RESEARCH DESIGN AND METHODS: The prevalence of diabetes, defined as self-reported or fasting glycemia ≥7 mmol/L, was documented in 119,666 adults from three high-income (HIC), seven upper-middle-income (UMIC), four lower-middle-income (LMIC), and four low-income (LIC) countries. Relationships between diabetes and its risk factors within these country groupings were assessed using multivariable analyses.

RESULTS: Age- and sex-adjusted diabetes prevalences were highest in the poorer countries and lowest in the wealthiest countries (LIC 12.3%, UMIC 11.1%, LMIC 8.7%, and HIC 6.6%; P < 0.0001). In the overall population, diabetes risk was higher with a 5-year increase in age (odds ratio 1.29 [95% CI 1.28-1.31]), male sex (1.19 [1.13-1.25]), urban residency (1.24 [1.11-1.38]), low versus high education level (1.10 [1.02-1.19]), low versus high physical activity (1.28 [1.20-1.38]), family history of diabetes (3.15 [3.00-3.31]), higher waist-to-hip ratio (highest vs. lowest quartile; 3.63 [3.33-3.96]), and BMI (≥35 vs. <25 kg/m(2); 2.76 [2.52-3.03]). The relationship between diabetes prevalence and both BMI and family history of diabetes differed in higher- versus lower-income country groups (P for interaction < 0.0001). After adjustment for all risk factors and ethnicity, diabetes prevalences continued to show a gradient (LIC 14.0%, LMIC 10.1%, UMIC 10.9%, and HIC 5.6%).

METHODS: In this pooled analysis, we studied 133,118 individuals (63,559 with hypertension and 69,559 without hypertension), median age of 55 years (IQR 45-63), from 49 countries in four large prospective studies and estimated 24-h urinary sodium excretion (as group-level measure of intake). We related this to the composite outcome of death and major cardiovascular disease events over a median of 4.2 years (IQR 3.0-5.0) and blood pressure.

FINDINGS: Increased sodium intake was associated with greater increases in systolic blood pressure in individuals with hypertension (2.08 mm Hg change per g sodium increase) compared with individuals without hypertension (1.22 mm Hg change per g; pinteraction<0.0001). In those individuals with hypertension (6835 events), sodium excretion of 7 g/day or more (7060 [11%] of population with hypertension: hazard ratio [HR] 1.23 [95% CI 1.11-1.37]; p<0.0001) and less than 3 g/day (7006 [11%] of population with hypertension: 1.34 [1.23-1.47]; p<0.0001) were both associated with increased risk compared with sodium excretion of 4-5 g/day (reference 25% of the population with hypertension). In those individuals without hypertension (3021 events), compared with 4-5 g/day (18,508 [27%] of the population without hypertension), higher sodium excretion was not associated with risk of the primary composite outcome (≥ 7 g/day in 6271 [9%] of the population without hypertension; HR 0.90 [95% CI 0.76-1.08]; p=0.2547), whereas an excretion of less than 3 g/day was associated with a significantly increased risk (7547 [11%] of the population without hypertension; HR 1.26 [95% CI 1.10-1.45]; p=0.0009).

INTERPRETATION: Compared with moderate sodium intake, high sodium intake is associated with an increased risk of cardiovascular events and death in hypertensive populations (no association in normotensive population), while the association of low sodium intake with increased risk of cardiovascular events and death is observed in those with or without hypertension. These data suggest that lowering sodium intake is best targeted at populations with hypertension who consume high sodium diets.

METHODS: We assessed fruit and vegetable consumption using data from country-specific, validated semi-quantitative food frequency questionnaires in the Prospective Urban Rural Epidemiology (PURE) study, which enrolled participants from communities in 18 countries between Jan 1, 2003, and Dec 31, 2013. We documented household income data from participants in these communities; we also recorded the diversity and non-sale prices of fruits and vegetables from grocery stores and market places between Jan 1, 2009, and Dec 31, 2013. We determined the cost of fruits and vegetables relative to income per household member. Linear random effects models, adjusting for the clustering of households within communities, were used to assess mean fruit and vegetable intake by their relative cost.

