LECTURE 7 (Uncertainty of science) Flashcards
What is Statistics?
The science of uncertainty
What do Scientific studies produce?
Statistics which must be interpreted
What are the properties of Scientific answers?
- not definitive as people want them to be
- best answers that we have at any given time
Describe examples that answers provided by science are not always as definitive as people want them to be
The chance of a woman having unprotected intercourse and becoming infected with HIV is 1 in 1000 -> science cannot promise that they won’t get the virus
The chance of a woman getting lumpectomy instead of mastectomy and still getting breast cancer is 10% -> science cannot promise that the cancer won’t come back
Describe the 1997 controversy of whether women ages 40-49 should be screened for breast cancer
Studies shown that routinely testing women age 50 and older with breast x-rays could reduce BREAST CANCER MORTALITY in the population -> However, studies done on YOUNGER WOMEN had not demonstrated a life-saving benefit for this group
What are the reasons why breast cancer screening might not demonstrate a life-saving benefit?
- Routine screening increases radiation exposure -> raising risk of cancer
- Yields many false alarms -> necessary medical testing and major expense
- Follow-up testing may cause complications -> many women remain anxious even after cancer is ruled out
What happened after the 1997 issue of breast cancer screening came out?
The public and political response was heated -> A letter signed by 39 congresswomen stated that screening could save lives despite lack of evidence -> Pressure from politicians caused the director of NCI (National cancer Institute) to recommend that women in their 40s should be screened
[in 2007 the conclusion was that individual women in consultation with their doctor, should decide whether to be screened]
Why is the benefit of screening less for women in their 40s?
- Incidence of breast cancer is lower in women in their 40s
- Effectiveness of mammography is lower in denser breasts of the younger women
What does “Science is a work in progress” mean?
Today’s news contradicts yesterday’s reports
What is over-diagnosis?
Instances where the cancer would not otherwise have presented over the women’s lifetime, but the apparent discovery may lead to invasive treatment that could significantly reduce the woman’s quality of life
What is the P value?
Expresses the probability that the observed result could have occurred by chance alone
What is the P value for a criteria for a result to be considered statistically significant?
0.05 or less
What are the different ways to express the degree of probability of an experiment?
- P value
- Confidence interval
What is the Confidence interval?
A range of values within which the true result probably falls
Explanation: The narrower the confidence interval is, the lower the likelihood of a random error is
What are the reasons why a low p-value and a narrow confidence interval could lead to a wrong conclusion?
- Wrong statistical significance -> errors caused purely by chance
- Bias/confounding
What is statistical significance?
There is only a small statistical probability that the same result could have been found by chance and that the intervention had no real effect
What is Statistical power/sensitivity?
Likelihood of a significance test detecting an effect when there actually is one
What is Power?
The probability of finding an effect if there is an effect
High power -> indicated a large chance of a test detecting a true effect -> A study with a larger number of cases is more likely to be valid
What is the difference between false-negative and false-positive results?
False-negative = to find no effect when there actually is one
False-positive = occur when the study finds an effect that is not real (random variation appears to be a true effect)
What is the difference when tests are highly sensitive and highly specific?
Highly sensitive = they yield few false negatives
Highly specific = they yield few false positives
Why are most public health screening programs sensitive tests?
To avoid missing any individual with a serious disease who could be helped by some intervention
Explanation: some sensitive tests aren’t very specific -> false positive results -> many more tests conducted
What are the properties of Sensitivity and Specificity?
- Used to evaluate the validity of laboratory tests (not the results of the test)
- Determines whether or not to use a certain test or what situations a certain test would work best in
- Fixed as long as CUTOFF POINT is not changed -> not affected by changing prevalence
- Given as a % from 0% to 100%
What is Sensitivity?
- % of patients with the disease that receive a +ve result
- % chance that the test will correctly identify a person who actually has the disease
FORMULA: TP/Diseased
What is Specificity?
- % of patients without the disease that receive a -ve result
- % chance that the test will correctly identify a person who is disease free
FORMULA: TN/Not diseased
What is the difference between the Initial screening test and the Confirmatory test?
Initial screening test = high sensitivity
Confirmatory test = high specificity
Explanation: those that receive a +ve result on the first test will be given a second test (confirmatory test) -> if both tests are +ve -> definitive diagnosis
