A sudden death risk score specifically for hypertension: based on 25 648 individual patient data from six randomized controlled trials.

Le Hai-Ha, Subtil Fabien, Cerou Marc, Marchant Ivanny, Al-Gobari Muaamar, Fall Mor, Mimouni Yanis, Kassaï Behrouz, Lindholm Lars, Thijs Lutgarde, Gueyffier François

aLaboratoire de Biologie et Biométrie Evolutive, UMR 5558, Université Claude Bernard Lyon 1, Villeurbanne, France bDepartamento de Preclínicas, Escuela de Medicina, Universidad de Valparaíso, Valparaíso, Chile cInstitute of Social and Preventive Medicine, Cochrane, Lausanne University Hospital, Lausanne, Switzerland dLaboratoire de Pharmacologie and de Pharmacodynamie, Université Cheikh Anta Diop, Dakar, Sénégal eCentre d'Investigation Clinique, EPICIME (Epidémiologie, Pharmacologie, Investigation Clinique, Information Médicale, Mère-Enfant), Hospices Civils de Lyon/INSERM/UCB Lyon1/UMR5558 fService de Pharmacotoxicologie, EPICIME-CIC 1407 INSERM, Hospices Civils de Lyon, Lyon, France gDepartment of Public Health and Clinical Medicine, Målpunkt X Norrlands Universitetssjukhus 206, Umeå University, Umeå, Sweden hHypertension and Cardiovascular Epidemiology, Leuven, Belgium iService de Pharmacotoxicologie, Unité de Pharmacologie Clinique et essais thérapeutiques, Hospices Civils de Lyon, Lyon, France.

J Hypertens. 2017 Nov;35(11):2178-2184. doi: 10.1097/HJH.0000000000001451.

OBJECTIVE

To construct a sudden death risk score specifically for hypertension (HYSUD) patients with or without cardiovascular history.

METHODS

Data were collected from six randomized controlled trials of antihypertensive treatments with 8044 women and 17 604 men differing in age ranges and blood pressure eligibility criteria. In total, 345 sudden deaths (1.35%) occurred during a mean follow-up of 5.16 years. Risk factors of sudden death were examined using a multivariable Cox proportional hazards model adjusted on trials. The model was transformed to an integer system, with points added for each factor according to its association with sudden death risk.

RESULTS

Antihypertensive treatment was not associated with a reduction of the sudden death risk and had no interaction with other factors, allowing model development on both treatment and placebo groups. A risk score of sudden death in 5 years was built with seven significant risk factors: age, sex, SBP, serum total cholesterol, cigarette smoking, diabetes, and history of myocardial infarction. In terms of discrimination performance, HYSUD model was adequate with areas under the receiver operating characteristic curve of 77.74% (confidence interval 95%, 74.13-81.35) for the derivation set, of 77.46% (74.09-80.83) for the validation set, and of 79.17% (75.94-82.40) for the whole population.

CONCLUSION

Our work provides a simple risk-scoring system for sudden death prediction in hypertension, using individual data from six randomized controlled trials of antihypertensive treatments. HYSUD score could help assessing a hypertensive individual's risk of sudden death and optimizing preventive therapeutic strategies for these patients.