A case study

1
A case study

• Investigation What causes colon cancer? What is
  the effect of diet on colon cancer?

2
Observational studies

• 1971 Dr. Denis Burkitt, British surgeon, studies
  disease patterns in poor, rural, African sites
• Rural Africans Much less colon cancer, diet rich
  in higher fiber
• Dont Forget Fibre in Your Diet (1979)
• Mounting evidence in 1970s, 1980s
• Societies/cultures with high fiber diets have low
  rates of colon cancer

3
Observational evidence

• Observation leads to hypothesis
• Biological reasons to suggest fiber
• Fiber adds bulk in intestines to help absorb and
  excrete cancer-causing substances
• Experiments with animals
• Rats fed carcinogens get less colon cancer if fed
  lots of fiber
• Observational studies of humans inconsistent
• Observational study of 88,000 nurses failed to
  detect effect of diet
• Nurses chose their own diet and were observed.
  Were those who chose high fiber different in
  other ways from those who did not?

4
How to find the truth?

• Based on the evidence would you advocate a high
  fiber diet to those at risk for colon cancer?
• Does the evidence suggest that a high fiber diet
  leads to lower rates of colon cancer?

5
Confounding variables
High Fiber
Reduced Cancer
High Fiber
Reduced Cancer
Confounding Variables ??? Pollution Low
sugar Socio-economic Genetic
6

• Experiments needed to settle the issue
• Two clinical trials in 2000
• Both trials are well conceived, well designed,
  well implemented, and clearly presented. (New
  England Journal of Medicine)

7
Clinical Trial 1

• Alberts et al. Lack of effect of a high-fiber
  cereal supplement on the recurrence of colorectal
  adenomas. N Engl J Med 2000 342 1156-1162
• 2,079 subjects
• Lasted 4 years
• Subjects randomly assigned to either eat low-fat,
  high fiber diet or to follow their usual diet

8
Clinical Trial 2

• Polyp Prevention Trial
• Schatzkin et al. Lack of effect of a low-fat,
  high-fiber diet on the recurrence of colorectal
  adenomas. N Engl J Med, 20003421149-1155.
• 1,429 subjects
• Lasted 3 years
• Subjects randomly assigned to either eat
  high-fiber cereals and food bars or to eat cereal
  and food bars that looked and tasted the same but
  were low in fiber.

9
Clinical trials

• Results
• Trial 1 (7 cases in fiber group, 2 in control
  group)
• Trial 2 (10 cases in fiber group, 4 in control
  group)
• What is disappointing . . . is that the findings
  of both trials are negative. (New England
  Journal of Medicine)
• 2 Fiber Studies Find No Benefit For The Colon
  (New York Times headline, April 20, 2000)
• I think weve definitely disproved the fiber
  hypothesis for colon cancer, Dr. David Alberts

10
Some room for caution, but . . .

• All subjects had had a benign polyp removed
• Perhaps different results for clean subjects
• Length of study at most 4 years
• Perhaps different results over longer time frame
• Perhaps fiber works earlier or later
• In any case, experiments show there is no
  evidence that fiber lowers cancer rates

11
Conclusion

• Randomized controlled trials have now shown us
  that the use of some of the diets and nutritional
  supplements thought to lower the risk of
  colorectal cancer has no short-term benefits with
  respect to preventing adenomas. There may be many
  reasons to eat a diet that is low in fat and high
  in fiber, fruits, and vegetables . . . but
  preventing colorectal adenomas, at least for the
  first three to four years, is not one of them.
  With regard to questions about diet and
  colorectal cancer, though, definitive answers
  still seem to be beyond the reach of both
  observational epidemiologic studies and
  randomized, controlled trials. (New England
  Journal of Medicine, April 20, 2000)

12
Randomized, controlled experiment

• The three key ingredients
• Compare two or more variables (often a treatment
  versus placebo)
• Control for lurking, confounding variables
• Isolate the causative factor
• Randomize Assign subjects to treatment by chance
• Goal Make groups as similar as possible except
  for the treatment
• Repetition Use enough subjects to reduce chance
  variation in the results

13
Language of experiments

• Response variable A variable that measures an
  outcome or result of study
• Explanatory variable A variable we think explains
  or causes changes in the response variable
• Subjects Individuals studied in an experiment
• Treatment Experimental condition applied to
  subjects
• Two variables are confounded when their effects
  on response variable cannot be distinguished from
  each other
• Statistical significance An observed effect so
  large that it would rarely occur by chance