FINDINGS: Of 143 305 participants who reported plausible energy intake in the food frequency questionnaire, mean fruit and vegetable intake was 3·76 servings (95% CI 3·66-3·86) per day. Mean daily consumption was 2·14 servings (1·93-2·36) in low-income countries (LICs), 3·17 servings (2·99-3·35) in lower-middle-income countries (LMICs), 4·31 servings (4·09-4·53) in upper-middle-income countries (UMICs), and 5·42 servings (5·13-5·71) in high-income countries (HICs). In 130 402 participants who had household income data available, the cost of two servings of fruits and three servings of vegetables per day per individual accounted for 51·97% (95% CI 46·06-57·88) of household income in LICs, 18·10% (14·53-21·68) in LMICs, 15·87% (11·51-20·23) in UMICs, and 1·85% (-3·90 to 7·59) in HICs (ptrend=0·0001). In all regions, a higher percentage of income to meet the guidelines was required in rural areas than in urban areas (p<0·0001 for each pairwise comparison). Fruit and vegetable consumption among individuals decreased as the relative cost increased (ptrend=0·00040).

INTERPRETATION: The consumption of fruit and vegetables is low worldwide, particularly in LICs, and this is associated with low affordability. Policies worldwide should enhance the availability and affordability of fruits and vegetables.

METHODS: HGS was measured using a Jamar dynamometer in 125,462 healthy adults aged 35-70 years from 21 countries in the Prospective Urban Rural Epidemiology (PURE) study.

RESULTS: HGS values differed among individuals from different geographic regions. HGS values were highest among those from Europe/North America, lowest among those from South Asia, South East Asia and Africa, and intermediate among those from China, South America, and the Middle East. Reference ranges stratified by geographic region, age, and sex are presented. These ranges varied from a median (25th-75th percentile) 50 kg (43-56 kg) in men <40 years from Europe/North America to 18 kg (14-20 kg) in women >60 years from South East Asia. Reference ranges by ethnicity and body-mass index are also reported.

METHODS: A cross-section of 163,397 adults aged 35 to 70 years were recruited from 661 urban and rural communities in selected low-, middle- and high-income countries (complete data for this analysis from 151,619 participants). Using blood pressure measurements, self-reported health and household data, concentration indices adjusted for age, sex and urban-rural location, we estimate the magnitude of wealth-related inequalities in the levels of hypertension awareness, treatment, and control in each of the 21 country samples.

RESULTS: Overall, the magnitude of wealth-related inequalities in hypertension awareness, treatment, and control was observed to be higher in poorer than in richer countries. In poorer countries, levels of hypertension awareness and treatment tended to be higher among wealthier households; while a similar pro-rich distribution was observed for hypertension control in countries at all levels of economic development. In some countries, hypertension awareness was greater among the poor (Sweden, Argentina, Poland), as was treatment (Sweden, Poland) and control (Sweden).

SETTING: 545 communities from 17 high-income, upper-middle, low-middle and low-income countries (HIC, UMIC, LMIC, LIC) involved in the Environmental Profile of a Community's Health (EPOCH) study from 2009 to 2014.

PARTICIPANTS: Community audits and surveys of adults (35-70 years, n=12 953).

PRIMARY AND SECONDARY OUTCOME MEASURES: Summary scores of tobacco policy implementation (cost and availability of cigarettes, tobacco advertising, antismoking signage), social unacceptability and knowledge were associated with quit ratios (former vs ever smokers) using multilevel logistic regression models.

