|
 
 |
| EDITORIAL |
|
| Year : 2021 | Volume
: 12
| Issue : 4 | Page : 169-173 |
|
Blinding Assessment: One Step Forward
Jeehyoung Kim1, Jongbae J Park2, Heejung Bang3, Jafar Kolahi4
1 Department of Orthopedic Surgery, Seoul Sacred Heart General Hospital, Seoul, South Korea
2 Department of Anesthesiology, Duke University School of Medicine, Durham, NC, USA
3 Division of Biostatistics, Department of Public Health Sciences, University of California, Davis, CA; Center for Healthcare Policy and Research & Clinical and Translational Science Center, Davis School of Medicine, University of California, Sacramento, CA, USA
4 Independent Research Scientist, Founder of Dental Hypotheses, Isfahan, Iran
| Date of Submission | 25-Jul-2021 |
| Date of Decision | 17-Nov-2021 |
| Date of Acceptance | 25-Nov-2021 |
| Date of Web Publication | 21-Dec-2021 |
Correspondence Address:
Jeehyoung Kim
Department of Orthopedic Surgery, Seoul Sacred Heart General Hospital, Seoul, Postal Code: 02488
South Korea
 Source of Support: None, Conflict of Interest: None  | Check |
DOI: 10.4103/denthyp.denthyp_99_21
How to cite this article:
Kim J, Park JJ, Bang H, Kolahi J. Blinding Assessment: One Step Forward. Dent Hypotheses 2021;12:169-73 |
There is a current trend employing blinding or masking in entertainment television; namely, several popular music/voice contests in which musicians or singers perform without their face being observed by the judges. On the other hand, fans, fellow artists, or judges may believe that they could identify a performer’s musical or vocal style even if they were blindfolded or the performer hidden behind a mask or curtain, they are often completely surprised when the individual’s identity is revealed.
In medicine, blinding or masking (let us denote by “B”) could be an important component of randomized controlled (or clinical) trials (RCTs), where “R” signifies randomization (“a coin flip”), “C” stands for control (using a sham, placebo, or other comparator), and “T” suggests a treatment or intervention study (as opposed to an observational study). Blinding could be implemented in an RCT to reduce bias by withholding information relative to treatment allocation from subjects, investigators, outcome assessors, and/or data analysts involved in the study. “R” and “B” are complementary or interdependent; randomization and blinding both shield patients and researchers from the contamination of knowledge. Blinding is possible when randomization is employed, and randomization is successful under blind conditions.[1] Complete blinding can be difficult or infeasible to achieve due to logistics or cost, or is not always necessary, depending on study aim, design, comparators, or context/setting. However, partial blinding may be arguably preferable to no blinding; in fact, some consider blinding the default standard, modus operandi, or the perfect and necessary corollary for the RCTs.[1],[2] Using the example of the entertainment contest, in which a masked singer can see the judges whereas singer’s face is not visible to them, demonstrates that partial or single blinding may be appropriate and not compromise the outcome; the same might be claimed regarding “mortality” outcome in RCTs.
In this article, we briefly review the definition(s) and history of blinding in RCTs, current guidelines for its application, some important recent findings, and possible future directions.
The terms “blinding” or “masking” suggest that information is withheld from view, such as can be accomplished by a coded treatment assignment that is only revealed upon study completion or when strongly justified (e.g., based on study protocol or data safety issues/recommendations). As true blinding/masking relates to all five senses (such that treatment assignment cannot be known by sight, smell, taste, hearing, or touch), sight generally serves as a proxy for all the senses in the context of research. In the TV show of “the masked singer,” we mask the identity of the singers as a way of blinding the judges; so the objects are masked and the judges are blinded.
