Colorectal Cancer Risk: Relative vs. Absolute Data
This page breaks down Professor Bart Kay's analysis of the EPIC-Oxford epidemiology study regarding cancer rates in vegetarians.
The Reported Finding
The study (published in the AJCN, 2009) reported a 49% increase in the risk of colorectal cancer for vegetarians compared to meat-eaters.
Relative vs. Absolute Risk
The "49%" figure is a Relative Risk ratio. To understand the actual impact, we must look at the Absolute Incidence.
The Calculation
Based on the study's raw data:
| Group | Person-Years | Diagnoses | Absolute Rate |
|---|---|---|---|
| Non-Vegetarians | 320,625 | 166 | ~0.0005 (5 in 10k) |
| Vegetarians | 153,729 | 62 | ~0.0004 (4 in 10k) |
Key Observation: The raw incidence was actually lower in the vegetarian group (4 in 10,000) than the non-vegetarian group (5 in 10,000).
Why the "49% Increase"?
The 49% increase only appears after statistical adjustment. Researchers adjusted the data for factors like smoking and BMI. Professor Kay argues this practice is misleading because:
- It manipulates observed data based on correlations that are not proven to be causal.
- It allows for "shock" headlines that do not reflect the physical reality of the study's findings.
Educational Takeaway
When reviewing health headlines:
- Ignore the Relative Risk: Percentages like "50% increase" are often used to make small differences look large.
- Find the Absolute Differential: Look for the actual number of cases per population size.
- Question Adjustments: Be wary of statistics that have been "controlled" or "adjusted" to the point where they contradict the raw observations.
In this specific cohort, the difference in colorectal cancer incidence between vegetarians and non-vegetarians was approximately 1 case per 10,000 person-years, which is statistically and practically negligible.