Ger 1000 Tut 1 Design of Studies

...

BLINDING

Single Blinding. If the participants, but not the evaluators, of an experiment do not know whether each participant has been assigned to the treatment or the control group, then it is single-blinded. This removes Placebo effect

Double-Blinding. If both the participants and the evaluators do not know whether each participant has been assigned to the treatment or the control group, then it is double-blinded. This removes any bias from the evaluator’s side

No Blinding. However, if both the participants and the evaluators know whether each participant has been assigned to the treatment or the control group, then it is not blinded.

Placebo Effect - a beneficial effect produced by a placebo drug or treatment, which cannot be attributed to the properties of the placebo itself, and must therefore be due to the patient's belief in that treatment.

This is useful if control group are not blinded. For e.g., if the control group knows the treatment group are getting vaccination for a certain illness, they may take precaution to prevent themselves from falling ill.

- While subjects can be assigned at random to treatment and control, adherence cannot be similarly controlled.

(People don’t usually go home, flip a coin, and use the coin to decide if they should take their medicine. Some amount of bias is involved based on a person’s habits and character, and the investigator cannot control that.)

- To test of blind is broken, can test adherence rates (namely find the rates among adherers). If similar, bind not broken,

Observational Studies.

In observational studies, researchers do not assign participants to either a treatment or a control group. Rather, participants assign themselves to either one of these two groups.

In observational studies, confounding is a real problem.

- Investigator has NO power to regulate factors (only can observe events happening)

- Subjects assign themselves into “Control Group” and “Treatment Group”

- Note: Technically not correct to use the terms ‘control group’ and ‘treatment group’

- Bias may arise [Observational Studies are prone to Confounding]

- What is the confounding factor?

Cohort Studies. In cohort studies, researchers select participants based on their exposure status. If the samples are selected at random, cohort studies can be used to estimate population risks of disease, and risk ratios of disease.

Case-Control Studies. In case-control studies, researchers select participants based on their disease status. Even if the samples are selected at random, case-control studies cannot be used to estimate population risks of disease, and risk ratios of disease.

Historical controls

An Experiment that spans a long period of time, might have differences between its control and treatment group due to changes over time. Also, individuals may not have been randomly assigned into a “control” and “treatment” group, hence they have differences.

Thus, if we use historical controls, we must take into account any possible confounding factor that might have occurred due to a long change in time

Types of Studies

There are many ways to conduct a study.

EG: If we are interested in the association between smoking (Exposure) and cancer (response)

A. Invite 100 smokers and 100 non-smokers Cohort study [pic 2]

• Ask each one if they have cancer

B. Invite 100 cancer patients and 100 non-cancer patients Case and Control Study[pic 3]

• Ask each one if they smoke

C. Researcher knows who the smokers are before knowing Prospective Study[pic 4]

who the cancer patients are

• Looking into the future

ii. Researcher knows who the cancer patients are before Retrospective Study[pic 5]

knowing who the smokers are

• Looking into the past

[pic 6]

[pic 7]

...