Intern Emerg Med DOI 10.1007/s11739-014-1054-y
IM - ORIGINAL
Infective and non-infective endocarditis in critically ill patients: a clinical–pathological study Giorgio Berlot • Cristina Calderan • Cristina Fiorenza • Davide Cappelli • Stefano Addesa • Rossana Bussani
Received: 23 October 2013 / Accepted: 23 January 2014 Ó SIMI 2014
Abstract The aims of this study are to estimate the incidence, the outcome and the associated risk factors of infective and non-infective endocarditis (IE and NIE, respectively) in intensive care unit (ICU) patients. We studied the post-mortem findings and the clinical data of the patients who died in our ICU between 1996 and 2010. Of the 765 reviewed autopsies, 21 patients (2.7 %) presented cardiac vegetations. These cases consisted of 12 IEs and 9 NIEs. Three patients with IE had a mechanical prosthetic valve, and in 11 cases invasive devices had been used. Multiple peripheral embolisms were discovered at autopsy. In particular, the brain appeared to be more affected in patients with IE, while pulmonary embolisms were commonly associated with NIE. Blood cultures were positive in nine patients with IE. The imaging diagnostics (transthoracic and transesophageal echocardiography) which were seldom performed in both groups, proved to be of little help. As a consequence, an IE was correctly diagnosed before death in three patients (25 %) and suspected in two other cases (17 %), while a NIE was diagnosed before death in one patient alone. In conclusions, critically ill patients admitted to general ICUs, multiple factors related both to the underlying conditions and to performed procedures can facilitate the occurrence of IE and NIE making, at the same time, their diagnosis challenging. Many cases, in fact, are diagnosed only at autopsy. Yet again, post-mortem examination proves to be an G. Berlot (&) C. Calderan C. Fiorenza D. Cappelli S. Addesa Department of Anaesthesia and Intensive Care, Cattinara Hospital, Strada di Fiume 447, 34149 Trieste, Italy e-mail: [email protected] R. Bussani Department of Pathology, University of Trieste, Trieste, Italy
invaluable tool for the evaluation of diagnostic accuracy in critical care. Keywords Infective endocarditis Non-infective endocarditis Autopsy Clinical errors Missed diagnoses
Introduction The estimated incidence of infective endocarditis (IE) ranges between 30 and 100 episodes per million patientyears [1] with a mortality rate exceeding 30 %. The main risk factors for its occurrence include underlying congenital or degenerative valvular abnormalities, the presence of prosthetic valves, bloodstream infections and intravenous drug use [2, 3] (Fig. 1). Different causes account for non-infective endocarditis (NIE), including neoplasms, multifactorial pro-coagulative states, long-term hypoxia, chronic low flow states and cachexia [4, 5] (Fig. 2). The very same factors are responsible for the occurrence of both forms of endocarditis, in the hospital and in critically ill patients admitted to intensive care unit (ICU). All ICU patients are particularly prone to bloodstream infection-related IE as compared to regular ward patients, mainly due to the widespread use of invasive devices such as central venous catheters (CVC), pulmonary artery catheters (PAC) and intra-arterial lines (IAL). On the other hand, the ever-increasing age of these patients with the related burden of disease, the procoagulative state associated with sepsis and sepsis-related conditions and the above-enlisted risk factors set the stage for NIE [6]. The current diagnoses of both IE and NIE are based on the detection of valvular or mural vegetations by means
123
Intern Emerg Med
those acquired in the regular wards from those originating in the ICU. Therefore, (a) to estimate the incidence in a population of patients admitted to a general ICU for causes different than IE and NIE; (b) to evaluate the appropriateness of the diagnostic investigations and the correctness of the final diagnosis; (c) to individually characterize the underlying clinical conditions and possible risk factors for IE and NIEcorrelated; and finally (d) to identify ICU-acquired IE and NIE, we took advantage of the high rate of autopsies performed in our hospital. We examined all the autopsy records of patients deceased in our ICU, and cross-checked those in whom endocarditis was found with their respective clinical records.
Patients and methods
Fig. 1 Infective endocarditis of the aortic valve; Hematoxylin–eosin, 91
Fig. 2 Non-infective endocarditis of the aortic valve; Hematoxylin– eosin, 91
of transthoracic (TTE) and transesophageal (TEE) echocardiography. By necessity, these investigations are usually obtained on the basis of symptoms that can go undetected in ICU patients, thus making this diagnosis elusive. Several authors studied the clinical course and the complications of patients admitted to the ICU with an already established diagnosis of IE or NIE [7–10]. Fewer data are available about these disorders in patients admitted for other reasons; at the same time, as far as nosocomial IE/ NIE are concerned, it is of relevant interest to separate
123
We retrospectively reviewed the post-mortem examinations of patients who died in our 13-bed general ICU between January 1, 1996 and December 31, 2010. As the study was retrospective and did not imply any intervention or violation of the privacy, the consent of the local ethical committee was waived. Our ICU is located inside a 750-bed University hospital, and admits about 1,000 patients/year. Patients with acute cardiac conditions not requiring mechanical ventilation as well as cardiac surgery patients who were treated in different units were not included in this study. Similarly, we excluded patients who had died after trauma as, according to the current Italian Law, autopsy findings cannot be disclosed to the physicians in charge. We also excluded obstetric and pediatric patients, as they are treated in another University maternal and child health hospital. The autopsies were performed by a single experienced pathologist with a particular expertise in critical and cardiovascular diseases. The diagnosis of endocarditis was based on the presence of vegetations on the cardiac valves or on the endocardium. Moreover, the differentiation between IE and NIE was based on the results of microscopic examination of the vegetations. An endocarditis was considered hospital or ICU-acquired if it occurred C72 h post ward or ICU admission, respectively [7, 9]. Subsequently, the medical records of each positive case were retrieved, and a number of data were collected, including (a) the presence of the diagnosis of proven or suspected endocarditis and the related investigations; (b) all the relevant personal information (age, gender and history); (c) the length of stay (LOS) in hospital before ICU admission (LOSpre) and the length of stay in ICU (LOSICU); (d) the type of surgical procedures performed before or during the ICU admission; (e) the presence of congenital valvular abnormalities and artificial valves;
Intern Emerg Med
(f) the occurrence of SIRS, sepsis and sepsis-related complications, defined according to the SCCM criteria [11]; (g) the bacteriological results; (h) the presence of indwelling CVC, PAC and IAL lines; and (i) the cardiac and extracardiac complications of IE and NIE. The vegetations of NIE were subdivided into five different types according to their macroscopic appearance [12] (Table 1). We used the free software R for statistical analysis (R Development Core Team, R Foundation for Statistical Computing, Vienna, Austria 2008). The statistical analysis was performed with the Fisher’s exact t test; a p \ 0.05 was considered statistically significant.
