Subcostal TAPSE: a retrospective analysis of a novel right ventricle function assessment method from the subcostal position in patients with sepsis

This is a retrospective observational study performed in Nepean Hospital Intensive Care Unit. Septic patients admitted to the ICU between August 2012 and March 2014 were identified for inclusion in this study.

The primary outcome measure was the comparative accuracy of the TAPSE measurement obtained from the subcostal window compared with the conventional TAPSE measure from the apical four-chamber view. Secondary measures included 90-day mortality.

Patients without echocardiograms, inadequate acoustic windows for analysis, and those who did not meet sepsis criteria were excluded. This study was approved by the Nepean Blue Mountains Local Health District Human Research Ethics Committee (study ref: 14/34) and did not require individual patient consent due to its retrospective and observational nature.

This is intended to serve as a pilot study for future prospective research into the role of RV dysfunction in sepsis outcomes. As such, it was anticipated that approximately 50 participants would be required. The data were obtained from a computerised database.

Images were evaluated for conventional measures of RV function: RV dimension, A4C TAPSE, velocity of tricuspid annular systolic motion (TV S′), fractional area change (FAC), tricuspid valve regurgitation peak velocity, estimated RA–RV pressure difference, and IVC diameter in accordance with the latest American Society of Echocardiography guidelines (Table 1) [17]. The subcostal views were given to a different assessor in a blinded fashion, from which the subcostal TAPSE values were measured (S.O.). The subcostal TAPSE was performed using B-mode imaging, in contrast to A4C TAPSE, which utilises M-mode (Fig. 1a). The tricuspid annulus was identified in diastole, and a cursor placed over it. Having evaluated the longitudinal axis, the image was then progressed to systole where a second cursor was placed over the tricuspid annulus. The distance between the two points is the subcostal TAPSE (Fig. 1b).

Statistical analysis was performed using JMP 10.0.0 software (SAS Institute Inc., Cary, NC, USA). Continuous data are described using mean ± standard deviation (SD) or median ± interquartile range (IQR) and were analysed between groups using analysis of variance and t test. Categorical data are expressed as number of patients and percentage of group and comparisons are made by Pearson’s Chi-squared test or Fisher’s exact test if less than five patients were in a specific group. Probability values are considered two-sided and a p value < 0.05 was considered significant. The correlation between subcostal and A4C TAPSE values was assessed using a Bland–Altman plot and paired t test. Categorisation of RV function: “normal” versus “abnormal” function was compared by McNemar’s test, confidence intervals are “exact” Clopper–Pearson confidence intervals. Inter- and intra-observer subcostal TAPSE variability was performed on a random 40% of the study population (23/58 patients) and assessed by Bland–Altman plots as absolute difference and expressed as a percentage of their mean. Mortality was assessed using univariate logistic regression for continuous data and Chi-squared tests for categorical data.

TAPSE from A4C and subcostal views showed a mean difference of 1.2 mm (CI 0.04–2.36). However, the Bland–Altman plot (Fig. 3) demonstrates most data points outside this confidence interval, with fanning at high TAPSE values.

A difference between subcostal TAPSE and A4C TAPSE of less than 3 mm is present in 92.1% of data pairs when TAPSE < 24 mm; 96.7% of data pairs when TAPSE < 22 mm; and 100% of data pairs when TAPSE < 19 mm. At TAPSE values greater than or equal to 24 mm, 63.6% of values had a difference < 3 mm. Intra-observer variability was good with mean absolute difference (± SD) of 2.3 mm (± 2.7) and expressed as a percentage of the mean 11.4% (± 9). Inter-observer variability was adequate with mean absolute difference of 3.4 mm (± 2.4) and expressed as percentage of the mean 15.8% (± 11).

When subcostal TAPSE is compared with A4C TAPSE as the gold standard, it identifies “Abnormal” TAPSE with a sensitivity of 97.8% (CI 88.2–99.9%) and specificity of 87.5% (47.4–99.7%).

No correlation was seen between either TAPSE measures of RV dysfunction and 90-day mortality (A4C TAPSE OR = 1.05, CI 0.92–1.20; p = 0.48; subcostal TAPSE OR = 0.98, CI 0.86–1.11 p = 0.74). However, when categorised, “abnormal” TAPSE was associated with higher mortality compared with “normal” TAPSE, although it did not reach statistical significance (57.1% vs 23.9% p = 0.09). Increased RV dimensions at the basal and mid-levels were associated with increased 90-day mortality (basal dimensions OR = 1.16, CI 1.05–1.32, p = 0.004; mid-dimensions OR = 1.12 CI 1.02–1.25, p = 0.012). No other measures of RV dysfunction were associated with increased mortality.

In this retrospective, observational study, we demonstrated that subcostal TAPSE correlated well with the conventional A4C TAPSE at low and moderate TAPSE values, with most differences less than 3 mm. The negative prognostic significance of TAPSE values has been reported at values < 18 mm [23], in sepsis, increased mortality has been found at TAPSE values less than 24 mm [17]. It is in this region of interest where subcostal (SC) TAPSE had reasonable accuracy compared with A4C TAPSE. The difference of less than 3 mm is considered to be a clinically acceptable difference. SC TAPSE maintained accuracy when categorising “abnormal” and “normal” TAPSE values.

