Long-term outcomes of initial thoracic endovascular repair versus medical therapy in acute uncomplicated type B aortic dissection: real-world evidence from a nationwide claims database in Japan – a retrospective cohort study

Abstract

Objectives

To compare the long-term outcomes of initial thoracic endovascular aortic repair (TEVAR) versus initial medical therapy (iMT) in patients with acute uncomplicated type B aortic dissection (uTBAD), using real-world evidence from a nationwide claims database in Japan. This study aligns with stage 4 of the Idea, Development, Exploration, Assessment, and Long-term Study (IDEAL) framework for surgical innovation.

Design

A retrospective cohort study using propensity score matching (PSM) to balance baseline characteristics.

Setting

Japanese nationwide health insurance claims database, between 1 January 2015 and 31 December 2023.

Participants

Among 40 229 cases with tentative codes for aortic dissection (International Classification of Diseases-10: I71.0), 4995 met all eligibility criteria for acute uTBAD. Among these patients, 96 underwent TEVAR in the subacute phase (15–90 days post diagnosis), while 4899 were managed with iMT. After PSM, 96 TEVAR cases were matched to 480 iMT cases in a 1:5 ratio.

Main outcome measures

The primary outcomes were aorta-related events and all-cause mortality. The secondary outcome was the frequency of follow-up CT imaging every year.

Results

After PSM, the baseline characteristics of both groups were balanced. Median age was 56 years (IQR: 50–62 years) in both groups, and follow-up duration was similar (TEVAR: 31 months; iMT: 28 months, p=0.84).

At 60 months, Kaplan-Meier estimates showed an aorta-related event rate of 21.9% (95% CI: 12.6% to 36.4%) for TEVAR and 19.9% (95% CI: 15.6% to 25.2%) for iMT (p=0.99).

All-cause mortality was 4.4% (95% CI: 1.4% to 13.6%) for TEVAR and 6.6% (95% CI: 4.0% to 10.6%) for iMT (p=0.70). No significant differences were observed.

Conclusions

While aorta-related events accumulated steadily in the crude iMT group, no survival benefit was observed for subacute TEVAR. These findings support ongoing randomized controlled trials and show the utility of claims-based analyses in IDEAL Stage 4.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Thoracic endovascular aortic repair (TEVAR) is the standard treatment for complicated type B aortic dissection (TBAD), but its role in acute, uncomplicated TBAD (uTBAD) remains to be determined. Previous randomized controlled trials (RCTs), such as INSTEAD-XL and ADSORB, have shown mixed results, highlighting the need for further evidence. Therefore, several RCTs are ongoing in Europe and the USA, while real-world data are increasingly valued to enhance the generalizability of trial findings.

WHAT THIS STUDY ADDS

• This study is the first to report the feasibility of using Japan’s nationwide claims database to evaluate long-term outcomes of TEVAR in the subacute phase compared with initial medical therapy (iMT) for acute uTBAD. The findings revealed a steady rate of aorta-related events in the iMT group, highlighting the unmet needs in uTBAD management. However, as no significant difference in all-cause mortality was observed, the design and conduct of the ongoing RCT to compare iMT and TEVAR are further justified.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• In this study, we highlighted the feasibility of using claims databases to support long-term follow-up within the IDEAL framework, contributing to global quality improvement initiatives in vascular surgery. Real-world insights will support healthy competition between TEVAR for enhancing device technology and surgical technique and iMT for improving CT follow-ups and beta-blocker use.

Introduction

Thoracic endovascular aortic repair (TEVAR) has become the standard treatment for complicated type B aortic dissection (cTBAD). However, the optimal management of acute uncomplicated TBAD (uTBAD) remains a subject of ongoing debate.¹ Traditionally, initial medical therapy (iMT), aimed at heart rate and blood pressure control, has been the mainstay of uTBAD management. Increasing evidence shows that pre-emptive TEVAR may offer additional benefits by promoting favorable aortic remodeling and preventing long-term complications, such as rupture or aneurysm formation.²

Studies, including the Investigation of Stent Grafts in Aortic Dissection with Extended Lengh of Follow-up (INSTEAD-XL),³ the Acute Dissection Stent Grafting or Best Medical Treatment (ADSORB),⁴ and the International Registry of Acute Aortic Dissection (IRAD),^{5 6} have reported the role of TEVAR in uTBAD. However, important questions remain regarding patient selection, timing of intervention, and long-term outcomes.⁷,¹⁰ Recent studies on real-world data (RWD)¹¹,¹⁴ and ongoing randomized controlled trials (RCTs) in Scandinavia, the USA, and the UK¹⁵,¹⁸ aim to address these gaps.