RESULTS: Average tobacco control policy score was greater in communities from HIC. Overall 56.1% (306/545) of communities had >2 outlets selling cigarettes and in 28.6% (154/539) there was access to cheap cigarettes (<5cents/cigarette) (3.2% (3/93) in HIC, 0% UMIC, 52.6% (90/171) LMIC and 40.4% (61/151) in LIC). Effective bans (no tobacco advertisements) were in 63.0% (341/541) of communities (81.7% HIC, 52.8% UMIC, 65.1% LMIC and 57.6% LIC). In 70.4% (379/538) of communities, >80% of participants disapproved youth smoking (95.7% HIC, 57.6% UMIC, 76.3% LMIC and 58.9% LIC). The average knowledge score was >80% in 48.4% of communities (94.6% HIC, 53.6% UMIC, 31.8% LMIC and 35.1% LIC). Summary scores of policy implementation, social unacceptability and knowledge were positively and significantly associated with quit ratio and the associations varied by gender, for example, communities in the highest quintile of the combined scores had 5.0 times the quit ratio in men (Odds ratio (OR) 5·0, 95% CI 3.4 to 7.4) and 4.1 times the quit ratio in women (OR 4.1, 95% CI 2.4 to 7.1).

METHODS: We studied 125 287 participants from 18 countries in North America, South America, Europe, Africa, and Asia in the Prospective Urban Rural Epidemiology (PURE) study. Habitual food intake was measured with validated food frequency questionnaires. We assessed the associations between nutrients (total fats, saturated fatty acids, monounsaturated fatty acids, polyunsaturated fatty acids, carbohydrates, protein, and dietary cholesterol) and cardiovascular disease risk markers using multilevel modelling. The effect of isocaloric replacement of saturated fatty acids with other fats and carbohydrates was determined overall and by levels of intakes by use of nutrient density models. We did simulation modelling in which we assumed that the effects of saturated fatty acids on cardiovascular disease events was solely related to their association through an individual risk marker, and then compared these simulated risk marker-based estimates with directly observed associations of saturated fatty acids with cardiovascular disease events.

FINDINGS: Participants were enrolled into the study from Jan 1, 2003, to March 31, 2013. Intake of total fat and each type of fat was associated with higher concentrations of total cholesterol and LDL cholesterol, but also with higher HDL cholesterol and apolipoprotein A1 (ApoA1), and lower triglycerides, ratio of total cholesterol to HDL cholesterol, ratio of triglycerides to HDL cholesterol, and ratio of apolipoprotein B (ApoB) to ApoA1 (all ptrend<0·0001). Higher carbohydrate intake was associated with lower total cholesterol, LDL cholesterol, and ApoB, but also with lower HDL cholesterol and ApoA1, and higher triglycerides, ratio of total cholesterol to HDL cholesterol, ratio of triglycerides to HDL cholesterol, and ApoB-to-ApoA1 ratio (all ptrend<0·0001, apart from ApoB [ptrend=0·0014]). Higher intakes of total fat, saturated fatty acids, and carbohydrates were associated with higher blood pressure, whereas higher protein intake was associated with lower blood pressure. Replacement of saturated fatty acids with carbohydrates was associated with the most adverse effects on lipids, whereas replacement of saturated fatty acids with unsaturated fats improved some risk markers (LDL cholesterol and blood pressure), but seemed to worsen others (HDL cholesterol and triglycerides). The observed associations between saturated fatty acids and cardiovascular disease events were approximated by the simulated associations mediated through the effects on the ApoB-to-ApoA1 ratio, but not with other lipid markers including LDL cholesterol.

INTERPRETATION: Our data are at odds with current recommendations to reduce total fat and saturated fats. Reducing saturated fatty acid intake and replacing it with carbohydrate has an adverse effect on blood lipids. Substituting saturated fatty acids with unsaturated fats might improve some risk markers, but might worsen others. Simulations suggest that ApoB-to-ApoA1 ratio probably provides the best overall indication of the effect of saturated fatty acids on cardiovascular disease risk among the markers tested. Focusing on a single lipid marker such as LDL cholesterol alone does not capture the net clinical effects of nutrients on cardiovascular risk.

