Journal Club (méthodologie) Introduction à la lecture d’articles médicaux (Partie 4)

Introduction à la lecture d’articles médicaux (4)

« How to Read a Clinical Trial Paper »

Dr. Ioannis KOKKINAKIS

Chef de Clinique

Service de Médecine Interne - Hôpital du Valais – Sion

05.10.2017

Page 2

m7da6csj6idm7xi/AAAKCvNlMtX2xsGS9xjmBqcHa

Qu’est-ce qu’on a déjà vu?

1. Grille de Lecture pour ERC

2. Autres Grilles des Lecture

3. Recommandations Int.

4. Types d’études

5. Etudes observationnels

6. Etudes expérimentales

7. Métanalyses

8. Rappel courte de la grille de lecture

9. PICOTS

10. Consort Statement

11. Rappels statistiques de base

•

Lien pour télécharger les présentations:

https://www.dropbox.com/sh/

?dl=0

Page 3

Qu’est-ce qu’on va voir …

• Comment lire un article médical

• Les pièges à éviter

Page 4

•

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3380258/

Page 5

Page 6

Are These Subjects Like My Patients?

• Subjects in the study are representative of the patients a

clinician sees in his practice?

• Were the study subjects sicker, or did they have more

complicated disease?

• Did they fail more prior medications?

• Did they have longer disease durations?

Shail, Higgins, Gastroenterology & Hepatology Volume 8, Issue 4 April 2012

Page 7

Page 8

Recruitment methods also influence the make-up of the study population and

can lead to different response rates.

Shail, Higgins, Gastroenterology & Hepatology Volume 8, Issue 4 April 2012

Page 9

Oxygen Therapy in Suspected Acute Myocardial Infarction N Engl J Med 2017; 377:1240-1249

Never mesure p-value for the baseline characteristics

Page 10

Multiplicity of analysis

Page 11

Acute

What Happened to the Subjects?

Did They Drop Out? Why?

Reasons drop out:

•

Lack of benefit (they were too sick

to stay in the study)

•

Side effects (which made them

want to leave the study)

•

Personal beliefs

•

Bias

Oxygen Therapy in Suspected

Myocardial Infarction N Engl J Med 2017; 377:1240-1249

Page 12

http://www.consort-statement.org

Page 13

Is the Study Design Biased?

• Focus:

– Comparator group

– Allocation of subjects to treatment arms

– Blinding

•

Comparator: standard-of-care therapy.

•

Functional disease states: placebo group and/or a comparator that has

proven to be effective (placebo effect measured)

•

Studies of rifaximin (Xifaxan, Salix Pharmaceuticals) versus placebo for

treatment of IBS,

•

32% of the placebo-treated patients - relief of symptoms

•

41% of subjects in the treatment arm

•

Without the placebo comparator, the effect of rifaximin on IBS symptoms

would appear quite large.

Shail, Higgins, Gastroenterology & Hepatology Volume 8, Issue 4 April 2012

Page 14

Is the Study Design Biased?

• Focus:

– Comparator group

– Allocation of subjects to treatment arms

– Blinding

•

CONSORT diagram should be easy to follow and similarly detailed for each

arm.

•

ALLOCATION CONCEALMENT

•

Investigators cannot anticipate which arm the next subject will enter.

•

If the investigators knew a very sick patient would be likely to receive

placebo, they could avoid enrolling that patient, which could result in biased

enrollment.

•

Randomization

– population level

– clustered at the site level (all the subjects at 1 site are in a single arm)

Shail, Higgins, Gastroenterology & Hepatology Volume 8, Issue 4 April 2012

Page 15

Is the Study Design Biased?

• Focus:

– Comparator group

– Allocation of subjects to treatment arms

– Blinding

•

Single-blind - subject unaware of the allocation

•

Doubleblind - investigators are also unaware of the patient’s allocation.

•

Patients who are aware of their intervention may have lower compliance,

prone to reporting bias.

•

Investigators who are aware of the intervention may overestimate a

treatment effect.

•

Studies comparing outcomes with and without blinding have shown a

significant overestimation of the treatment effect in studies conducted without

blinding

Shail, Higgins, Gastroenterology & Hepatology Volume 8, Issue 4 April 2012

Page 16

Study investigators rarely report whether the blinding process was successful.

•

In a 2007 analysis of 1,599 studies, only 2% reported the success of the

blinding process; of those, almost half determined that blinding was

unsuccessful (ie, patients could determine or guess whether they were

receiving active drug or placebo)

•

If patients in a placebo-controlled trial can deduce that they are in the placebo

arm, they could be less likely to report improvement, leading to underestimation

of the placebo effect and overestimation of the relative treatment effect.

Shail, Higgins, Gastroenterology & Hepatology Volume 8, Issue 4 April 2012

Page 17

Page 18

Does the Study Include an Intention-to-Treat Analysis?

Per protocol analysis… biais?

Page 19

Is This a Test of Superiority? Equivalence?

Noninferiority?

•

In cases where a new therapy is more convenient or cheaper, an equivalence

or noninferiority study can be appropriate.

•

The sample size needed to determine noninferiority or equivalence is always

much higher than the size needed to determine superiority.

•

If a study is initially designed to test for superiority but this outcome is not

found, switching to an equivalence study would be inappropriate for 2 reasons:

– the acceptable difference would not have been defined in advance

– the sample size would be too small.

Shail, Higgins, Gastroenterology & Hepatology Volume 8, Issue 4 April 2012

Page 20

Other

•

considerations…

Does the Measurement Matter?

Is It Reproducible? Accurate?

Dichotomous Outcomes, Continuous Outcomes, Correlations, Time-to-Event

Endpoints

How Are Missing Data Addressed?

– Nonresponder imputation: consider any subject who has missing data as

one who would have failed to meet the endpoint

– Lastobservation-carried-forward approach

– Imputation: other data are used to estimate what the missing data would

have been

Do the Design and Methods Conform to the Prestudy Protocol?

– Protocol registration

How big the effect is?

–

NNT

Page 21

Page 22