Table 3 Reason for admission to hospital and to ICU, and clinical cause(s) of death Reason for admission to hospital
Reason for admission to ICU
Cause of death
1
Chronic obstructive pulmonary disease
Respiratory failure
Multiple organ failure
2
Acute abdomen
Post-surgery monitoring
Septic shock
3
Acute abdomen
Post-surgery monitoring
Cardiac arrest
4
Cardiac arrest
Cardiac arrest
Post-anoxic coma
5
Bronchopneumonia
Sepsis, suspected EI
Cerebral infarction
6
C-ANCA associated vasculitis
Suspected pulmonary embolism
Multiple organ failure
7
Abdominal aortic aneurysm
Septic shock, suspected IE
Septic shock
8
Rectal bleeding
Sepsis
Septic shock
9
Bronchopneumonia
Respiratory failure
IE, heart failure
10
Cerebral hemorrhage, suspected IE
Cerebral hemorrhage, suspected EI
Cerebral hemorrhage, IE
11
Suspected IE, myopericarditis
Cardiogenic/ septic shock
Septic shock
12
Diarrhea
Cerebral hemorrhage
ARDS, IE
1
Acute abdomen
Suspected pulmonary embolism
Sepsis
2
Bronchopneumonia
Respiratory failure
Cardiocirculatory shock
3
Bronchopneumonia
Respiratory failure
Cardiac arrest
4
Bronchopneumonia
Respiratory failure
Pulmonary embolism
5
Asthenia
Cardiac arrest
Cardiocirculatory shock
6 7
Abdominal pain Perineal fistula
Acute abdomen Respiratory failure
Septic shock Septic shock
8
Decompensated diabetes mellitus
Sepsis
Septic shock
9
Obliterant arteriopathy of the lower limbs
Post-surgery monitoring
Acute myocardial infarction
IE
Results In the study period, 765 autopsies were performed. At gross examination of the heart, 21 patients (2.7 %) presented vegetations that were subsequently divided into IE and NIE on the basis of the criteria listed above (Table 1). The hospital and ICU admission diagnoses and the cause of death reported on the medical charts are described in Tables 2 and 3. In three patients, all diagnosed with IE, one or two mechanical prosthetic valves were present, whose location and type are described in Table 4. In three cases of IE, a concomitant myocarditis was present. Similar longstanding conditions were present in both groups (Table 5). Table 1 Macroscopic classification of the NIE [10] Type
Findings
1
Small, \3 mm, unverrucal, firmly attached to the valve
2
Large, [3 mm, unverrucal, adherent to the valve
3
Small, 1–3 mm, multiverrucal, friable
4
Large, [3 mm, multiverrucal, friable
5
‘‘Healed type’’, similar in consistency to the valve attached
Table 2 Characteristics of patients with endocarditis
N (%)
IE
NIE
12
9
p
Gender \0.05
Male
12
4
Female
0
5
Age (years)
66 (47.5–73)
68 (64–77)
ns
LOSICU (days) LOSPRE-ICU (days)
10 (7.75–17.5) 10.5 (5.75–22.25)
3 (3–14) 8 (3–23)
ns ns
IE infective endocarditis, ARSD acute respiratory distress syndrome
Legend data are numbers except for age, LOSICU and LOSPRE-ICU which are medians (interquartile range), LOS length of stay
NIE
The underlying pathological conditions and the procedures performed during the ICU admission are reported in Tables 5 and 6. Notably, 11 out of 12 patients with IE and all patients with NIE had an indwelling CVC or PAC and an IAL was present in all patients.
123
Intern Emerg Med Table 4 Location of endocarditis
IE
NIE
Native valve
9
9
Aortic
2
3
Mitral
3
5
Pulmonary
1
0
Tricuspid
2
0
Aortic ? mitral
1
0
Left ventricular wall
0
1
Prosthetic valve
3
0
Aortic
0
0
Mitral
2
0
Pulmonary
0
0
Tricuspid
0
0
Aortic ? mitral
1
0
Left ventricular wall
0
0
Variable
IE
NIE
p
1
ns
Aortic stenosis and insufficiency
0
0
ns
Previous valvular surgery
3
0
ns
Drug addiction
1
0
ns
Previous radiotherapy
1
1
ns
1
3
ns
Abdominal
3
3
ns
Vascular
1
1
ns
Abdominal ? vascular
1
0
ns
Neurosurgical
1
0
ns
Thoracic
0
1
ns
Indwelling CVC
11
9
ns
IAL Mechanical ventilation
12 12
9 9
ns ns
Heart pacing
2
0
ns
Acute heart failure
4
1
ns
Ischemic heart disease
1
0
ns
Decompensated dilated cardiomyopathy
1
9
\0.05
Decompensated hypertensive cardiomyopathy
2
0
ns
Severe sepsis/septic shock
9
4
ns
RRT renal replacement therapy, CVC central venous catheter, IAL intra-arterial line
Table 7 Location of peripheral embolism IE
NIE
p
ns
Brain
5
1
ns
1
0
ns ns
1
1
ns
Heart
4 4
0 2
ns ns
Spleen
2
0
Kidney ? spleen
1
1
ns
Heart ? spleen ? kidney
1
0
ns
Previous non-cardiac surgical procedures Thoracic
1
Location
Renal failure
Abdominal
3
Acute conditions
3
Acute, RRT Chronic anticoagulant therapy
p
RRT
Mitral insufficiency
Chronic, RRT
NIE
Non-cardiac surgical procedures
Preexisting valvular abnormality
Chronic, no RRT
IE
Procedures and interventions
Table 5 Concomitant long-standing pathological conditions Conditions
Table 6 Procedures performed during the admission, and concomitant acute condition
1
3
ns
Heart ? kidney
1
0
ns
Lungs
0
5
\0.05
1
0
ns
Diabetes mellitus
2
3
ns
Solid tumors
1
0
ns
Chronic lung disease
1
1
ns
Others
Immune disorders
1
2
ns
Hepatitis C
1
0
ns
Tuberculosis
1
2
ns
RRT renal replacement therapy
Multiple peripheral embolisms were frequently discovered both in patients with IE and NIE: the brain appeared to be particularly affected in patients with IE, while pulmonary embolisms were commonly associated with NIE (Table 7). Blood cultures were positive in nine patients with IE (Table 8).
