Research Methodology: Blinding Bias
1 Savović et al. 2012 — BRANDO: Blinding Bias in 1,973 Randomised Trials
Full Citation:: Savović J, Jones HE, Altman DG, et al. Influence of reported study design characteristics on intervention effect estimates from randomized, controlled trials. Annals of Internal Medicine. 2012;157(6):429–38. (Savović et al. 2012) DOI:: 10.7326/0003-4819-157-6-201209180-00537 Key Findings::
- Meta-epidemiological analysis (BRANDO collaboration): 1,973 RCTs within 234 meta-analyses
- Trials lacking double-blinding exaggerate intervention effects by average 13% (ROR 0.87)
- Bias driven entirely by trials with subjective outcomes; negligible in objective/mortality outcome trials
- Subjective-outcome trials without blinding: ~23% exaggeration (ROR ~0.77)
Relevance:: Primary BRANDO project publication. Establishes the quantitative basis for critiquing all unblinded ME/CFS behavioral intervention trials. Directly applicable to the 2025–2026 exercise meta-analysis cluster. Certainty Assessment::
- *Quality:* High — *Annals of Internal Medicine*; 7-dataset combined synthesis
- *Sample:* 1,973 RCTs across 234 meta-analyses
- *Replication:* Confirmed by Hróbjartsson 2014 and Pitre 2023
2 Hróbjartsson et al. 2014 — Within-Trial Quantification of Patient Blinding Bias
Full Citation:: Hróbjartsson A, Emanuelsson F, Thomsen ASS, Hilden J, Brorson S. Bias due to lack of patient blinding in clinical trials. International Journal of Epidemiology. 2014;43(4):1272–83. (Hróbjartsson et al. 2014) DOI:: 10.1093/ije/dyu115 Key Findings::
- 12 trials (n=3,869) using internal blinded/unblinded sub-study design
- Unblinded patients exaggerated treatment effect by mean SMD 0.56 (95% CI 0.41--0.71) for patient-reported outcomes
- No comparable exaggeration for observer-reported outcomes
Relevance:: Gold-standard quantification of patient-blinding bias. Zhao 2026’s fatigue SMD 0.85 is within the range explainable by this bias alone. Certainty Assessment::
- *Quality:* High — internal sub-study design; *International Journal of Epidemiology*
- *Sample:* n=3,869 across 12 trials
- *Limitations:* Acupuncture-heavy sample (10/12 trials)