Developing guidance for a risk-proportionate approach to blinding statisticians within clinical trials: a mixed methods study

Table 5 Primary analysis—odds ratios (OR) describing the influence of the blinding status of the statistician and other selected study design features on whether a statistically significant finding is reported

Blinding status of the statistician known
N = 152	OR (95% CI)^a	p-value
Statistician not blinded	1.02 (0.49, 2.13)	0.958
Multiple comparisons^b	1.34 (0.63, 2.86)	0.450
Target sample size achieved	1.15 (0.51, 2.62)	0.736
Trial not blinded^c	3.00 (1.16, 7.78)	0.024
Imputation of missing blinding status of the statistician^d
N = 179	OR (95% CI)^a	p-value
Statistician not blinded	1.04 (0.52, 2.08)	0.907
Multiple comparisons^b	1.38 (0.69, 2.78)	0.360
Target sample size achieved	1.40 (0.66, 2.95)	0.383
Trial not blinded^c	2.76 (1.14, 6.69)	0.025

^aOdds ratio (and 95% confidence interval) from logistic regression model comparing the influence of the listed study design features on the likelihood of a statistically significant finding
^bA trial is defined as having multiple comparisons if it contains more than two treatment groups or multiple co-primary outcomes
^cA trial is defined as not blinded if any of the participants, clinicians and outcome assessors are not blinded. To facilitate inclusion in the model wherever the blinding status is unclear, this was assumed as not blinded
^dWhere blinding status is unclear it was assumed the statistician was not blinded

ISSN: 1745-6215