123
A TTE was performed in five patients with IE and in four patients with NIE. TTE findings were suggestive of infective endocarditis, but not conclusive in two patients of the IE group. In addition, it showed a valve vegetation in one patient with NIE. A TEE was obtained in three patients with IE, two were positive. It was not performed in the NIE group. Independently from the technique used, in no case was the size of the observed abnormalities reported. Overall, an IE was correctly diagnosed before death in three patients (25 %) and suspected in two other cases (17 %). As far as the NIE is concerned, it was diagnosed before death in one patient only (11 %). In NIE patients, two cases were classified as Type 1, six as Type 3 and the last one belonged to Type 5 (Table 8).
Intern Emerg Med Table 8 Results of blood cultures in patients with brain embolism Germ
N
MSSA
2
MRSA
4
Bacteroides fragilis
1
Streptococcus agalactiae
1
MRSA ? enterobacter cloacae
1
MSSA multiple drug-sensible Staphylococcus aureus, MRSA multiple drug-resistant Staphylococcus aureus
Discussion A number of severe, lethal conditions may go unnoticed throughout the entire hospital stay, and eventually be discovered only at the autopsy. Such pathologies are often unrelated to the cause of admission, hospital- or ICUacquired, but may often be potentially treatable. Some investigators have demonstrated by post-mortem studies among patients admitted both to regular wards and ICU that the rate of undetected major pathological conditions ranges from 12.6 to 38 %, and that their clinical relevance can vary from nil to being the main cause of death [13–15]. Critically ill patients are particularly vulnerable to the risk of missed diagnosis due to different factors, including the use of sedatives, the concomitant action of different disorders causing similar symptoms, immunosuppression and the presence of sepsis-induced systemic disturbances [14–16]. Actually, either IE or NIE can be particularly difficult to recognize in this setting since (a) the related cardiovascular symptoms may be attributed to other conditions that determined the admission to the ICU or complicated the clinical course, including heart failure, septic shock, etc., and (b) the neurologic signs or symptoms can go undetected due to the use of sedatives and neuromuscular blocking agents. Since there is a paucity of data regarding the incidence of either IE and NIE among critically ill patients admitted to the ICU for other reasons, we moved backward, initiating from the postmortem findings to arrive at the clinical data. The crosscheck of medical and autoptical records supported some interesting findings. First, although in our study the raw incidence of both IE and NIE was lower than other difficult to diagnose potentially lethal conditions, including pulmonary embolisms (PE) and acute myocardial infarction (AMI), the overall rate of missed diagnosis of both IE and NIE was higher than that reported in these conditions: in previous studies performed on a similar population of patients, we find that roughly 16 % of PE and 27 % of AMI are diagnosed only at autopsy [17, 18]. The actual rate of missed diagnosis of endocarditis is difficult to estimate, as it varies widely
among different studies. Saad et al. [19] demonstrate the diagnosis of IE is missed in 27 % of patients who had died in a cardiac hospital; conversely, in a very recent longitudinal study performed in patients who died in a general ICU, IE was discovered at the autopsy in 1 % of cases. Surprisingly, in this latter investigation, the rate of missed diagnosis of IE remained quite stable during the whole 20-year-long study period [16], thus underscoring the difficulties encountered in the diagnosis of this disorder despite the impressive improvement in the imaging techniques. The high rate of false-negative echocardiograms observed in our study confirms this observation, strengthening the concept that endocarditis can go unnoticed even with the most sophisticated diagnostic procedures. Little is known about the incidence of NIE in ICU patients: our observed rate of 1 % is in agreement with other investigations performed in patients admitted to regular wards, which demonstrate an incidence ranging 1–1.65 % [4, 20, 21]. Second, the location of the two forms differed, IE having been present on both sides and NIE limited to the left side. Other investigators also demonstrate a clear prevalence of the left-sided IE in non-IV drug abusers, who, on the other hand, present a more pronounced involvement of the tricuspid valve [1, 22]. The preferential involvement of NIE for left-sided valves is observed also by other investigators [4, 20–22]. Third, the brain was the organ most affected by fragments originating from the valvular and mural vegetations. Neurologic symptoms are the most common complication of IE and often represent the cause of ICU admission and death [7, 8]. Finally, in our study as well as in others, gram-positive strains were isolated from blood cultures drawn shortly before death. As no patient was admitted to the ICU with an established diagnosis of endocarditis or directly related symptoms, using the indicated cut-off time value of 72 h [7, 9] it is likely that (a) most of the cases of IE should be considered hospital if not ICU-acquired; and (b) the probable cause of IE was a bloodstream infection most likely originating from the CVC, PAC or IAL. Naturally, on the basis of the available data, we cannot exclude that some patients had an IE at the ICU admission but that the symptoms went unnoticed due to the aforementioned conditions. Conversely, we cannot draw a similar conclusion for NIE, as no time limits have been used to separate community from nosocomial cases; however, we cannot exclude that in patients with sepsis, the related pro-coagulative state might have contributed to their formation, making them an ICU-acquired disease [6]. Worldwide, the rate of autopsies performed in patients dying in hospital is declining due to a number of reasons, including the need for cost containment, the fear of
123
Intern Emerg Med
litigation and the continuous improvement of imaging techniques. Thus, one could argue that autopsy does not play any role in the confirmation of either the diagnostic accuracy, at least in patients who die after a reasonable time frame during which all the necessary investigations have been carried out. However, several authors have demonstrated that despite all the technological improvements, a number of life-threatening, potentially treatable disorders are still detected only at the post-mortem examination. As reported in the literature, this diagnostic discrepancy applies particularly to critical patients, in whom a host of circumstances can make the diagnosis of potentially treatable disorders, such as IE, NIE, PE and AMI particularly challenging [23, 24]. Consistent with the unquestionable value of post-mortem examination, our institution has maintained its tradition of a high autopsy rate.
