Statistical methods for assessing treatment effects on ordinal outcomes using observational data

In this article, we propose a marginal structural ordinal logistic regression model (MS-OLRM) to assess treatment effects on ordinal outcomes. Many statistical methods have been developed to estimate average treatment effect (ATE) when the outcome is continuous or binary. The methodology for assessing the effect of treatment for an ordinal outcome is less studied. To address this, we propose utilizing a superiority score as a measure of treatment effect, assessing whether the outcome under treatment is stochastically larger than the outcome under control. Our approach involves employing MS-OLRM in conjunction with Inverse Probability of Treatment Weighting (IPTW) to estimate the superiority score under treatment compared to the control. This methodology adjusts for confounding factors between treatment and outcome by utilizing IPTW, ensuring that all covariates are balanced among different treatment groups in the weighted sample. To assess the performance of the proposed method, we conduct extensive simulation studies. Finally, we apply the developed method to assess the treatment effects of medications and behavioral therapies on patients’ recovery from alcohol use disorders using the Kentucky Medicaid 2012–2019 database.