Table 3.

Proportion of participants experiencing at least one grade 3–5 adverse event by 52 weeks that was possibly, probably, or definitely related to study treatment using a modified intention-to-treat analysis for a model based on the raw dataset and imputed dataset

	Raw dataset	Imputed dataset
Prednisolone (n/N, %)	41/113 (36%)	40·0%*
Doxycycline (n/N, %)	22/121 (18%)	22·5%*
Adjusted† difference in proportions (%, 95% CI)	19·0% (7·9–30·1); p=0·001	18·4% (6·0–30·8); p=0·004
Unadjusted difference in proportions (%, 95% CI)	18·1% (6·9–29·3); p=0·002	17·5% (4·8–30·1); p=0·007

^*Estimates from unadjusted regression model on imputed dataset.

^†Estimates are from a regression model adjusted for baseline severity of bullous pemphigoid; however, Karnofsky score and age were omitted from the model due to model non-convergence with it present.