This study including patients with suspected PH referred to a referral care centre, echocardiographic evaluation demonstrated a high discriminatory power for diagnosis of PH. Compared with the gold-standard invasive method, there are disagreements between sPAP and RAP measurements, which is similar to the results of other studies within controlled research contexts [5–7].
The diagnostic threshold of mPAP for the definition of chronic PH has been lowered from ≥ 25 mmHg to > 20 mmHg [1, 15]. This change is based on an increasing body of evidence demonstrating that even mild elevations of pressure have prognostic value. Echocardiography remains the most appropriate approach for the prediction of PH. Gall and colleagues affirm that decreasing the lower limit of TRV does not improve the diagnostic yield of echocardiography for PH and provides key evidence reinforcing the clinical value offered by composite echocardiographic parameters in patients inclusive of mild PH .
The population presented in this study represents all patients evaluated at the referral center Pulmonary Circulation Unit of Hospital das Clínicas of Federal University of Minas Gerais who were indicated to undergo RHC for diagnosis of PAH and CTEPH consecutively in the period of this study. The sample had a larger number of patients with PAH (62.9%) than those with CTEPH (37.1%). Differently from the international reports, schistosomiasis was the most prevailing etiology, as already expected [16–19]. It results from the high prevalence of schistosomiasis in some regions of Brazil, particularly in the state of Minas Gerais, where its occurrence is endemic . The average age of the patients and most of them being women, similarly to the available records from other countries, suggests that this prevalence does not depend on location and gender [18, 19]. Similarly to international reports, most of the patients were in FC III and IV (52.6%), demonstrating a delay in the recognition of this disease yet, which morbimortality is high [17, 18]. As there is a higher screening of the cases for which the RHC is indicated in the PH reference centers, the number of patients who were not definitely diagnosed with PH was very small (5,2%).
Comparison between echocardiographic and invasive hemodynamic measurements of pulmonary pressure
The accuracy of the TTE in providing the diagnosis of PH has been evaluated since the 80’s, as it is a largely available, low-cost, non-invasive procedure. Yock and Popp (1984) demonstrated a good correlation of the measurement of sPAP by modified Bernoulli’s equation between the TTE and the RHC in 54 patients (r = 0.93, SEE = 8mmHg) . Other authors have also demonstrated a strong correlation between these measurements [22–24].
However, the agreement between two measurements evaluated by the usual correlation methods can be inaccurate . Fisher et al used the Bland-Altman method that is considered more appropriate for accuracy measurements and found a significant bias of the TTE to estimate the sPAP (-0,6mmHg; range of 95% agreement: -40.0 to 38.8mmHg) in 65 patients within a maximum one-hour interval between TTE and RHC . Other two studies have reported similar results with the bias ranging from 2.2mmHg to 8mmHg in the estimation of sPAP by TTE (range of 95% agreement: -34.2 to 38.6mmHg and − 28.4 to 44.4mmHg) as compared to the RHC measures [6, 7]. Data from the REVEAL registry showed a low accuracy of the initial TTE in 57.4% of sPAP estimates (> 10mmHg higher or lower than RHC) and in 36.5% of RAP estimates (> 5mmHg higher or lower than RHC) in examinations made within a maximum 12-month interval .
Other authors have demonstrated a better agreement between measurements of sPAP, but the level of agreement on the difference on the mean rather was large, indicating just moderate precision of the echocardiographic measurements [25–28]. D’Alto M et al found no significant bias (-0,5mmHg) between TTE and RHC estimates of sPAP with wide limits of agreement (-19mmHg to 18mmHg) . Differently from other studies, RHC was indicated for reasons other than PAH (PAH: 36%; pulmonary venous hypertension: 40%; lung disease PH: 16%). Greiner et al have made a retrospective, single-center study including a larger sample of patients (n = 1695) and a maximum of 5 days interval between the examinations indicated due left heart disease (59%), valve disease (27%) and PAH (6%) . They found a mean sPAP of 45.3 ± 15.5mmHg by TTE and 47.4 ± 16.4mmHg by RHC, showing a strong correlation between measurements of sPAP (r = 0.87; p < 0.0001) and RAP (r = 0.82; p < 0.0001). The Bland-Altman analysis has shown a − 2mmHg bias for sPAP (95% agreement limits: -18.1 for 14.1mmHg) and a 1mmHg bias for RAP (95% agreement limits: 0.1 for 1.9mmHg) . Doutreleau et al have studied prospectively 106 patients with suspected or confirmed PH, simultaneously submitted to both methods . PH was not confirmed in 16.9% of the patients, 10.4% were diagnosed with post-capillary PH and 72.7% with pre-capillary PH. The correlations were strong (for sPAP: r = 0.84; for RAP: r = 0.70) and the Bland-Altman analysis showed a significant bias of 1.4mmHg for sPAP (range of 95% agreement: -22.6 to 25.4mmHg) and 1.9mmHg for RAP (range of 95% agreement: -6.1 to 9.9mmHg) . Another study included consecutive patients with indication of RHC (PAH and CTEPH: 40%; heart failure: 42%) and a maximal of 3 hours interval between TTE and invasive measurements, revealed too minimal bias (mean bias = + 2,4mmHg) between echocardiographic and invasive sPAP measure, but with wide limits of agreement (-20 to + 25mmHg) .
