The Magnitude of Black/Hispanic Disparity in COVID-19 Mortality Across United States Counties During the First Waves of the COVID-19 Pandemic.

Objectives: To quantify the Black/Hispanic disparity in COVID-19 mortality in the United States (US). Methods: COVID-19 deaths in all US counties nationwide were analyzed to estimate COVID-19 mortality rate ratios by county-level proportions of Black/Hispanic residents, using mixed-effects Poisson regression. Excess COVID-19 mortality counts, relative to predicted under a counterfactual scenario of no racial/ethnic disparity gradient, were estimated. Results: County-level COVID-19 mortality rates increased monotonically with county-level proportions of Black and Hispanic residents, up to 5.4-fold (≥43% Black) and 11.6-fold (≥55% Hispanic) higher compared to counties with <5% Black and <15% Hispanic residents, respectively, controlling for county-level poverty, age, and urbanization level. Had this disparity gradient not existed, the US COVID-19 death count would have been 92.1% lower (177,672 fewer deaths), making the rate comparable to other high-income countries with substantially lower COVID-19 death counts. Conclusion: During the first 8 months of the SARS-CoV-2 pandemic, the US experienced the highest number of COVID-19 deaths. This COVID-19 mortality burden is strongly associated with county-level racial/ethnic diversity, explaining most US COVID-19 deaths.

Corrigendum: The Magnitude of Black/Hispanic Disparity in COVID-19 Mortality Across United States Counties During the First Waves of the COVID-19 Pandemic.

[This corrects the article DOI: 10.3389/ijph.2021.1604004.].

A new look at risk patterns related to coronary heart disease incidence using survival tree analysis: 12 Years Longitudinal Study.

We identified risk patterns associated with incident coronary heart disease (CHD) using survival tree, and compared performance of survival tree versus Cox proportional hazards (Cox PH) in a cohort of Iranian adults. Data on 8,279 participants (3,741 men) aged ≥30 yr were used to analysis. Survival trees identified seven subgroups with different risk patterns using four [(age, non-HDL-C, fasting plasma glucose (FPG) and family history of diabetes] and five [(age, systolic blood pressure (SBP), non-HDL-C, FPG and family history of CVD] predictors in women and men, respectively. Additional risk factors were identified by Cox models which included: family history of CVD and waist circumference (in both genders); hip circumference, former smoking and using aspirin among men; diastolic blood pressure and lipid lowering drug among women. Survival trees and multivariate Cox models yielded comparable performance, as measured by integrated Brier score (IBS) and Harrell's C-index on validation datasets; however, survival trees produced more parsimonious models with a minimum number of well recognized risk factors of CHD incidence, and identified important interactions between these factors which have important implications for intervention programs and improve clinical decision making.

Risk Factors for Incidence of Cardiovascular Diseases and All-Cause Mortality in a Middle Eastern Population over a Decade Follow-up: Tehran Lipid and Glucose Study.

BACKGROUND: To examine the association between potentially modifiable risk factors with cardiovascular disease (CVD) and all-cause mortality and to quantify their population attributable fractions (PAFs) among a sample of Tehran residents. METHODS: Overall, 8108 participants (3686 men) aged≥30 years, were investigated. To examine the association between risk factors and outcomes, multivariate sex-adjusted Cox proportional hazard regression analysis were conducted, using age as time-scale in two models including general/central adiposity: 1)adjusted for different independent variables including smoking, education, family history of CVD and sex for both outcomes and additionally adjusted for prevalent CVD for all-cause mortality 2)further adjusted for obesity mediators (hypertension, diabetes, lipid profile and chronic kidney disease). Separate models were used including either general or central adiposity. RESULTS: During median follow-up of >10 years, 827 first CVD events and 551 deaths occurred. Both being overweight (hazard ratio (HR), 95%CI: 1.41, 1.18-1.66, PAF 13.66) and obese (1.51, 1.24-1.84, PAF 9.79) played significant roles for incident CVD in the absence of obesity mediators. Predicting CVD, in the presence of general adiposity and its mediators, significant positive associations were found for hypercholesterolemia (1.59, 1.36-1.85, PAF 16.69), low HDL-C (1.21, 1.03-1.41, PAF 12.32), diabetes (1.86, 1.57-2.27, PAF 13.87), hypertension (1.79, 1.46-2.19, PAF 21.62) and current smoking (1.61, 1.34-1.94, PAF 7.57). Central adiposity remained a significant positive predictor, even after controlling for mediators (1.17, 1.01-1.35, PAF 7.55). For all-cause mortality, general/central obesity did not have any risk even in the absence of obesity mediators. Predictors including diabetes (2.56, 2.08-3.16, PAF 24.37), hypertension (1.43, 1.11-1.84, PAF 17.13), current smoking (1.75, 1.38-2.22, PAF 7.71), and low education level (1.59, 1.01-2.51, PAF 27.08) were associated with higher risk, however, hypertriglyceridemia (0.83, 0.68-1.01) and being overweight (0.71, 0.58-0.87) were associated with lower risk. CONCLUSIONS: Modifiable risk factors account for more than 70% risk for both CVD and mortality events.

