Abstract
Background
Multicenter, standardized, long-term data are essential for assessing outcomes of patients following aortic (valve) treatment. To facilitate this, the international Heart Valve Society (HVS) Aortic Valve (AV) Database was established. This paper provides an overview of the HVS AV Database.
Methods
The HVS AV Database includes adult patients with aortic valve disease (regurgitation and/or stenosis), with or without ascending aorta aneurysm, undergoing surgical or transcatheter intervention. It has an ambispective design, and centers interested in interventional outcomes following aortic (valve) intervention can participate free of charge.
Results
The variables collected in the database are harmonized with the International Consortium for Health Outcomes Measurement (ICHOM) standard Set of Patient-Centered Outcome Measures for Heart Valve Disease (HVD). In total, 33 outcome variables, including 12 clinical and 21 echocardiographic, and 286 case-mix variables are collected. Currently, the database includes 10,893 patients from 78 participating centers worldwide, with 7910 patients having undergone aortic valve repair and 1968 aortic valve replacement. Participating centers can extract their own data at any time, and multicenter research proposals are encouraged. The database is compliant with GCP, FDA, GDPR and HIPAA regulations, supporting data transfer across continents.
Conclusions
The HVS AV Database is aimed at improving outcomes for patients requiring intervention for aortic (valve) disease. Through uniform scientific reporting, it contributes to global efforts to standardize data capture, improve patient outcomes and shift the focus from device to global patient outcomes.
Keywords
Data resource basics
Background
Aortic valve disease (AVD), including aortic valve stenosis (AS) and aortic valve regurgitation (AR), is the most common form of heart valve disease (HVD), and its prevalence continues to increase as the global population ages.1,2 Moreover, AVD is inevitably intertwined with thoracic aortic aneurysms (TAA). 3 Over the past few decades, the treatment of patients with aortic (valve) disease has seen a wide range of options, varying from medical treatments to surgical or transcatheter interventions. Medical treatment is often temporary and serves mainly as a bridge-to-intervention, 4 and even though a plethora of intervention techniques for AVD or TAA exists, there is significant global variation in their use. Most studies reporting on outcomes after surgical treatment for AVD or TAA are often single-centered with a lack of long-term follow-up.
Many knowledge gaps still remain surrounding the field of surgical or transcatheter intervention for AVD or TAA, and in order to address these, standardized, multicenter data are needed. This has led to the launch of the international Heart Valve Society (HVS) Aortic Valve (AV) Database, one single platform that allows for an evaluation of outcomes after all aortic (valve) treatment.
The HVS AV Database
In August 2013, the HVS AV Database was established as part of the Aortic Valve Research Network (VRN) of the HVS. 5 The database was initially named Aortic Valve Insufficiency and ascending aorta Aneurysm InternATiOnal Registry, AVIATOR. AVIATOR was a longitudinal observational cohort study enrolling patients with ascending aorta aneurysm and/or AR. 5 Patients operated on for type A aortic dissection could be included as well. In 2019, AVIATOR merged with the Long-Term Evaluation Of Prosthetic Aortic Valve Replacement (LEOPARD) registry, enrolling patients with AS and surgical or transcatheter intervention. Since the merge of AVIATOR with the LEOPARD registry, the database is named the HVS AV Database.
Governance
The HVS AV Database project is officially declared as a VRN of the HVS. It is governed by a group of core leaders with a rotating chair, a Scientific Committee (SC), a database management team and a VRN coordinator, all overseen by the HVS board (Figure 1). The core leaders serve as the executive body of the Aortic VRN, leading its general direction, dealing with financial issues and interactions with other HVS working groups. The database management team, consisting of a database coordinator and a database manager, are responsible for maintaining the database, including guiding application processes of new centers, maintaining the data in the online database platform, data storage, data usage, data quality monitoring and data quality enhancement initiatives. The SC consists of six members, including two cardiologists, two surgeons and two basic scientists, one of whom is chair. Members of the SC serve for three years, after which they can reapply for another three years of service or the position of chair of the SC. The SC reviews research proposals and scientific output (i.e. abstracts, papers and presentations) originating from the registry.