In the setting of RCTs, we might informally or loosely distinguish between these terms by using “masking” to refer to treatment distinguishability at randomization or at a short follow-up, and “blinding” to reflect what happened throughout a trial. Intuitively, the term “mask” may be preferable in studies of medical conditions involving vision, such as blindness; similarly, “blind” may be preferable in studies with mask as intervention (e.g., RCTs with coronavirus disease 2019 or other infectious diseases). There are reports that allocation concealment is more achievable and may be more important than blinding,[3] and that beyond subjects’ knowledge of their treatment assignment, their perceptions and beliefs can still bias reported outcomes which, in turn, can bias the estimated treatment effect.[4]
Adding to the discussion of standardized terms relative to blinding/masking are the concepts of “single-blind” and “double-blind”; as these terms are generally considered imperfect, too simplistic, or even misleading, they are very likely to remain in the lexicon in the RCT community and practice and beyond, despite repeated calls for improvement/revision. A number of articles addressed the need for better guidance on this topic.[3],[5],[6]
Various standard practice guidelines for clinical trials, including the Consolidated Standards of Reporting Trials (CONSORT), International Conference on Harmonization (ICH), and Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT), offer definitions of blinding and explain relevant issues around it. CONSORT treats blinding and masking almost synonymously, whereas ICH primarily uses the term “blinding” more broadly.[7]
In Appendix, key points regarding blinding are excerpted from CONSORT.
According to Wikipedia, the first blinded experiment could be the one conducted by the French Academy of Sciences in 1784 to investigate the claims of Mesmerism More Details as proposed by Franz Mesmer.[8]
In the experiment, researchers blindfolded mesmerists and asked them to identify objects that the experimenters had previously filled with “vital fluid.” The subjects were unable to do so. In 1817, the first blinded experiment recorded to have occurred outside of a scientific setting compared the musical quality of a Stradivarius violin to one with a guitar-like design. A violinist played each instrument while a committee of scientists and musicians listened from another room so as to avoid prejudice.
Interestingly, famous historic figures such as A. Lavoisier and B. Franklin utilized blinding to test therapeutic claims made for mesmerism.[8],[9]
According to Bhatt, the first double-blind controlled trial in modern medicine could be patulin for the common cold (1943).[10]
The Medical Research Council (MRC) in the UK carried out a trial in 1943-4 to investigate patulin treatment (an extract of Penicillium patulinum) for the common cold. This was the first double blind comparative trial with concurrent controls in the general population at that time. It was one of the last trials with non-randomized or quasi-randomized allocation of subjects. The MRC Patulin Clinical Trials Committee (1943) was chaired by Sir Harold Himsworth, and its statisticians were M. Greenwood and W.J. Martin. This nationwide study enrolled over one thousand British office and factory workers suffering from colds. This was quite a challenging endeavor in wartime.
The study was rigorously controlled by keeping the physician and the patient blinded to the treatment. The treatment allocation was done using an alternation procedure. A nurse allocated the treatment in strict rotation in a separate room. The nurse filed the record counterfoil separately and detached the code label for the appropriate bottle before asking the patient to visit the doctor. The statisticians considered this an effective random concurrent allocation. However, the outcome of the trial was disappointing as the analysis of trial data did not show any protective effect of patulin.
Kaptchuck provided a thorough and precious historical account on blinding and placebo control.[1] There is extensive literature on blinding techniques, and many trialists and researchers are reasonably familiar with methods used in their fields, such as independent person generating randomization schedule, use of opaque envelopes, applying the same color/appearance/smell to study treatment, and possibly generating similar side effects, for example, active placebo, especially when blinding is critical. Thus, the role and potential effects of placebo, placebo/nocebo, and blinding are highly interrelated [Figure 1].[11],[12],[13] | Figure 1 (a) Justice symbol and (b) Placebol/Nocebol. Source: Justice Goddess was downloaded from https://pixabay.com/, which provides images freely to the general public without requiring permission or copyright.
