Kinds of Studies:
-Case Study: n=1, Observation write up, no comparision or controls.
-Case Report: Not really science, but good at generating hypothesis.
-Case Series: n is about 5
- These can be useful with new diseases
-These are really unconvincing when they discuss the acuracy of the treatment.
-Studies of new treatment efficacy needs controls.
1.) Historical controls: Want all your patients to get the treatment.
-Get data from medical records of people who already had the disease at an earlier time. (ex: 1 yr. before and they use these records to compare the outcomes of untreated patients with your treated patients treatment.)
-Matched fro age, sex, socioeconomic status, etc…(However, you can only match on traits you think are important.)
-Stimultaneous Nonrandomized controls
2.) Randomization: Randomly assign patients to treatment or negative control group.
-Best: Double-blinded: Where neither the physician nor the patient knows what group the patient is in.
-Single-Blind: Patient does not know, but the researcher or physician does.
It is good for the patient not to know because of the placebo effect. On the other hand, if the researchers do not know, then they will not over observe.
3.) Crossover Study: Randomize: Treatment vs. contol in which the groups are switched. This means that the patients in the treatment group will go to the control and those in the control go to the treatment group. You could not get a population that is not representative of the entire population. The more data that is obtained, the more representative it is.
4.) Stratified Randomization: What if: A patient's trait strongly influences the outcome. (ex: age) First, you would sort your study participants based on the trait you think will influence the outcome of your study. Then, after you have the demographic breakdown of the treatment vs. control goups determined you will assign your participants to groups randomly.
5.) Factorial Design: Allow you to compare two treatments to each other and in combination. (ex: in the book on aspirin and beta carotene)
Physicians Health Study: Showed that Aspirin greatly reduced heart-attacks in older adults. They used physicians for this study because doctors are more likely to stay in the study and be complient and follow protocol.
6.) Noncompliance: Tendency for people who are enrolled in a study to not follow the protocol they have been assigned to.
2 Types: a.) Dropouts: Voluntary disenrollment from a study. (No data from them) Some may stay in the study, but stop doing treatment. (often w/out the researchers knowing)
b.) Drop-ins: Patients in one group may take assigned agent of the other group. A participant in the study (-) control group, but starts taking the experimental groups treatment. (researchers may not know)
(ex: Hypothesis: Echinacea daily decreases chance of getting colds and flu: have risk or noncompliance.)
