We conducted a multicenter, nationwide, long-term cohort study of a population with CTEPH diagnosis in Spain. Patients’ data were prospectively included in the Spanish REHAP registry from January 1, 2007, to December 31, 2018, and retrospectively analyzed. Because the BPA program was implemented in Spain in 2013, we considered this year the reference date for comparisons.

All patients met CTEPH diagnostic criteria: mean pulmonary artery pressure (mPAP) estimated by right heart catheterization> 25mmHg, pulmonary capillary wedge pressure <15mmHg, pulmonary vascular resistance (PVR)> 240 din/s/cm5, perfusion defects on lung scintigraphy, and radiologic signs of pulmonary thromboembolic disease on computed tomography or pulmonary angiography. Patients <18 years old were excluded.

The day of the first right heart catheterization was considered the date of CTEPH diagnosis. Each hospital had its own process for assessing the therapeutic approach, including whether or not to refer the patient to an expert PH center. Residual PH at least 6 months after PEA or the last BPA procedure was considered significant if the final PVR exceeded 416 din/s/cm5 and/or if the final mPAP exceeded 30mmHg.8,9 Patients were followed up until death or study end (December 31, 2018). The study was approved by the appropriate research ethics committee.

Baseline characteristics were compared using a t test for continuous variables and the chi-square or ANOVA tests of proportions for categorical variables. Analyses were conducted by using SPSS version 22.0 software (SPSS Inc, Chicago, IL, United States).

A propensity score model was used to predict the probability of interventional treatment (PEA and/or BPA). Propensity scores facilitated a similar distribution of baseline characteristics between interventional treatment and medical therapy groups, minimizing the selection bias of observational studies. We selected baseline variables that were significantly associated with PEA/BPA in the univariate analysis (age, treatment with prostanoid analogs, PVR at diagnosis, period of diagnosis). Patients who underwent BPA or PEA were matched 1:1 with those who did not, based on the propensity scores calculated using the nearest-neighbor technique (standardized difference <0.05%). The internal validation error of the constructed model was 0.03. This matching process selected 2 cohorts of 294 patients. These patients were divided into 2 subgroups according to the treatment strategy selected (table 1 of the supplementary data) to further compare differences related to their characteristics and prognoses.

We first described survival in the general population. Overall mortality was defined as death from any cause during the study period. In a subsequent analysis, we evaluated prognostic data from the matched population. Cox regression analysis and Kaplan-Meier curves were used to assess mortality. The final multivariate regression model for the matched sample included all variables showing a significant association in the univariable analysis (table 2 of the supplementary data) and those considered clinically relevant (age and sex).

A total of 1019 patients were included during the study period, 451 (44.3%) from 2007 to 2012. The mean age was 61±15 years and 56.5% were women. In total, 97 patients (9.6%) were selected for percutaneous treatment with BPA and 350 (34.3%) for surgery (table 1). The median duration of follow-up for the patients in this study was 1095 days [0-7665 days].

Most patients (55.8%) received medical therapy exclusively with specific PH drugs. These patients were older than those who were selected for PEA or BPA (table 1). Patients included in the BPA program had the worst hemodynamic condition at diagnosis. Interestingly, 7 patients underwent lung transplantation, 6 of whom were diagnosed before 2013.

Of the 1019 patients, 656 (64.4%) were referred to an expert CTEPH center. There were significant differences between the 2 groups (table 2). Patients who were evaluated at an expert CTEPH center were younger and had more advanced disease at presentation (higher PVR, signs of severe right ventricular dysfunction). They also benefited from interventional treatment more than those who were exclusively evaluated at their local centers.

The incidence rate described here expresses the number of new cases of CTEPH reported to the REHAP registry in the Spanish adult population, as defined by the Spanish census, within the selected period of time. The CTEPH incidence based on Spanish adult population data10 was 0.9 cases per year per million individuals in 20076 and rose to 1.7 cases per year per million population in 2018. Similarly, the recorded CTEPH prevalence (based on cases reported to the REHAP registry) increased from 3.2 in 200710 to 22.5 per million population in 2018. Changes concerning CTEPH treatment during the study period are displayed in table 3 of the supplementary data.

We prospectively analyzed data from the Spanish REHAP registry. Unlike other countries,11–14 the assessment and prescription of PH treatment in Spain is not strictly limited to expert PH centers and can be made by any specialist in the public health system. Therefore, the Spanish REHAP registry is open to voluntary participation by all physicians who treat patients with PH (subgroups 1 and 4), all over the country. This approach enables a more comprehensive picture, accurately reflecting the national CTEPH landscape.