Conclusions In critically ill patients admitted to general ICUs, multiple factors related both to the underlying conditions and to the performed procedures facilitate the occurrence of both infective and non-infective endocarditis. At the same time, the diagnosis can be extremely challenging due to concomitant factors that might mask the related symptoms. As a matter of fact, in our experience, both forms went largely under-diagnosed and their presence was detected only at autopsy, thus proving to be a good tool for the evaluation of the diagnostic accuracy as well as for its consequent didactic role. Conflict of interest
None.
References 1. Thury F, Grisoli D, Collart F, Habib G, Raoult D (2012) Management of infective endocarditis: challenges and perspectives. Lancet 379:965–975 2. Moreillon P, Que YO (2004) Infective endocarditis. Lancet 363:139–149 3. Hoen B, Duval X (2013) Infective endocarditis. N Engl J Med 368:1425–1433 4. Truskinowsky AM, Hutchins GM (2001) Association between nonbacterial thrombotic endocarditis and hypoxigenic pulmonary diseases. Virchows Arch 438:357–361 5. Lee V, Gilbert JD, Byard RW (2012) Marantic endocarditis—a not so benign entity. J Forensic Leg Med 19:312–315 6. Della Valle P, Pavani G, D’AAnelo A (2012) The protein C pathway and sepsis. Thromb Res 29:296–300
123
7. Goue¨llo JP, Asfar P, Brenet O, Kouatchet A, Berthelot G, Alquier P (2000) Nosocomial endocarditis in the intensive care unit: an analysis of 22 cases. Crit Care Med 28:377–382 8. Sonneville R, Mirable M, Hajage D et al (2011) Neurologic complications and outcome of infective endocarditis in critically ill patients: the Endocardite er Reanimation prospective multicenter study. Crit Care Med 36:1474–1481 9. Mattı`n-Da´vila P, Fortu´n J, Navs E et al (2005) Nosocomial endocarditis in a tertiary hospital—an increasing trend in native valve cases. Chest 128:772–779 10. Karth G, Koreny M, Binder T et al (2002) Complicated infective endocarditis necessitating ICU admission: clinical course and prognosis. Crit Care 6:149–154 11. Bone RC, Balk RA, Cerra FB et al (2009) ACCP/SCCM consensus conference committee. Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. The ACCP/SCCM consensus conference committee. American College of Chest Physicians/Society of Critical Care Medicine 1992. Chest 136(5 Suppl):e28 12. Asopa S, Patel A, Khan OA, Sharma R, Ohri SK (2007) Non bacterial thrombotic endocarditis. Eur J Cardio-thoracic Surg 32:696–701 13. Shojania KG, Burton EC, McDonald KM, Goldman L (2003) Changes in rates of autopsy detected diagnostic errors over time—a systematic review. JAMA 289:2849–2856 14. Combes A, Mokhtari M, Couvelard et al (2004) Clinical and autopsy diagnoses in the intensive care unit. Arch Intern Med 164:392 15. Nadrous HF, Afessa B, Pfeifer EA, Peters SG (2003) The role of autopsy in the intensive care unit. Mayo Clin Proc 78:947–950 16. Tejerina E, Esteban E, Fernandez-Segoviano P et al (2012) Clinical diagnosis and autopsy findings: discrepancies in critically ill patients. Crit Care Med 40:842–846 17. Berlot G, Vergolini A, Calderan C et al (2010) Acute myocardial infarction in non-cardiac critically ill patients: a clinical–pathological study. Monaldi Arch Chest Dis 74:164–171 18. Berlot G, Calderan C, Vergolini A et al (2011) Pulmonary embolism in critically ill patients receiving antithrombotic prophylaxis. A clinical–pathological study. J Crit Care 26:28–33 19. Saad R, Yamada AT, Pereira da Rosa FH, Gutierrez PS, Mansur AJ (2007) Comparison between clinical and autopsy diagnoses in a cardiology hospital. Heart 93:1414–1419 20. Ferlan G, Fiorella A, De Pasquale C, Tunzi F (2010) Primary coronary embolism as an unusual manifestation of nonbacterial thrombotic endocarditis in a patient with gastric cancer. Cardiol Res Pract 2010:319732 21. Llenas-Garcia J, Guerra-Vales J, Montes-moreno S, Lopez-Rios S, Castelbon-frenadez S, Chimeno-garcia J (2007) Nonbacterial endocarditis: clinicopathologic study of a necropsy series. Rev Esp Cardiol 60:493–500 22. Mourvillier B, Trouillet JL, Timsit JF et al (2004) Infective endocarditis in the intensive care unit: clinical spectrum and prognostic factors in 228 consecutive patients. Intensive Care Med 30:2046–2052 23. Burton JL, Underwood J (2007) Clinical, educational and epidemiological value of autopsy. Lancet 369:1471–1480 24. De Vlieger GY, Mahieu EM, Meersseman W (2010) Clinical review: what is the role for autopsy in the ICU? Crit Care 14:221–228
IM - ORIGINAL
Infective and non-infective endocarditis in critically ill patients: a clinical–pathological study Giorgio Berlot • Cristina Calderan • Cristina Fiorenza • Davide Cappelli • Stefano Addesa • Rossana Bussani
Received: 23 October 2013 / Accepted: 23 January 2014 Ó SIMI 2014
Abstract The aims of this study are to estimate the incidence, the outcome and the associated risk factors of infective and non-infective endocarditis (IE and NIE, respectively) in intensive care unit (ICU) patients. We studied the post-mortem findings and the clinical data of the patients who died in our ICU between 1996 and 2010. Of the 765 reviewed autopsies, 21 patients (2.7 %) presented cardiac vegetations. These cases consisted of 12 IEs and 9 NIEs. Three patients with IE had a mechanical prosthetic valve, and in 11 cases invasive devices had been used. Multiple peripheral embolisms were discovered at autopsy. In particular, the brain appeared to be more affected in patients with IE, while pulmonary embolisms were commonly associated with NIE. Blood cultures were positive in nine patients with IE. The imaging diagnostics (transthoracic and transesophageal echocardiography) which were seldom performed in both groups, proved to be of little help. As a consequence, an IE was correctly diagnosed before death in three patients (25 %) and suspected in two other cases (17 %), while a NIE was diagnosed before death in one patient alone. In conclusions, critically ill patients admitted to general ICUs, multiple factors related both to the underlying conditions and to performed procedures can facilitate the occurrence of IE and NIE making, at the same time, their diagnosis challenging. Many cases, in fact, are diagnosed only at autopsy. Yet again, post-mortem examination proves to be an G. Berlot (&) C. Calderan C. Fiorenza D. Cappelli S. Addesa Department of Anaesthesia and Intensive Care, Cattinara Hospital, Strada di Fiume 447, 34149 Trieste, Italy e-mail: [email protected] R. Bussani Department of Pathology, University of Trieste, Trieste, Italy
