Class Notes (1,100,000)
CA (650,000)
SFU (10,000)
BPK (800)
BPK 140 (100)
Lecture 3

kins 140 Lecture 3 Prof Arnold


Department
Biomedical Physio & Kines
Course Code
BPK 140
Professor
Anne- Kristina Arnold
Lecture
3

Page:
of 6
kins140 lecture 3
see assignment #1
Learning Objectives
1. Evaluate sources of health information for credibility and describe the characteristics of
credible health information
2. Explain the components of the scientific method
4. State the limitations of human experiments and
suggest alternative solutions Explain how epidemiology works
5. Pinpoint the weakness of epidemiological evidence
6. Identify whether a particular example represents experimental, epidemiological, clinical or
anecdotal evidence
The credibility of health claims
-how do you decide what to believe?
-if you read something what would make it more or less believable to you?
Health information is confusing
Where to start: evidence
-this is a science course.
-scientist are trained to look at the evidence when evaluating health claims.
-there are several types of evidence, some better than others
1. Experimental evidence
a. the preferred type of scientific evidence
-hypotheses
-experimental group
-control group
-individual variability
-random sample from population
-random assignment to control or exp group
-statistical significance
problems with experiments:
-time (long latent period)
-Ethics (animals, tissue cultures [skin cells], computer models)
2. Epidemiological evidence
-”epi”=epidemic, “demo”=people
-epidemiology: study of the distribution & determinants of a condition in a population”
-it is scientific evidence, but not as high quality as experimental evidence
-like medical detective work.
Epidemiological studies
-correlating who gets a disease with what they do or what they are exposed to
-smokers get lung cancer (and other other cancers and diseases) more often
than non smokers
-women who are sexually active develop cervical cancer more often than women
who are not
-women who do not give birth to any children develop breast cancer more often
than women who have given birth
-correlation-association
-observational (observe but don’t intervene)
-examples: framingham study, cholera in london ~1800, young men with a rare
skin cancer and pneumonia in NY and San francisco (1980’s)
Epidemiology vs. Experiment
-experiment manipulates variables under controlled conditions
-epidemiological studies observe without interfering
Association =/= Causation
-just because two things are statistically associated or correlated does not ,mean that
one thing causes the other
-cities with more churches have higher rates of crime
-there are criteria to help decide if the association is just an accident or coincident or
whether it may be casual
Criteria for association to be casual
1. strength (non smoker risk 1.0 smoker risk 10-20x) -(but many small risks are real!)
2. Dose-response -(risk of cancer increases with amount smoked
3. Consistency -(studies can be repeated)
4. Temporally correct -(the subjects smoked, THEN developed lung cancer
5. Specificity -(about 80% of lung cancer patients are smokers or former smokers) -(only
15% of smokers get lung cancer (though some get COPD, etc)
6. Biological plausibility -(cigarette smoke contains compounds that are known
carcinogens)
other kinds of experimental support for an association result
-animal studies
eg. Correlate a particular DNA sequence variant in a DNA repair gene with
lymphoma
-make a “knock-out” mouse lacking the gene
-cell culture studies studies
-make a cell line lacking the gene of interest
-demonstrate that compared to a normal cell line, DNA is repaired less
effectively
Consideration in scientific evidence
1. bias
2. confounding variables
3. statistical significance
4. replication
5. sample size
6. ethics
1. Reducing bias
Bias: Systematic errors in collecting or interpreting data
-random assignment to test or control group (for a prospective study
-blind study
-placebo (given to the control group in a blind study
-double blind (neither the sibject nor the people administering the test know who is
receiving drug vs placebo
2. Avoiding Confounding Factors
-confounding factors=other factors that could acount for the result
-eg: association between gambling and lung cancer
-e.g., in a case-control study, the case and control groups should be similar in
every way, EXCEPT that the case group is affected and the control group is not
Same proportion of sexes in the two groups Same ethnicities in the two
groups Same geographic region Same age range
see sex differences in health status
Androcentricity: male viewpoint
-overgeneralization to one sex
-gender insensitivity: overlooking sex as an important varibles
Ethnicity- not a simple concept
-has several components
-origins or ancestry
-culture
-language
-identity
-the term race is falling out of use is now seen by some to be an over simplification of
human diversity into four groups