Essential Clinical Epidemiology Randomized Controlled Trial Anna - - PowerPoint PPT Presentation

Essential Clinical Epidemiology Randomized Controlled Trial

Anna Keski-Rahkonen Pyry Sipilä Clinicum, Department of Public Health

Randomly generated celebrity faces

Video

Some solutions to homework puzzles

“God doesn't play dice with the universe.”

Albert Einstein Want to know more

about God playing dice with the universe?

RCT, Basic Study Design

Slide adapted from Lorraine Alexander & Karen Yeatts Treatment Group Control Group

Outcome YES NO Outcome YES NO

Randomization

• Sound randomization will on average

distribute all known and unknown confounding factors equally to different arms (groups) of the study.

• This can be affected by chance, of course.

The Cult of the RCT

RCT – common problems

• Something goes wrong with the

randomization

• Intention to treat analysis not

conducted

Allocation concealment

DATA CENTER

Making sure that randomization will occur

• > Removing selection bias

Who is Blind?

Participant Researcher Data Analyst

Intention to Treat, Basic Principle:

ITT

• Anything nonrandom (completion of study,

problems in allocaion concelment etc. etc.) will make the RCT an observational study.

• An observational study in the disguise of RCT

has all the pitfalls of any other observational study - and potentially worse.

• RCTs have usually not been designed to

adequately control for confounding.

RCTs, Advantages & Disadvantages

• Advantages:
• unbiased distribution of confounders;
• blinding more likely;
• randomisation facilitates statistical analysis.
• Disadvantages:
• expensive: time and money;
• volunteer bias;
• ethically problematic at times.

Slide: Oxford University, Centre for Evidence-Based Medicine

“All RCTs do is show that what you’re dealing with is not snake oil.” Ben Williams

PI CO Que stio ns

• P = Pa tie nt / Pro b le m
• I

= I nte rve ntio n

• C = Co mpa riso n/ Co ntro l
• O = Outc o me

What is a Large Effect? Effect Sizes: Cohen’s d

• Cohen classified effect sizes as
• small (d = 0.2),
• medium (d = 0.5), and
• large (d ≥ 0.8).5
• “A medium effect of .5 is visible to the naked eye
• f a careful observer. A small effect of .2 is

noticeably smaller than medium but not so small as to be trivial. A large effect of .8 is the same distance above the medium as small is below it.”

https://youtu.be/0sSHqWzqq-I

Visualize with your pair using bit.do/effectsize

How the distributions in the intervention and control group differ

• small (d = 0.2),
• medium (d = 0.5), and
• large (d ≥ 0.8).
• “A medium effect of .5 is visible to the naked eye of a

careful observer. A small effect of .2 is noticeably smaller than medium but not so small as to be trivial. A large effect of .8 is the same distance above the medium as small is below it.”

Power Calculations

https://youtu.be/P2Z6u8hwl2U

Practical Problem

→

cancer↓? cardiovascular disease↓?

Observational studies: Low levels of serum Vitamin D3 -> more cancer & cardiovascular disease