Speaking Notes

PADM 5324

September 28, 2009

Dr. Neubauer

WHERE WE ARE

• The Midterm Exam. is due October 6, 2009.
• Your homework 1 assignment is available on the web and is being distributed this evening.
• How is the use of the Google Docs pages coming?
• I attended a "tabletop planning" session last week regarding ASU plans for an H1N1 event. I learned that there is a plan to designate one wing of one residence hall for students who may have the H1N1 flu.
• In other news, the first doses of vaccine will apparently be nasal doses that contain small amounts of the live virus.

Chapter 8 -- Randomized Trials: Some Further Issues

REVIEW -- the benefit of a randomized trial (double blinded) is that the outcomes of a treatment group can be compared with the outcomes of a control group to better ascertain the apparent consequences of treatment.

Given that you have a "sample population" of people with the condition (and "n" is quite large) the "gold standard" is to be able to randomly assign the participants to two groups and give one group the new treatment.

THE WHOLE POINT is to possibly collect evidence regarding the SAFETY and EFFICACY of the new treatment. We don't know if the new treatment is safe and effective (and for "whom") or not.

We assume that the people in the control group receive a existing treatment rather than no treatment at all.

WE DON'T KNOW "REALITY" BUT IF WE DID we could identify Type 1 and Type II errors.

TYPE I ERROR -- The outcomes with the new treatment are no different than outcomes with the old treatment (or a placebo) but we conclude from the research that the new treatment is better (or worse).

TYPE II ERROR -- The outcomes with the new treatment are different than outcomes with the old treatment (or a placebo) but we conclude from the research that there are no differences in outcomes.

The other two possibilities are that the new treatment is no better or worse than the existing treatment (or placebo) and we come to that conclusion; and that the new treatment is better (or worse) and we come to that conclusion.

The problem is, WE JUST DON'T KNOW FOR SURE, even after having done the study.

SO, HOW CAN WE REDUCE OUR UNCERTAINTY?

1) Other things being equal, LARGER SAMPLES ARE BETTER THAN SMALLER SAMPLES.

2) If our HYPOTHESIS IS BASED ON A SOUND THEORY OF CAUSATION we probably should have a greater faith in the outcome of our study, assuming the outcome is what we expected.

3) FINANCIAL INCENTIVE SHOULD NOT be a cause for faith in the outcome we expected. Nor should it cast doubt assuming a good RESEARCH DESIGN and a DOUBLE BLIND method of assignment participants to groups and to assessing individual outcomes.

4) PHASE TESTING can contribute to certainty regarding our findings.

PHASE TESTING (page 156) involves a SEQUENCE of research involving increasingly larger sample populations and finally CONTINUED SURVEILLANCE once the new treatment (drug, whatever) is actually marketed to large numbers of patients. In every phase there are two major issues -- SAFETY and EFFICACY. Safety includes the identification of CONTRAINDICATIONS.

http://en.wikipedia.org/wiki/Contraindicated

5) It is likely that multiple researchers may be doing similar studies at the same time. If they all report similar findings that should increase our faith (confidence) in the findings.

A BIG PROBLEM is that "failed" research is less likely to be published than "successful" research. FINDING NO DIFFERENCE is not as "exciting" as finding a difference. "Boring" journals are less likely to be read (and less likely to sell).

So, the editors of the major medical journals have agreed to create a REGISTRY of all clinical trials, such that if the researcher did not register his/her intent to do research PRIOR to involving participants he or she will be UNABLE TO PUBLISH FINDINGS.

I have three QUESTIONS about this.

1) Are the findings available in the registry?

2) Is this an obstacle to research?

3) If some researcher DISCOVERS SOMETHING REALLY IMPORTANT will the editors really prevent it being published? Would it be ethical to do so?