In parallel, the Idea, Development, Exploration, Assessment, and Long-term Study framework (IDEAL) provides structured guidance for evaluating surgical innovation.¹⁹ While RCTs typically correspond to stage 3 (assessment), RWD studies play a complementary role in stage 4 (long-term study), enabling broader population analyses and extended follow-up.²⁰

The complexity of evaluating initial TEVAR for acute uTBAD—clinically and methodologically—makes this condition particularly well suited for assessment using the IDEAL framework.

This study leverages nationwide claims data from Japan,²¹,²⁴ focusing on the subacute phase of TEVAR (15 days–3 months post diagnosis)^{25 26} with comprehensive follow-up enabled by Japan’s universal healthcare system. We aimed to generate real-world evidence (RWE) to inform clinical decision-making, support ongoing RCTs, and demonstrate the feasibility of claims-based long-term follow-up within the IDEAL framework.

Methods

Study design and data source

This retrospective cohort study was conducted using data from the Japan Medical Data Centre (JMDC) claims database.²² As of 2023, JMDC covered approximately 19.3 million individuals, accounting for 16.1% of the Japanese population. The database contains anonymized health insurance claims data from hospitals, clinics, and pharmacies.

The JMDC database provides comprehensive information on diagnoses, medical procedures, prescriptions, surgeries, medical devices, and laboratory tests, coded using internationally standardized systems such as the 10th revision of the International Classification of Diseases (ICD-10) and the Anatomical Therapeutic Chemical (ATC) classification.²²

A unique advantage of the JMDC is its ability to longitudinally track patients across different healthcare institutions, enabling assessment of preindex risk factors and long-term outcomes.^{23 24}

Data access and quality control

Investigators received a preprocessed, deidentified dataset comprising eligible study participants according to predefined inclusion criteria. Direct access to the entire JMDC database was not available. The dataset had undergone standardized data cleaning and validation by JMDC before being released for research use.

Patient selection

In the JMDC database, 40 229 cases with tentative diagnostic codes for aortic dissection were initially obtained. To identify a clinically relevant cohort, we applied a multistep selection algorithm combining diagnosis and treatment codes—previously validated in our prior work.^{23 24} This approach aimed to increase specificity while maintaining sufficient sensitivity in claims-based data.

Patients identified as acute uTBAD between 1 January 2015 and 31 December 2023 were eligible for inclusion. Uncomplicated cases were defined by the absence of rupture, malperfusion, or hypertension (HT) requiring emergency intervention, or aortic-related death within 3 months of diagnosis, using claims-based criteria. These definitions are consistent with international clinical trials and registry criteria, such as INSTEAD-XL and IRAD.

The index date was defined as the date of the first hospitalization or chest CT scan associated with the aortic dissection diagnosis. To ensure a robust baseline assessment, we applied a 6-month lookback period, requiring continuous enrollment. Comorbidities were confirmed based on the co-occurrence of diagnosis and prescription codes within the same calendar month. To ensure comparability and avoid immortal time bias, patients who underwent TEVAR within 14 days of diagnosis were excluded, as early intervention is typically conducted in response to complications and reflects a different clinical scenario. A detailed summary of inclusion and exclusion criteria and code mappings is available in online supplemental table S3, with full code listings in online supplemental table S1.

The final cohort (n=4995) was stratified into two groups as follows:

• TEVAR group: patients who underwent TEVAR during the subacute phase (15–90 days after index date).^{25 26}

• iMT group: patients who received iMT without TEVAR within 90 days.

All patients were longitudinally followed across healthcare institutions using the JMDC’s cross-institutional linked claims system, which enables patient tracking across multiple healthcare facilities over time.

