Background Machine learning-based risk prediction models may outperform traditional statistical models in large datasets with many variables, by identifying both novel predictors and the complex interactions between them. This study compared deep learning extensions of survival analysis models with Cox proportional hazards models for predicting cardiovascular disease (CVD) risk in national health administrative datasets. Methods Using individual person linkage of administrative datasets, we constructed a cohort of all New Zealanders aged 30–74 who interacted with public health services during 2012. After excluding people with prior CVD, we developed sex-specific deep learning and Cox proportional hazards models to estimate the risk of CVD events within 5 years. Models were compared based on the proportion of explained variance, model calibration and discrimination, and hazard ratios for predictor variables. Results First CVD events occurred in 61 927 of 2 164 872 people. Within the reference group, the largest hazard ratios estimated by the deep learning models were for tobacco use in women (2.04, 95% CI: 1.99, 2.10) and chronic obstructive pulmonary disease with acute lower respiratory infection in men (1.56, 95% CI: 1.50, 1.62). Other identified predictors (e.g. hypertension, chest pain, diabetes) aligned with current knowledge about CVD risk factors. Deep learning outperformed Cox proportional hazards models on the basis of proportion of explained variance (R2: 0.468 vs 0.425 in women and 0.383 vs 0.348 in men), calibration and discrimination (all P <0.0001). Conclusions Deep learning extensions of survival analysis models can be applied to large health administrative datasets to derive interpretable CVD risk prediction equations that are more accurate than traditional Cox proportional hazards models.

Maori priMary health care treasuresPounamu (greenstone) is the most precious of stones to Māori. 'ahakoa he iti, he pounamu'(Although it is small, it is valuable) pounamu J PRIM HEALTH CARE 2016;8(1):60-66.

Introduction: Cardiovascular disease remains the leading cause of premature death and disability for all New Zealanders. Māori, the Indigenous people of New Zealand, are disproportionately affected. The New Zealand Māori Health Strategy recognises that "health and wellbeing are influenced and affected by the 'collective' … and the importance of working with people in their social contexts, not just with their physical symptoms" (Ministry of Health, 2002, p. 1). In a Māori worldview, a holistic approach to health is innate. Objectives: This project piloted a kaupapa Māori approach within an existing 12-week clinical exercise and lifestyle management programme. The aims of the study were to determine the effectiveness of a kaupapa Māori 12-week exercise and lifestyle management programme on parameters of cardiac risk and quality of life. Methods: 12 Māori participants attended, 3 times per week over a 12-week period, for monitored, supervised, and individualised exercise. Participants performed a progressive aerobic-only programme for 6 weeks and then a combined aerobic and resistance training programme from weeks 7 through 12. Education sessions were chosen by participants. Results: There was a statistically significant improvement in waist circumference (-3.7 cm; p = .05), hip circumference (-4.6 cm; p = .03), systolic blood pressure (-22 mm Hg; p = .01), and HDL cholesterol (0.22 mmol/L; p = .01). In addition, physical (p = .05) and overall (p = .03) quality of life improved. Conclusion: A kaupapa Māori approach within a structured lifestyle management programme modifies cardiac risk parameters in Māori.

ObjectivesHigh-sensitivity cardiac troponin testing is used in the diagnosis of acute coronary syndromes but its role during convalescence is unknown. We investigated the long-term prognostic significance of serial convalescent high-sensitivity cardiac troponin concentrations following acute coronary syndrome.MethodsIn a prospective multicentre observational cohort study of 2140 patients with acute coronary syndrome, cardiac troponin I concentrations were measured in 1776 patients at 4 and 12 months following the index event. Patients were stratified into three groups according to the troponin concentration at 4 months using the 99th centile (women>16 ng/L, men>34 ng/L) and median concentration of those within the reference range. The primary outcome was cardiovascular death.ResultsTroponin concentrations at 4 months were measurable in 99.0% (1759/1776) of patients (67±12 years, 72% male), and were ≤5 ng/L (median) and >99th centile in 44.8% (795) and 9.3% (166), respectively. There were 202 (11.4%) cardiovascular deaths after a median of 4.8 years. After adjusting for the Global Registry of Acute Coronary Events score, troponin remained an independent predictor of cardiovascular death (HR 1.4, 95% CI 1.3 to 1.5 per doubling) with the highest risk observed in those with increasing concentrations at 12 months. Patients with 4-month troponin concentrations >99th centile were at increased risk of cardiovascular death compared with those ≤5 ng/L (29.5% (49/166) vs 4.3% (34/795); adjusted HR 4.9, 95% CI 3.8 to 23.7).ConclusionsConvalescent cardiac troponin concentrations predict long-term cardiovascular death following acute coronary syndrome. Recognising this risk by monitoring troponin may improve targeting of therapeutic interventions.Trial registration numberACTRN12605000431628;Results.

Introduction Implant rates for cardiac implantable electronic devices (CIED), including permanent pacemakers (PPM) and implantable cardioverter defibrillators (ICD), have increased globally in recent decades. This is the first national study providing a contemporary analysis of national CIED implant trends by sex-specific age groups over an extended period. Methods Patient characteristics and device type were identified for ten years (2009 to 2018) using procedure coding in the National Minimum Datasets, which collects all New Zealand (NZ) public hospital admissions. CIED implant rates represent implants/million population. Results New PPM implant rates increased by 4.6%/year (p<0.001), increasing in all age groups except patients <40 years. Males received 60.1% of new PPM implants, with higher implant rates across all age groups compared to females. The annual increase in age-standardised implant rates was similar for males and females (3.4% vs 3.0%, p=0.4). By 2018 the overall PPM implant rate was 538/million. New ICD implant rates increased by 4.2%/year (p<0.001), increasing in all age groups except patients <40 and [?]80 years. Males received 78.1% of new ICD implants, with higher implant rates across all age groups compared to females. The annual increase in age-standardised implant rates was higher in males compared to females (3.5% vs 0.7%, p<0.001). By 2018 the overall ICD implant rate was 144/million population. Conclusion CIED implant rates have increased steadily in NZ over the past decade but remain low compared to international benchmarks. Males had substantially higher CIED implant rates compared to females, with a growing gender disparity in ICD implant rates.