Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Section B

Assessing the Risk of Bias in Clinical Trials:


Selection Bias

The material in this video is subject to the copyright of the owners of the material and is being provided for educational purposes under
rules of fair use for registered students in this course only. No additional copies of the copyrighted work may be made or distributed.
Sources of Bias in an RCT

2
Random Sequence Generation

! Occurs at the start of a trial before allocation of participants

! Determines a random order of assigning people into intervention and control groups

! Accounts for known and unknown confounders

! Prevents systematic differences between groups

! " Prevents selection bias

3
Studies of Coronary Artery Bypass Graft Surgery

Pooled differences in mortality between medically and surgically treated patient,


by study design
Mean
Mean % Mean %
Number of difference
Type of study mortality mortality S.D.
studies in %
(medical) (surgical)
mortality
Randomized
controlled 8 12.0% 7.6% 4.4% 5.5%
trials
Quasi- 13 28.5% 14.7% 13.8% 16.9%
experimental

Source: Wortman, P.M. & Yeaton, W.H. (1983). Eval Stud Review Ann. 4
Does Randomization Protect against Bias?

! In studies comparing randomized and non-randomized trials of the same intervention,


it was not possible to predict the magnitude or direction of findings based on study
design

Source: (2011). Odgaard-Jensen CDSR. 5


Random Sequence Generation

! Low risk of bias—unpredictable


! Random numbers table
! Computer random number generator
! Stratified or block randomization
! Coin toss, shuffling cards or envelopes, throwing dice

! High risk of bias—predictable


! Quasi-random—date of birth, day of visit, ID or record number, alternate allocation
! Non-random—choice of clinician or participant, test results, availability

See Section 8.9 of the Cochrane Handbook 6


Allocation Concealment

! Occurs at the start of the trial during allocation of


participants

! When a person is recruited to a study, no one is able to


predict which group the person will be allocated to

! Ensures strict implementation of the randomly generated


sequence
! Prevents changing the order
! Prevents selecting who to recruit

! " Prevents selection bias

7
Trials Where Randomization Is Inadequately Concealed Are More Likely to
Show a Beneficial Effect

Source: Pildal, J. et al. (2008). Int. J. Epidemiol, 37:422; doi:10.1093/ije/dyn046 8


Comparison No. of Ratio of Ratio of odds P value of Variability in
(No of meta-analyses trials* odds ratios ratios (95% CI) test of bias¶ (P value)
interaction
Overall (76) 314 v 432
0.93 (0.83 to 1.04) -- 0.11 (<0.001)
All cause mortality (18) 79 v 121
Other outcomes (58) 235 v 311 1.04 (0.95 to 1.14) 0.011 0.01 (0.27)
0.83 (0.70 to 0.98) 0.18 (<0.001)
Objective outcomes (44) 210 v 227
Subjective outcomes (32) 104 v 205 1.01 (0.92 to 1.10) 0.01 0.08 (<0.001)
0.75 (0.61 to 0.82) 0.14 (0.001)
Drug intervention (57) 250 v 372
Other intervention (19) 64 v 60 0.92 (0.81 to 1.05) 0.66 0.10 (<0.001)
1.00 (0.71 to 1.39) 0.22 (0.003)
0.5 0.75 1 1.5 2
*Non-blinded v blinded Non-blinded Non-blinded
¶ Between-meta-analysis more less
heterogeneity variance beneficial beneficial

Ratios of odds ratios comparing intervention effect estimates in 314 non-blinded trials versus 432
blinded trials. Columns on the right show P values from tests of interaction, and estimates of the
variability between meta-analyses in the size of bias
Adapted from Wood, L. et al. (Mar 3 2008). BMJ. 9
Allocation Concealment

! Low risk of bias—unpredictable


! Central allocation (phone, web, pharmacy)
! Sequentially numbered, sealed, opaque envelopes
! Sequentially numbered, identical drug containers

! High risk of bias—predictable


! Random sequence known to staff in advance
! Envelopes or packaging without all safeguards
! Non-random, predictable sequence

See Section 8.10 of the Cochrane Handbook 10

You might also like