Patient demographics

Baseline demographic characteristics were obtained, including age, sex, and cardiovascular risk factors (CVRF), such as HT, dyslipidemia (DL), and diabetes mellitus. Additional comorbidities were assessed, including chronic obstructive pulmonary disease, cerebrovascular disease (CVD), chronic kidney disease (CKD), atrial fibrillation, congestive heart failure (CHF), ischemic heart disease (IHD), and Marfan syndrome.

Comorbidities were identified using a validated claims-based algorithm developed in our previous work.^{23 24} Specifically, each condition was confirmed by the co-occurrence of relevant ICD-10 diagnosis codes and prescription records (ATC codes) within the same calendar month to improve diagnostic specificity in administrative data.

Additionally, we extracted prescription data for medications commonly used to control heart rate and blood pressure in patients with aortic dissection, including renin-angiotensin system inhibitors (RASis), calcium channel blockers (CCBs), and beta-blockers (BBs). These medications were identified by ATC codes and assessed within the first 3 months following the index date. Adherence could not be directly measured; however, prescription records in this early phase were used as a proxy for therapeutic intent and treatment intensity. These variables were incorporated as covariates in the propensity score model.

Outcomes

The primary outcomes were: (1) aortic-related events and (2) all-cause mortality.

Aorta-related events were defined as a composite of:

• Aortic rupture (identified via emergency surgery codes or in-hospital death within 72 hours of diagnosis).

• Aortic reintervention (open surgical repair or additional TEVAR), aortic-related death (based on in-hospital death with a related diagnosis code).

• Progression to cTBAD requiring intervention.

These definitions align with prior observational studies and were adapted to the Japanese claims data structure (see online supplemental table S2).

All-cause mortality was defined using a validated algorithm combining discharge status, terminal care charges, and death certificate billing codes.

Secondary outcomes included:

• The annualized frequency of follow-up chest CT imaging (total CT claims divided by follow-up duration in years).

• The incidence of complications during follow-up (type A aortic dissection, CVD, paraplegia, venous thromboembolism, IHD, CHF, CKD, and disseminated intravascular coagulation).

• The use of antiplatelet agents or oral anticoagulants as proxies for cardiovascular risk management.

All outcomes were ascertained by linking diagnostic codes with procedural or prescription records within the same calendar month (see online supplemental table S1 for code lists).

Outcome evaluation began in the fourth month after the index date and continued until death, withdrawal from the insurance system, or 31 December 2023—whichever occurred first—to avoid immortal time bias and align with the subacute TEVAR exposure definition.

Since aortic rupture is rarely captured reliably in claims data and postmortem confirmation is unavailable, it was not analyzed as a standalone endpoint. Instead, we prioritized composite aortic events and all-cause mortality as more reliable and clinically meaningful long-term outcomes.

Statistical analysis

Baseline characteristics were summarized using means and SD for continuous variables and frequencies (percentages) for categorical variables.

To address potential sources of bias inherent in observational studies, several mitigation strategies were employed. Selection bias was addressed through propensity score matching (PSM) using prespecified covariates known to influence treatment selection. Information bias was minimized by using standardized diagnostic and procedure codes with established validity in previous validation studies. Surveillance bias, which could arise from TEVAR patients potentially having more frequent follow-ups, was addressed by including CT follow-up frequency as a matching variable. Immortal time bias was mitigated by defining a fixed exposure window (subacute phase: 15–90 days post diagnosis) and requiring survival to this period of cohort inclusion.

Age was treated as a continuous variable in PSM to enable precise matching across the age spectrum. CT follow-up frequency was calculated as an annualized rate (total CT scans divided by years of follow-up) to standardize for varying follow-up durations. Time-to-event outcomes were analyzed using Kaplan-Meier methods with censoring at death, study end, or withdrawal from the insurance system.

Propensity scores were generated using logistic regression, with TEVAR as the dependent variable and the following covariates: age, sex, CVRFs (HT, DL, diabetes), year of hospitalization, and early-phase pharmacological treatment (prescriptions for RASis, CCBs, and BBs).

To approximate clinical decision-making, the propensity score model included baseline CVRFs and postindex prescriptions for RASis, BBs, and CCBs—representing the quality and intensity of medical therapy administered in the early phase.

PSM was conducted using a greedy nearest-neighbor algorithm in a 1:5 ratio without replacements, with a caliper width of 0.2 SD of the logit of the propensity score.²⁴ Kaplan-Meier curves were used to estimate cumulative incidence, and the log-rank test assessed differences between groups.

