INTERHEART

In Review

Yusuf S, et al. "Effect of potentially modifiable risk factors associated with myocardial infarction in 52 countries (the INTERHEART study)". Lancet. 2004. 364:937-952.
PubMed • Full text • PDF

Clinical Question

Among a global population of patients, what proportion of risk of first myocardial infarction is attributable to modifiable risk factors?

Bottom Line

Nine modifiable risk factors, including dyslipidemia, smoking, hypertension, diabetes, abdominal obesity, psychosocial factors, consumption of fruits and vegetables, alcohol use, and regular physical activity account for over 90% of the risk of first myocardial infarction.

Major Points

Modifiable cardiovascular risk factors have been well established in populations of European origin in high-income countries. However, 80% of the global burden of cardiovascular disease are in low and middle income countries. It is unclear whether these results can be applied worldwide across different geographic regions and ethnicities. Not only may risk factors for coronary artery disease vary between populations, prevalence of risk factors may differ as well.

Published in 2004, The Effect of Potentially Modifiable Risk Factors Associated with Myocardial Infarction (INTERHEART) case-control study enrolled 15,152 cases and 14,820 age and sex matched controls worldwide across all six habitable continents. Cases were defined as patients with a first myocardial infarction, and controls as patients without a history of exertional chest pain or history of coronary artery disease. Nine risk factors, including smoking, raised ApoB/ApoA1 ratio, hypertension, diabetes, abdominal obesity, psychosocial factors, daily fruits and vegetables, and regular physical activity collectively accounted for 90% of the population attributable risk (PAR) in men and 94% of the PAR in women. These associations remained present across ethnicities and geographic regions. This landmark epidemiological study suggested that the majority of first acute myocardial infarctions can be prevented with modification of these nine risk factors.

The three most important cardiovascular risk factors globally were a high ApoB/ApoA1 ratio (PAR 49.2%), a history of smoking (PAR 35.7%), and psychosocial factors (PAR 32.5%). The former two risk factors alone account for two thirds of the PAR alone, and risk for both variables increase in a dose-dependent manner. Protective risk factors included alcohol intake, daily fruits and vegetables, and exercise. Lifestyle modifications, including exercise, daily fruits and vegetables, alcohol intake, and smoking are responsible for half of the PAR (55%).

While most associations between the nine risk factors and cardiovascular disease was strong (p<0.0001), it was significantly weaker for alcohol intake(p=0.03). There was only a small protective benefit of regular alcohol consumption (defined as consumption more than three times a week), which may be more prominent in women. A subsequent analysis done by the INTERHEART investigators showed that heavy drinking was associated with an increased risk of myocardial infarction in the subsequent 24 hours.

Of notable interest, family history has a PAR of 10% but only adds 1% to the overall PAR, suggesting that the effect of family history is mostly mediated through the other modifiable risk factors rather than through an independent mechanism.

Design

• Global (six continents) matched case-control study
• N=29,972
  • First acute myocardial infarction (n=15,152)
  • Age and sex matched controls (n=14,820)
• Setting: 262 sites in 52 countries across six continents
• Primary outcome: Population attributable risk (PAR) and odds ratios (OR) of modifiable risk factors
  • Demographics
  • Socioeconomic status
  • Lifestyle: smoking, leisure, physical activity, diet
  • Personal and family history of cardiovascular disease
  • Risk factors (Hypertension, diabetes)
  • Psychosocial factors (depression, locus of control, perceived stress, life events)
  • Height, weight, waist & hip circumference, heart rate

