Multivariate imputation by chained equations (MICE) is commonly used for imputing missing data in epidemiologic research. The “true” imputation model may contain nonlinearities which are not included in default imputation models. Random forest imputation is a machine learning technique which can accommodate nonlinearities and interactions and does not require a particular regression model to be specified. We compared parametric MICE with a random forest-based MICE algorithm in 2 simulation studies. The first study used 1,000 random samples of 2,000 persons drawn from the 10,128 stable angina patients in the CALIBER database (Cardiovascular Disease Research using Linked Bespoke Studies and Electronic Records; 2001–2010) with complete data on all covariates. Variables were artificially made “missing at random,” and the bias and efficiency of parameter estimates obtained using different imputation methods were compared. Both MICE methods produced unbiased estimates of (log) hazard ratios, but random forest was more efficient and produced narrower confidence intervals. The second study used simulated data in which the partially observed variable depended on the fully observed variables in a nonlinear way. Parameter estimates were less biased using random forest MICE, and confidence interval coverage was better. This suggests that random forest imputation may be useful for imputing complex epidemiologic data sets in which some patients have missing data.

SummaryBackgroundInternational research for acute myocardial infarction lacks comparisons of whole health systems. We assessed time trends for care and outcomes in Sweden and the UK.MethodsWe used data from national registries on consecutive patients registered between 2004 and 2010 in all hospitals providing care for acute coronary syndrome in Sweden and the UK. The primary outcome was all-cause mortality 30 days after admission. We compared effectiveness of treatment by indirect casemix standardisation. This study is registered with ClinicalTrials.gov, number NCT01359033.FindingsWe assessed data for 119 786 patients in Sweden and 391 077 in the UK. 30-day mortality was 7·6% (95% CI 7·4–7·7) in Sweden and 10·5% (10·4–10·6) in the UK. Mortality was higher in the UK in clinically relevant subgroups defined by troponin concentration, ST-segment elevation, age, sex, heart rate, systolic blood pressure, diabetes mellitus status, and smoking status. In Sweden, compared with the UK, there was earlier and more extensive uptake of primary percutaneous coronary intervention (59% vs 22%) and more frequent use of β blockers at discharge (89% vs 78%). After casemix standardisation the 30-day mortality ratio for UK versus Sweden was 1·37 (95% CI 1·30–1·45), which corresponds to 11 263 (95% CI 9620–12 827) excess deaths, but did decline over time (from 1·47, 95% CI 1·38–1·58 in 2004 to 1·20, 1·12–1·29 in 2010; p=0·01).InterpretationWe found clinically important differences between countries in acute myocardial infarction care and outcomes. International comparisons research might help to improve health systems and prevent deaths.FundingSeventh Framework Programme for Research, National Institute for Health Research, Wellcome Trust (UK), Swedish Association of Local Authorities and Regions, Swedish Heart-Lung Foundation.

Aims The population with stable coronary artery disease (SCAD) is growing but validated models to guide their clinical management are lacking. We developed and validated prognostic models for all-cause mortality and non-fatal myocardial infarction (MI) or coronary death in SCAD. Methods and results Models were developed in a linked electronic health records cohort of 102 023 SCAD patients from the CALIBER programme, with mean follow-up of 4.4 (SD 2.8) years during which 20 817 deaths and 8856 coronary outcomes were observed. The Kaplan–Meier 5-year risk was 20.6% (95% CI, 20.3, 20.9) for mortality and 9.7% (95% CI, 9.4, 9.9) for non-fatal MI or coronary death. The predictors in the models were age, sex, CAD diagnosis, deprivation, smoking, hypertension, diabetes, lipids, heart failure, peripheral arterial disease, atrial fibrillation, stroke, chronic kidney disease, chronic pulmonary disease, liver disease, cancer, depression, anxiety, heart rate, creatinine, white cell count, and haemoglobin. The models had good calibration and discrimination in internal (external) validation with C-index 0.811 (0.735) for all-cause mortality and 0.778 (0.718) for non-fatal MI or coronary death. Using these models to identify patients at high risk (defined by guidelines as 3% annual mortality) and support a management decision associated with hazard ratio 0.8 could save an additional 13-16 life years or 15-18 coronary event-free years per 1000 patients screened, compared with models with just age, sex, and deprivation. Conclusion These validated prognostic models could be used in clinical practice to support risk stratification as recommended in clinical guidelines.

BackgroundNeutrophil counts are a ubiquitous measure of inflammation, but previous studies on their association with cardiovascular disease (CVD) were limited by small numbers of patients or a narrow range of endpoints.ObjectivesThis study investigated associations of clinically recorded neutrophil counts with initial presentation for a range of CVDs.MethodsWe used linked primary care, hospitalization, disease registry, and mortality data in England. We included people 30 years or older with complete blood counts performed in usual clinical care and no history of CVD. We used Cox models to estimate cause-specific hazard ratios (HRs) for 12 CVDs, adjusted for cardiovascular risk factors and acute conditions affecting neutrophil counts (such as infections and cancer).ResultsAmong 775,231 individuals in the cohort, 154,179 had complete blood counts performed under acute conditions and 621,052 when they were stable. Over a median 3.8 years of follow-up, 55,004 individuals developed CVD. Adjusted HRs comparing neutrophil counts 6 to 7 versus 2 to 3 × 109/l (both within the ‘normal’ range) showed strong associations with heart failure (HR: 2.04; 95% confidence interval [CI]: 1.82 to 2.29), peripheral arterial disease (HR: 1.95; 95% CI: 1.72 to 2.21), unheralded coronary death (HR: 1.78; 95% CI: 1.51 to 2.10), abdominal aortic aneurysm (HR: 1.72; 95% CI: 1.34 to 2.21), and nonfatal myocardial infarction (HR: 1.58; 95% CI: 1.42 to 1.76). These associations were linear, with greater risk even among individuals with neutrophil counts of 3 to 4 versus 2 to 3 × 109/l. There was a weak association with ischemic stroke (HR: 1.36; 95% CI: 1.17 to 1.57), but no association with stable angina or intracerebral hemorrhage.ConclusionsNeutrophil counts were strongly associated with the incidence of some CVDs, but not others, even within the normal range, consistent with underlying disease mechanisms differing across CVDs. (White Blood Cell Counts and Onset of Cardiovascular Diseases: a CALIBER Study [CALIBER]; NCT02014610)