Click here to view |
Fast forward to 2000, CONSORT 2001 recommended reporting how the success of blinding was evaluated, and then this item was removed in the CONSORT 2010 guidance.[14] Quite unusually but understandably, a stand-alone paper was published on the single issue of this removal, which generated substantial controversy.[15],[16],[17] Some difficulties can arise when interpreting blinding data − including bidirectionality [between treatment effect size (ES) and proportion/degree of correct guess (or unblinding), or confounding]; the fact that correct guess is not always a bad thing; and when to assess, say, at the beginning and/or end of the trial.[18]
Meanwhile, over two decades blinding protocols and statistical methods such as the Blinding Index (BI) have been developed, popularized, and elucidated.[18],[19],[20],[21],[22],[23] The Template for Intervention Description and Replication (TIDieR)-Placebo guidelines in 2020 provide a reporting checklist, including blinding assessments and use of BIs in Item 13, designed to supplement CONSORT, SPIRIT, and TIDieR (all available at https://www.equator-network.org/).[12]
Conflicting meta-analyses on the topic of blinding have been recently published in two journals: the British Medical Journal (BMJ) (2020, accompanied by the Editor’s commentary); and Clinical Trials (CT; 2021).[24],[25],[26] They reached somewhat opposite conclusions regarding the importance and impact of blinding on clinical outcomes, with the former paper suggesting that blinding may not be as important as we claimed/believed, whereas the latter paper retained a belief in its value. The BMJ article used a meta-epidemiologic approach (also known as “meta of meta”) to evaluate various objective and subjective outcomes based on 142 meta-analyses (1153 trials). The CT article used more granular or raw data (individual participants’ guesses of the treatment they received), from 3899 participants in 40 back pain trials published in 2000 to 2019, utilizing BI and ES (e.g., Cohen-d) as a summary measure of potential (un)blinding and treatment effect, respectively. Thus, the dissimilar conclusions may be explained by differences in study aims, design, inclusion/exclusion criteria, and data collection, such that findings from both studies may be correct in their own right.
Given the disparate results in the recent BMJ and CT papers, perhaps the most appropriate stance is that blinding might be less important for objective outcomes, reducing the temptation to label unblinded trials as being of low quality/less rigorous (as is the case in some RCT quality scoring systems), but that it remains essential for subjective outcomes. Although our authors do not have supporting empirical data, it may be reasonable to consider blinding as important when substantial conflict(s) of interest (financial or intellectual) exists. In addition, this may be an opportune time to discuss the importance of “blinding for data analysts” who are commonly unblinded and of “blinding to study hypothesis,” whenever feasible. Nowadays, in modern RCTs that employ advanced or complex (e.g., adaptive) trial designs or test high-tech devices (including e-health or m-health), blinding is easily waived or bypassed, often with a weak control treatment selected, which may lead to “easy win.”
In any comparative evaluation, we recommend blinding whenever feasible and assessing whenever willing. The element of voluntariness or willingness may be the salient recommendation about blinding, such that randomization and control can be in place without blinding; indeed, that is why “R” and “C” are in RCT, not “B.” Honest and willing reporters of trial results (particularly, not looking-good data) should not be punished. Correct guess would be an ideal phenomenon if treatment really works and beats a well-chosen control, and more blinding data (collected in standardized format) could facilitate subsequent meta-analyses. Recently developed packages or codes in Stata, Excel, and R could be useful to those volunteerers.[27],[28],[29],[30],[31] [Figure 2] illustrates recent research activities on blinding assessment through published articles and bibliography − active researchers, jointly used terminologies, journals, and international collaborations. | Figure 2 Recent research activities on blinding assessment through published articles and bibliography: (a) active researchers, (b) hot topics, (c) journals, and (d) international collaborations. SCOPUS search queries were “blinding index” OR “blinding success” OR “blinding assessment”.
