Global Burden of Disease Ecological Data Analysis of Noncommunicable Disease Early Deaths in the World's Highest Animal Food Consuming 1 Billion People to Derive a Multiple Regression Risk Factor Formula with Coefficients Equated to Population Attributable Risk Percents

Abstract

Background: Population attributable risk percents for multiple risk factors for the same health outcome have been very difficult to derive in ecological cohort studies or other observational studies.

Methods: In this ecological cohort study with Institute of Health Metrics and Evaluation (IHME) Global Burden of Disease Study (GBD) data, we correlated noncommunicable disease deaths /100k /year in males and females 15-69 years old from 1990-2017 (NCD) with dietary and other risk factors. We used 7846 population-weighted cohorts from 195 countries worldwide, each comprising about 1 million people, representing about 7.8 billion people. The goal was to derive a multiple regression formula with NCD (dependent variable) and dietary and other risk factors (independent variables) with coefficients equated to their population attributable risk percents (PAR%s).

Findings: To facilitate the analysis, we grouped seven animal food risk factors (all in kilocalories/day (kcal/day)): animal foods 7=processed meat +red meat +fish +milk +poultry +eggs +added saturated fatty acids (added SFA)). In a subset of 500 male/female pairs (n=1000 cohorts, representing about 1 billion people with the world’s highest mean animal food consumption: animal foods 7≥467 kcal/day), we derived a multiple regression formula with risk factor coefficients (“+” signs increased NCD risk and “-“ signs decreased NCD risk) equated to their PAR%s. High animal foods NCD risk factor formula=0.68*Processed meat +5.30*Red meat -1.22*Fish -0.73*Milk +2.54* Poultry -0.91*Eggs +2.18*Added SFA +1.44*Added PUFA -0.91*Added TFA + 4.90*Alcohol +5.62*Sugary beverages -0.95*Potatoes -1.00*Sweet potatoes +0.18* Corn -1.36*Fruit -2.79*Vegetables -1.81*Nuts and seeds -1.99*Whole grains -0.54* Legumes -0.91*Rice +11.69*Physical inactivity +1.18*Stop breast feeding<6 months +2.54*Ambient air pollution + 1.09*Smoking prevalence (0-1) +0.73*Sublingual tobacco +6.71*BMI kg/M 2 +2.54*Fasting Plasma Glucose + 8.34*Systolic Blood Pressure -11.69*Sex (male=1 and female=2) Sum of PAR% of all risk factors=84.46%, mean NCD=1055 deaths/100k /year, n=1000 cohorts. The worldwide mean animal foods 7=242.0 kcal/day, mean NCD=1428 death/100k/year, n=7846 cohorts.

InterpretationUsing risk factor and NCD data in a high animal food consuming GBD subset, we derived an NCD versus 29 risk factors formula with risk factor coefficients equated to their PAR%s. These risk factor PAR%s for NCD were specific to the high animal food subset of mostly highly developed countries. NCD risk factor PAR%s worldwide and in other population subsets would be very different. Multiple regression analysis of GBD data offers a potentially useful tool to probe large population health databases. These findings may have implications for clinical practice and public health policies.FundingNone for the analysis. The Bill and Melinda Gates Foundation funded the acquisition and provision of the GBD data.

Funding Information: This research received no grant from any funding agency in the public, commercial or not-for-profit sectors. The Bill and Melinda Gates Foundation funded the acquisition of the data by the IHME for this analysis.

Declaration of Interests: None reported.

Keywords: Global health, public health, noncommunicable diseases, dietary risk factors, multiple regression mathematical modeling