Observational registries represent a feasible way to deepen knowledge about the clinical history of uncommon diseases such as CTEPH and to assess their long-term outcomes. Accordingly, it is crucial to determine the primary source of information, which depends on each country's health care system. In this regard, Gibbs et al.15 have described the role and structure of PH networks in France and the United Kingdom. Their registries are both based on data from expert PH centers, which usually organize the patients’ entire medical care. Likewise, the ASPIRE registry16 and the German13 and Swiss17 national series share the same design. A major strength of our study is that we also provide epidemiological information related to patients who were assessed and treated at nonreferral hospitals.

National registries are usually used to calculate the incidence and prevalence rates for entire populations.18 Because the REHAP data do not consider the entire Spanish National Health System, we offer the most approximated epidemiological information we could obtain. We found a growing trend in reported incident cases throughout the study period. Although our findings are still far from those of German registries (reported incidence of 5.7/million in 201613), an increasing awareness of the disease has resulted in a substantial growth in reported CTEPH cases in Spain over the last decade (from 0.9 in 2007 to 1.7/million in 2018).

According to Skride et al.,19 who compared different CTEPH registries in Europe, the lowest CTEPH prevalence was reported in Spain (2012) and the highest in Sweden. However, the current Spanish CTEPH prevalence has increased to 22.5/million population, data more consistent with the Swedish series. This dramatic increase in prevalence (from 3.2/million in 2007 to 22.5/million in 2018) could be explained by the reduction in disease-specific mortality and the standardization of patient registry data fostered by the REHAP.

There are 2 national CTEPH centers in Spain (Hospital 12 de Octubre, Madrid, and Hospital Clínic, Barcelona). Each local hospital evaluates and provides PH treatment according to its own criteria. In our series, 64.4% of the patients were referred to expert centers for a more thorough evaluation. Although the Spanish referral rate for PEA evaluation is above average (44% in Europe19), this decentralized health care model results in nonhomogeneous medical assistance: patients who were assessed at the national CTEPH centers were much more likely to benefit from PEA or BPA techniques (table 2).

This reinforces the argument that access to expert centers ensures optimal care. As has been previously stated by Escribano et al.,20 patients who were not assessed in a CTEPH referral hospital were less likely to undergo surgery. Consistent with these results, an international prospective registry found that low-volume PEA centers (< 10 PEA/y) reported a significantly higher rate of “nonoperable” patients,21 also suggesting that accessibility to high-volume specialized PEA centers might influence the final therapeutic approach.

The rise in the reported incidence of CTEPH in Spain was associated with a major shift in the clinical profile of CTEPH patients, with the hemodynamic status at diagnosis improving with time. Patients diagnosed between 2007 and 2012 had a worse clinical and hemodynamic condition at inclusion (table 3 of the supplementary data). The diagnosis date was also related to a poorer prognosis (figure 4). Changes in CTEPH treatment over the last decade may have influenced these outcomes. Remarkably, even with the implementation of BPA programs, the number of patients referred for PEA consistently increased from 2007 to 2017 (P <.0001). On the other hand, medical therapy with prostacyclin analogs drastically decreased from one period to the other (P <.0001), probably due to the wider pool of resources for treating these patients. Similar to the national series from the United Kingdom,12 a small number of patients underwent transplantation.

Our mortality rate was slightly better than those reported by other European CTEPH registries.8,13,19 Compared with a national study conducted before 2013, CTEPH prognosis in Spain has considerably improved.20 Mortality in the PEA subgroup was comparable to that reported by Delcroix et al.24 for 679 patients included in an international European registry. In the BPA subgroup, the survival rates of percutaneous therapy intervention were remarkably high, despite the particularly unfavorable hemodynamic conditions at diagnosis. Our results were similar to those reported in a Japanese series.25 The safety and efficacy of this procedure make this technique a favorable choice for CTEPH management.

Multivariable Cox regression analysis of the propensity-matched cohort showed that, ultimately, prognosis mainly depended on disease severity and the chosen therapeutic strategy. Those who were not selected for any interventional procedure (PEA or BPA) had a particularly somber outcome. We cannot exclude the possibility that some of the patients of the medical subgroup were suitable candidates for PEA or determine if BPA was considered an alternative for every CTEPH patient declared inoperable. It is therefore important to underscore the relevance of reference PH centers to raise the overall quality of care and improve long-term outcomes.

Because REHAP is a voluntary observational registry, incomplete reporting may lead to missing data. The BPA program was not finalized for every patient by the end date. As such, the results are incomplete in terms of the ultimate improvement. Because we compared historical cohorts, changes introduced from 2013 onward may have biased the results (the Will Rogers phenomenon). Although observational investigations are not the most appropriate method to evaluate survival improvements, the propensity score-matching analysis method aims to reduce the statistical bias by distributing the patient's baseline characteristics as uniformly as possible.