Click here to view |
A take-home message for providers as well as consumers of medical/healthcare product or information may be: whenever you invent something or evaluate something/someone, you might ask yourself, “Can I pass a blind test (real or hypothetical)?” A world class pianist who did not fully succeed in distinguishing famous/expensive pianos versus a lot cheaper ones in a blind test admitted honestly and humorously “Since I failed in this blind test, one can buy this piano which must be very cost-effective.”[32] Indeed, the Paris Double-Blind Violin Experiment in 2012 of old Italian violins (including Stradivarius) versus new violins yielded BI = −0.09 (95% confidence interval: −0.43 to 0.25) for the new and 0.14 (−0.16 to 0.44) for the old![33],[34] This blinding data may reflect “Random guess” or “Wishful thinking” scenario…[35]
Acknowledgments
The authors thank Caron Modeas, Evolved Editing, LLC, for English editing and Drs Phil Pfeifer, David Spence, and Allan Jaffe for reading our manuscript and stimulating and insightful discussions and advice.
Financial support and sponsorship
HB was partly supported by the National Institutes of Health through grants UL1 TR001860 and R01 AR076088. The funding source had no role in the study design or implementation.
Conflicts of interest
There are no conflicts of interest.
Supplementary Material
Appendix: CONSORT on blinding
http://www.consort-statement.org/checklists/view/32–consort-2010/93-blinding
11a. Blinding
If done, who was blinded after assignment to interventions (for example, participants, care providers, those assessing outcomes) and how
11b. If relevant, description of the similarity of interventions
The term “blinding” or “masking” refers to withholding information about the assigned interventions from people involved in the trial who may potentially be influenced by this knowledge. Blinding is an important safeguard against bias, particularly when assessing subjective outcomes.
http://www.consort-statement.org/resources/glossary
Blinding (masking) − The practice of keeping the trial participants, care providers, those collecting data, and sometimes even those analyzing data unaware of which intervention is being administered to which participant. Blinding is intended to prevent bias on the part of study personnel. The most common application is “double-blinding,” in which participants, caregivers and those assessing outcome are blinded to intervention assignment. The term “masking” may be used instead of blinding.
http://www.consort-statement.org/checklists/view/62786-social-and-psychological-interventions/1232-randomisation-allocation-concealment-mechanism
Mechanism used to implement the random allocation sequence, describing any steps taken to conceal the sequence until interventions were assigned
In addition to generating a truly random sequence (Item 8a), researchers should conceal the sequence to prevent foreknowledge of the intervention assignment by persons enrolling and assigning participants. Otherwise, recruitment and allocation could be affected by knowledge of the next assignment. Authors should report whether and how allocation was concealed. When allocation was concealed, authors should describe the mechanism and how this mechanism was monitored to avoid tampering or subversion (e.g. centralised or ’third–party’ assignment, automated assignment system, sequentially numbered identical containers, sealed opaque envelopes). While masking (blinding) is not always possible, allocation concealment is always possible.
http://www.consort-statement.org/checklists/view/648-non-pharmacologic-treatment/679-blinding
11a. If done, who was blinded after assignment to interventions (e.g., participants, care providers, those administering co-interventions, those assessing outcomes) and how
11c. If blinding was not possible, description of any attempts to limit bias
| References | |  |
| 1. | Kaptchuck TJ. Intentional ignorance: a history of blind assessment and placebo controls in medicine. Bull Hist Med 1998;72:389-433.  |
| 2. | Drucker A, Chan A-W. Blindsided: challenging the dogma of masking in clinical trials. BMJ 2020;368:m229. |
