Spin, Blinding and Dropout as Red Flags of Bias in Trials

Media stories favor dramatic findings, university press releases can overstate results, and researchers naturally hunt for positive results. Isabel Boutron's group showed that spin, concluding positively despite no effect on the primary outcome, is common in trial reports. Since many readers stop at the last line of the abstract, where spin tends to sit, Neil urges reading the methods and results, the sections people admit to reading least. Trial biases tend to produce falsely positive or inflated results. Check that allocation was truly random and concealed from those recruiting participants. Blinding therapists is rarely possible outside interventions such as detuned ultrasound, so effect sizes include therapist and patient enthusiasm, but outcome assessors can usually be blinded. Dropout is not random: people leave because of harm, side effects or a treatment that does not fit their lives. Look for how many dropped out, whether it was balanced and whether an intention-to-treat analysis was used. Watch for selective outcome reporting by comparing the paper with its trial registration or published protocol; an unregistered trial is reasonable to set aside.