invaluable tool for the evaluation of diagnostic accuracy in critical care. Keywords Infective endocarditis Non-infective endocarditis Autopsy Clinical errors Missed diagnoses
Introduction The estimated incidence of infective endocarditis (IE) ranges between 30 and 100 episodes per million patientyears [1] with a mortality rate exceeding 30 %. The main risk factors for its occurrence include underlying congenital or degenerative valvular abnormalities, the presence of prosthetic valves, bloodstream infections and intravenous drug use [2, 3] (Fig. 1). Different causes account for non-infective endocarditis (NIE), including neoplasms, multifactorial pro-coagulative states, long-term hypoxia, chronic low flow states and cachexia [4, 5] (Fig. 2). The very same factors are responsible for the occurrence of both forms of endocarditis, in the hospital and in critically ill patients admitted to intensive care unit (ICU). All ICU patients are particularly prone to bloodstream infection-related IE as compared to regular ward patients, mainly due to the widespread use of invasive devices such as central venous catheters (CVC), pulmonary artery catheters (PAC) and intra-arterial lines (IAL). On the other hand, the ever-increasing age of these patients with the related burden of disease, the procoagulative state associated with sepsis and sepsis-related conditions and the above-enlisted risk factors set the stage for NIE [6]. The current diagnoses of both IE and NIE are based on the detection of valvular or mural vegetations by means
123
Intern Emerg Med
those acquired in the regular wards from those originating in the ICU. Therefore, (a) to estimate the incidence in a population of patients admitted to a general ICU for causes different than IE and NIE; (b) to evaluate the appropriateness of the diagnostic investigations and the correctness of the final diagnosis; (c) to individually characterize the underlying clinical conditions and possible risk factors for IE and NIEcorrelated; and finally (d) to identify ICU-acquired IE and NIE, we took advantage of the high rate of autopsies performed in our hospital. We examined all the autopsy records of patients deceased in our ICU, and cross-checked those in whom endocarditis was found with their respective clinical records.
Patients and methods
Fig. 1 Infective endocarditis of the aortic valve; Hematoxylin–eosin, 91
Fig. 2 Non-infective endocarditis of the aortic valve; Hematoxylin– eosin, 91
of transthoracic (TTE) and transesophageal (TEE) echocardiography. By necessity, these investigations are usually obtained on the basis of symptoms that can go undetected in ICU patients, thus making this diagnosis elusive. Several authors studied the clinical course and the complications of patients admitted to the ICU with an already established diagnosis of IE or NIE [7–10]. Fewer data are available about these disorders in patients admitted for other reasons; at the same time, as far as nosocomial IE/ NIE are concerned, it is of relevant interest to separate
123
We retrospectively reviewed the post-mortem examinations of patients who died in our 13-bed general ICU between January 1, 1996 and December 31, 2010. As the study was retrospective and did not imply any intervention or violation of the privacy, the consent of the local ethical committee was waived. Our ICU is located inside a 750-bed University hospital, and admits about 1,000 patients/year. Patients with acute cardiac conditions not requiring mechanical ventilation as well as cardiac surgery patients who were treated in different units were not included in this study. Similarly, we excluded patients who had died after trauma as, according to the current Italian Law, autopsy findings cannot be disclosed to the physicians in charge. We also excluded obstetric and pediatric patients, as they are treated in another University maternal and child health hospital. The autopsies were performed by a single experienced pathologist with a particular expertise in critical and cardiovascular diseases. The diagnosis of endocarditis was based on the presence of vegetations on the cardiac valves or on the endocardium. Moreover, the differentiation between IE and NIE was based on the results of microscopic examination of the vegetations. An endocarditis was considered hospital or ICU-acquired if it occurred C72 h post ward or ICU admission, respectively [7, 9]. Subsequently, the medical records of each positive case were retrieved, and a number of data were collected, including (a) the presence of the diagnosis of proven or suspected endocarditis and the related investigations; (b) all the relevant personal information (age, gender and history); (c) the length of stay (LOS) in hospital before ICU admission (LOSpre) and the length of stay in ICU (LOSICU); (d) the type of surgical procedures performed before or during the ICU admission; (e) the presence of congenital valvular abnormalities and artificial valves;
Intern Emerg Med
(f) the occurrence of SIRS, sepsis and sepsis-related complications, defined according to the SCCM criteria [11]; (g) the bacteriological results; (h) the presence of indwelling CVC, PAC and IAL lines; and (i) the cardiac and extracardiac complications of IE and NIE. The vegetations of NIE were subdivided into five different types according to their macroscopic appearance [12] (Table 1). We used the free software R for statistical analysis (R Development Core Team, R Foundation for Statistical Computing, Vienna, Austria 2008). The statistical analysis was performed with the Fisher’s exact t test; a p \ 0.05 was considered statistically significant.