METHODS: The Prospective Urban Rural Epidemiology (PURE) study is a large, epidemiological cohort study of individuals aged 35-70 years (enrolled between Jan 1, 2003, and March 31, 2013) in 18 countries with a median follow-up of 7·4 years (IQR 5·3-9·3). Dietary intake of 135 335 individuals was recorded using validated food frequency questionnaires. The primary outcomes were total mortality and major cardiovascular events (fatal cardiovascular disease, non-fatal myocardial infarction, stroke, and heart failure). Secondary outcomes were all myocardial infarctions, stroke, cardiovascular disease mortality, and non-cardiovascular disease mortality. Participants were categorised into quintiles of nutrient intake (carbohydrate, fats, and protein) based on percentage of energy provided by nutrients. We assessed the associations between consumption of carbohydrate, total fat, and each type of fat with cardiovascular disease and total mortality. We calculated hazard ratios (HRs) using a multivariable Cox frailty model with random intercepts to account for centre clustering.

FINDINGS: During follow-up, we documented 5796 deaths and 4784 major cardiovascular disease events. Higher carbohydrate intake was associated with an increased risk of total mortality (highest [quintile 5] vs lowest quintile [quintile 1] category, HR 1·28 [95% CI 1·12-1·46], ptrend=0·0001) but not with the risk of cardiovascular disease or cardiovascular disease mortality. Intake of total fat and each type of fat was associated with lower risk of total mortality (quintile 5 vs quintile 1, total fat: HR 0·77 [95% CI 0·67-0·87], ptrend<0·0001; saturated fat, HR 0·86 [0·76-0·99], ptrend=0·0088; monounsaturated fat: HR 0·81 [0·71-0·92], ptrend<0·0001; and polyunsaturated fat: HR 0·80 [0·71-0·89], ptrend<0·0001). Higher saturated fat intake was associated with lower risk of stroke (quintile 5 vs quintile 1, HR 0·79 [95% CI 0·64-0·98], ptrend=0·0498). Total fat and saturated and unsaturated fats were not significantly associated with risk of myocardial infarction or cardiovascular disease mortality.

INTERPRETATION: High carbohydrate intake was associated with higher risk of total mortality, whereas total fat and individual types of fat were related to lower total mortality. Total fat and types of fat were not associated with cardiovascular disease, myocardial infarction, or cardiovascular disease mortality, whereas saturated fat had an inverse association with stroke. Global dietary guidelines should be reconsidered in light of these findings.

METHODS: We did a prospective cohort study (Prospective Urban Rural Epidemiology [PURE] in 135 335 individuals aged 35 to 70 years without cardiovascular disease from 613 communities in 18 low-income, middle-income, and high-income countries in seven geographical regions: North America and Europe, South America, the Middle East, south Asia, China, southeast Asia, and Africa. We documented their diet using country-specific food frequency questionnaires at baseline. Standardised questionnaires were used to collect information about demographic factors, socioeconomic status (education, income, and employment), lifestyle (smoking, physical activity, and alcohol intake), health history and medication use, and family history of cardiovascular disease. The follow-up period varied based on the date when recruitment began at each site or country. The main clinical outcomes were major cardiovascular disease (defined as death from cardiovascular causes and non-fatal myocardial infarction, stroke, and heart failure), fatal and non-fatal myocardial infarction, fatal and non-fatal strokes, cardiovascular mortality, non-cardiovascular mortality, and total mortality. Cox frailty models with random effects were used to assess associations between fruit, vegetable, and legume consumption with risk of cardiovascular disease events and mortality.