All variables were derived from routinely collected administrative claims, which do not contain missing values for coded diagnoses, procedures, and prescriptions. Patients with incomplete data—such as less than 3 months of follow-up or an insufficient look-back period—were excluded during cohort construction. Therefore, no imputation was conducted.

All hypothesis tests were two-sided, and p<0.05 was considered statistically significant. No correction for multiple testing was applied, and results should be interpreted with appropriate caution.

All analyses were conducted using SAS V.9.4 (SAS Institute, Cary, North Carolina, USA).

Results

Data source and patient selection

The patient selection process is illustrated in figure 1. Between 1 January 2015 and 31 December 2023, 4995 patients met the inclusion criteria for acute uTBAD and were included in the final analysis. Among these patients, 96 underwent TEVAR during the subacute phase (15–90 days post diagnosis), while 4899 were initially managed with medical therapy (iMT).

Figure 1. Patient flow diagram. This flowchart details the patient selection process, outlining inclusion and exclusion criteria applied to obtain the final matched cohort. Overall, 4995 patients with uncomplicated type B aortic dissection from January 2015 to December 2023 were included (see onlinesupplemental tables S1S5). BBs, beta blockers; CCBs, calcium channel blockers; cTBAD, complicated type B aortic dissection; DL, dyslipidemia; DM, diabetes mellitus; HT, hypertension; ICD, International Classification of Diseases; iMT, initial medical therapy; JMDC, Japan Medical Data Centre; PS, propensity score; TEVAR, thoracic endovascular aortic repair; uTBAD, uncomplicated type B aortic dissection.

This analytical cohort was derived from a source population of approximately 20 million insured individuals included in the JMDC claims database, which primarily covers working-age beneficiaries under Japan’s national health insurance system. To construct the cohort, we applied a previously validated claims-based algorithm linking diagnostic, imaging, and procedural codes. The broader context of 40 229 patients with tentative codes for aortic dissection (ICD-10: I71.0) underscores the scalability and reproducibility of our method, as well as the generalizability of our findings to the real-world population undergoing conservative or interventional treatment for uTBAD in Japan.

It is important to note that the JMDC database does not include individuals aged ≥75 years, who are covered by the Japanese Medicare System for the Elderly Age 75 and Over (JMSE75). To address this limitation, we have secured access to a complementary national claims database from the JMSE75 and plan to apply the same methodology in future research to enable population-wide coverage across the full age spectrum.

Baseline characteristics

Before PSM, the median age was 55.0 years (IQR: 48.0–61.0 years) in iMT and 56.0 years (IQR: 50.0–61.5 years) in TEVAR. The proportion of females was slightly higher in iMT (26.5%) than it was in TEVAR (14.6%). Regarding medication use, RASis were prescribed in 52.0% of iMT patients and 65.6% of TEVAR patients (p=0.01), CCBs in 56.5% of iMT patients and 85.4% of TEVAR patients (p<0.01), and beta blockers (BBs) in 47.2% of iMT patients and 86.5% of TEVAR patients (p<0. 01).

After PSM using CVRFs before the index month and medical therapy following the index month as matching factors, both groups were well matched: the median age was 56.0 years (IQR: 50.0–62.0 years) in iMT and 56.0 years (IQR: 50.0–61.5 years) in TEVAR. Female representation was similar after matching, with 14.8% in iMT and 14.6% in TEVAR. These baseline characteristics are summarized in table 1 and online supplemental table S4.

Table 1. Baseline characteristics.

Variables	After the PS match n=576
	iMT		TEVAR		P value
	n=480		n=96
Age, years	56.0	(50.0, 62.0)	56.0	(50.0, 61.5)	0.97*
Female sex	71	14.8%	14	14.6%	0.96†
CVRF: HT	88	18.3%	18	18.8%	0.92†
CVRF: DL	30	6.3%	6	6.3%	1.00†
CVRF: DM	12	2.5%	1	1.0%	0.38†
COPD	0	0.0%	0	0.0%	-†
CVD	3	0.6%	0	0.0%	0.44†
CKD	32	6.7%	3	3.1%	0.18†
Atrial fibrillation	32	6.7%	4	4.2%	0.36†
CHF	30	6.3%	6	6.3%	1.00†
IHD	11	2.3%	1	1.0%	0.43†
Marfan syndrome	12	2.5%	0	0.0%	0.12†
Antiplatelets	66	13.8%	6	6.3%	0.04†
Oral anticoagulants	43	9.0%	5	5.2%	0.22†
Diuretics	146	30.4%	24	25.0%	0.29†
Rx: RASis	326	67.9%	63	65.6%	0.66†
Rx: BBs	419	87.3%	83	86.5%	0.82†
Rx: CCBs	419	87.3%	82	85.4%	0.62†

