Supporting Information

Content list:
Detailed Methods:
Description of study patients (including Table S1)
Medical interventions
Primary End-point
Ascertainment of cause of death
Clinical measures collected

Ultrasound data
Other clinical data
Description of ECG Risk-score
(including Table S2 Calculation of ECG Risk-score)
Detailed Statistics
Supplementary Results
Supplementary Figures S1 and S2

Detailed Methods
Study patients Patients attending all Swedish regional pediatric cardiology centers with a
diagnosis of hypertrophic cardiomyopathy (HCM) presenting <19 years of age since 1972
were identified retrospectively in 1990-1999 using the diagnostic registry of each center,
and. original patient records were reviewed.1, 2 From 1999 cases were collected
prospectively from all Swedish pediatric cardiology centers. Included patients had HCM as
defined by a maximal wall thickness >+2SD using Detroit-score (using Cardio Z-app),3 and a
diastolic septum-to-cavity ratio (sepcavr) or diastolic posterior left ventricular wall-to-cavity
ratio (lvcavr) >99th percentile for age,4 with a non-dilated ventricle with normal or increased
systolic contractility in the absence of another cardiac or systemic disease.5 HCM-
phenocopies secondary to maternal gestational diabetes, endocrine disorders, Friedreich´s
ataxia, mitochondrial disease or storage-disorders such as Danon´s disease were not
included. Included in this analysis are patients diagnosed up to December 2016, with at least
five years of follow-up in all survivors (except one Noonan syndrome patient not included in
Kaplan-Meier analysis), but patients who presented with sudden cardiac death (SCD) as first
disease manifestation (n=12), or with no in-vivo cardiac wall thickness measurements are
excluded. In order to be able to also evaluate both the HCMRisk-Kids, 2011and 2020AHA
criteria, and the ESC2014 criteria, we required in vivo cardiac wall thickness measurements,
and case-notes were surveyed for details of family history and unexplained syncopal
episodes. 151 patients in the national cohort, follow-up mean (SD) 13.3y (9.3), had all the
necessary information, out of whom there were 33 patients with sudden death or
resuscitated malignant arrhythmia (21 sudden deaths, six resuscitated cardiac arrests and
six appropriate ICD-interventions), referred to as SCD/CA-group. Forty-one patients (27%)
had stigmata of Noonan, Leopard or Cranio-Facial-Cutaneous syndrome and are grouped as

1
RASopathy-HCM (RAS-HCM), mean follow-up 12.7y (10.6). There were 110 patients without
syndrome association (ns-HCM), mean follow-up 13.4y (8.5), 72 familial HCM, and 39 cases
without family history.

Table S1 Base-line clinical characteristics of national cohort, with comparisons

between
Non-syndrome HCM and RASopathy associated HCM
(median [IQR], or percent of total)

Non-syndrome HCM RASopathy HCM p-value

N=111 N=41

Age at diagnosis, years 10.9 [4.2-14.9] 0.2 [0.1-1.2] <0.00001

Male sex 65% 61% n.s
Duration of follow-up (median) 11.6 [8.0-15.9] 13.6 [2.9-18.2] n.s.
FH of HCM 68% 20% <0.0001
FH of SCD 34% 8% 0.0023
Unexplained syncope 14% 11% n.s.
NSVT early follow-up 11% 4% n.s.
NSVT anytime 22% 24% n.s.
Max wall thickness (mm) at D 15 [12-22] 10.5 [8.0-16.0] 0.00002
Detroit Z-score (MWth) 4.3 [2.6-5.8] 4.6 [3.3-5.8] n.s.
Boston wt only Z-score (MWT) 9.6 [5.9-16.5] 11.6 [5.9-18.3]* n.s.
LA:ao ratio 1.3 [1.2-1.5] 1.4 [1.1-1.7] n.s.
LVOTO at rest 36% 73% <0.0001
HCM-Risk-Kids at diagnosis 4.8 [3.2-7.4]
HCM-Risk-Kids age 7plus 4.7 [3.0-8.2] 5.7 [3.3-8.0] n.s.
ECG Risk-score at diagnosis 3 [2-8] 5 [3-9] 0.034
ECG Risk-score age 7plus 4 [2-8] 6 [3-8] 0.08
Age at SCD 15.9 [11.8-23.0] 14.4 [12.3-42] n.s.
Early medical management
Beta-blocker 68% 53% 0.08
Early prop eqv mg/kg in treated 6.0 [3.4-9.2] 5.5 [3.0-8.0] n.s.
Calcium-channel blocker 5% 22% 0.006
No therapy 15% 10% n.s.
Addition of disopyramide 30% 24% n.s.
Addition of amiodarone 2% 5% n.s.
Subsequent myectomy 10% 24% 0.028
Beta-bloc at latest FU, treated 5.5 [3.4-9,2] 6.2 [2.9-13.3] n.s.