Population

Inclusion Criteria

• Cases
  • First myocardial infarction
  • Admission to CCU or cardiology ward
  • Presentation within 24h of symptom onset
  • Characteristic symptoms
  • ECG changes indicative of MI
    • New pathological Q waves
    • mm 1ST elevation in 2 or more contiguous limb leads
    • new LBBB
    • new persistent ST-T wave changes
    • raised troponin 2x ULN
• Controls
  • First control per case
    • Visitor or relative of a patient from a non-cardiac ward, or non first-degree relative visitor of cardiac patient
  • Second control per case
    • Preferred: Hospital or outpatient clinic attendance for: refraction and cataracts, physical check-up routine Papanicolaou smear, routine breast examination, elective minor surgery unrelated to heart disease or risk factors, elective orthopedic surgery
    • Acceptable: Hospital or outpatient clinic attendance for: outpatient fractures, arthritic complaints, plastic surgery, hemorrhoids, hernias, hydroceles, routine colon cancer screening, endoscopy, minor dermatologic disorders

Exclusion Criteria

• Cases
  • Cardiogenic shock
  • Significant chronic medical illness (liver disease, renal disease, malignant disease, pregnancy, other)
• Controls
  • Cardiogenic shock
  • Significant chronic medical illness
  • Previous diagnosis of heart disease
  • History of exertional chest pain

Source of Controls

• Hospital-based controls (58%): no disorders related to known or potential risk factors for acute MI
• Community-based controls (36%): attendants or relatives of patients from non-cardiac ward, or non first-degree relative attendant of a cardiac patient
• WHO MONICA Study (3%)
• Unknown source of recruitment (3%)

Geographic Origin

• Western Europe: 5%
• Central and Eastern Europe: 14%
• North America: 2.4%
• South America and Mexico: 10%
• Australia and New Zealand: 5%
• Middle East: 13%
• Africa: 5%
• South Asia: 14%
• China and Hong Kong: 24%
• Southeast Asia and Japan: 8%

Ethnic Origin

• European: 27%
• Chinese: 25%
• South Asian: 17%
• Other Asian: 7%
• Arab: 10%
• Latin American: 9%
• Black African: 1%
• Coloured African: 2%
• Other: 0.5%

Outcomes

Comparisons are cases vs. controls. 99% confidence intervals are used unless otherwise specified.

Primary Outcomes

All risk factors were significantly associated with acute myocardial infarction (p<0.0001), except for alcohol which had a weaker association (p=0.03).

• Current and former smoking: 65% vs. 48%
  • OR 2.04 [99% CI 1.86-2.25]; PAR 35.7% [99% CI 32.5-39.1]
• Diabetes: 18% vs. 8%
  • OR 2.37 [99% CI 2.07-2.71]; PAR 9.9% [99% CI 8.5-11.5]
• Hypertension: 39 vs. 22%
  • OR 1.91 [99% CI 1.74-2.10]; PAR 18% [99% CI 15.7-20.4]
• Abdominal Obesity, top versus bottom tertile: 46% vs. 33%
  • OR 1.62 [99% CI 145-1.80]; PAR 20.1% [99% CI 15.3-26.0]
• Psychosocial: Prevalence unavailable
  • OR 2.67 [99% CI 2.21-2.32]; PAR 32.5% [99% CI 25.1-40.8]
• Daily fruits & vegetables: 36% vs. 42%
  • OR 0.70 [99% CI 0.62-0.79]; PAR 13.7% [99% CI 9.9-18.6]
• Exercise: 14% vs. 19%
  • OR 0.86 [99% CI 0.76-0.97]; PAR 12.2% [99% CI 5.5-25.1]
• Alcohol intake: 24%.0 vs. 24.5%
  • OR 0.91 [99% CI 0.82-1.02]; PAR 6.7% [99% CI 2.0-20.2]
• ApoB/ApoA1 ratio; top versus bottom quintile: 33% vs. 20%
  • OR 3.25 [99% CI 2.81-3.76]; PAR 49.2% [99% CI 43.8-54.5]