Different Combinations of Glucose Tolerance and Blood Pressure Status and Incident Diabetes, Hypertension, and Chronic Kidney Disease.

BACKGROUND: The impact of different combinations of glucose tolerance and blood pressure status on the development of type 2 diabetes mellitus (T2DM), hypertension (HTN), and chronic kidney disease (CKD) still needs to be investigated. METHODS AND RESULTS: A total of 12 808 Iranian adults aged ≥20 years were included in 3 separate analyses to investigate incidence of T2DM, HTN, and CKD. Multivariate Cox proportional hazard models were used to calculate hazard ratios (95% CI). During a median follow-up of >10 years, the overall incidence rate for T2DM, HTN, and CKD was 12.2, 29.8, and 24.8 per 1000 person-years. For incident T2DM, considering normal glucose tolerance/normal blood pressure as reference, prediabetes (PreDM)/HTN had the highest risk (hazard ratio: 7.22 [5.71-9.12]) while PreDM/normal blood pressure also showed a significant risk (5.58 [4.41-7.05]). Furthermore, risk of PreDM/HTN was higher than PreDM/normal blood pressure (P<0.05). For incident HTN, normal glucose tolerance/prehypertension was a strong predictor (3.28 [2.91-3.69]); however, addition of PreDM or T2DM did not increase the risk. For incident CKD, every category that included HTN and/or T2DM showed significant risk; this risk was marginally significant for the PreDM/HTN group (1.19 [0.98-1.43], P=0.06). In addition, PreDM/ normal blood pressure was a marginally significant risk factor for incident HTN while normal glucose tolerance/prehypertension was a significant predictor of T2DM. CONCLUSIONS: Presence of HTN was associated with increased risk of T2DM among the PreDM population; however, dysglycemia did not increase the risk of HTN among individuals with prehypertension. For incident CKD, intensive management of HTN and T2DM, rather than their predisease states, should be considered.

A tutorial on variable selection for clinical prediction models: feature selection methods in data mining could improve the results.

OBJECTIVES: Identifying an appropriate set of predictors for the outcome of interest is a major challenge in clinical prediction research. The aim of this study was to show the application of some variable selection methods, usually used in data mining, for an epidemiological study. We introduce here a systematic approach. STUDY DESIGN AND SETTING: The P-value-based method, usually used in epidemiological studies, and several filter and wrapper methods were implemented to select the predictors of diabetes among 55 variables in 803 prediabetic females, aged ≥ 20 years, followed for 10-12 years. To develop a logistic model, variables were selected from a train data set and evaluated on the test data set. The measures of Akaike information criterion (AIC) and area under the curve (AUC) were used as performance criteria. We also implemented a full model with all 55 variables. RESULTS: We found that the worst and the best models were the full model and models based on the wrappers, respectively. Among filter methods, symmetrical uncertainty gave both the best AUC and AIC. CONCLUSION: Our experiment showed that the variable selection methods used in data mining could improve the performance of clinical prediction models. An R program was developed to make these methods more feasible and visualize the results.