Epidemiological Studies: Correlation vs Cause

Chapter summary

Epidemiological studies are great at identifying trends and overall effects, but they can mask other information.
They rely heavily on statistics, the populations studied and confounding factors. They almost always show correlation, not causation.

It studies patterns, causes and effects of health and disease in defined populations.
It is the cornerstone of public health. It identifies risk factors and targets for prevention, and informs policy.
For many diet questions, experiments on people are impossible or unethical, so epidemiology is often the best evidence available.

Diabetes affects about 4% of Australians, up from 1.5% in 1989. Over 65% of Australians are overweight or obese, and 85% of diabetes is type 2.
Romaguera (Imperial College London) used eight European cohorts in the EPIC-InterAct study, about 350,000 participants. Each 12-oz can of sugar-sweetened soft drink a day was linked to 22% higher diabetes risk, and two cans to 44%.
Pure fruit juice and nectar were not linked to diabetes risk.
A meta-analysis of earlier, mostly North American studies found a 25% increase in type 2 diabetes for each daily sugary drink.

Researchers’ reading: Consistent results across many studies and countries make a real effect plausible.
The author’s caution: There are many correlations but no definitive causation. Is it the drink, or higher BMI? Do overweight people choose artificially sweetened drinks? Studies rely on subjects’ memory of what they ate, and on questionnaires.
Why it matters: Ethically, researchers cannot assign people to drink three cans a day for ten years. Animal and mechanistic studies fill the gap.

Keys proposed that different dietary fats affect health differently. His 1980 Seven Countries Study showed cholesterol strongly related to coronary heart disease deaths.
The author says Keys presented a subset of the data. With seven countries the correlation was linear, but with all 14 countries there was no correlation. The study nevertheless became the basis for fat and cholesterol reduction and statins.
Keys’s defenders: The Seven Countries Study followed individuals over time and found links in its own cohorts. Some scientists argue the criticism of country selection is overstated.

Value epidemiology, but do not mistake it for proof of cause.
Be most careful when science reaches the media. Qualifications are seldom mentioned, and causation is implied where there is none.

Wikipedia. Epidemiology.
Australian Institute of Health and Welfare (2013) The latest statistics on diabetes in Australia.
The InterAct Consortium (2013) Consumption of sweet beverages and type 2 diabetes incidence in European adults: results from EPIC-InterAct. Diabetologia. doi:10.1007/s00125-013-2899-8
Keys, A. (1980) Seven Countries: A Multivariate Analysis of Death and Coronary Heart Disease. Harvard University Press.
Kromhout, D. (1999) Serum cholesterol in cross-cultural perspective: the Seven Countries Study. Acta Cardiol 54:155-158.
Katan, M. and Beynen, A. (1981) Linoleic acid consumption and coronary heart disease in USA and UK. Lancet 2(8242):371.