| 3. | Schulz KF, Chalmers I, Altman DG. The landscape and lexicon of blinding in randomized trials. Ann Intern Med 2002;136:254-59. |
| 4. | Mathieu E, Herbert R, McGeechan K et al. A theoretical analysis showed that blinding cannot eliminate potential for bias associated with beliefs about allocation in randomized clinical trials. J Clin Epidemiol 2014;67:667-71. |
| 5. | Park J, White AR, Stevinson C et al. Who are we blinding? A systematic review of blinded clinical trials. Perfusion 2001;14:296-304. |
| 6. | Lang T, Stroup D. Who knew? The misleading specificity of “double-blind” and what to do about it. Trials 2020;21:697. |
| 7. | International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use (ICH). ICH Harmonised Guideline. Integrated Addendum to ICH E6(R1): Guideline for Good Clinical Practice E6(R2), Current Step 4 Version Dated July 29, 2016. |
| 8. | Donaldson IML. Mesmer’s1780 proposal for a controlled trial to test his method of treatment using “animal magnetism”. J R Soc Med 2005;98:572-5. |
| 9. | |
| 10. | Bhatt A. Evolution of clinical research: a history before and beyond James Lind. Perspect Clin Res 2010;1:6-10. [ PUBMED] [Full text] |
| 11. | Demasi M, Jefferson T. Placebo—the unknown variable in a controlled trial. JAMA Int Med 2021;181:577-8. |
| 12. | Howick J, Webster R, Rees J et al. TIDieR-Placebo: a guide and checklist for reporting placebo and sham controls. PLoS Med 2020;17:e1003294. |
| 13. | Gowdey C, Hamilton J, Philp R. A controlled clinical trial using placebos in normal subjects: a teaching exercise. CMAJ 1967;96:1317-22. |
| 14. | Schulz KF, Altman DG, Moher D et al. CONSORT 2010 changes and testing blindness in RCTs. Lancet 2010;375:1144-46. |
| 15. | Kolahi J, Bang H, Park J et al. CONSORT 2010 and controversies regarding assessment of blindness in RCTs. Dent Hypotheses 2010;1:99-105. [Full text] |
| 16. | Hopton AK, Macpherson H. Assessing blinding in randomised controlled trials of acupuncture: challenges and recommendations. Chin J Integr Med 2011;17:173-6. |
| 17. | Colagiuri B. Testing for blinding is the only way to determine whether a trial is blind. BMJ 2010;340:c332. |
| 18. | Bang H, Flaherty SP, Kolahi J et al. Blinding assessment in clinical trials: a review of statistical methods and a proposal of blinding assessment protocol. Clin Res Regul Aff 2010;27:42-51. |
| 19. | Bang H, Ni L, Davis CE. Assessment of blinding in clinical trials. Control Clin Trials 2004;25:143-56. |
| 20. | James KE, Bloch DA, Lee KK et al. An index for assessing blindness in a multi-centre clinical trial: disulfiram for alcohol cessation − a VA cooperative study. Stat Med 1996;15:1421-34. |
| 21. | Landsman V, Landsman D, Li C et al. Overdispersion models for correlated multinomial data: applications to blinding assessment. Stat Med 2019;38:4963-76. |
| 22. | Landsman V, Fillery M, Vernon H et al. Sample size calculations for blinding assessment. J Biopharm Stat 2018;28:857-69. |
| 23. | Kolahi J, Abrishami M. Multiple-blind: towards a new blinding protocol for future generations of clinical trials. Med Hypotheses 2009;73:843-5. |
| 24. | Moustgaard H, Clayton G, Jones H et al. Impact of blinding on estimated treatment effects in randomised clinical trials: meta-epidemiological study. BMJ 2020;368:16802. |
| 25. | Godlee F. Blinding may be unnecessary, but please divest. BMJ 2020;368:m255. |
| 26. | Freed B, Williams B, Situ X et al. Blinding, sham, and treatment effects in randomized controlled trials for back pain in 2000–2019: a review and meta-analytic approach. Clin Trials 2021;18:361-70. |
| 27. | |
| 28. | Chen J. BLINDING: Stata module to compute blinding indexes. Statistical Software Components S456898, Boston College Department of Economics; 2008. |
| 29. | |
| 30. | |
| 31. | |
| 32. | |
| 33. | |
| 34. | Fritz C, Curtin J, Poitevineau J et al. Soloist evaluations of six old Italian and six new violins. PNAS 2014;111:7224-29. |
| 35. | Bang H. Random guess and wishful thinking are the best blinding scenarios. Contemp Clin Trials Commun 2016;3:117-21. |
[Figure 1], [Figure 2]
|
Search Pubmed for
