This prospective study yielded several important findings. First, TomTec, VVI, and Epsilon were able to measure RA strain on both GE and Philips scans; however, EchoPAC could only measure RA strain on GE but not Philips scans. Means and LLNs with 95% CIs for each type of RA strain by strain vendor were reported to determine abnormal from normal strain. Mean RA strains by age groups, sex, and scanner vendor were also reported. Finally, key factors, especially age and strain vendor, were found to significantly affect RA strain measurements, which have important clinical implications.

Establishing the normal ranges for continuous variable biomarkers, such as physiological, laboratory, and imaging parameters, is essential for their clinical applications. A commonly employed method for this purpose involves measuring these biomarkers in a sample of healthy individuals and determining either the 5th and 95th or the 1st and 99th percentiles to demarcate the lower and upper limits of normal. In the case of strain, particularly, the LLN holds greater clinical significance as lower strain values are indicative of abnormality. This threshold is arguably more crucial than assessing the mean value. Additionally, considering the error and uncertainty associated with every calculated estimate, including LLN, it is critical to report the 95%CI around the LLN. Applying this method allows the LLN to be clinically interpreted in two ways. For example, in the case of RA strain measured by TomTec in this study, the LLN with its associated 95%CI for RASr is 29.6 (26.5, 32.7). This is traditionally interpreted as RASr higher than 29.6% is normal and lower than 29.6% is abnormal. Taking into account the 95%CI and errors with all biological measurements, however, means that RASr should be interpreted as values above 32.7% are normal, below 26.5% is abnormal and between 26.5–32.7% is borderline, as there is uncertainty around where the LLN is but most likely within that 95%CI [10, 11].

The current American Society of Echocardiography (ASE) guidelines dating back to 2015 suggested using 25% as the LLN for RA strain. However, this is not well established[1]. A prior subject-level meta-analysis reported that 15 studies in 2469 healthy subjects were selected for analysis. The normal range values for RA strain were reservoir 42.7% (95% CI, 39.4 to 45.9%), conduit 23.6% (95% CI, 20.7 to 26.6%), and contractile16.1% (95% CI, 13.6 to 18.6%) [9]. Another meta-analysis including 4,111 subjects from 21 studies showed that the pooled mean RA strains were reservoir 44% (95% CI 25%-63%), conduit 18% (95% CI 7%-28%), and contractile 17% (95% CI 2%-32%)[8]. For mean RASr, our TomTec and VVI values are similar to prior meta-analyses, while EchoPAC and Epislon are slightly lower in magnitude. A subsequent study from the World Alliance Societies of Echocardiography reported a mean RASr of 46 ± 12% (45 ± 12% in men and 48 ± 13% in women), a slightly higher magnitude than our findings[12]. Variations in reported means and LLN of RA strains across studies leading to significant heterogeneity seen in meta-analyses are likely associated with differences in subject demographics and echocardiography equipment and software. The observed differences in RA strain measurements among vendors are contributed by variations in proprietary speckle-tracking methodologies, software algorithms, and strain calculation definitions unique to each vendor. These also include differences in myocardial border detection, region of interest selection, frame rate optimization, how peak strain is defined and measured, and spatial–temporal resolution. Our findings further emphasize the need that until all techniques have standardized method of RA strain, the same software should be used for serial monitoring of RA strain, and caution and LLNs be accounted for when comparing strain measured on different strain vendor software. Prospectively establishing means and LLNs in local populations when newer strain measurement techniques and software are applied for wider clinical use should be considered rather than solely relying on thresholds reported in prior guidelines and studies.

In our regression analyses, we found that strain software and age were key parameters affecting two of three types of RA strain measurements each. It has been shown previously that as we age, the reservoir and conduit phase functions of both atria decrease in magnitude, which we also found, while the contractile phase function is more variable [13, 14]. Similarly, smaller studies have also shown age as seen in our results, blood pressure, and heart rate variability as influencers of RA deformation parameters [15, 16]. Aging is associated with atrial myopathy, characterized by structural and electrical remodeling, including interstitial fibrosis and altered ion channel expression, which can reduce atrial strain [17]. Additionally, aging is associated with regional conduction slowing and increased prevalence of low-voltage areas in the atria [18]. Hemodynamic changes in conditions like PH or left heart failure and physiological states like exercise may also impact right-sided mechanics, leading to altered strain, contractility, or abnormal flow. Associations between RA strain and various factors, including right-ventricular systolic function, coronary artery disease, PH, and response to cardiac resynchronization therapy, have been observed [19,20,21,22]. In PH, RA strain provides valuable prognostic information. Hasselberg et al. demonstrated that RA peak longitudinal strain was independently associated with survival in patients with precapillary PH, adding prognostic value beyond traditional measures like RA area and pressure [23]. Similarly, D’Alto et al. found that RA reservoir function, measured by strain rate, was an independent predictor of clinical worsening in idiopathic pulmonary arterial hypertension [24]. Further, in heart failure, RA strain also holds prognostic significance. Jain et al. reported that RA reservoir and conduit strain were independent predictors of all-cause mortality in patients with heart failure with preserved and reduced ejection fraction [25]. This suggests that RA strain can provide additional prognostic information beyond the conventional measures of right-ventricular function. Interestingly, we found VVI to have the highest magnitude of mean RASr and RASct across vendors, while the opposite has been observed in studies for left- and right-ventricular systolic strain [26,27,28,29]. On the other hand, EchoPAC had lower magnitude RASr and RASct measurements than TomTech. These findings highlight the importance of still using the same vendor software (and, in the case of EchoPAC, using the same GE scanner) in the serial monitoring or comparisons of RA strain measurements. Further, the ASE, along with the European Association of Echocardiography and the Canadian Society of Echocardiography, recommends that serial strain assessments be performed using the same software to minimize variability and ensure consistency in longitudinal patient monitoring [30]. This recommendation is crucial for clinical practice, as inconsistent strain measurements can lead to misinterpretation of cardiac function and potentially impact clinical decision-making. In addition, the use of separate reference ranges of RA strain by strain software and possibly age group are suggested, and these parameters should be taken into account when interpreting RA strain values.

This study has some limitations. It is an observational cohort study, which inherently carries biases, although it is prospective in design. The sample size of 100 participants is relatively modest, especially when stratified across different age groups, genders, and scanners for subgroup analyses; therefore, external validation in larger healthy cohorts is warranted. In addition, left atrial strain methods were employed for RA strain measurements in this study similar to prior studies because of lack of standardized RA strain-specific modules by the main strain vendors, and this may also introduce accuracy and variability in RA strain measurements. The cohort comprised healthy individuals (majority Caucasian) from a United States tertiary cardiology center, which might limit the generalizability of the results to other populations with different demographic profiles worldwide. Each strain software used in the study has its own limitations in terms of measurement accuracy, leading to significant intra- and inter-reader agreement reported. As this study was cross-sectional and focused on healthy subjects, we did not evaluate outcomes during follow-up and prognostic utility of RA strain, which warrants future research.