Study design

Question 1

how can research lead to wrong conclusion? (4 ways)

Answer 1

• chance
• bias
• confounding
• fraud

Question 2

3 basic principles of emperic research

Answer 2

1) ceteris paribus: counterfactual model, je moet een zo gelijk mogelijke vergelijking maken (controle groep). 2) Occams Razor: choose the hypothesis with the fewest assumption, 3) Hume’s problem: emperical proof is not the same as mathematical proof, it is impossible in emperical science.

Question 3

experimental vs observational studies

Answer 3

experimental: research manipulate exposure. observational studies: only observerd by researchers

Question 4

case control vs cohort

Answer 4

CC: subjects are characterised by outcome and exposure status is determined

cohort: opposite.

Question 5

Why randomize?

Answer 5

esscence: no bias by differential allocation and groups are equal in relevant factors, ie, that determine outcome

Guarantees: no subjectivity in treatment allocation, all differneces due to random variaton

Question 6

why double blind?

Answer 6

avoids post randomisation differences

co-medication, life style, compliance, endpoint evaluation, endpoint reporting

Question 7

why controlled

Answer 7

Zonder controle groep: subjective effects (placebo effect), objective effects(circadian rythm), regression tot the mean, blinding is difficult

Question 8

what is regression to the mean?

Answer 8

natural course, natuurlijk beloop. Ongeacht van interventie zullen meesten zaken zichzelf oplossen

Question 9

Randomiseren in etiologische studies niet makkelijk of nodig want..

Answer 9

• randomiseren niet mogelijk: genetica (vastgelegd bij geboorte)
• niet haalbaar/ethish: bijwerkingen
• onnodig: evaluaren unintended effects (genetica en bijwerkingen)

NB: voor bestuderen van intended effect moet je altijd gerandomiseerde studie doen.

Question 10

Studies naar bijwerking bijzonderheden

Answer 10

• niet makkelijk in gerandomiseerde studies: bw te zeldzaam, patient selectie strict
• randomiseren is niet nodig: in therapeutische studies moet de link tussen behandeling en prognose verbroken worden. deze is er niet bij etiologie (unexpected outcome en unintended outcome)

Question 11

wat is risico van het gebruike van “prevalent” cases ipv “incident cases”

Answer 11

incidence - prevalence bias. those who live longer have a higher chance to be included. they may be different. no problem if exposure not related to disease duration and death risk

Question 12

Case control: indications and contra-indications

Answer 12

• exposition status known before disease occurs
• efficient in time and mony
• only option for rare diseases and transient risk factors
• NOT APPRoPRIATE for rare exposures
• it has all the limitations of observational studies for theapeutic research questions
• no absoulte risks: cummulative incidence or incidens rates

Question 13

When a study is not a randomised experiment ..

Answer 13

.. then it is invariably an observational study

Question 14

experimental studies are always

Answer 14

prospective cohort studies