This study is, to our knowledge, the largest prospective evaluation of LV diastolic dysfunction by TTE in patients admitted to the ICU for severe sepsis or septic shock. LV diastolic dysfunction is relatively common during severe sepsis and septic shock with 34.6% and 61.8% of the cohort showing evidence of LV diastolic dysfunction on admission or at some point during their clinical course of sepsis, respectively. Notably, grade I diastolic dysfunction (generally felt to indicate a relatively low left atrial pressure) was associated with worse outcome when compared with grade II or III diastolic dysfunction. This novel finding was not consistent with our original hypothesis.
The current study is in conflict with a single previous study that suggested worse outcome with more severe LV diastolic dysfunction
. The reason for this discrepancy may be related to patient population, treatment algorithms, timing of the TTE, or definition of diastolic dysfunction employed. Our study specifically suggests that among patients presenting to the ICU with impaired diastolic function, those with low filling pressures as assessed by TTE have worse outcome than patients with LV diastolic dysfunction and higher filling pressures. The epidemiology and implications of diastolic dysfunction during critical illness appear to be quite different from that among community dwellers with stable disease or patients with acute cardiovascular disease (decompensated congestive heart failure or myocardial infarction). A similar effect has been observed in LV systolic function with a lower left ventricular ejection fraction associated with better survival in sepsis
[1, 4], although there was no difference in mortality in our cohort of patients with normal versus low LV systolic function.
Rivers and colleagues' pivotal study of early aggressive optimization of oxygen delivery in patients with severe sepsis or septic shock suggested that initial resuscitation was crucial to outcome
. That pivotal study, however, relied upon CVP to assess adequacy of volume expansion, a measurement that has been criticized on grounds that CVP poorly predicts an increase in cardiac output after volume expansion
[33–35]. In our study, the mean CVP among those patients with grade I diastolic dysfunction was the same as that among patients with grade II or III diastolic dysfunction. Our study suggests that there may exist a group of patients with grade I diastolic dysfunction on TTE who would benefit from further volume expansion despite an elevated CVP. While small numbers in the subgroup of patients with grade I diastolic dysfunction limit generalizability, the higher mortality and lesser volume expansion prior to ICU admission suggest that further volume resuscitation in this group may improve survival. This study may provide equipoise for an interventional study of echocardiographic versus CVP-driven volume expansion among patients with early severe sepsis or septic shock.
Our study also highlights the need to establish a consistent, reproducible definition of LV diastolic dysfunction among critically ill patients, especially those with severe sepsis or septic shock. Prior studies report a varied incidence of LV diastolic dysfunction in patients with severe sepsis or septic shock, which may reflect a difference in patient populations and different definitions of diastolic dysfunction. In a retrospective review of 94 general ICU patients, Sturgess et al. reported an incidence of LV diastolic dysfunction equal to 67%
. Applying Sturgess and colleagues' definition to our cohort, we found a relatively consistent incidence of LV diastolic dysfunction on the initial echocardiogram of 61.7%. In a small study of 35 patients with septic shock requiring mechanical ventilation, Etchecopar-Chevreuil et al. found that LV diastolic dysfunction occurred in 20% of patients
. Bouhemad et al. reported a 20% incidence of isolated LV diastolic dysfunction in 54 post-operative patients with septic shock using transesophageal echocardiography
. Left atrial size is often used as a criterion for diagnosing LV diastolic dysfunction
. However, left atrial enlargement is most likely a slowly developing adaptation to chronically elevated LV filling pressures
. In the case of sepsis, diastolic abnormalities are frequently acute, and LA enlargement is not expected to occur rapidly. Therefore, LA size is not likely to be a reliable marker of diastolic filling abnormalities in this specific condition. Because of this, we chose to rely on the Doppler assessment of LV function and filling in this study.
The natural history of LV diastolic dysfunction in our cohort from the time of ICU admission to resolution of sepsis was variable, as some patients developed diastolic dysfunction over the course of their sepsis and others resolved their diastolic dysfunction. This finding is incongruent with other studies reporting that diastolic dysfunction in sepsis is transient and reversible
[17, 18]. Due to the continually changing physiologic profile of patients with severe sepsis and septic shock, the clinical importance of this finding is unclear. Future studies with classification of patients by age and grade of diastolic dysfunction should help clarify the evolution and clinical implication of LV diastolic dysfunction in severe sepsis and septic shock.
The limitations of this study include its observational nature. We used both research and clinical TTEs for our analyses. Clinicians were blinded to the research TTEs, but the clinical TTEs may have influenced clinical management. However, fluid management after the initial echocardiogram, which was similar regardless of degree of diastolic dysfunction, does not suggest that knowledge of the TTE changed management. Additionally, we chose to define diastolic dysfunction by the ASE 2009 guidelines, which were not developed for patients with sepsis. Furthermore, we specifically measured tissue Doppler of the septal annulus which may overestimate the severity of diastolic dysfunction
. While other measures of diastolic function are available, such as the indexed left atrial volume and/or the ratio of the diastolic reversal in pulmonary inflow to mitral A inflow duration
[37, 38], pulmonary inflow is difficult to obtain reliably in critically ill patients, and in our cohort, left atrial dilatation was extremely rare. The relatively small number of patients in patient subgroups underscores the importance of much larger sample sizes for studies in critical care echocardiography than have been reported to date. In addition, the study investigators are level-II echocardiographers. Whether the same observations would obtain in healthcare settings with less-trained critical care echocardiographers is not known, although the markers of diastolic dysfunction investigated in this study are generally easy to learn.
Two strengths of this study are its prospective nature and its size: it is one of the largest studies to date of critical care echocardiography. We were able to obtain acceptable images and quantify LV diastolic function in 95% of these critically ill patients, many of whom are generally considered to have poor windows for TTE, suggesting that TTE is useful in critically ill patients. TTE has major advantages over transesophageal echocardiography in terms of risk to the patient and ease in performing the exam, particularly for serial studies.