1. FMS1204S: Fraud, deception and data
Liew Xuan Qi (A N)
Cheong Hui Ping (A W)
Hong Chuan Yin (A M)

3. TITLE Current State of the Art in Medical Research Web Trials
Difficulties with Blinding
More on Blindness: Considering the Shangri-La Diet
Back to Web Trial

4. Current State of the Art in Medical Research
Moderately large randomized clinical trials
Published in a journal
Meta-analysis of these trials

5. However,
sample sizes in clinical trials seem to be simultaneously too small or to large
Too small: just barely statistically significant, hence the need for meta-analysis.
Too large: takes a huge amount of effort, and doesn’t allow for much learning and experimentation during the study.

6. Moreover,
High cost: people have to think seriously and justify what they want to do.
On the other hand, within any particular research plan, it would seem to limit the possibility for innovation.

7. Web trials

8. What is it?
Intermediate between controlled randomized trials and recording of observational data
If volunteer randomized and they follow the protocol, it would be a clean randomized experiment

9. Is it better? Lots of selection, dropout, measurement error, etc.
Moves it toward an observational study.
The data collection is dispersed
While the database is centralized.
More regularity—less variation

10. Difficulties with blinding

11. Seth Robert
web trial would be better than the conventional double-blind clinical trial,
if the goal is to guide practice.
in real life—patients are not blinded.
Blinding is a tool to equate expectations.
Better to equate expectations by comparing different treatments both believed to be effective.

12. Analysing Data from Web Trials
complex and interesting in new ways and accessible to everyone.
the observational study can be misleading.
Eg: Nurses Health Study
experiment actually contradicted the observational study—
statistically significant negative effect for one and statistically significant positive effect for the other.
It wasn’t just that there was significance for the experiment and no significance for the observational study

13. blind versus don’t blind.
Andrew Gelman
believe any experiment where subjects are not blind to the treatment has a problem
Knowledge of treatment could affect outcome
Seth Robert
Is there useful effect?
equate expectations across conditions and that blinding is just one way to do this.

14. More on Blindness: Considering the Shangri-La Diet
Andrew Gelman
Motivation to lose weight
Selection problem: trying new things to lose weight
Seth Robert
Assumes that people who are always ready to try something new and lose weight.
Participants try his diet are unusual early adopter types
Believe that merely changing what you eat (to foods with unfamiliar or at least less familiar flavors) should lower your set point.

15. Web Trials early in the research chain and are relatively practical
don’t worry a lot about mechanism. Worry about efficacy instead
trying to get inside this sort of study, rather than just relying on the “intent to treat” or explicit randomization.

16. NANI?
Treatments actually chosen by the subjects, not just at the treatments to which they were assigned.
learn about the process of selection.
use methods such as principal stratification to estimate the effects of the treatment among different subpopulations

17. So…. Web trials Potential for gathering lots of data
getting people’s active participation motivates them to randomize, to apply the treatment, and to record results.
have the potential to get people involved in the project as participants, not just ‘subjects.’

18. Sample size??
Too small? (because they don’t have the power to give definitive results)
Too big (because they don’t allow for innovation during the time of the study)?
Different goals need different tools—and different-sized experiments.
One goal is to come up with new ideas worth testing; another is to test those ideas.

19. “if the effect is strong, you don’t need a big study.”
Richard Doll, a famous epidemiologist

20. Any Question?