Abbreviations: p-value refers to Fishers exact test for proportions, and Mann-Whitney U-test
for measures; FH = Family history; NSVT = non-sustained ventricular tachycardia; at D =at
diagnosis; MWth = maximal wall thickness; wt = weight; * = Z-score at same time as HCM-
Risk-Kids7plus; LA:ao ratio = ratio between left atrial and aortic root diameter on long-axis
M-mode measurement; LVOTO = ventricular outflow-tract obstruction; prop eqv = beta-
blocker dose measured in propranolol equivalents; FU=follow-up

The cohort was predominantly Caucasian in ethnicity (89% Northern European Caucasian,
6.6% Mediterranean Caucasian, 1.3% South American Caucasian, 1.3% Asian and 0.7%

2
African). From 2005 onwards patients with presentation of HCM in childhood were routinely
screened for causative mutations, and in the West Götaland Region there was an active out-
reach family screening programme. Among patients with non-syndrome associated
childhood presentation in West Götaland region 40% had an identified MYBPC3 mutation,
and 30% an MYH7 mutation (nearly double the prevalence in cases with adult presentation
in the same region). There were further cases with ACTC-mutation (5%), MYPN-mutation
(2.6%) and MYL2 mutation (2.6). A further 10.6% had an identified VUS who could not be
further categorised because of insufficient size of family to assess if the VUS segregated
with disease.
Comparing this geographical cohort with the HCM-Risk-Kids cohort they have virtually
identical age at diagnosis, but slightly less marked hypertrophy, and among non-syndromal
HCM-cases, a higher proportion of familial HCM, a higher proportion with family history of
SCD and probably a higher proportion with significant outflow-tract obstruction at rest
(details in Table 1 in main manuscript).
Medical interventions Medical treatments given, and interventional procedures have been
documented. Among ns-HCM patients 68% received early beta-blocker therapy (defined as
beta-blocker dose reached within six months of diagnosis; mostly propranolol or metoprolol,
and only 2% received atenolol). Beta-blocker dose being taken at latest follow-up was also
recorded, at which point 80% received beta-blockers. Other early therapy: 30% received
disopyramide combined with beta-blocker, 5% had calcium-blocker therapy, and 2%
amiodarone. All patients receiving high-dose propranolol or metoprolol received slow-release
preparations twice daily. For most of the period beta-blockers were the first choice for
therapy for symptomatic patients, but in a period from approximately 1980 and most of that
decade most pediatricians instead used verapamil as first choice, switching back to
propranolol as first choice around 1990. From about year 2000 a greater proportion of
patients were treated with beta-blockers, and a higher target-dose was aimed for, and the
addition of disopyramide when control of gradient was inadequate was introduced. This
resulted in a lower need for surgical myectomy in this later period. In order to be able to
assess confounding effects of beta-blocker dose all doses were converted to equivalents in
propranolol dose, since that was the most commonly used beta-blocker, according to our
previously published conversion: propranolol 80 mg = metoprolol 100mg6 = bisoprolol 5 mg
= atenolol 50 mg.
Primary end-point Sudden cardiac death/survived cardiac arrest/appropriate ICD-
intervention (SD/CA). A larger proportion of these deaths occurred in the part of the cohort
diagnosed before 2000, when more patients were diagnosed because of symptoms, and
fewer through family screening, and beta-blocker therapy if given was in low doses.

