In this multicenter cohort study, we identified all adults with BAV from the outpatient clinics of three major departments of cardiology in Denmark during three department-specific accrual periods: 2018–2020 (Zealand University Hospital), 2019–2020 (Aarhus University Hospital (AUH)), and 2020–2022 (Herlev-Gentofte Hospital). Participating departments have dedicated outpatient clinics for patients with simple congenital heart disease, providing both pre-surgical clinical surveillance and post-surgical follow-up. Furthermore, AUH is a tertiary hospital with highly specialized cardiology functions, including heart surgery and a complex congenital heart disease clinic for both pediatric and adult patients. To identify possible BAV cases, we screened electronic health charts of patients with active affiliation to outpatient clinics for congenital heart disease and patients with an ICD-10 code of BAV or congenital aortic valve disease. Next, we reviewed echocardiograms in our imaging databases to visually confirm the BAV diagnosis. We defined BAV by the presence of a fish mouth–shaped orifice or only two distinct cusps.

To identify a prospective cohort from this mixed population of BAV outpatients, we included all patients with an available echocardiogram performed after 31st December 2017, since this was the beginning of the accrual period. Exclusion criteria were prior surgery of the aortic valve or ascending aorta, genetic syndromes, concomitant congenital heart malformations except unrepaired ventricular or atrial septal defects, and age < 18 years. Prior surgical correction of aortic coarctation was not an exclusion criterion. We also excluded patients with a unicuspid valve and patients with an indeterminate anatomical classification. Lastly, we excluded 12 patients with partial fusion BAV.

The primary outcome was a composite endpoint including any interventional procedure—both endovascular and surgical—targeting the aortic valve or ascending aorta. This included aortic valve replacement or repair, and aortic reconstruction, but excluded procedures exclusively correcting aortic coarctation. Procedures with concomitant coronary artery bypass grafting or percutaneous coronary intervention were included. This primary outcome will be referred to as surgery of the aortic valve or ascending aorta. In each patient, we registered the time from the date of the first available echocardiogram in the study period (index TTE) until the date of surgery of the aortic valve or ascending aorta, death, or end of observation, whichever came first. The last observation for each patient was the date of the most recent review of their electronic health chart. For the whole study population, the last observation was during the period March–December 2022.

We retrieved data by reviewing electronic health journals and imaging databases (EchoPAC and Viewpoint (GE Healthcare), Impax (AGFA)) and collected data in a secure RED-Cap database hosted by Region Zealand [14, 15]. Since the electronic health journals in Denmark cover all hospital contacts within the same region and receive discharge letters from all other hospitals, we expect a highly sensitive detection of clinical events. Furthermore, the electronic health journal is linked to the Danish Civil Registration System and automatically registers all deaths.

BAV morphology was categorized both according to major BAV type (fused vs. two-sinus BAV) and by specific phenotypes as defined in the Consensus Classification [13]. For a comprehensive introduction to the Consensus Classification and how it relates to the universally used classification as suggested by Sievers and Schmidtke, we refer to the original statement [13, 16]. In brief, the major BAV types differ by the shape of the aortic root and the symmetry and angulation of the two cusps. A fused BAV has a three-sinus-shaped aortic root, asymmetric cusps, and possibly a visible raphe. The aortic root of the two-sinus BAV is nearly oval with only two distinct sinuses, and the cusps are more symmetric. Compared to the Sievers and Schmidtke classification, the major changes in the Consensus Classification are that Sievers type 2 with two raphes is no longer considered a bicuspid valve. Furthermore, a visible raphe, which is a prerequisite for Sievers type 1 classification, is not essential for the morphology to be considered a fused BAV. As a result, all two-sinus BAVs are expected to be classified as Sievers type 0, while the fused BAV group likely includes both patients previously classified as Sievers type 0 and type 1. The major BAV types of the Consensus Classification may be further subdivided into specific phenotypes based on the spatial orientation of the commissures (see Figure 2). To ensure correct anatomical classification, we reviewed all available transthoracic and transesophageal echocardiograms of each patient. We further noted the visibility of the raphe.