Comparison of demographic and clinical characteristics between the TEVAR and iMT groups after 1:5 propensity score matching (n=576). Comparison in the crude cohort (n=4995) is summarized in online supplemental table S4. Continuous variables are presented as medians with IQR, and categorical variables as counts and percentages. P values are calculated using the Wilcoxon rank-sum test for continuous variables and Fisher’s exact test for categorical variables. No missing values were observed for baseline characteristics, as all data were derived from mandatory claims entries.

^*Wilcoxon rank sum test.

^†Fisher’s exact test.

BBs, beta blockers; CCBs, calcium channel blockers; CHF, congestive heart failure; COPD, chronic obstructive pulmonary disease; CVD, cerebrovascular disease; CVRF, cardiovascular risk factors at the index month (HT, DL, DM); DL, dyslipidemia; DM, diabetes mellitus; HT, hypertension; IHD, ischemic heart disease; iMT, initial medical therapy; PS, propensity score; RASis, renin-angiotensin system inhibitors; Rx, medication use after index month (RASis, BBs, CCBs); TEVAR, thoracic endovascular aortic repair.

Primary outcomes: mortality and aorta-related events

In the crude cohort, the median follow-up period after the index month was 29 months for the iMT group (IQR: 14–53) and 31 months for the TEVAR group (IQR: 11–50.5). After PSM, follow-up duration remained similar: 28 months (IQR: 15–53) for the iMT group and 31 months (IQR: 11–50.5) for the TEVAR group (p=0.84).

The Kaplan-Meier estimate for aorta-related interventions (figure 2) showed a 60-month cumulative event rate of 13.5% (95% CI: 12.2% to 14.9%) in the crude iMT cohort and 21.9% (95% CI: 12.6% to 36.4%) in the crude TEVAR cohort, with no statistically significant difference observed between the groups (log-rank p=0.084). In the PSM cohort, the 60-month cumulative incidence was 19.9% (95% CI: 15.6% to 25.2%) in the iMT group and 21.9% (95% CI: 12.6% to 36.4%) in the TEVAR group, with no statistically significant difference observed (log-rank p=0.99).

Figure 2. Kaplan-Meier curve for aorta-related events. This Kaplan-Meier curve shows the cumulative incidence of aortic-related events in the TEVAR and iMT groups. (A) Crude cohort (n=4995): iMT (n=4899) had a 60-month cumulative incidence of 13.5% (95% CI: 12.2% to 14.9%), and TEVAR (n=96) had 21.9% (95% CI: 12.6% to 36.4%), with no significant difference (log-rank p=0.084). (B) PSM cohort (n=576): iMT (n=480) had a 60-month cumulative incidence of 19.9% (95% CI: 15.6% to 25.2%), and TEVAR (n=96) had 21.9% (95% CI: 12.6% to 36.4%), with no significant difference (log-rank p=0.99). Note: estimates are based on Kaplan-Meier survival analysis at 60 months. iMT, initial medical therapy; PSM, propensity score-matching; TEVAR, thoracic endovascular aortic repair.

For all-cause mortality (figure 3), the crude 60-month cumulative incidence was 5.3% (95% CI: 4.5% to 6.3%) in the iMT group and 4.4% (95% CI: 1.4% to 13.6%) in the TEVAR group, with no significant difference observed (log-rank p=0.89). In the PSM cohort, the 60-month mortality was 6.6% (95% CI: 4.0% to 10.6%) in the iMT group and 4.4% (95% CI: 1.4% to 13.6%) in the TEVAR group, with no statistically significant difference observed (log-rank p=0.70).