Three meta-analysis and a systematic review evaluated the accuracy of the estimation of sPAP by TTE [29–32]. However, the findings were not consistent because a significant heterogeneity among these studies due to inclusion of participants without disease, other with left and right heart disease, and to the use of correlation analysis by most of them, which is less appropriate as an agreement measurement.
Although the non-invasive evaluation by TTE is recommended by current guidelines, the agreement between these methods is still debated. As far as we know, no published Brazilian study has made this evaluation in adult patients with suspected diagnosis of PAH and CTEPH in the context of daily clinical practice. The present study showed a high discriminatory power of the sPAP and TRV measured by the TTE to diagnosis of PH, although the cut-off points of 48mmHg for sPAP and 3.08m/s for TRV have low NPV, which would limit them as screening values. Besides that, it showed a statistically significant bias for sPAP (8.0; limits of agreement:-34.9 to 50.9) and for RAP (-3.3; limits of agreement:-15.9 to 9.3). There was lesser variability between the duplicated measurements for sPAP (CV = 26%) than between those for RAP (CV = 57%). By using a clinically acceptable difference of 10mmHg for sPAP and 5mmHg for RAP, only 33.4% of the echocardiographic estimate of sPAP and 55.1% of that for RAP were accurate, similarly to the results found by Fisher et al  (sPAP: 52%), Rich et al  (sPAP: 49.4%) and REVEAL analysis  (sPAP: 42.6% and RAP: 63.5%). This was in disagreement with the prospective study developed by Doutreleau et al  and Venkateshvaran et al , in which the measurements of sPAP were accurate in 68% and 62% of the patients, respectively and the measurements of RAP in 79% of the patients . Additionally, in present study the TTE underestimated the sPAP values and overestimated the RAP values with higher frequency (sPAP: 41.5% versus 25.1%; RAP: 33.7% versus 11.2%) and magnitude (sPAP: -30.4 ± 10.2 versus 15.2 ± 8.9mmHg; RAP: 11.3 ± 4.8 versus-8.4 ± 3.7mmHg), with disagreeing values higher than 10mmHg and 5mmHg, respectively. Fisher et al have found no difference in this frequency, but the sPAP values have been more underestimated than overestimated (-30 ± 16 versus19 ± 11mmHg; p = 0.03) . Data from Rich et al  and REVEAL registry  have also demonstrated that sPAP values were underestimated by 30% and 20.6% in the TTE and by 34.8% and 22.5%, respectively, and overestimated RAP values (26.3% versus 12.4%) . These results illustrate how difficult it is to evaluate the severity of the PH and to stratify its risk using TTE, since the invasive RAP obtained by RHC is a well known prognostic variable [3, 28, 33].
The limitations of this study should be taken into consideration. As it is an observational cross-sectional study, the TTE and RHC examinations have not been performed simultaneously, but at a 3.3month average interval between them, within the context of the clinical practice regulated by the local public health authority. This way, the expected variability between equipment used and examiners should be taken into account, once the TTE examinations have been performed in different SUS’s cardiology units. However, this study had the purpose of evaluating the consistency of the TTE as a screening test for the diagnosis of PH out of the controlled research environment. In view of that, a result lower than those reported in the literature was expected. So, these results are important as they express the differences between measurements by both methods and reflect limitations of the methods themselves, patient characteristics and the expertise of operators in the local current practice of diagnosis of PH [4, 10, 32].
As a conclusion, the initial echocardiographic evaluation of patients with suspected of PH in the context of clinical practice showed a high discriminatory power for diagnosis of PH and the disagreements between sPAP and RAP measurements reinforces TTE as a valid screening tool and the need of RHC measurements to confirm the diagnosis. The technical improvement of these procedures and a better interaction among the professionals involved in the evaluation of these patients may contribute to an even better accuracy of the TTE for an earlier recognition of this condition.