Governance. HVS: Heart Valve Society; AV: Aortic valve. Created in BioRender. 6
Patient inclusion criteria
Patients eligible for the database are those undergoing surgery because of AR (including congenital mixed AVD) and/or aortic aneurysm (root or ascending aorta) and are above the age of 18 years old at the time of surgery. Moreover, patients with a surgical or transcatheter intervention because of AS can be included if they are between the age of 18 and 65 years old at the time of intervention. Patients who had surgery for type A aortic dissection and who are above the age of 18 years old at the time of surgery are eligible as well.
Database design
The HVS AV Database has a hybrid ambispective design, with patients included entirely retrospectively, retrospectively with both retrospective and prospective components, or entirely prospectively. Patient inclusion in the database before the official entry of a center into the Aortic VRN is considered the retrospective part of the ambispective design. The ambispective design of the database allows for the assessment of multiple short- and long-term outcomes.
Participating centers & included patients
The HVS AV Database originated in Europe, and the majority of centers participating are European. However, the database covers patient populations from almost every continent, with a total of 78 centers participating in the HVS AV Database (Figure 2) of which 36 centers are actively enrolling patients. The database currently includes 10,893 patients: 9983 patients (91.2%) from 62 centers across Europe, 675 patients (6.2%) from six centers across North America, 163 patients (1.5%) from seven centers across Asia, 43 patients (0.4%) from one center across Africa and 29 patients (0.3%) from two centers across South America. Table 1 provides an overview of patients currently included in the database, baseline characteristics and peri-operative outcomes, stratified for reason for referral to intervention. For the purpose of illustration, only complete cases for reason for referral to intervention were included in the analysis, leading to 9920 patients.
Participating centers. Created in BioRender. 7
Baseline characteristics and peri-operative outcomes of HVS AV Database.
Continuous data are presented as means with standard deviations (normally distributed data) or medians with interquartile ranges (non-normally distributed data), and categorical data are presented as proportions. The Anderson-Darling test was used to test for normality of the data. In the case of normally distributed data, the One-Way ANOVA was used to compare three or more groups, and the Kruskal-Wallis test was used for comparison of non-normally distributed data. The Chi-square test was used for categorical variables. P-values <0.05 were considered statistically significant. A Bonferroni test was used for post hoc analyses. HVS AV: Heart Valve Society Aortic Valve; AV: aortic valve; n: number of patients; y: years; IQR: interquartile range; NYHA: New York Heart Association; h: hours; min: minutes; AC: aortic cross-clamping.
Median length of follow-up currently is 2.6 years (IQR: 2.1-6.5), encompassing 44,634 patient-years. One-year follow-up completeness is 73.1%, and two-year follow-up completeness is 67.5%, as calculated by the method proposed by Wu et al. 8 Overall follow-up completeness is 49.3%. Early mortality across all procedures is 1.5% and linearized occurrence rate for late mortality is 1.2%, where early mortality is defined as ≤ 30 days from intervention and late mortality >30 days.
Data linkage
External data are not linked by default, but the structure of the HVS AV Database is designed to enable data linkage to other databases. Specifically, the HVS coordinated a multi-society task force that developed a standardized set of outcomes for patients with heart valve disease, called the International Consortium for Health Outcomes Measurement (ICHOM) standard Set of Patient-Centered Outcome Measures for Heart Valve Disease (HVD). 9 The goal of ICHOM is to promote alignment of outcome measurements globally using standardized outcome measurement. The variables described in ICHOM's standardized set of outcomes for HVD provide the basis for the variables collected in the HVS AV Database. As such, data of other registries or cohorts that use the ICHOM standard Set of Patient-Centered Outcome Measures for HVD can be externally linked to data of the HVS AV Database.
Ethics
Each participating center is responsible for ensuring that the use and import of data in the database comply with local ethical standards. The study is conducted according to the principles of the World Medical Association (WMA) Declaration of Helsinki on Ethical Principles for Medical Research Involving Human Subjects (19 October 2013) and the WMA Declaration of Taipei on Ethical Considerations regarding Health Databases and Biobanks (October 2016).
Funding
The HVS AV Database is supported by industry funding (Edwards Lifesciences and Vascutek Limited, trading as Terumo Aortic). Funders do not have access to data and are not involved in scientific output. However, data can be provided anonymously for quality improvement of their own products, for example in the case of the European Union Medical Device Regulation (EU MDR).