Combination Reported ORs and PARs

• All risk factors: OR 129 (99% CI 90-185); PAR 90.4% (99% CI 88.1-92.4)
• All risk factors, current smokers only (no ex-smokers), extremes of obesity and ApoB/A1 ratio: OR 333.7 (99% CI 230.2-483.9)
• ApoB/A1 ratio and smoking: PAR 66.8% (99% CI 62.8-70.6)
• All lifestyle: PAR 54.6%
• Smoking, Hypertension, Diabetes: OR 13.1 (99% CI 10.69-15.83); PAR 53%
• Smoking, Hypertension, Diabetes, ApoB/ApoA1 Ratio: OR 42.3 (99% CI 33.2-54.0); PAR 75.8% (99% CI 72.7-78.6)
• Smoking, Hypertension, Diabetes, ApoB/ApoA1 Ratio, Abdominal Obesity: PAR 80.2% (77.5-82.7)

Dose-Dependent Risk Factors

• Smoking: Linear relationship between OR and number of cigarettes smoked per day. When >40 cigarettes are smoked daily, OR = 9.16
• ApoB/ApoA1 ratio: A linear relationship between the deciles of ApoB/A1 ratio and OR is demonstrated. At the highest decile, the OR is 1.28. At the lowest decile, the OR is 0.43.

Subgroup Analysis

• Gender, Male vs. female (PARs)
  • Smoking: 42.7% vs. 14.8%
  • Fruits and vegetables: 11.7% vs. 19.1%
  • Exercise: 9.3% vs. 27.1%
  • Alcohol: 5.1% vs. 22.1%
  • All lifestyle: 56.5% vs. 60.6%
  • Hypertension: 14.9% vs. 29.0%
  • Diabetes: 8.0% vs. 16.1%
  • Abdominal obesity: 19.7% vs. 18.7%
  • All psychosocial: 28.8% vs. 45.2%
  • Lipids: 49.5% vs. 47.1%
  • All nine risk factors: 90% vs. 94%

• Age, young (<=55y) vs old (>55y) (ORs)
  • Only statistically significant (P<0.001) results are presented. 99% CI's used.
  • Smoking: 3.33 (2.86-3.87) vs. 2.44 (2.10-2.84)
  • Diabetes: 2.96 (2.40-3.64) vs. 2.05 (1.71-2.45)
  • High ApoB/ApoA1 ratio: 4.35 (3.49-5.42), vs. 2.50 (2.05-3.05)
  • All risk factors except smoking: 101.86 (61.22-169.46) vs. 42.24 (26.96-69.37)
  • All nine risk factors: 216.47 (126.67-369.94) vs. 81.99 (50.02-134.40)

• Geographic Region
  • PAR was 86-93% using all nine risk factors across ethnic groups
  • Dyslipidemia, smoking, and psychosocial are the most important risk factors worldwide
  • Abdominal obesity has a higher PAR in high and mid income countries
  • Smoking has a lower PAR in high and mid income countries
  • In China, obesity is a far less important risk factor (PAR 5.5% vs. 20.1% worldwide) compared to other geographic regions

Other

• Family History
  • OR 1.45 (1.31-1.60)
  • PAR 9.8% (7.6-12.5)
  • Overall PAR increases from 90.4% to 91.4% with addition of family history
    • Young: PAR 14.8% (11.7-18.5)
    • Old: PAR 10.4% (8.3-13.0)

Criticisms

• Case-control design prone to confounding
• Inconsistency with repeated measures of risk factors, with only moderate agreement rates for alcohol, fruit, vegetable intake, and exercise
• Post-myocardial infarction changes in risk factor measurement (treatment of hypertension post-MI, hyperglycemia post-MI)
• Cases and controls over-represent patients from urban areas
• Hospitalized controls are unlikely to accurately represent the population prevalence of risk factors
• Low recruitment of patients of African origin

Funding

• Canadian Institutes of Health Research
• Heart and Stroke Foundation of Ontario
• International Clinical Epidemiology Network
• Various national bodies in different countries (available in web appendices)
• Pharmaceutical companies (unrestricted grants): AstraZenca, Novartis, Hoechst Marion Roussel, Bristol Myers Squibb, King Pharma, Sanofi-Sythelabo