Causes of death All Swedes have a unique personal identification number, and vital status
was last ascertained in May 2021. Where causes of death were not recorded in hospital
notes, they were obtained from death certificates from National Board of Health and Welfare,
and validated by post mortem and histology from forensic pathology departments.2 Apart
from the 12 presenting with SCD and the 33 classed as SCD/CA-group, there were 13
additional cardiac deaths, one among ns-HCM, 12 among RAS-HCM. No patient was lost to
follow-up.

3
Clinical measures collected

Ultra-sound measurements
Original ultrasound measurements, M-mode, and where available, 2-D ultrasound
measurements were re-measured where available (IÖS and EF), or otherwise reports in
hospital notes were accepted. Detroit Z-scores for maximal wall thickness have been
calculated using the Cardio Z app.3 However we found that the choice of Z-score had very
great influence on values, and whereas we got a normal distribution of Z-score values within
±2 on our normal patients with the Detroit Z-score, we obtained very much higher Z-score
values with the Kampmann Z-score7, with 19% of normal patients having Z-score values >2.
Accordingly, only Detroit Z-score values are used in our study. However, as the live
calculation sheet for HCM-Risk-Kids also displayed the body weight-based Z-score8 used in
the HCM-Risk-Kids algorithm we have also documented those Z-scores in Table 1 in the
main manuscript in order to be able to carry out direct comparisons with the HCM-Risk-Kids
tertiary center-cohort. An LVOT gradient >20 mm Hg at rest was considered significant
obstruction (LVOTO).9 Presence of obvious systolic anterior movement of mitral-valve
apparatus (SAM) plus a systolic murmur was judged evidence of LVOTO in era before
Doppler measurements available.

Other clinical data: 24h-Holter ECG-recordings were not available in the 1970-ies, and
during that period recorded triplets or longer VTs on clinical monitoring, resting ECGs or at
the time of exercise-testing was recorded as equivalent to nsVT on Holter-recordings.
Exercise-testing on a bicycle ergometer has been available from the outset, but blood
pressure response was not always recorded. Hospital records were scrutinized for
information about syncope and family history of sudden death.

ECG risk-score: In all HCM-patients ECG-voltages and other ECG-measures were re-
measured in a standardized fashion at entry to the national registry by the registry-holder
(IÖS),1 or by EF. These measurements were done retrospectively 1990-1999, and
prospectively from 2000, as part of studies originally focusing on treatment effects.1, 10 Thus
bias was not possible as the ECG risk-score was not published until 2010,11 and 87% of
survivor ECG-measurements, and 84% of SD/CA-ECG-measurements were entered into the
registry before 2010. ECG risk-score was later calculated from entered measurements as
first described.11

4
Table S2. Calculation of the ECG risk-score
Morphological features
Deviation of QRS-axis present Yes 1 point
Pathological T-wave-inversion ≥1 mm present:
In limb-lead Yes 1 point
In precordial lead Yes 2 points
In both limb-lead and precordial lead Yes 2 points
ST-segment depression ≥ 2 mm present Yes 2 points
S-wave greater than R-wave in lead V4 Yes 2 points
ECG measurements
Six limb-lead QRS-amplitude sum in mV ≥7.7 mV 1 point
≥10.0 mV 2 points
≥12.0 mV 3 points
12-lead QRS-amplitude x duration product mV.sec ≥2.2 mV.sec 1 point
≥2.5 mV.sec 2 points
≥3.0 mV.sec 3 points
QTc (Bazetts formula) ≥440 msec 1 point
Maximal total score 14 points

Technical points: QRS-axis deviation is judged compared to normal range for age, both
right- or left deviation gives a point. With a “circular” axis= R and S-waves similar size in all
limb-leads, the initial vector i.e. biggest R-waves are used to indicate QRS-axis.
T-wave inversion in lead III only is not counted as pathological
T-wave –inversion in V1 and V2 is accepted, but not in V4
Clearly pathological T-wave progression, for example upright T in V1, but clearly inverted T
in V2 or V3, and then upright T in V4 again counts as pathological precordial T-wave.
QRS-amplitude is measured as R-wave plus Q-wave or S-wave whichever is the deepest,
i.e if ruler used, from top to bottom of complex.