Data collected
In order to be able to address a broad scope of research questions, the HVS AV Database tracks multiple clinical or imaging outcome variables and case-mix variables, which are, together with their respective definitions, harmonized with the ICHOM standard Set of Patient-Centered Outcome Measures for HVD. 9 Patient-level information is collected from the hospital's medical records and entered into an electronic case report form (eCRF) containing the variables through a secure data-entry system. Data can be uploaded in the eCRF manually (i.e. per patient) or using batch uploads. Collected data include baseline patient characteristics, procedural information, in-hospital outcomes, longitudinal measurement of echo parameters and (valve-related) events during at least yearly follow-up.
Outcome variables
The HVS AV Database tracks 33 outcome variables, of which 12 clinical and 21 echocardiographic outcome variables. Echocardiographic repeated measurements are collected longitudinally. Additionally, the database is designed to record repeated time-to-event outcomes, facilitating the analysis of repeating events (Table 2). All tracked outcome variables are specified together with their respective definitions in the data dictionary published online (https://heartvalvesociety.org/Aortic/Registry/documents.cgi).
Overview of collected outcome data of HVS AV Database.
* Including: in-hospital mortality and 30-days mortality.
** Including: in-hospital AV and/or aortic reoperation and valve-related events, in-hospital reoperation for bleeding/tamponade or mediastinitis, in-hospital non-cardiac or other cardiac reoperation, new unplanned PM implantation, peri-procedural myocardial infarction, acute kidney injury and blood transfusion.
*** Including: aortic annulus, sinuses of Valsalva, sino-tubular junction and tubular/ascending aorta diameters.
HVS: Heart Valve Society Aortic Valve; AV: aortic valve; TIA: transient ischemic attack; PM: pacemaker; NYHA: New York Heart Association; MV: mitral valve.
Case-mix variables
A total of 286 case-mix variables are tracked in the HVS AV Database in a hierarchical structure. This means that if, for example, a patient had an aortic valve repair, variables detailing aortic valve replacement are not applicable, or vice versa. All tracked case-mix variables, as well as their respective definitions, are specified in the data dictionary published online (https://heartvalvesociety.org/Aortic/Registry/documents.cgi).
Privacy considerations
Participating centers have a joint controllership agreement, and all participating centers have responsibilities to protect the personal data captured in the database. The database was established according to Good Clinical Practice (GCP) guidelines and is fully compliant with the General Data Protection Regulation (GDPR), the United States Food and Drug Administration (FDA) and the Health Insurance Portability and Accountability Act (HIPAA).
When centers enter data into the eCRF, a unique patient code is automatically generated. This code is specific to each patient and contains no identifying information. The identifying key file is not stored in the database and remains with the participating centers. Individual center data are not visible for any other center.
Data quality
Long-term follow-up is an essential part of epidemiologic studies, but often contains missing (follow-up) data. Missing data, and specifically missing follow-up, potentially leads to misestimation of event rates, introducing bias and compromising the validity of studies.10,11–12 Accurate and valid data are also essential to derive valid conclusions from databases. 13 Consequently, the HVS AV Database incorporates various data checks, procedures and feedback mechanisms to maintain high data quality and adequate follow-up completeness. These include the use of a data dictionary and monitoring timely and complete entry by feedback and benchmark reports. Utilizing a data dictionary with variable definitions that are monosemic will minimize center heterogeneity. As yearly follow-up of patients is one of the requirements for the database, centers that cannot meet this criteria will be put on ‘silent’, and their data will not be used for future data extractions. The annual feedback and benchmark reports are generated for each center by the database management team, which assess the completeness as well as the correctness of the submitted data and compare them with data from the whole database. Centers are requested to adjust their data according to these reports. Additionally, multiple validation steps are built into the eCRF in order to prevent incompatible responses.
Data resource use
The HVS AV Database provides a comprehensive platform that allows for standardized comparison of outcomes after aortic (valve) treatment. Each center has the ability to extract data from their own center at any point in time. In order to conduct multicenter research using data from the database, a research proposal must be submitted to the SC, which will review the proposal. Data from a center will be used automatically in a research proposal through a non-opposition procedure, giving centers the opportunity to opt out from research proposals. As soon as the data extraction for the research proposal is provided, the research team has a deadline of six months to perform the analysis and write a draft publication. If the deadline is exceeded, the topic can be passed on to another group. In all research endeavors, centers that contributed data to the specific project are eligible for co-authorship of its scientific publication.