FINDINGS: Participants were enrolled into the study between Jan 1, 2003, and March 31, 2013. For the current analysis, we included all unrefuted outcome events in the PURE study database through March 31, 2017. Overall, combined mean fruit, vegetable and legume intake was 3·91 (SD 2·77) servings per day. During a median 7·4 years (5·5-9·3) of follow-up, 4784 major cardiovascular disease events, 1649 cardiovascular deaths, and 5796 total deaths were documented. Higher total fruit, vegetable, and legume intake was inversely associated with major cardiovascular disease, myocardial infarction, cardiovascular mortality, non-cardiovascular mortality, and total mortality in the models adjusted for age, sex, and centre (random effect). The estimates were substantially attenuated in the multivariable adjusted models for major cardiovascular disease (hazard ratio [HR] 0·90, 95% CI 0·74-1·10, ptrend=0·1301), myocardial infarction (0·99, 0·74-1·31; ptrend=0·2033), stroke (0·92, 0·67-1·25; ptrend=0·7092), cardiovascular mortality (0·73, 0·53-1·02; ptrend=0·0568), non-cardiovascular mortality (0·84, 0·68-1·04; ptrend =0·0038), and total mortality (0·81, 0·68-0·96; ptrend<0·0001). The HR for total mortality was lowest for three to four servings per day (0·78, 95% CI 0·69-0·88) compared with the reference group, with no further apparent decrease in HR with higher consumption. When examined separately, fruit intake was associated with lower risk of cardiovascular, non-cardiovascular, and total mortality, while legume intake was inversely associated with non-cardiovascular death and total mortality (in fully adjusted models). For vegetables, raw vegetable intake was strongly associated with a lower risk of total mortality, whereas cooked vegetable intake showed a modest benefit against mortality.

INTERPRETATION: Higher fruit, vegetable, and legume consumption was associated with a lower risk of non-cardiovascular, and total mortality. Benefits appear to be maximum for both non-cardiovascular mortality and total mortality at three to four servings per day (equivalent to 375-500 g/day).

Methods: Cross-sectional data from 21 countries in the Prospective Urban and Rural Epidemiology study were collected covering 61 229 hypertensive individuals aged 35-70 years, their households and the 656 communities in which they live. Outcomes include whether hypertensive participants have their condition detected, treated and/or controlled. Multivariate statistical models adjusting for community fixed effects were used to assess the associations of three social capital measures: (1) membership of any social organisation, (2) trust in other people and (3) trust in organisations, stratified into high-income and low-income country samples.

Results: In low-income countries, membership of any social organisation was associated with a 3% greater likelihood of having one's hypertension detected and controlled, while greater trust in organisations significantly increased the likelihood of detection by 4%. These associations were not observed among participants in high-income countries.

METHODS: Using measures of discrimination and calibration, we tested the performance of the NL-IHRS (n=100 475) and FC-IHRS (n=107 863) for predicting incident CVD in a community-based, prospective study across seven geographic regions: South Asia, China, Southeast Asia, Middle East, Europe/North America, South America and Africa. CVD was defined as the composite of cardiovascular death, myocardial infarction, stroke, heart failure or coronary revascularisation.

RESULTS: Mean age of the study population was 50.53 (SD 9.79) years and mean follow-up was 4.89 (SD 2.24) years. The NL-IHRS had moderate to good discrimination for incident CVD across geographic regions (concordance statistic (C-statistic) ranging from 0.64 to 0.74), although recalibration was necessary in all regions, which improved its performance in the overall cohort (increase in C-statistic from 0.69 to 0.72, p<0.001). Regional recalibration was also necessary for the FC-IHRS, which also improved its overall discrimination (increase in C-statistic from 0.71 to 0.74, p<0.001). In 85 078 participants with complete data for both scores, discrimination was only modestly better with the FC-IHRS compared with the NL-IHRS (0.74 vs 0.73, p<0.001).

METHODS: The Prospective Urban Rural Epidemiology study is ongoing in 21 countries. Here we report an analysis done in 18 countries with data on clinical outcomes. Eligible participants were adults aged 35-70 years without cardiovascular disease, sampled from the general population. We used morning fasting urine to estimate 24 h sodium and potassium excretion as a surrogate for intake. We assessed community-level associations between sodium and potassium intake and BP in 369 communities (all >50 participants) and cardiovascular disease and mortality in 255 communities (all >100 participants), and used individual-level data to adjust for known confounders.