Table is modified from Östman-Smith et al. Eur Heart J 2010; 31:439-49.11 The score can
be used both with pediatric precordial lead positioning: V4R, V1,V2,V4,V5,V6, and for adult
lead positioning: V1,V2,V3,V4,V5,V6. Age-appropriate lead-positioning should be used.
From digital ECGs it is recommended that both QRS-axis, and QTc are reviewed, and
calculated also manually, if border-line.

Statistics

Statistical analyses were carried out using Statgraphics Centurion XVI, GraphPad Prism v.
6.02, and IBM SPSS Statistics v. 22 software. Imputed values were not inserted for missing
data. Categorical data were compared with two-tailed Fisher´s exact test, and continuous

5
variables with Mann-Witney U-test as the majority of parameters were not normally
distributed. Univariate and multivariate Cox-hazard analysis to assess risk factors was carried
out for measures in Table 2. Only the ns-HCM group had sufficient number of end-points for
multi-variate analysis. For the uni-variate analysis of risk-factors with Cox proportional
hazards method all ns-HCM-patients with ECG- and ultrasound data (n=110) were included,
even those who had presented with a re-suscitated cardiac arrest and those had died another
type of cardiac death. Sudden death/cardiac arrest/appropriate ICD discharge was still used
as primary end-point, in order to study how specific various risk factors were for arrhythmia-
associated death. In order to ensure direct comparability with HCMRisk-Kids the multivariate
analysis excluded the subjects that had presented with a cardiac arrest, so n=107. Multi-
variate analysis was started with the two most significant at univariate level, and
subsequently all other variables were added into the model one at a time, and kept or
discarded depending on if it was significant in the multi-variate model. The number of
variables was initially restricted to three at a time in order to have adequate statistical power,
based on statistician advice, and were analyzed with backward selection. O’Mahony et al,
201412 have suggested that one risk-factor per 10 end-points would be appropriate for risk-
factor assessment in adult HCM, but they had an annual event rate of 0.81% whereas we
had an event rate of 2.0% during first five years of follow-up giving us substantially greater
statistical power, and explaining divergence of statistical advice. The two types of ECG Risk-
score and the two types of HCMRisk-Kids score were tested against each other to assess
which was going into the multi-variate model. Once we had a three variable multi-variate
model, we tested to add possible additional confounders one by one, and somewhat to our
surprise ended up with a five variable multi-variate model, four highly statistically significant
factors. A fifth possibly significant confounder, disopyramide p=0.072, was not included in the
multi-variate model as there is some linkage with use of particularly high doses of beta-
blockers. Variables were analyzed with backward Wald selection.

Survival was analyzed by Kaplan-Meier survival curves as freedom from SCD/CA compared
with log-rank test. The five-year proportion with SCD/CA was calculated with Kaplan-Meier
survival analysis. Comparisons of predictive power were carried out using the C-statistic
from ROC-curves for various risk factors individually. Significance of difference between
ROC-curves in ns-HCM and in RAS-HCM was calculated with formula on
http://vassarstats.net/roc_comp.html. The cut-offs compared were those used in previous
publications, and in the case of the sum of ECG Risk-score and HCMRisk-Kids score we
used the Youden index to select a cut-off to maximize specificity without losing much
sensitivity. Sensitivity, specificity, positive and negative predictive values were calculated
using only those survivors that had reached at least five years of follow-up, i.e. those
censored alive before then were not included in the calculation, as we did not want to impute
an outcome. Similarly 10y-positive predictive value only included survivors with at least 10yr
of follow-up.

6
Supplemental Results

Figure S1

Figure S1 legend: Kaplan-Meier survival analysis of freedom from SCD/CA in ns-HCM

patients presenting with left ventricular outflow-tract obstruction at rest, stratified for
maintenance beta-blocker dose. Forty-seven percent of this patient sub-group received ≥4.5
mg/kg/day of propranolol equivalents, chosen as divider from earlier multi-centre study.1

7
Figure S2 illustrates the actual ROC-curves for firstly HCM-Risk-Kids, secondly the sum of
HCM-Risk-Kids and the ECG Risk-score, and thirdly for the sum of HCM-Risk-Kids score
closest to 7th birthday or at diagnosis if later (HCM-Risk-Kids7pl), and the ECG Risk-score
closest to 7th birthday or at diagnosis if later (ECG Risk-score7pl).

