short-term mortality. Long-term mortality refers to the death of patients 30 days and up to two years post-discharge – when several periods were reported, we used the value with the longest duration of time; 5) if more than one publication was identified for the same hospital center, we included the most recent full article.

To improve the quality of our findings only series with ≥25 patients were included. Also, in an effort to homogenize the sample under study, specific settings such as exclusive cardiac device related, prosthetic, tricuspid valve or renal patients IE series were excluded.

Finally, only studies published after 1990 were considered. Details of the inclusion and exclusion criteria are presented in Table 2.

Eighteen studies were finally included in the systematic analysis (Table 3). An example of the search conducted in PubMed is displayed in Figure 1. From this search in PubMed, a total of 149 unique articles were identified, of which 13 were selected for full text review and finally three articles were included in this systematic review.

Figure 1.

Flow chart indicating an example of the selection process for the selected studies.

(0.21MB).

Data on demographics such as age, gender, hospital setting, and study design were extracted. Medical background, clinical presentation, type of valve affected, microbiology profile, intrahospital complications (embolization, HF, refractory infection, acute renal failure, surgical intervention) and primary outcomes (short- and long-term mortality) were also retrieved.

The quality of studies was analyzed using the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) scale.14,15

The study was approved by the Ethics Committee of the University of Lisbon Academic Center (reference number 349/19).

A total of eighteen studies from Portuguese centers were identified, of which fourteen were full texts and four were conference articles, published between 1998 and 2020. Fifteen studies included both medical and surgical interventions for patients and three studies were exclusively surgical series. For the purpose of analysis, medical and surgical series will be discussed separately.

Most studies were from hospitals in the Lisbon area (eight studies), Oporto (five studies) and Coimbra (two studies). More than half of the included centers had an on-site cardiac surgery department available (61%).

Patients were enrolled from the cardiology department (eight studies), from internal medicine department (five studies), internal medicine and cardiology (two studies), cardiology and cardiac surgery (one study) and cardiac surgery (two studies).

All studies were retrospective and cohort-based and only one was multicenter.16 The number of years of analysis varied from two to 14 (mean 8.8 years). The sample size ranged from 25 to 320 patients, and a total of 1872 patients were included. All studies used the Duke's criteria or modified Duke's criteria for case selection.

All 18 studies reported short-term mortality and only five studies addressed long-term mortality (Table 3).

Eleven studies reported partial/complete information on intrahospital complications of IE.

The quality of the studies was variable, ranging from 36 to 91% of items on the checklist and no study met all the criteria from the STROBE list.

We discuss “Medical Series” and “Surgical Series” separately.

Patients demographics and characteristics

The 15 medical series studies included 1279 patients (Table 4). The mean age of patients was 55.5±12.1 years. The male gender predominated in all studies, a mean of 72.2% (from 62.1 to 82.1%).

The source of IE was predominantly community-related (56-79%) – this classification was obtained in five studies.17–21

Native valves were more predominantly involved (average 68.6%), followed by prosthetic valves (22.6%) and finally CDRIE (6.1%). Seven studies identified the location of IE, and the left cardiac valves were most commonly affected, except in two studies22,23 from late 1990s, in which tricuspid valve IE was more relevant, as the population presented a high prevalence of IV drug use (68 and 72% respectively). Multivalvular involvement was mentioned in three series, from 1.5% in Proença's series23 to 15% and 25% in more recent series, from Santos et al.17 in 2014 and Melo et al.24 in 2017.

Fever and constitutional symptoms were the most frequent presentation symptoms reported.

Blood cultures were positive in 73% of the patients, with proportion varying among studies (56-79%). Staphylococcus and streptococcus were the most frequent pathogens in our analysis (Figure 2).

Figure 2.

Microorganisms (%) involved in Portuguese IE series.

(0.19MB).

Five studies16,17,21,23,24 reported the presence of enterococcus, in proportions ranging from 1.5 to 13%.

In all 15 series reviewed, echocardiography was the mainstay imaging technique used in patients diagnosed with IE in Portuguese hospitals. Marques et al.21 mentioned the use of 18F-FDG PET/CT to assess one patient who had undergone a prosthetic valve implantation in the previous year.

Eight studies (53%) reported partial/complete information about in-hospital complications (Table 3 and Table 4). HF was present at admission or during hospitalization in 44.7%, whereas uncontrolled infection was reported in 17.2%. In addition, 15 to 55% of the patients had an embolic event during hospitalization. Acute kidney failure was very prevalent, affecting more than 50% of patients (from 56 to 58%).

Regarding valve complications, vegetations were the most prevalent major echo finding (48.5 to 93%).

In-hospital, parenteral (IV) antibiotic administration was referred in all series. Neves et al.25 presented the only series with completion of the parenteral antibiotic regimen in an outpatient setting (in 52% of the patients in their series). No specific data was provided in the different series regarding the type of antibiotic and duration of administration. Also, no information regarding the choice of the antibiotics used was given (e.g. national or international guidelines, intrahospital or department protocols).