FINDINGS: 95 767 participants in 369 communities were assessed for BP and 82 544 in 255 communities for cardiovascular outcomes with follow-up for a median of 8·1 years. 82 (80%) of 103 communities in China had a mean sodium intake greater than 5 g/day, whereas in other countries 224 (84%) of 266 communities had a mean intake of 3-5 g/day. Overall, mean systolic BP increased by 2·86 mm Hg per 1 g increase in mean sodium intake, but positive associations were only seen among the communities in the highest tertile of sodium intake (p<0·0001 for heterogeneity). The association between mean sodium intake and major cardiovascular events showed significant deviations from linearity (p=0·043) due to a significant inverse association in the lowest tertile of sodium intake (lowest tertile <4·43 g/day, mean intake 4·04 g/day, range 3·42-4·43; change -1·00 events per 1000 years, 95% CI -2·00 to -0·01, p=0·0497), no association in the middle tertile (middle tertile 4·43-5·08 g/day, mean intake 4·70 g/day, 4·44-5.05; change 0·24 events per 1000 years, -2·12 to 2·61, p=0·8391), and a positive but non-significant association in the highest tertile (highest tertile >5·08 g/day, mean intake 5·75 g/day, >5·08-7·49; change 0·37 events per 1000 years, -0·03 to 0·78, p=0·0712). A strong association was seen with stroke in China (mean sodium intake 5·58 g/day, 0·42 events per 1000 years, 95% CI 0·16 to 0·67, p=0·0020) compared with in other countries (4·49 g/day, -0·26 events, -0·46 to -0·06, p=0·0124; p<0·0001 for heterogeneity). All major cardiovascular outcomes decreased with increasing potassium intake in all countries.

INTERPRETATION: Sodium intake was associated with cardiovascular disease and strokes only in communities where mean intake was greater than 5 g/day. A strategy of sodium reduction in these communities and countries but not in others might be appropriate.

METHODS: In the Prospective Urban Rural Epidemiology (PURE) study, participants aged 35-70 years (n=156 625) were recruited from 110 803 households, in 604 communities and 22 countries; availability (presence of any dose of medication in the pharmacy on the day of audit) and medicine cost data were collected from pharmacies with the Environmental Profile of a Community's Health audit tool. Our primary analysis was to describe the availability and affordability of metformin and insulin and also commonly used and prescribed combinations of two medicines for diabetes management (two oral drugs, metformin plus a sulphonylurea [either glibenclamide (also known as glyburide) or gliclazide] and one oral drug plus insulin [metformin plus insulin]). Medicines were defined as affordable if the cost of medicines was less than 20% of capacity-to-pay (the household income minus food expenditure). Our analyses included data collected in pharmacies and data from representative samples of households. Data on availability were ascertained during the pharmacy audit, as were data on cost of medications. These cost data were used to estimate the cost of a month's supply of essential medicines for diabetes. We estimated affordability of medicines using income data from household surveys.

FINDINGS: Metformin was available in 113 (100%) of 113 pharmacies from high-income countries, 112 (88·2%) of 127 pharmacies in upper-middle-income countries, 179 (86·1%) of 208 pharmacies in lower-middle-income countries, 44 (64·7%) of 68 pharmacies in low-income countries (excluding India), and 88 (100%) of 88 pharmacies in India. Insulin was available in 106 (93·8%) pharmacies in high-income countries, 51 (40·2%) pharmacies in upper-middle-income countries, 61 (29·3%) pharmacies in lower-middle-income countries, seven (10·3%) pharmacies in lower-income countries, and 67 (76·1%) of 88 pharmacies in India. We estimated 0·7% of households in high-income countries and 26·9% of households in low-income countries could not afford metformin and 2·8% of households in high-income countries and 63·0% of households in low-income countries could not afford insulin. Among the 13 569 (8·6% of PURE participants) that reported a diagnosis of diabetes, 1222 (74·0%) participants reported diabetes medicine use in high-income countries compared with 143 (29·6%) participants in low-income countries. In multilevel models, availability and affordability were significantly associated with use of diabetes medicines.

