Different Types Of Bias In RCT And How To Avoid Them?
Evidence Based Dentistry
Done by : Mai Mohamed Ibrahim
Endo S1
Under supervision :DR. Fatma KoranyDifferent types of bias in RCT and how to avoid them?
Bias is any deviation of results from the truth. An RCT has fewer tendencies to bias than other study designs for assessing therapeutic interventions. There are at least seven important sources of bias in RCTs,
During data allocation:
Selection bias:
All participants in the study with true randomization in are given the identical chance to be allocated to each of the study groups. selection bias occur both when individuals are accepted or rejected for participation in a trial, and when the interventions are allocated to individuals when they have been accepted into the study . Inadequate allocation concealment can increase the estimate of the effect size of interventions by 40%. Good methods of allocation concealment include sequentially numbered, opaque, sealed envelopes; tamper-proof, sequentially numbered containers; pharmacy controlled lists; and telephone, fax, email, or internet contact with a central randomization office.
Ascertainment bias:
It Introduced by the one who administering the interventions, the person receiving the interventions, the investigator assessing or analyzing the outcomes, and even by those people who record the report describing the trial. Trials can be designed in order that the patient, the treatment team, the outcome assessor, and even the trial statistician and any data monitoring committee are all blinded to the allocated treatment. The more blinding that is achieved, the less biased the trial results should be.
Best services for writing your paper according to Trustpilot
During planning phase of RCT:
Choice of question bias:
One of forms of bias in an RCT is hidden is the choice of the question that the trial wish to answer.
Regulation bias:
It occurs when institutional review boards are overly restrictive, and block the study of important questions. It also occurs when they are overly permissive and allow or even encourage studies that may not be scientifically or socially valid, but may bring either funding or prestige to the institution.
Wrong design bias:
The wrong research design can produce misleading answers.
Missing data:
The more information from randomized patients that is missing, the more alert one should be of the trial results.
One way that trials can minimize the problem of missing patient information is to use central randomization and follow up.
Biases can occur during course of RCT:
Population choice bias:
The sample population can have a serious result on the generalizability of an RCT. Population choice may be limited when possibility of participants are approached or during registration of participants.
Intervention choice bias:
The stage at which an intervention is studied can be very important . This holds particularly true for surgical trials where there can be a learning curve bias for new operators or improvements in the techniques or contexts in which they are used. Similar concerns may hold for medical interventions, when dose or timing of a medication may be important determinants of the outcome.
Comparison choice (or control group) bias:
If a study compares an experimental intervention with a placebo control, the results will only tell us whether the intervention has a specific effect or not. It will not imply that the experimental intervention has a different or better effect than existing alternatives.
Outcome choice bias:
Sometimes RCTs evaluate outcomes that are easy to measure, rather than the outcomes that are relevant. Short-term outcomes are measured rather than the important long-term outcomes.
Counting death as a good outcome:
The trial may measure the proportion of patients who were disabled at follow up, using all patients randomized as the denominator. In this case, the trial is really comparing the proportion of patients who were alive and disabled at follow up to the proportion of patients who were not disabled at follow up, and this latter group includes both those who were alive and not disabled, and those who were dead. Thus death has been included as a good outcome. It would be more sensible to measure the proportion of patients who were alive and not disabled at follow up.
Biases can occur during Reporting of trail:
Withdrawal bias: bias introduced by inappropriate handling of withdrawals, drop outs, and protocol violations:
All participants in a trial should finish the study, follow the protocol, and afford data on all the outcomes of interest at all-time points. Data can be lost since some of the participants dismiss out before the end of the trial, as participants do not follow the protocol , or because some outcomes are not measured accurately or cannot be measured at all at one or more time points.
Selective reporting bias:
The investigator may even unconsciously be attracted more to certain outcomes than others. These have been called the social desirability bias in which the items that are desired, or the optimism bias in which the items hoped for, are more likely to be reported.
Fraud bias:
Intentional fraud is perhaps the most important, most serious, and most difficult to find out the source of bias. We wish that it is rare, but the extent to which fraudulent results are reported may be underestimated, and may be increasing under the pressure to produce results.
References:
S C Lewis, C P Warlow ,:How to spot bias and other potential problems in
randomized controlled trials. J Neurol Neurosurg Psychiatry 2004;75:181–187.
Alejandro R Jadad, Murray W Enkin, Randomized controlled trials : questions, answers, and musings, 2nd ed,2007. ISBN 978-1-4051-3266-4 (pbk. : alk. paper).
