We report the results of a retrospective multicenter study on 248 patients with severe, active, definite and left-sided IE requiring ICU admission. Main causative pathogens are equally represented by streptococci and staphylococci. During ICU stay, surgery was indicated for 75 % of patients but only 53 % of them underwent surgical procedures during ICU stay. Overall in-hospital mortality was 41.5 %. Independent factors associated with prognosis were SAPS II on ICU admission >35, SOFA on ICU admission >8, IE due to methicillin resistant Staphylococcus aureus and native IE.
Current data suggest that staphylococci are the most common causative pathogens of IE. In the overall population of adults with definite IE admitted to hospital, S. aureus accounted for 26.6–36.2 % of causal agents [2, 14]. In series including only adult patients admitted to ICU with infective endocarditis, S. aureus represented 45–56 % of identified causative organisms [4, 7]. Our results could appear surprising since even if staphylococci are involved in 43.4 % of patients, they are less frequent causative organisms than streptococci involved in 45.6 % of patients. However, they are similar to those reported more than 10 years ago by Hoen et al. in France and Hasbun et al. in USA [15, 16]. In these series having included patients in 1999 and between 1990 and 2000, streptococci are involved in 48 and 58 %, respectively. We have no clear explanation about these microbiological differences between our study and those reporting data from patients admitted in ICU [4, 7]. Nevertheless, studied patients could be a little different. In the study from Mourvillier and colleagues, prosthetic valve IE are more frequent than in our series (40.6 vs. 21.4 %) and, if we focused only on native IE, streptococci and staphylococci are equally involved as causative organisms [4]. In the study reported by Samol and colleagues, 31 % of patients had right-sided endocarditis and it is well known that S. aureus is then the most common pathogen [7, 17].
In our study, a surgical procedure required by IE was performed during hospitalization for 125 patients (50.4 %). In series focusing on patients admitted to ICU for IE, 35–52 % of patients underwent surgery [4, 6, 7]. Rather than these gross percentages, an important point is, in our mind, the percentage of performed surgical procedures among patients for which indications for surgery emerge during ICU stay. Literature data are unfortunately scarce. In our series, surgery was indicated for 186 (75 %) patients but only 99 (53 %) of them underwent surgical procedures during ICU stay. In the study reported by Mirabel et al., the percentage appears higher since 100 of 158 patients with recommended surgical procedure underwent surgery [6]. Unfortunately, in this series, the timing of surgical procedure (emergency, urgent and elective) was only reported for patients undergoing surgery. So, it was not possible to determine the adequacy of surgery according to the timing, and consequently, to compare these results with ours. It’s a shame because it would have been interesting to known if the low percentage of adequate surgery observed in our series when timing was considered as emergency or urgent was also observed in other studies. In our series, it could be also noticed that among the 85 patients with contra-indications to surgery, 15 underwent nonetheless surgery and that 13 of them survived. Finally, surgery was more often adequate for patients admitted in a tertiary care hospital with a cardiac surgery department than for patients in a general hospital without any cardiac surgery department. Such a result reinforces recent recommendations for referring complicated IE patients to tertiary care hospitals in which a collaborative approach of IE involving notably a cardiac surgeon is possible [18].
The impact of surgery on IE prognosis was the subject of numerous studies. Despite some conflicting results, surgical therapy appears most often associated with an improved early and late survival both in the overall population of patients than in patients admitted to ICU [4, 6, 7, 19–25]. In our series, in-hospital mortality was 41.5 %. In similar series, mortality rates varied between 30 and 45 %, and apart surgery, identified independent prognostic factors were septic shock, cerebral emboli, immunosuppression, neurological failure, severe comorbidities, S. aureus IE and SAPS II [4, 5, 7]. Most of these factors appear in our series as significant prognostic factors in bi variate analysis. Among them, we could notice that IE due to Streptococcus spp. were associated with a lower mortality than IE due to Staphylococcus spp. and that adequacy of antimicrobial and of surgical treatment also appeared as factors associated with survival. However, in our study, we willingly chose to not enter in multivariate analysis the significant factors about surgery identified in bivariate analysis (surgery during ICU stay, surgery and adequate surgery) since the overall population was not affected by these prognostic factors. Our multivariate analysis identified 4 independent factors. They were SAPS II > 35 (AOR = 2.604), SOFA > 8 (AOR = 3.327), IE due to methicillin resistant Staphylococcus aureus (AOR = 4.981) and native IE (AOR = 0.345). The fact that scores assessing severity and/or organ failure on ICU admission are independent prognostic factors is not surprising since they are usually found in all studies focusing on prognostic of ICU patients. The protective role of the native character of endocarditis is not, in our opinion, surprising since the deleterious role of the prosthetic character of endocarditis is well known [3]. In example, in the study reported by Murdoch et al. including 2781 patients from the International Collaboration on Endocarditis–Prospective Cohort Study, prosthetic valve involvement appears as an independent factor associated with mortality [2]. The deleterious role of an infection due to methicillin-resistant Staphylococcus aureus could appear more surprising since it has not yet been found in previous studies. However, in our opinion, it was not really studied. In the study reported by Murdoch, 869 patients exhibited a S. aureus IE but no data about sensitivity to methicillin was reported [2]. In a French study reporting data about 497 adults with Duke-Li–definite IE, 180 patients had a S. aureus IE [14]. Resistance to methicillin was observed in 13.6 % of S. aureus. Unfortunately, no data about impact of resistance to methicillin were provided. Finally, to the best of our knowledge, the study reported by Fowler et al. is one of the few studies providing prognostic data according to sensitivity to methicillin of S. aureus [26]. Among 1779 patients from the International Collaboration on Endocarditis–Prospective Cohort Study, the authors identified 424 patients with definite S aureus IE and no history of active IDU. Among them, 141 (33.3 %) were infected with methicillin resistant S. aureus. These patients tended to have higher mortality (29.8 vs. 23.3 %; p = 0.14) than those infected with a methicillin susceptible strain.
Our study has several limitations. First, all data were collected retrospectively. Second, it was a multicenter study. As a consequence of these 2 points, diagnostic methods, screening for complications and therapeutic measures were not standardized. Moreover, only one of the seven hospitals participating in the study had cardiac surgery units. It could thus be assumed that the access to cardiac surgery has not been the same for all patients, the most distant patients from surgical units being the least likely to benefit from emergency or urgent surgery. Similarly, a multidisciplinary approach could not be optimal for these later patients. Third, independent prognostic factors were identified by a stepwise logistic regression analysis. No case–control analysis was performed to evaluate the performance of identified factors. Fourth, our analysis was unable to establish a causal relationship between some therapeutic measures such as adequate antimicrobial treatment and survival. In a previous work, we demonstrated that such a treatment was an independent prognostic factor associated with survival [27]. Finally, we have only information on in-hospital mortality and long-term outcome was unknown.
In conclusion, mortality in patients admitted to ICU for left-sided IE remains high, especially in cases of endocarditis due to methicillin resistant Staphylococcus aureus and when organ failures occur and ICU scores are high.