Figure 3. Kaplan-Meier curve for all-cause mortality. This Kaplan-Meier curve illustrates cumulative all-cause mortality for the TEVAR and iMT groups. (A) Crude cohort (n=4995): iMT (n=4899) had a 60-month mortality of 5.3% (95% CI: 4.5% to 6.3%), and TEVAR (n=96) had 4.4% (95% CI: 1.4% to 13.6%), with no significant difference (log-rank p=0.89). (B) PSM cohort (n=576): iMT (n=480) had a 60-month mortality of 6.6% (95% CI: 4.0% to 10.6%), and TEVAR (n=96) had 4.4% (95% CI: 1.4% to 13.6%), with no significant difference (log-rank p=0.70). Note: Estimates are based on Kaplan-Meier survival analysis at 60 months. iMT, initial medical therapy; PSM, propensity score-matching; TEVAR, thoracic endovascular aortic repair.

Detailed event counts and proportions are provided in online supplemental table S2.

Secondary outcomes

In the PSM cohort, the annual frequency of chest CT follow-up was similar between the TEVAR and iMT groups: 1.8 scans yearly (IQR: 1.3–2.4) in TEVAR versus 1.7 scans yearly (IQR: 1.1–2.7) in iMT (p=0.24). No statistically significant differences were observed in the incidence of major complications during follow-up. Medication usage patterns were also comparable between the groups. Detailed outcome data are summarized in table 2 and online supplemental table S5.

Table 2. Clinical outcomes.

Variables	After the PS match n=576
	iMT		TEVAR		P value
	n=480		n=96
Follow-up months	28.0	(15.0, 53.0)	31.0	(11.0, 50.5)	0.84
CT follow-ups / year	1.7	(1.1, 2.7)	1.8	(1.3, 2.4)	0.24
Death	19	4.0%	3	3.1%	0.70
Aortic events	70	14.6%	14	14.6%	1.00
Type-A AD	8	1.7%	1	1.0%	0.65
CVD	10	2.1%	1	1.0%	0.50
Paraplegia	6	1.3%	1	1.0%	0.87
VTE	180	37.5%	37	38.5%	0.85
IHD	57	11.9%	6	6.3%	0.11
CHF	105	21.9%	20	20.8%	0.82
CKD	68	14.2%	18	18.8%	0.25
DIC	96	20.0%	27	28.1%	0.08
Antiplatelets	136	28.3%	18	18.8%	0.05
Oral anticoagulants	82	17.1%	12	12.5%	0.27

Summary of patient outcomes in the PSM cohort (n=576), including follow-up duration, frequency of chest CT imaging, all-cause mortality, aortic-related events, and other complications. Summary in the crude cohort (n=4995) is summarized in Supplementary Table S5online supplemental table S5. Values are reported as medians with IQR for continuous variables and as counts with percentages for categorical variables.

Categorical values are reported as total numbers (%) and continuous variables as medians (IQR).

AD, aortic dissection; CHF, congestive heart failure; CKD, chronic kidney disease; CVD, cerebrovascular disease; DIC, disseminated intravascular coagulation; IHD, ischemic heart disease; iMT, initial medical therapy; PS, propensity score; PSM, PS matching; TEVAR, thoracic endovascular aortic repair; VTE, venous thromboembolism.

Discussion

In this study, we demonstrated the feasibility of using health insurance claims data to evaluate long-term outcomes of subacute-phase TEVAR vs iMT in patients with acute uTBAD.

In the crude cohort, the iMT group showed a steady, time-independent accumulation of aortic-related events over a 60-month follow-up period. This trend shows the persistent long-term risk associated with conservative management and supports the classification of uTBAD as an ongoing unmet medical need.²⁷,³³

After PSM, no statistically significant differences were observed between the TEVAR and iMT groups in either all-cause mortality or aorta-related events. These results affirm the clinical equipoise that underpins ongoing RCTs, including the IMPROVE-AD, SUNDAY, and EARNEST studies.¹⁵,¹⁸

Collectively, these findings validate the robustness of our outcome definitions and show that carefully designed claims-based analyses can yield RWE that supports clinical decision-making.^{34 35} Within the IDEAL framework, this approach represents a feasible, scalable and cost-effective strategy for the long-term evaluation of surgical innovation.^{19 20}

Claims-based data analysis

This claims-based study underscores the unique value of RWD in providing a propensity-matched cohort for comparing TEVAR and iMT in uTBAD. It addresses the limitations of surgical registries that typically focus on periprocedural outcomes without long-term iMT data. Our PSM analysis showed comparable rates of aorta-related events in the TEVAR group but no significant differences in overall mortality. Outcomes may further improve with continued advancements in device technology and surgical technique. However, there remains room for improvement in real-world adherence to medical therapy in iMT patients as well.