Since the start in January 2013, seven articles have already been published, with 13 ongoing multicenter research projects.5,14151617181920–21 Most recently, the predictive value of AV cusp retraction, calcification and fenestration for feasibility of aortic valvuloplasty was investigated, 21 and in another study, the postoperative survival based on an early versus class-I indication for surgery was compared in patients undergoing surgery for severe AR. 20 Ongoing research projects include for example a propensity-matched comparison of aortic valvuloplasty and prosthetic AV replacement, short and mid-term results of AV repair and replacement in patients affected by infective endocarditis, or the type of cusp fusion and commissural orientation as a predicting factor for failure of AV repair in patients with bicuspid aortic valves. Moreover, aside from multicenter research outputs, two informative articles reporting on the initiative and structure of the database have been published as well.5,14 An up-to-date list of all published studies using data of the HVS AV Database can be found on the website (https://heartvalvesociety.org/Aortic/Research/finished.cgi), as well as an up-to-date list of all ongoing research projects and their respective research proposals (https://heartvalvesociety.org/Aortic/Research/ongoing.cgi).
Strengths and weaknesses
With almost 11,000 patient records as of 2024, the HVS AV Database forms a large platform from which multicenter research on surgical outcomes of the AV and/or aorta can be conducted. The particular strength of the database lies within the harmonization with the ICHOM standard Set of Patient-Centered Outcome Measures for HVD. 9 By using a standardized set of outcomes, the database facilitates uniform scientific reporting, which is the first step to globally standardize data capture of outcomes after aortic (valve) intervention. Ultimately, the rich uniform data collection of the HVS AV Database minimizes unobserved confounding factors and supports multiple subgroup analyses as well as the assessment of longitudinal trends, which will provide a deeper insight into the outcomes after aortic (valve) intervention. 22 Furthermore, global patient inclusion ensures better representation of patients with aortic (valve) disease. Additionally, the ability of the HVS AV Database to follow patients over time as well as its design to record repeated events allow for repeated time-to-event analyses.
As with other databases with long-term follow-up of patients, there is always the challenge of incomplete data, especially incomplete follow-up data. Moreover, the largest proportion of AV repair in the HVS AV Database is performed by high volume expert participating centers. Fostering active participation of lower volume centers is essential in order to acquire results that are generalizable. The installment of a dedicated database management team, which facilitates the monitoring of data quality and completeness, is key to achieving these goals.
Data resource access
Researchers, cardiologists and cardio-thoracic surgeons interested in conducting research using data from the HVS AV Database are highly encouraged to join. All centers performing aortic (valve) interventions can join the database free of charge. Collaboration with other registries is highly encouraged. The only requisite is that at least one member of each center is a member of the HVS. Application documents, protocol templates, research proposal templates, the data dictionary and the database hosting system can be found online on the website of the database: https://heartvalvesociety.org/Aortic/. Data from the HVS AV Database can be imported into any statistical software package by allowing researchers to request their preferred data format. Any queries can be directed to the HVS AV Database management team (aorticvalvedatabase@heartvalvesociety.org).
Footnotes
Acknowledgements
None.
Author contributions
Funding
The HVS AV Database is supported by funding from Edwards Lifesciences and Vascutek Limited, trading as Terumo Aortic.
Declaration of conflicting interests
The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Carlijn C.E.M. van der Ven, Jolanda Kluin, Johanna J.M. Takkenberg, Ismail El-Hamamsy, Bart Meuris, Pouya Youssefi and Kevin M. Veen have nothing to disclose.
Peter Verbrugghe has received consulting fees from Corcym; his institution has received payment or honoraria for educational events by Atricure and Artivion.
Philippe Pibarot has received institutional funding with no direct compensation from Novartis, Edwards Lifesciences, Medtronic and Pi-Cardia.
Emmanuel Lansac has received royalties or licences with no payment from Coroneo Inc: patent Extra Aortic ring. He has received direct payment from Gettinge for a lecture.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