All echocardiograms were routinely performed during clinical check-ups in the out-patient clinics. For the purpose of this study, examinations were re-reviewed off-line. Assessors were primarily consultant cardiologists working in one of the BAV outpatient clinics or one cardiology resident with approximately 2 years of echocardiography training and practical experience. In cases of uncertainty, consensus was reached between at least two assessors. To assess differences between the anatomical subgroups at baseline, we report the presence and grade of valve dysfunction, aortic dilatation, and left ventricular ejection fraction (LVEF) at index TTE. The grade of aortic stenosis (AS) was assessed using the mean/max pressure gradients across the valve and the aortic valve area (AVA) as calculated by the continuity equation. If calculation by the continuity equation was not accessible, we estimated AVA by planimetry, if a high-quality transesophageal echocardiogram was available. Since pressure gradients and AVA could not always be measured at the same echocardiogram, we categorized AS severity as either moderate-to-severe stenosis or absent-to-mild stenosis. We defined moderate-severe stenosis as the presence of at least one of the following: mean pressure gradient ≥ 20 mmHg, max pressure gradient ≥ 40 mmHg, or AVA < 1.5 cm2. Likewise, no more than mild AS was confirmed if at least one of the variables (pressure gradients or AVA) was available, but measurements were below the threshold for moderate-severe AS. If both pressure gradients and AVA measures were unavailable, the AS variable was registered as missing. The presence and grade of aortic regurgitation (AR) were re-assessed semi-quantitatively in accordance with current guidelines [17]. Left ventricular ejection fraction (LVEF) was estimated using the Simpsons biplane method and categorized as follows: normal (LVEF > 55%), mildly reduced (LVEF 40-55%), moderate-severely reduced (LVEF < 40%). The ascending aorta was measured in the parasternal long-axis view at end-diastole from leading-edge to leading-edge. We measured the cross-sectional diameter at sinus of Valsalva, sinotubular junction, and the broadest part of the visible tubular ascending aorta. We defined aortic dilatation by a measurement of at least 39 mm at any segment.

Finally, we registered date of birth, sex, aortic coarctation, and concomitant heart defects. We noted whether patients reported a family history of confirmed or suspected BAV, sudden cardiac death, or aortic dissection. From the procedural reports, we noted the primary indication for the procedure on the aortic valve or proximal aorta and the type of intervention.

After assessing their distribution, continuous variables are summarized using median [interquartile range] or mean [SD] as appropriate and compared using Wilcoxon-rank-sum test or t test as appropriate. Categorical or binary variables are presented as counts [percentage] and compared using chi-square tests. The frequency of the primary outcome is described using the incidence rate per 100 person years overall and according to major BAV type.

To assess the association between BAV anatomy and the primary outcome, we estimated the hazard ratio with 95% confidence intervals (HR [95%CI]) in both simple and sex-adjusted cause-specific Cox regression models using age as timescale. Since study inclusion was dependent on patients surviving until their age at index TTE, we accounted for this delayed entry in the Cox models. Consequently, each patient only contributed observation time to the models starting from their age at the index TTE. Observation continued until either the age at time of surgery or the age at censoring (death or end of observation). In all analyses, we used the largest subgroup, major BAV type (fused), or phenotype (fused R-L phenotype), as the reference for comparison. The null hypothesis of no difference in hazards between subgroups was assessed by Wald’s test. The proportional hazards assumption (i.e., that the hazard ratio remains constant across all ages in our study) was assessed by testing the independence between time and Schönfeld residuals for each variable, as well as for the global multivariate model. Furthermore, estimated cumulative hazard was visualized using Nelson-Aalen curves. To evaluate whether the association between major BAV type and timing of surgery differed by sex, we tested for the presence of interaction between BAV type and sex using the likelihood ratio test.

We conducted several secondary analyses to investigate potential factors that may mediate the association between BAV morphology and timing of surgery. First, we hypothesized that any observed association would be mediated by development of valve dysfunction or aortic dilatation at an earlier age. Therefore, we visualized the prevalence of moderate-to-severe valve dysfunction and aortic dilation > 39 mm at index TTE across three age strata (18–40 years, 40–60 years, and > 60 years) and by BAV morphology. Associations between BAV type and moderate-severe valve dysfunction or aortic dilatation were assessed in three logistic regression analyses adjusting for age. Secondly, we conducted several univariate Cox regression analyses to evaluate potential associations between the age-specific hazard of surgery and the following risk factors: male sex, aortic coarctation, and the following findings on index echocardiogram: moderate-severe valve dysfunction, aortic dilatation > 39 mm at any segment, and visible raphe. In all hypothesis tests, we considered a two-sided p value < 0.05 as significant. All data processing, statistical analyses, and visualization were performed using R version 4.4.2 [18].