Compared with the Duke University study by Weissler et al,¹⁴ both studies used insurance claims data and propensity scores to reduce bias between TEVAR and iMT groups, though important differences exist. The Duke study focused on TEVAR within 30 days (acute phase), while our study targeted the subacute phase (15–90 days), aligning with international reporting standards^{25 26} that may allow for more stable conditions and potentially improved procedural outcomes.¹³ Another key distinction is patient demographics: the Duke cohort included patients aged ≥65 years, whereas our cohort included patients aged ≤75, reflecting differences in healthcare systems and patient demographics. Together, these studies show the robustness and practicality of claims-based analysis for real-world insights within the IDEAL framework.

Global quality initiatives

International collaborations are essential in advancing RWE for aortic dissection management, helping to establish standards for patient care and quality improvement in vascular surgery. Initiatives such as the European Vascular Registry Network (VASCUNET), the International Consortium of Vascular Registries (ICVR), and the Vascular Quality Initiative (VQI) have provided detailed benchmarking data, driving improvements across regions.³⁶ For instance, the VQI registry has been pivotal in generating RWE,¹¹ including post-market studies mandated by the US Food and Drug Administration (FDA) for TEVAR.³⁷ More recently, the Vascular Implant Surveillance and Interventional Outcomes Coordinated Registry Network ³⁸ has integrated registry data with Medicare and other longitudinal datasets, offering a cost-effective and scalable model for continuous quality improvement.^{39 40}

However, these initiatives have predominantly focused on Western populations, often underrepresenting Asian ethnic groups. In this study, we addressed this gap by contributing RWE from Japan, thereby enriching the global understanding of uTBAD management and highlighting the importance of including more ethnically diverse populations. By adding data from an Asian context, we support broader international collaboration and advocate for more inclusive quality improvement efforts in vascular care.

Japanese perspectives

Our findings reveal key similarities and nuanced differences between Japan and Western countries in the management of uTBAD, highlighting Japan’s potential to make meaningful contributions to global quality improvement initiatives. TEVAR was approved in Japan for acute cTBAD in 2015 and chronic cTBAD in 2019,¹⁰ positioning our study (2015–2023) as a timely evaluation of TEVAR versus iMT in the context of uTBAD.

Japan has adopted systematic reviews to develop evidence-based clinical guidelines, with recent updates in 2022 aligning more closely with international standards.^{10 41} In parallel, the use of large-scale administrative databases has been increasing, mirroring global research trends.⁴²

The growing adoption of TEVAR in Japan has been documented in several national databases, including reports from the Japanese Society for Vascular Surgery,⁴³ the Japan Cardiovascular Surgery Database,⁴⁴ and the National Clinical Database. The Japanese Committee for Stentgraft Management reported over 22 250 TEVAR procedures between 2006 and 2016, including 4259 cases of TBAD.⁴⁵ Additionally, the Japanese Registry of All Cardiac and Vascular Diseases - Diagnosis Procedure Combination database recorded TEVAR use in 4.4% of acute TBAD cases.⁴⁶

Previous Japanese studies have primarily focused on perioperative outcomes; however, our study is the first to provide a long-term, multiinstitutional analysis using anonymized health claims data. Since TEVAR is regulatory indicated for cTBAD but not explicitly approved for uTBAD in Japan,¹⁰ we distinguished between the traditional definition of cTBAD (life-threatening conditions) per the Society for Vascular Surgery (SVS)/Society of Thoracic Surgeons (STS) guidelines,²⁶ and the broader IRAD definition, which encompasses less acute presentations.⁴⁷

Japan’s high utilization of CT imaging and strong follow-up adherence^{23 46} likely enhance the quality of long-term data, supporting alignment with SVS/STS reporting standards.^{25 26} By adhering to these standards while addressing contextual nuances, Japan’s approach offers valuable insights and a meaningful contribution to international quality improvement efforts in vascular care.