Figure legend Figure S2

Area under the curve (AUC) of ROC-curves as continuous functions compared in 92 patients
where all three measurements available. HCMRisk-Kids score (pale-blue, AUC=0.83
[95%CI 0.70-0.96], HCMRisk-Kids score plus ECG Risk-score as continuous function (red,
AUC=0.92 [0.86-0.99], and HCMRisk-Kids7pl score plus ECG Risk-score7pl (green,
AUC=0.92 [0.86-0.98].

SUPPLEMENT METHODS REFERENCES

1. Östman-Smith I, Wettrell G and Riesenfeld T. A cohort study of childhood hypertrophic

cardiomyopathy: improved survival following high-dose beta-adrenoceptor antagonist
treatment. J Am Coll Cardiol. 1999;34:1813-22.
2. Östman-Smith I, Wettrell G, Keeton B, Holmgren D, Ergander U, Gould S, et al. Age- and
gender-specific mortality rates in childhood hypertrophic cardiomyopathy. Eur Heart J.
2008;29:1160-7.
3. Pettersen MD, Du W, Skeens ME and Humes RA. Regression equations for calculation of
z scores of cardiac structures in a large cohort of healthy infants, children, and adolescents:
an echocardiographic study. J Am Soc Echocardiogr. 2008;21:922-34.

8
4. Östman-Smith I and Devlin AM. A simple method for assessing the regression or
progression of ventricular hypertrophy in the growing child and adult: the value of left
ventricular wall-to-cavity ratios. Eur J Echocardiography. 2001;2:22-30.
5. Gersh BJ, Maron BJ, Bonow RO, Dearani JA, Fifer MA, Link MS,et al. 2011 ACCF/AHA
Guideline for the Diagnosis and Treatment of Hypertrophic Cardiomyopathy: a report of the
American College of Cardiology Foundation/American Heart Association Task Force on
Practice Guidelines. J Am Coll Cardiol. 2011;58:e212-60.
6. van Herwaarden CL, Binkhorst RA, Fennis JF and van't Laar A. Effects of propranolol and
metoprolol on haemodynamic and respiratory indices and on perceived exertion during
exercise in hypertensive patients. Br Heart J. 1979;41:99-105.
7. Kampmann C, Wiethoff CM, Wenzel A, Stolz G, Betancor M, Wippermann CF, et al.
Normal values of M mode echocardiographic measurements of more than 2000 healthy
infants and children in central Europe. Heart. 2000;83:667-72.
8. Norrish G, Ding T, Field E, Ziolkowska L, Olivotto I, Limongelli G, et al. Development of a
Novel Risk Prediction Model for Sudden Cardiac Death in Childhood Hypertrophic
Cardiomyopathy (HCM Risk-Kids). JAMA cardiology. 2019.
9. Decker JA, Rossano JW, Smith EO, Cannon B, Clunie SK, Gates C, et al. Risk factors
and mode of death in isolated hypertrophic cardiomyopathy in children. J Am Coll Cardiol.
2009;54:250-4.
10. Östman-Smith I, Wettrell G, Keeton B, Riesenfeld T, Holmgren D and Ergander U.
Echocardiographic and electrocardiographic identification of those children with hypertrophic
cardiomyopathy who should be considered at high-risk of dying suddenly. Cardiol Young.
2005;15:632-42.
11. Östman-Smith I, Wisten A, Nylander E, Bratt EL, de-Wahl Granelli A,et al. Oulhaj A
Electrocardiographic amplitudes: a new risk factor for sudden death in hypertrophic
cardiomyopathy. Eur Heart J. 2010;31:439-49.
12. O'Mahony C, Jichi F, Pavlou M, Monserrat L, Anastasakis A, Rapezzi C,et al. A novel
clinical risk prediction model for sudden cardiac death in hypertrophic cardiomyopathy (HCM
risk-SCD). Eur Heart J. 2014;35:2010-20.