INTERPRETATION: Availability and affordability of essential diabetes medicines are poor in low-income and middle-income countries. Awareness of these global differences might importantly drive change in access for patients with diabetes.

METHODS: The Prospective Urban Rural Epidemiology (PURE) study is a large multinational cohort study of individuals aged 35-70 years enrolled from 21 countries in five continents. Dietary intakes of dairy products for 136 384 individuals were recorded using country-specific validated food frequency questionnaires. Dairy products comprised milk, yoghurt, and cheese. We further grouped these foods into whole-fat and low-fat dairy. The primary outcome was the composite of mortality or major cardiovascular events (defined as death from cardiovascular causes, non-fatal myocardial infarction, stroke, or heart failure). Hazard ratios (HRs) were calculated using multivariable Cox frailty models with random intercepts to account for clustering of participants by centre.

FINDINGS: Between Jan 1, 2003, and July 14, 2018, we recorded 10 567 composite events (deaths [n=6796] or major cardiovascular events [n=5855]) during the 9·1 years of follow-up. Higher intake of total dairy (>2 servings per day compared with no intake) was associated with a lower risk of the composite outcome (HR 0·84, 95% CI 0·75-0·94; ptrend=0·0004), total mortality (0·83, 0·72-0·96; ptrend=0·0052), non-cardiovascular mortality (0·86, 0·72-1·02; ptrend=0·046), cardiovascular mortality (0·77, 0·58-1·01; ptrend=0·029), major cardiovascular disease (0·78, 0·67-0·90; ptrend=0·0001), and stroke (0·66, 0·53-0·82; ptrend=0·0003). No significant association with myocardial infarction was observed (HR 0·89, 95% CI 0·71-1·11; ptrend=0·163). Higher intake (>1 serving vs no intake) of milk (HR 0·90, 95% CI 0·82-0·99; ptrend=0·0529) and yogurt (0·86, 0·75-0·99; ptrend=0·0051) was associated with lower risk of the composite outcome, whereas cheese intake was not significantly associated with the composite outcome (0·88, 0·76-1·02; ptrend=0·1399). Butter intake was low and was not significantly associated with clinical outcomes (HR 1·09, 95% CI 0·90-1·33; ptrend=0·4113).

INTERPRETATION: Dairy consumption was associated with lower risk of mortality and major cardiovascular disease events in a diverse multinational cohort.

METHODS: The PURE study is a prospective, population-based cohort study of individuals aged 35-70 years who have been enrolled from 21 countries across five continents. The key outcomes were the incidence of fatal and non-fatal cardiovascular diseases, cancers, injuries, respiratory diseases, and hospital admissions, and we calculated the age-standardised and sex-standardised incidence of these events per 1000 person-years.

