Methods
Design and study population From February 2012 to May 2013, 14 667 children and young-adults (aged 40 years-old or less) of the central region of Portugal participated in the SCD-SOS survey (NCT01845909). It included a 12-lead ECG and a previously validated (5) questionnaire about symptoms, personal and family history that was filled by 11878 patients willing to provide more information (Supplementary Material 1 ). The SCD-SOS protocol was approved by the local Ethics Committee (355/Sec/10/03/2011), Comissão de Ética do Centro Hospitalar de Coimbra, Instituto Português do Ritmo Cardíaco e Comissão Nacional de Protecção de Dados.
SCD-SOS Survey and etiologic evaluation This was a detailed, digital-based, self-administered questionnaire, mainly composed of multiple-choice questions and few boxes allowing participants to provide additional details on their answers. It deemed to explore relevant clinical characteristics in the setting of SCD risk, and detect syncope related to potential causes of SCD. Based on the answers provided (Supplementary Material 1 ), we were able to determine the most probable etiology of the TLOC episodes reported. According to the 2018 ESC syncope guidelines, (7) reflex syncope (RS), either vasovagal or situational, was assumed in individuals who reported an episode precipitated by pain, emotion, fear, warm environment, or standing, with or without at least one typical progressive prodrome (pallor, sweating, and/or nausea). Orthostatic hypotension was presumed if syncope after or while standing was described in the absence of any of the previous prodromes, or if volume depletion was presumed based on the specifications provided (post-surgery, after donating blood). The remaining causes were ascertained based on the features ticked by the participants and US was an exclusion diagnosis if any of the previously described criteria were not satisfied (Figure 1 ). In addition, we were able to perform a risk stratification of the syncopal event based on the presence of 4 high-risk features (7) that were included in the multiple-choice questionnaire: 2 minor - no warning symptoms and a family history of SCD at young age; and 2 major – syncope during exertion and palpitations immediately preceding syncope. Participants could specify whether there were contexts or precipitating factors other than the ones listed in the questionnaire in a blank field. This allowed to classify the cases in which head trauma or other diseases were the causes of TLOC. We did not consider TLOC related to other disease when only “fever” was specified in the blank field.
Participants also provided information on whether they were physically active, the type of sports and number of hours practiced per week. There were questions on whether they had already practiced competitive or semi-professional sports, and on the timing and duration of competition. The type of exercise was specified in a box by each participant, thus allowing the classification of sports into 9 levels of static and/or dynamic components, based on previously-described hemodynamics and cardiac adaptations of athletes who compete in each type of exercise. (8) In the case of individuals who specified more than one modality (maximum of three), all were individually considered for the purpose of this analysis.
Finally, we studied potential predictors related to past medical history. Besides from screening for past personal history of palpitations, the questionnaire included information about relevant personal and family history of cardiac disease, as well as specific cardiac diagnosis in a multiple-choice format.
12-lead ECG analysis A 12-lead ECG was performed in supine position using a Mortara ELI 10 Portable Resting ECG machine (Mortara Instrument, Milwaukee, Wisconsin, USA) with a paper speed of 25 mm/s and amplification of 0.1 mV/mm. The heart rate, QRS duration, PR and QT interval were registered using the recorder’s automatic measurement software (VERITAS ECG algorithm, Mortara Instrument). The QT corrected (QTc) interval was also automatically obtained using theFridericia correction, which was preset in the device and was previously validated, as described in a paper published by our group. (5) And the Bazett formula (QTc=QT/square root of RR interval) was calculated based on the automatic QT and RR intervals. The ECG of individuals who reported a TLOC episode were manually revised by 4 authors (MC, DB, JP and RP) and the following features were registered: atrioventricular (AV) block, incomplete and complete right and left bundle branch block (RBBB and LBBB), left (-30º to -90º) and right (>90º) axis deviation, ventricular preexcitation and number of premature ventricular complexes (PVCs). Left ventricular hypertrophy (LVH) was defined according to the presence of either Sokolow-Lyon (S in V1 + R in V5 or V6 ≥35mm) or Cornell criteria (S in V3 + R in aVL ≥20mm in men and ≥28mm in woman) and analysis for associated pathologic ECG findings was performed in the cases that fulfilled the voltage criteria. (9)
Statistical analysis We performed statistical analysis using Stata 13.0 software. Categorical variables were described as numbers of cases and percentages, and continuous variables as means ± standard deviation or medians [interquartile ranges] (as appropriate). We used qui-squared tests to determine whether the presence of clinical and ECG features differed between individuals with US and the remaining causes of TLOC. To express the strength of these relations, we obtained the odds ratios (OR) with 95% confidence intervals (CI) through univariate logistic regression for the outcome US. To compare the distribution of continuous variables between individuals with US and the remaining causes of TLOC, we used Student’s t-test or a non-parametric alternative (Mann-Whitney) where appropriate. Multivariate logistic regression was used for establishing a US predictor model, using variables significantly associated with US on univariate analysis. A p-value < 0.05 was regarded as significant and two-tailed tests were applied.