The mean difference between measures was small; however, most data points fell outside of the confidence intervals, limiting the utility of that measure. There is increased variance at high TAPSE values. The reason for this may be due to translational movement in a hyperdynamic heart or inaccuracies in identifying the axis of contraction with vigorous movement. Although this limits the accuracy of the absolute value of subcostal TAPSE in these circumstances, it does not impede the utility of subcostal TAPSE altogether, as at high values TAPSE is considered to be normal and is unlikely to influence prognostication or interventions and would support an assessment of normal RV function.

Although this data set is small and retrospective, it supports the feasibility of using subcostal TAPSE. Further research is required before implementing its use in clinical practice, particularly ensuring minimal inter-observer variability and discrimination of clinically relevant outcomes.

With greater understanding of the importance of RV dysfunction in critical illness, accurate transthoracic echo assessment of the RV is a crucial tool in managing intensive-care patients. Subjective assessment of RV function has been shown to only have fair reliability when performed by ICU specialists adequately trained in echo, but only for initial exclusion of significant RV pathology. Quantification of RV function is important to ensure accurate estimation of performance [24].

TAPSE is a commonly used tool in cardiac function analysis; however, its utility is limited in the critically ill due to difficulty obtaining A4C views. Inability to obtain TAPSE has been reported to range from 6 to 25% of cases [25‐27]. Other measures of RV function such as Fractional Area of Change have been described as “tedious” with TAPSE comparably easier [28]. Subcostal views have previously been shown to be comparable to A4C views in the assessment of RV function [21]. Most TAPSE measurements take less than 30 s to perform. The relative ease of obtaining images from the subcostal window makes this an attractive alternative in the critically ill.

The use of the subcostal window to assess the RV has been increasingly considered, although with varied success [7, 29]. Subcostal echocardiographic assessment of tricuspid annular kick (SEATAK) has been purported as an alternative to TAPSE in critically ill patients [7]. SEATAK utilises the subcostal short-axis view; using M-mode, the cursor placed over the tricuspid annulus and measured in a similar way to A4C TAPSE measurements. This study confirmed the feasibility of using the subcostal view; however, the angle of M-mode may underestimate RV function. Subcostal TAPSE has the advantage of identifying and measuring in the longitudinal axis along which the RV contracts.

RV dysfunction in critical illness and sepsis, in particular, has been associated with increased mortality, [11, 30]; however, the association between specific echocardiographic measures and outcomes have not been consistently reported [3, 4, 9, 13, 31].

TAPSE has been demonstrated to accurately measure RV function and predict mortality in a variety of conditions [14, 18‐20, 25]. Reduced TAPSE has been associated with increased mortality and longer hospital length of stay [17], although this finding has not been replicated [4]. In this study, there was no association between TAPSE and 90-day mortality. There are a number of possible reasons for this: the numbers in this study are small and may not suffice to show any difference if present; there are a complex variety of factors that result in adverse outcomes in sepsis; TAPSE alone may not be a prognostically significant marker. In addition, the timing of the echocardiogram was not controlled for. As such, the findings may be affected by fluid resuscitation, inotropic and ventilator support or the natural course of the illness. Further investigation is required to assess the utility of subcostal TAPSE in assessing RV function in sepsis and its role in prognostication. In particular, standardisation of timing of the echocardiogram (such as when inotropic or vasoactive medications are commenced, or serially at predetermined times during the ICU stay) may aid in identifying an association with RV dysfunction and outcomes. Based on these data, a study to detect an association between TAPSE and mortality would require a sample size of at least 638 utilising a 1 mm difference in TAPSE with a standard deviation of 4.5, an alpha of 0.05 and 80% power.

Interestingly, there was an association between increased RV diameter and mortality in this population. The reason for this association is not clear; however, increased end-diastolic volume (EDV) has previously reported in sepsis, although it has not been associated with an increase in mortality [2, 8, 32]. Increase in RV diameter occurs in both pressure and volume overload [33‐35]. We hypothesise that EDV may be a surrogate for fluid overload. As excess fluid has been associated with increased mortality in sepsis [36‐38], this provides a possible link between the RV measurements and mortality outcomes. Other causes of right-ventricular dilatation, such as pulmonary hypertension associated with sepsis and high intrathoracic pressures with positive pressure ventilation, may contribute to RV dilatation, but are beyond the scope of this study. Further investigation is required to evaluate the relationship, if any, between fluid status, RV dimensions, and mortality.

The main strengths of this study are that it supports the feasibility of a novel echo technique and it generates further hypotheses regarding the role of the RV in sepsis and provides direction for further research in this area. This study is limited by its retrospective nature and small numbers. A number of patients were excluded, because there were no echocardiograms performed. In addition, the timing of the echo was not controlled for, which may influence the accuracy of using TAPSE as a prognostic tool. The exclusion of patients without echocardiograms would not impact our primary research aim, and the overall numbers excluded for inadequate subcostal views were small. Exclusion of patients may have reduced overall numbers, so that finding associations, if any exist, between adverse outcomes and RV dysfunction is less likely. Reassuringly, the association between A4C and subcostal TAPSE that has been found is likely to be robust, within the limits of a small study, as the patient characteristics are similar to patient groups in other published sepsis trials [38].

Another limitation of this trial was that one clinician performed all subcostal TAPSE measurements. Despite this, inter-observer variation was adequate, and intra-observer variation was good in a subset of measures. This method needs to be further assessed before general use ensuring accuracy persists with multiple clinicians.