Prevalence and prognostic value of various types of right ventricular dysfunction in mechanically ventilated septic patients

This study was an observational study conducted at a tertiary hospital’s intensive care unit (ICU). We retrospectively studied a cohort of adult septic patients who were on mechanical ventilation from 1 May 2018 to 1 August 2020.

We adopted the same definition of sepsis and septic shock as described in Sepsis-3 [14]. The exclusion criteria included the following: lack of CVP monitoring; lack of MV support via tracheal intubation; intra-abdominal pressure above 12 mmHg; new onset of acute coronary syndrome within 1 week; severe valvular disease or history of valvular surgery; history of chronic pulmonary hypertension; insufficient echocardiographic image; and withholding of life support.

Echocardiograms were recorded within the first 24 h of ICU admission. Two physicians (H Zhang and Q Zhang) with 10 years of echo experience obtained the images and they were blinded to the clinical states of the patients upon echo examination. The echo results were reported based on the PRICES statement [15]. At least three cardiac cycles were analysed and averaged. M-mode and Doppler echocardiographic measurements were taken according to standard protocols. The measurement of TAPSE, left ventricular ejection fraction (LVEF), averaged tissue Doppler velocity of lateral and medial mitral annuli at early diastole (e’), and tricuspid regurgitation (TR) were performed as previously described [12]. The ratio of RV end-diastolic area and LV end-diastolic area (R/LVEDA) was obtained at the end of ventricular diastole. FAC was defined as (end-diastolic area – end-systolic area)/end-diastolic area × 100 [6]. The E velocity was measured using pulsed wave Doppler with the sample volume placed between the tips of the mitral valve. The diameter of the left ventricular outflow tract (LVOT) was obtained at the parasternal long-axis view. The velocity–time integral (VTI) was obtained by positioning the sample volume at the LVOT approximately 0.5 cm below the aortic valve via pulsed Doppler imaging [16]. Cardiac output (CO) was calculated using the following formula: CO = π × (LVOT diameter/2)² × VTI × heart rate. The CO was then indexed to body surface area.

We collected the patients’ demographic information, Acute Physiology and Chronic Health Evaluation (APACHE) II score, and Sequential Organ Failure Assessment (SOFA) score at ICU admission. Each patient’s heart rate (HR), mean arterial pressure (MAP), CVP, positive end-expiratory pressure (PEEP), and plateau pressure (Pplat) were also collected at the time of the echo examination.

The primary outcome was 30-day survival, and the secondary outcomes included length of ICU stay, mechanical ventilation (MV) duration and cardiac index.

We performed statistical analysis using SPSS 13.0 (SPSS, Inc., Chicago, Illinois, USA). Continuous variables are expressed as the mean ± SD or as the median and the interquartile range. Categorical variables are presented as frequencies and percentages. The distributions of the continuous values were assessed for normality by the Kolmogorov–Smirnov test. Differences among groups were assessed by one-way ANOVA, the Kruskal–Wallis test, the Chi-squared test, or Fisher’s exact test, as appropriate. If necessary, a Dunn–Bonferroni test was performed for post hoc comparisons. Receiver operating characteristic (ROC) curves were generated and the areas under each respective curve were calculated. Prognostic factors for 30-day mortality were determined using the Cox regression model. The following variables were considered for the survival analysis: age, SOFA score, APACHE II score, PEEP, Pplat, LVEF, E/e’, RVF and RVSD. The variables that had p < 0.1 in the univariable model were included in the multivariable model and the hazard ratio was calculated, together with its 95% confidence interval. Cumulative survival curves of the 30-day follow-up were estimated with the Kaplan–Meier method. Sensitivity analyses were performed using different cut-off values for RVF, as well as incorporating LV systolic dysfunction, when investigating the association between RV function and 30-day mortality. Intraobserver and interobserver variabilities in LVEF, TR velocity and FAC were assessed in 20 randomly selected patients and were tested using intraclass correlation coefficients (ICCs). An ICC > 0.8 was considered excellent agreement. Two-tailed p < 0.05 was considered significant.

In all, 368 patients were screened for enrolment, and 215 patients were included in this study. The patients were divided into 4 cohorts based on the presence of RVF and RVSD: patients with normal RV function (normal, n = 101), patients with RVF but without RVSD (RVF only, n = 38), patients with RVSD but without RVF (RVSD only, n = 44), and patients with combined RVF–RVSD (RVF/RVSD, n = 32) (Fig. 1). The general characteristics are listed in Additional file 1: Tables S1 and S2.

The four groups had similar age and sex proportions. The RVF/RVSD group had higher APACHE II and SOFA scores than the RVF and normal groups (p < 0.05). The RVF/RVSD group had the highest PEEP level among all groups (p = 0.001) (Table 1).

The intraobserver variability analysis revealed that the ICCs for LVEF, TR velocity and FAC were 0.908 (95% CI: 0.782–0.962), 0.916 (95% CI: 0.801–0.966), and 0.851 (95% CI: 0.661–0.938), respectively. The interobserver variabilities for LVEF, TR velocity and FAC were 0.875 (95% CI: 0.712–0.949), 0.904 (95% CI: 0.758–0.962), and 0.827 (95% CI: 0.614–0.928), respectively.

The four groups had similar HR and MAP. Both the RVF/RVSD group and RVSD only groups had a lower LVEF than the normal group (p < 0.05) (Table 1, Fig. 2e).

The ROC analysis showed that the areas under the curve for CVP, R/LVEDA, TAPSE and FAC were 0.644 (p = 0.006); 0.525 (p = 0.634); 0.652 (p = 0.004) and 0.690 (p < 0.001), respectively (Fig. 3, Table 2).

We generated Kaplan–Meier curves for estimated survival at 30 days after ICU admission. The RVF/RVSD group had higher mortality than the RVF only and normal groups (RVF/RVSD vs. RVF only, log-rank:12.613, p < 0.001; RVF/RVSD vs. normal, log-rank:25.208, p < 0.001). The RVF/RVSD group also had higher mortality than the RVSD only group, but was not statistically significant (RVF/RVSD vs. RVSD only, log-rank:3.662, p = 0.057). The RVSD only group had higher mortality than the RVF only group and normal groups (RVSD only vs. RVF only, log-rank:3.995, p = 0.046; RVSD only vs. normal, log-rank: 7.376, p = 0.007). No difference was found between the RVF only group and normal groups (RVF only vs. normal, log-rank: 0.012, p = 0.912) (Fig. 4).

In a Cox regression survival analysis, after adjusting for APACHEII, SOFA, PEEP, Pplat, and E/e’, the presence of RVF/RVSD was independently associated with 30-day mortality (HR 3.004, 95% CI: 1.370–6.587, p = 0.006). In contrast, neither the presence of RVF only nor the presence of RVSD only was associated with 30-day mortality (HR 0.951, 95% CI: 0.305–2.960, p = 0.931; HR 1.912, 95% CI:0.853–4.287, p = 0.116) (Table 3).

We performed sensitivity analysis by using CVP ≥ 10 mmHg, CVP ≥ 12 mmHg, and R/LEDA ≥ 0.7 separately as cut-off values of RVF, and found that RVF was still not an independent predictor of 30-day mortality in these patients.

The RVF/RVSD and RVSD only groups had a lower cardiac index than the RVF only group and normal groups (p < 0.05). No significant difference was found regarding MV duration or length of ICU stay among the four groups (Table 1, Fig. 2f).