Regarding valve surgery, the presented data was very distinct. The surgical rate in all 15 series varied between 3.1 and 52%, with an average of 29.8% (Figure 3). Nevertheless, the more recent series reveal a tendency toward a higher surgical rate (Figure 4). Analyzing the surgical rate taking into account the presence of a cardiac surgery unit (CCT) in the hospital center, a non-significant a higher rate was observed in the centers that had an on-site CCT[35 (CI 95%: 28.7-41.6)% vs. 23(95% CI: 11.3-26.1)%].

Figure 3.

Cardiac valve surgery (%) in the Portuguese IE medical series.

(0.13MB).

Figure 4.

Cardiac valve surgery (%) in Portuguese IE series (the central year of each series was considered). Black dashes identify the mean value of cardiac surgery (in %) obtained from all series.

(0.08MB).

The indication for surgical intervention, was fully specified in two studies,21,26 HF was predominant (50 and 74%), followed by uncontrolled infection (27.3 and 46.8%) and embolization (13.6 and 15.6%). Cardiac surgery was considered a protective factor for patients with a clinical surgical indication.16,21,26

Regarding the reasons for not performing surgery, Marques et al.21 highlighted the presence of significant comorbidities, cognitive impairment and ischemic stroke with significant sequelae as impeding factors.

Timing of surgery was only addressed by Moreira et al.,26 concluding for their series that the timing of surgery (early versus late) was not associated with increased mortality. No information on the culture of surgical material was provided.

All studies reported short-term mortality, with an average of 21.9% (ranging from 3 to 37.2%) (Figure 5 and Table 3).

Figure 5.

Rate of short-term mortality (%) in Portuguese IE series.

(0.15MB).

Only four series evaluated long-term mortality.17,18,20,26 In these four series, more than a third of all patients died within 2 years despite hospitalization, antibiotic and surgical treatment (ranging from 29.1 to 34%) – Table 3.

The cause of death was only specified in two studies,16,21 septic shock being the first cause, followed by HF and embolism with similar rates.

In this analysis, we review the three surgical series27–29 described in Table 3 and two partial surgical cohorts included in the series validated by Moreira et al.26 and Ferreira et al.16 in their studies (Table 5).

A total of 736 patients, mostly men, in their sixth decade of life were identified.

A higher prevalence of native valve endocarditis was included (aortic valve prevailed, except in the Madeira et al.28 cohort with a similar rate in the mitral valve). Right heart valves, namely tricuspid valve, were rarely subject to intervention. Multivalvular involvement and intervention were not disclosed.

Regarding causal agents, staphylococcus spp. ranged from 19 to 45%, followed by streptococcus spp. (15 to 29.5% of patients); most patients had positive blood cultures. Enterococcus was present in 11.7 to 15% of patients. No information on microbiology and molecular testing of operative specimens or post-operative antibiotic regimen and duration was provided.

Surgery was considered emergent/urgent in the majority of cases. Two centers reported having performed elective surgeries (5.6% in Guiomar et al. series29 and 18% in Ferreira et al. cohort16); Madeira et al.28 on the other hand did not perform elective surgeries in their cohort.

All centers followed the scientific guidelines on indication for surgery, the first indication being refractory HF, followed by uncontrolled infection and embolization risk. Length of hospitalization was on average approximately four weeks (disclosed in two series).16,29 Surgical time or aortic cross-clamping times, on the other hand, was not addressed.

As for the major outcomes, post-operative complications were partially disclosed in three series,26,27,29 acute renal and respiratory failure, as well as sepsis, arrhythmias and ischemic stroke were identified. Short term mortality rate ranged from 13.1 and 16% in all five series. Guiomar et al.29 specified septic and cardiogenic shock as the main causes of death. Long-term mortality rate was only addressed by Moreira et al.26 with up to one third of patients dying after one year and up to 43% at five years of follow-up.

An analysis according to subgroups of patients hospitalized before and after 2009 (taking into consideration the different recommendations on surgical intervention in IE3) is only available in the Alves et al. study,27 concluding that the more recent group presented older patients, with less heart failure (HF) and a higher rate of embolization and enterococcus spp. infection.

This was a systematic review of mainly single institution series and therefore selection bias may be present as the studies may well not fully represent the entire population. Also, despite exclusion of very small series, the majority of series had a small to moderate number of patients which could lead to reporting bias. Most series were retrospective, purely observational, which led to differences in reporting epidemiological, clinical and technical data and measure outcomes. The significant heterogeneity of the number of patients, hospital settings and the time of the analyzed series also prevented the use of more complex statistical analyses. In particular, most series reported on patients included both before and after the 2009 ESC Guidelines on Infective Endocarditis,3 and for this reason we could not perform a comparative before-after analysis.

Regarding the quality of studies, most had up to 75% on the STROBE checklist which may indicate risk of bias.