Table 3 Reason for admission to hospital and to ICU, and clinical cause(s) of death Reason for admission to hospital
Reason for admission to ICU
Cause of death
1
Chronic obstructive pulmonary disease
Respiratory failure
Multiple organ failure
2
Acute abdomen
Post-surgery monitoring
Septic shock
3
Acute abdomen
Post-surgery monitoring
Cardiac arrest
4
Cardiac arrest
Cardiac arrest
Post-anoxic coma
5
Bronchopneumonia
Sepsis, suspected EI
Cerebral infarction
6
C-ANCA associated vasculitis
Suspected pulmonary embolism
Multiple organ failure
7
Abdominal aortic aneurysm
Septic shock, suspected IE
Septic shock
8
Rectal bleeding
Sepsis
Septic shock
9
Bronchopneumonia
Respiratory failure
IE, heart failure
10
Cerebral hemorrhage, suspected IE
Cerebral hemorrhage, suspected EI
Cerebral hemorrhage, IE
11
Suspected IE, myopericarditis
Cardiogenic/ septic shock
Septic shock
12
Diarrhea
Cerebral hemorrhage
ARDS, IE
1
Acute abdomen
Suspected pulmonary embolism
Sepsis
2
Bronchopneumonia
Respiratory failure
Cardiocirculatory shock
3
Bronchopneumonia
Respiratory failure
Cardiac arrest
4
Bronchopneumonia
Respiratory failure
Pulmonary embolism
5
Asthenia
Cardiac arrest
Cardiocirculatory shock
6 7
Abdominal pain Perineal fistula
Acute abdomen Respiratory failure
Septic shock Septic shock
8
Decompensated diabetes mellitus
Sepsis
Septic shock
9
Obliterant arteriopathy of the lower limbs
Post-surgery monitoring
Acute myocardial infarction
IE
Results In the study period, 765 autopsies were performed. At gross examination of the heart, 21 patients (2.7 %) presented vegetations that were subsequently divided into IE and NIE on the basis of the criteria listed above (Table 1). The hospital and ICU admission diagnoses and the cause of death reported on the medical charts are described in Tables 2 and 3. In three patients, all diagnosed with IE, one or two mechanical prosthetic valves were present, whose location and type are described in Table 4. In three cases of IE, a concomitant myocarditis was present. Similar longstanding conditions were present in both groups (Table 5). Table 1 Macroscopic classification of the NIE [10] Type
Findings
1
Small, \3 mm, unverrucal, firmly attached to the valve
2
Large, [3 mm, unverrucal, adherent to the valve
3
Small, 1–3 mm, multiverrucal, friable
4
Large, [3 mm, multiverrucal, friable
5
‘‘Healed type’’, similar in consistency to the valve attached
Table 2 Characteristics of patients with endocarditis
N (%)
IE
NIE
12
9
p
Gender \0.05
Male
12
4
Female
0
5
Age (years)
66 (47.5–73)
68 (64–77)
ns
LOSICU (days) LOSPRE-ICU (days)
10 (7.75–17.5) 10.5 (5.75–22.25)
3 (3–14) 8 (3–23)
ns ns
IE infective endocarditis, ARSD acute respiratory distress syndrome
Legend data are numbers except for age, LOSICU and LOSPRE-ICU which are medians (interquartile range), LOS length of stay
NIE
The underlying pathological conditions and the procedures performed during the ICU admission are reported in Tables 5 and 6. Notably, 11 out of 12 patients with IE and all patients with NIE had an indwelling CVC or PAC and an IAL was present in all patients.
123
Intern Emerg Med Table 4 Location of endocarditis
IE
NIE
Native valve
9
9
Aortic
2
3
Mitral
3
5
Pulmonary
1
0
Tricuspid
2
0
Aortic ? mitral
1
0
Left ventricular wall
0
1
Prosthetic valve
3
0
Aortic
0
0
Mitral
2
0
Pulmonary
0
0
Tricuspid
0
0
Aortic ? mitral
1
0
Left ventricular wall
0
0
Variable
IE
NIE
p
1
ns
Aortic stenosis and insufficiency
0
0
ns
Previous valvular surgery
3
0
ns
Drug addiction
1
0
ns
Previous radiotherapy
1
1
ns
1
3
ns
Abdominal
3
3
ns
Vascular
1
1
ns
Abdominal ? vascular
1
0
ns
Neurosurgical
1
0
ns
Thoracic
0
1
ns
Indwelling CVC
11
9
ns
IAL Mechanical ventilation
12 12
9 9
ns ns
Heart pacing
2
0
ns
Acute heart failure
4
1
ns
Ischemic heart disease
1
0
ns
Decompensated dilated cardiomyopathy
1
9
\0.05
Decompensated hypertensive cardiomyopathy
2
0
ns
Severe sepsis/septic shock
9
4
ns
RRT renal replacement therapy, CVC central venous catheter, IAL intra-arterial line
Table 7 Location of peripheral embolism IE
NIE
p
ns
Brain
5
1
ns
1
0
ns ns
1
1
ns
Heart
4 4
0 2
ns ns
Spleen
2
0
Kidney ? spleen
1
1
ns
Heart ? spleen ? kidney
1
0
ns
Previous non-cardiac surgical procedures Thoracic
1
Location
Renal failure
Abdominal
3
Acute conditions
3
Acute, RRT Chronic anticoagulant therapy
p
RRT
Mitral insufficiency
Chronic, RRT
NIE
Non-cardiac surgical procedures
Preexisting valvular abnormality
Chronic, no RRT
IE
Procedures and interventions
Table 5 Concomitant long-standing pathological conditions Conditions
Table 6 Procedures performed during the admission, and concomitant acute condition
1
3
ns
Heart ? kidney
1
0
ns
Lungs
0
5
\0.05
1
0
ns
Diabetes mellitus
2
3
ns
Solid tumors
1
0
ns
Chronic lung disease
1
1
ns
Others
Immune disorders
1
2
ns
Hepatitis C
1
0
ns
Tuberculosis
1
2
ns
RRT renal replacement therapy
Multiple peripheral embolisms were frequently discovered both in patients with IE and NIE: the brain appeared to be particularly affected in patients with IE, while pulmonary embolisms were commonly associated with NIE (Table 7). Blood cultures were positive in nine patients with IE (Table 8).