Limitations

This study has some limitations inherent to the use of claims-based data, which are not originally designed to answer specific clinical research questions.

First, the JMDC database enables cross-institutional patient tracking; nevertheless, it lacks detailed clinical information, such as imaging findings and anatomical characteristics. This may reduce the precision of diagnostic and procedural assessments. Additionally, diagnostic accuracy depends on coding practices, which may vary between institutions and over time, potentially introducing misclassification bias.

Second, despite applying PSM to reduce selection bias, unmeasured confounding cannot be excluded. Variables such as aortic diameter, dissection extent, entry tear location, symptom severity, and institutional or physician preferences may have influenced treatment allocation but are unavailable in the claims data. To mitigate this, we incorporated CVRFs and early-phase medication use as proxies for disease severity and treatment intent. In the PSM process, 480 of 4899 iMT patients (9.8%) were selected to match with 96 TEVAR patients. The higher event rates in the matched iMT group, compared with the crude cohort, indicate that PSM successfully identified iMT patients with risk profiles similar to TEVAR candidates, supporting the validity of the comparison.

Third, outcome ascertainment relied on healthcare utilization records within the claims database. Events occurring outside the healthcare system—such as sudden death at home or fatal aortic rupture without hospitalization—may have been missed. Aortic rupture, in particular, is rarely confirmed without postmortem investigation and is often underascertained in claims data and clinical registries. Accordingly, we focused on more robust endpoints, such as all-cause mortality and aorta-related interventions, which are more reliably recorded. However, claims-derived mortality data are known to be incomplete, especially for outpatient deaths. As reported by Sakai et al,⁴⁸ the sensitivity of death identification in the JMDC database is high for inpatients (94.3%) but considerably lower for outpatients (47.4%). Even when including terminal care charges and death certification codes, the overall sensitivity improved only modestly (to 54.7%). Nonetheless, we believe that comparing against a parallel control group using PSM helps minimize differential bias.

Fourth, as the study was conducted in Japan—where access to advanced imaging and structured follow-up is extensive—caution is warranted when generalizing these findings to healthcare systems with differing infrastructures.

Finally, given the relatively small number of TEVAR cases (n=96) and a modest number of outcome events—particularly in the matched cohort—the study may have been underpowered to detect small-to-moderate treatment differences. This limitation should be considered when interpreting the lack of statistically significant differences in all-cause mortality.

Despite these limitations, this study shows the feasibility of using routinely collected claims data to monitor real-world management and long-term outcomes of uTBAD in Japan. The current database scale is limited; however, its coverage continues to expand, and future analyses with larger datasets are expected to improve the precision of outcome estimates and support more definitive conclusions.

Conclusion

This study represents the first nationwide claims-based analysis in Japan comparing long-term outcomes between subacute-phase TEVAR and iMT in patients with acute uTBAD. While TEVAR may promote favorable aortic remodeling, no significant difference in long-term all-cause mortality was observed between the treatment groups, thereby reinforcing the clinical equipoise underpinning ongoing randomized trials.

Our findings also show that structured claims data can capture the natural history of uTBAD under iMT, providing a valid and scalable real-world comparator for the long-term evaluation of surgical innovations. This study aligns with stage 4 of the IDEAL framework and shows the feasibility of using RWE for comparative effectiveness research in vascular care.

As health insurance databases continue to expand, integrating claims data with registry-based and imaging datasets will be essential to refine patient selection, optimize intervention timing, and support global quality improvement initiatives in uTBAD management.

The views expressed in this study are those of the authors and do not necessarily reflect those of the Ministry of Health, Labour and Welfare (MHLW), Japan, Agency for Medical Research and Development (AMED), or their affiliated institutions. The funding organization had no role in considering the study design, collection, analysis, interpretation of data, writing of the manuscript, or decision to submit the manuscript for publication. The JMDC database is maintained independently by JMDC Inc. with no involvement from study funders. The funding organizations had no role in considering the study design, collection, analysis, interpretation of data, writing of the manuscript, or decision to submit the manuscript for publication.

Data availability statement

Data may be obtained from a third party and are not publicly available.