FINDINGS: This analysis assesses the incidence of events in 162 534 participants who were enrolled in the first two phases of the PURE core study, between Jan 6, 2005, and Dec 4, 2016, and who were assessed for a median of 9·5 years (IQR 8·5-10·9). During follow-up, 11 307 (7·0%) participants died, 9329 (5·7%) participants had cardiovascular disease, 5151 (3·2%) participants had a cancer, 4386 (2·7%) participants had injuries requiring hospital admission, 2911 (1·8%) participants had pneumonia, and 1830 (1·1%) participants had chronic obstructive pulmonary disease (COPD). Cardiovascular disease occurred more often in LICs (7·1 cases per 1000 person-years) and in MICs (6·8 cases per 1000 person-years) than in HICs (4·3 cases per 1000 person-years). However, incident cancers, injuries, COPD, and pneumonia were most common in HICs and least common in LICs. Overall mortality rates in LICs (13·3 deaths per 1000 person-years) were double those in MICs (6·9 deaths per 1000 person-years) and four times higher than in HICs (3·4 deaths per 1000 person-years). This pattern of the highest mortality in LICs and the lowest in HICs was observed for all causes of death except cancer, where mortality was similar across country income levels. Cardiovascular disease was the most common cause of deaths overall (40%) but accounted for only 23% of deaths in HICs (vs 41% in MICs and 43% in LICs), despite more cardiovascular disease risk factors (as judged by INTERHEART risk scores) in HICs and the fewest such risk factors in LICs. The ratio of deaths from cardiovascular disease to those from cancer was 0·4 in HICs, 1·3 in MICs, and 3·0 in LICs, and four upper-MICs (Argentina, Chile, Turkey, and Poland) showed ratios similar to the HICs. Rates of first hospital admission and cardiovascular disease medication use were lowest in LICs and highest in HICs.

INTERPRETATION: Among adults aged 35-70 years, cardiovascular disease is the major cause of mortality globally. However, in HICs and some upper-MICs, deaths from cancer are now more common than those from cardiovascular disease, indicating a transition in the predominant causes of deaths in middle-age. As cardiovascular disease decreases in many countries, mortality from cancer will probably become the leading cause of death. The high mortality in poorer countries is not related to risk factors, but it might be related to poorer access to health care.

METHODS: In this multinational, prospective cohort study, we examined associations for 14 potentially modifiable risk factors with mortality and cardiovascular disease in 155 722 participants without a prior history of cardiovascular disease from 21 high-income, middle-income, or low-income countries (HICs, MICs, or LICs). The primary outcomes for this paper were composites of cardiovascular disease events (defined as cardiovascular death, myocardial infarction, stroke, and heart failure) and mortality. We describe the prevalence, hazard ratios (HRs), and population-attributable fractions (PAFs) for cardiovascular disease and mortality associated with a cluster of behavioural factors (ie, tobacco use, alcohol, diet, physical activity, and sodium intake), metabolic factors (ie, lipids, blood pressure, diabetes, obesity), socioeconomic and psychosocial factors (ie, education, symptoms of depression), grip strength, and household and ambient pollution. Associations between risk factors and the outcomes were established using multivariable Cox frailty models and using PAFs for the entire cohort, and also by countries grouped by income level. Associations are presented as HRs and PAFs with 95% CIs.

FINDINGS: Between Jan 6, 2005, and Dec 4, 2016, 155 722 participants were enrolled and followed up for measurement of risk factors. 17 249 (11·1%) participants were from HICs, 102 680 (65·9%) were from MICs, and 35 793 (23·0%) from LICs. Approximately 70% of cardiovascular disease cases and deaths in the overall study population were attributed to modifiable risk factors. Metabolic factors were the predominant risk factors for cardiovascular disease (41·2% of the PAF), with hypertension being the largest (22·3% of the PAF). As a cluster, behavioural risk factors contributed most to deaths (26·3% of the PAF), although the single largest risk factor was a low education level (12·5% of the PAF). Ambient air pollution was associated with 13·9% of the PAF for cardiovascular disease, although different statistical methods were used for this analysis. In MICs and LICs, household air pollution, poor diet, low education, and low grip strength had stronger effects on cardiovascular disease or mortality than in HICs.

INTERPRETATION: Most cardiovascular disease cases and deaths can be attributed to a small number of common, modifiable risk factors. While some factors have extensive global effects (eg, hypertension and education), others (eg, household air pollution and poor diet) vary by a country's economic level. Health policies should focus on risk factors that have the greatest effects on averting cardiovascular disease and death globally, with additional emphasis on risk factors of greatest importance in specific groups of countries.