123
A TTE was performed in five patients with IE and in four patients with NIE. TTE findings were suggestive of infective endocarditis, but not conclusive in two patients of the IE group. In addition, it showed a valve vegetation in one patient with NIE. A TEE was obtained in three patients with IE, two were positive. It was not performed in the NIE group. Independently from the technique used, in no case was the size of the observed abnormalities reported. Overall, an IE was correctly diagnosed before death in three patients (25 %) and suspected in two other cases (17 %). As far as the NIE is concerned, it was diagnosed before death in one patient only (11 %). In NIE patients, two cases were classified as Type 1, six as Type 3 and the last one belonged to Type 5 (Table 8).
Intern Emerg Med Table 8 Results of blood cultures in patients with brain embolism Germ
N
MSSA
2
MRSA
4
Bacteroides fragilis
1
Streptococcus agalactiae
1
MRSA ? enterobacter cloacae
1
MSSA multiple drug-sensible Staphylococcus aureus, MRSA multiple drug-resistant Staphylococcus aureus
Discussion A number of severe, lethal conditions may go unnoticed throughout the entire hospital stay, and eventually be discovered only at the autopsy. Such pathologies are often unrelated to the cause of admission, hospital- or ICUacquired, but may often be potentially treatable. Some investigators have demonstrated by post-mortem studies among patients admitted both to regular wards and ICU that the rate of undetected major pathological conditions ranges from 12.6 to 38 %, and that their clinical relevance can vary from nil to being the main cause of death [13–15]. Critically ill patients are particularly vulnerable to the risk of missed diagnosis due to different factors, including the use of sedatives, the concomitant action of different disorders causing similar symptoms, immunosuppression and the presence of sepsis-induced systemic disturbances [14–16]. Actually, either IE or NIE can be particularly difficult to recognize in this setting since (a) the related cardiovascular symptoms may be attributed to other conditions that determined the admission to the ICU or complicated the clinical course, including heart failure, septic shock, etc., and (b) the neurologic signs or symptoms can go undetected due to the use of sedatives and neuromuscular blocking agents. Since there is a paucity of data regarding the incidence of either IE and NIE among critically ill patients admitted to the ICU for other reasons, we moved backward, initiating from the postmortem findings to arrive at the clinical data. The crosscheck of medical and autoptical records supported some interesting findings. First, although in our study the raw incidence of both IE and NIE was lower than other difficult to diagnose potentially lethal conditions, including pulmonary embolisms (PE) and acute myocardial infarction (AMI), the overall rate of missed diagnosis of both IE and NIE was higher than that reported in these conditions: in previous studies performed on a similar population of patients, we find that roughly 16 % of PE and 27 % of AMI are diagnosed only at autopsy [17, 18]. The actual rate of missed diagnosis of endocarditis is difficult to estimate, as it varies widely
among different studies. Saad et al. [19] demonstrate the diagnosis of IE is missed in 27 % of patients who had died in a cardiac hospital; conversely, in a very recent longitudinal study performed in patients who died in a general ICU, IE was discovered at the autopsy in 1 % of cases. Surprisingly, in this latter investigation, the rate of missed diagnosis of IE remained quite stable during the whole 20-year-long study period [16], thus underscoring the difficulties encountered in the diagnosis of this disorder despite the impressive improvement in the imaging techniques. The high rate of false-negative echocardiograms observed in our study confirms this observation, strengthening the concept that endocarditis can go unnoticed even with the most sophisticated diagnostic procedures. Little is known about the incidence of NIE in ICU patients: our observed rate of 1 % is in agreement with other investigations performed in patients admitted to regular wards, which demonstrate an incidence ranging 1–1.65 % [4, 20, 21]. Second, the location of the two forms differed, IE having been present on both sides and NIE limited to the left side. Other investigators also demonstrate a clear prevalence of the left-sided IE in non-IV drug abusers, who, on the other hand, present a more pronounced involvement of the tricuspid valve [1, 22]. The preferential involvement of NIE for left-sided valves is observed also by other investigators [4, 20–22]. Third, the brain was the organ most affected by fragments originating from the valvular and mural vegetations. Neurologic symptoms are the most common complication of IE and often represent the cause of ICU admission and death [7, 8]. Finally, in our study as well as in others, gram-positive strains were isolated from blood cultures drawn shortly before death. As no patient was admitted to the ICU with an established diagnosis of endocarditis or directly related symptoms, using the indicated cut-off time value of 72 h [7, 9] it is likely that (a) most of the cases of IE should be considered hospital if not ICU-acquired; and (b) the probable cause of IE was a bloodstream infection most likely originating from the CVC, PAC or IAL. Naturally, on the basis of the available data, we cannot exclude that some patients had an IE at the ICU admission but that the symptoms went unnoticed due to the aforementioned conditions. Conversely, we cannot draw a similar conclusion for NIE, as no time limits have been used to separate community from nosocomial cases; however, we cannot exclude that in patients with sepsis, the related pro-coagulative state might have contributed to their formation, making them an ICU-acquired disease [6]. Worldwide, the rate of autopsies performed in patients dying in hospital is declining due to a number of reasons, including the need for cost containment, the fear of
123
Intern Emerg Med
litigation and the continuous improvement of imaging techniques. Thus, one could argue that autopsy does not play any role in the confirmation of either the diagnostic accuracy, at least in patients who die after a reasonable time frame during which all the necessary investigations have been carried out. However, several authors have demonstrated that despite all the technological improvements, a number of life-threatening, potentially treatable disorders are still detected only at the post-mortem examination. As reported in the literature, this diagnostic discrepancy applies particularly to critical patients, in whom a host of circumstances can make the diagnosis of potentially treatable disorders, such as IE, NIE, PE and AMI particularly challenging [23, 24]. Consistent with the unquestionable value of post-mortem examination, our institution has maintained its tradition of a high autopsy rate.
