Lecture 11 - Problems And Limits Of Epidemiology Flashcards
The ultimate goal of many epidemiological studies is to…..
determine the causes of disease
What are the 3 steps in which this “ultimate goal” is accomplished?
- Observing a possible association between an exposure and an illness
- Developing a hypothesis about a cause and effect relationship
- Testing the hypothesis through a formal, epidemiological study
While a formal study can strongly support the conclusion that a certain exposure causes a certain disease……
there are many potential sources of error in drawing a conclusion
Studies of _______ are especially more prone to error. Why?
chronic diseases because they often have multiple determinants and develop over long periods of time
What is the key difference (that poses a potential error) between studying humans and studying animals?
With humans, it is unlikely that participants would remain on the appropriate diet over the necessary length of time - too many temptations
it is not realistic to expect to succeed at a randomized, double blind controlled trial that requires people to….
alter their behavior over a significant amount of time (unless if they are suffering from a serious disease and participation is the only way to have access to a new, potentially effective, treatment)
instead of a randomized, double blind trial to test a DIETARY hypothesis, what other alternative study could be conducted?
a cohort study - large group of people could be asked detailed questions about their diet and over the next 5 years the health of the 2 groups is compared
While a cohort study would NOT require people to change their current behavior, what are some cons associated with this?
people who have voluntarily chosen to eat a healthy diet may differ in other respects than those who eat the average diet (more exercise, less likely to smoke, etc)
CAUSES TOO MANY VARIABLES - HEALTHY LIVING IS THE REASON FOR NO HEART DISEASE - NOT NECESSARILY JUST A HEALTHY DIET
What is a 3rd type of study that could be conducted?
case control study
researchers choose a group of heart attack survivors and another group of the same age who HAVENT had a heart attack and question their diets over the past 5 years
What are some problems with using a case control study?
-people may forget what they’ve eaten in the past
-Patients might be embarrassed to admit
-Information may not be reliable
News reports of new health studies are often confusing and 2 different articles may show contradictory results- what accounts for this?
reported result could be a merely random variation due to chance
So that results aren’t contradictory, as a general rule - what is required?
statistical significance
-require a LARGE NUMBER of subjects in studies of chronic disease to draw valid conclusions
So that results aren’t contradictory, as a general rule - what is required?
statistical significance
-require a LARGE NUMBER of subjects in studies of chronic disease to draw valid conclusions
Why are chronic diseases so difficult to determine the cause?
causes of these diseases are usually complex and there are usually long periods between exposures to potential causes
TOO MANY VARIABLES
The weaker the relationship between exposure and disease, the ______ the group of people must be for the relationship to be evident
larger
Why were the results concerning smoking and lung cancer (Doll-Hill and Hammond-Horn) so convincing?
they involved a large number of subjects and the link was very strong
What are confounding variables?
factors that are associated with the exposure and may independently affect the risk of developing the disease
Give an example of a study that may have had some confounding variables?
1980s study that suggested coffee drinking could cause pancreatic cancer
many coffee drinkers are also smokers —> that could have been the cause
THERE HAS BEEN NO INDEPENDENT CONFIRMATION
coffee-cancer study —how could these errors be eliminated?
conduct a new study ONLY ON NONSMOKERS
Selection bias is a particular problem in choosing subjects for what kind of study?
a case control study (low fat vs high fat diet example. ppl who consume less fat are likely to live a healthier lifestyle in general)
Cohort studies extend over many years and are thus likely to suffer from a form of selection bias caused by…..
people dropping out or not responding
A ___ drop out rate casts doubts on the result of any epidemiological study
high
Reporting bias/recall bias is a problem in what kind of study?
case control study
(ex: birth defects - mother would be reluctant to admit her potential involvement. people who drink heavily tend to under report. underweight individuals may over report their fat intake)
What is the most important indication that an epidemiological result is valid? give an example
if the results are consistent with other investigations
ex- british case control study linking birth control to breast cancer only had a 2.3 odds ratio
nurses health cohort study had a 1.5 relative risk
odds ratio is low but it is amplified by the fact that the 2 independent studies yielded the same result
Besides different studies having the same result, what is another way to prove a cause and effect relationship?
dose-response relationship
-radiologists had the lowest life expectancy of the 3 groups and they’re ALSO exposed to radiation the most (highest dose)
What is another way that epidemiological evidence is more convincing?
if there is a known biological explanation for association between exposure and disease
Most epidemiologic studies are _____ and have little potential for ______. What is the exception?
observational and have little potential for harm
exception - intervention studies
These days, there are strict ethical limits in any study involving humans. why?
as a result of the abuses of Joseph Mengele who conducted medical experiments on concentration camp prisoners during WW2
What are the 3 groups who are most at risk of abuse in medical testing?
-poor patients
-prisoners
-patients of mental institutions
What is the purpose of the Institutional Review Board and why was it founded?
any study to be conducted must be approved by the Institutional Review Board. They must agree that it is:
1. well designed
2. its benefits outweigh its risks
3. Subjects are truly given informed consent
Founded because of the syphilis case
When does the IRB have the authority to halt a clinical trial?
if the treatment group is clearly showing better or worse results than the control group
Give a specific example in which a clinical trial was stopped?
The Physician’s Health study - obvious that subjects taking aspirin were suffering fewer heart attacks than those in the placebo group
Explain why the AIDS epidemic was different in the case of ethics
people who had AIDS knew that they had a fatal disease and no cure - they were desperate for any access to a new drug even if a clinical trial was not conducted
What is the issue when a study is being conducted in which there is already an effective treatment?
people do not want to risk being randomized to the placebo group if they suspect that the already active therapy is effective
Give an example TODAY of a study in which there is already an effective treatment?
bone marrow transplants. The National Cancer Institute is sponsoring a trial that requires women to be randomly placed in the transplant group or conventional chemo (current)
-many are reluctant to participate because they want the treatment that currently works (NEW TREATMENT HAS 5% CHANCE OF DEATH FROM THE TREATMENT ALONE)
