Skip to main content
. 2018 Jan 22;21(1):34–38. doi: 10.1136/eb-2017-102859

Table 5.

Comparing relapse rate under treatment B with that under treatment A using different analytical approaches to estimate causal ORs while accounting for confounding by sex, age and severity of symptoms

Analytical approach OR for relapse (B vs A) and 95% CI
Logistic regression for relapse including only hospital (no adjustment) 2.14 (1.45 to 3.17)
Logistic regression for relapse including sex, age and severity independently 0.98 (0.64 to 1.52)
Logistic regression for relapse including sex, age, severity with all 2-way interactions between sex, age and severity 1.0 (0.65 to 1.55)
IPTW weighted analysis with weights constructed with sex, age and severity independently in the model for the defining the weights 0.99 (0.65 to 1.51)
IPTW weighted analysis with weights constructed with all two-way interactions between sex, age and severity in the model for the defining the weights 1.0 (0.65 to 1.53)

IPTW, inverse probability of treatment weight.