Conclusions In critically ill patients admitted to general ICUs, multiple factors related both to the underlying conditions and to the performed procedures facilitate the occurrence of both infective and non-infective endocarditis. At the same time, the diagnosis can be extremely challenging due to concomitant factors that might mask the related symptoms. As a matter of fact, in our experience, both forms went largely under-diagnosed and their presence was detected only at autopsy, thus proving to be a good tool for the evaluation of the diagnostic accuracy as well as for its consequent didactic role. Conflict of interest
None.
References 1. Thury F, Grisoli D, Collart F, Habib G, Raoult D (2012) Management of infective endocarditis: challenges and perspectives. Lancet 379:965–975 2. Moreillon P, Que YO (2004) Infective endocarditis. Lancet 363:139–149 3. Hoen B, Duval X (2013) Infective endocarditis. N Engl J Med 368:1425–1433 4. Truskinowsky AM, Hutchins GM (2001) Association between nonbacterial thrombotic endocarditis and hypoxigenic pulmonary diseases. Virchows Arch 438:357–361 5. Lee V, Gilbert JD, Byard RW (2012) Marantic endocarditis—a not so benign entity. J Forensic Leg Med 19:312–315 6. Della Valle P, Pavani G, D’AAnelo A (2012) The protein C pathway and sepsis. Thromb Res 29:296–300
123
7. Goue¨llo JP, Asfar P, Brenet O, Kouatchet A, Berthelot G, Alquier P (2000) Nosocomial endocarditis in the intensive care unit: an analysis of 22 cases. Crit Care Med 28:377–382 8. Sonneville R, Mirable M, Hajage D et al (2011) Neurologic complications and outcome of infective endocarditis in critically ill patients: the Endocardite er Reanimation prospective multicenter study. Crit Care Med 36:1474–1481 9. Mattı`n-Da´vila P, Fortu´n J, Navs E et al (2005) Nosocomial endocarditis in a tertiary hospital—an increasing trend in native valve cases. Chest 128:772–779 10. Karth G, Koreny M, Binder T et al (2002) Complicated infective endocarditis necessitating ICU admission: clinical course and prognosis. Crit Care 6:149–154 11. Bone RC, Balk RA, Cerra FB et al (2009) ACCP/SCCM consensus conference committee. Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. The ACCP/SCCM consensus conference committee. American College of Chest Physicians/Society of Critical Care Medicine 1992. Chest 136(5 Suppl):e28 12. Asopa S, Patel A, Khan OA, Sharma R, Ohri SK (2007) Non bacterial thrombotic endocarditis. Eur J Cardio-thoracic Surg 32:696–701 13. Shojania KG, Burton EC, McDonald KM, Goldman L (2003) Changes in rates of autopsy detected diagnostic errors over time—a systematic review. JAMA 289:2849–2856 14. Combes A, Mokhtari M, Couvelard et al (2004) Clinical and autopsy diagnoses in the intensive care unit. Arch Intern Med 164:392 15. Nadrous HF, Afessa B, Pfeifer EA, Peters SG (2003) The role of autopsy in the intensive care unit. Mayo Clin Proc 78:947–950 16. Tejerina E, Esteban E, Fernandez-Segoviano P et al (2012) Clinical diagnosis and autopsy findings: discrepancies in critically ill patients. Crit Care Med 40:842–846 17. Berlot G, Vergolini A, Calderan C et al (2010) Acute myocardial infarction in non-cardiac critically ill patients: a clinical–pathological study. Monaldi Arch Chest Dis 74:164–171 18. Berlot G, Calderan C, Vergolini A et al (2011) Pulmonary embolism in critically ill patients receiving antithrombotic prophylaxis. A clinical–pathological study. J Crit Care 26:28–33 19. Saad R, Yamada AT, Pereira da Rosa FH, Gutierrez PS, Mansur AJ (2007) Comparison between clinical and autopsy diagnoses in a cardiology hospital. Heart 93:1414–1419 20. Ferlan G, Fiorella A, De Pasquale C, Tunzi F (2010) Primary coronary embolism as an unusual manifestation of nonbacterial thrombotic endocarditis in a patient with gastric cancer. Cardiol Res Pract 2010:319732 21. Llenas-Garcia J, Guerra-Vales J, Montes-moreno S, Lopez-Rios S, Castelbon-frenadez S, Chimeno-garcia J (2007) Nonbacterial endocarditis: clinicopathologic study of a necropsy series. Rev Esp Cardiol 60:493–500 22. Mourvillier B, Trouillet JL, Timsit JF et al (2004) Infective endocarditis in the intensive care unit: clinical spectrum and prognostic factors in 228 consecutive patients. Intensive Care Med 30:2046–2052 23. Burton JL, Underwood J (2007) Clinical, educational and epidemiological value of autopsy. Lancet 369:1471–1480 24. De Vlieger GY, Mahieu EM, Meersseman W (2010) Clinical review: what is the role for autopsy in the ICU? Crit Care 14:221–228
