Substance Use
Impact of Pre-admission Opioid Abuse on In-Hospital Mortality in Traumatic Brain Injury Patients: A Counterfactual and Mediation Analysis Sepideh Saroukhani* Sepideh Saroukhani Jie Liang Fatemeh Movaghari Pour Hongyin Lai Yuan Li Syeda Amna Kazmi Kelly West Rabail A. Chaudhry George W. Williams Vahed Maroufy
Pre-admission opioid abuse and dependence (POAD) have acute and chronic impacts on outcomes of patients with Traumatic brain injuries (TBIs). Based on traditional association models, previous research suggested an inverse association between POAD and in-hospital mortality in TBI patients. However, these models are limited in establishing unbiased estimates of direct effects, especially from retrospective data due to unmeasured or residual confounding. On the other hand, conducting prospective studies such as randomized clinical trials (RCTs), is neither ethical nor feasible for this type of research questions that involve opioid abuse as the exposure. Using a large retrospective observational dataset (N=455,843) from the National Inpatient Sample (NIS) between 2004-2015, we conducted counterfactual distribution analysis and mediation analysis with multiple mediators (MMA) to discern the direct and indirect effects of POAD on in-hospital post-TBI mortality. While POAD had no significant direct effect [Quantile estimate (95% confidence interval): – 0.07 (-0.15, 0.02), P = 0.15], other covariates and mediators had significant effects (P < 0.001) on in-hospital post-TBI mortality. Specifically, smoking, alcohol abuse, depression and anxiety showed the relative mediatory effect on in-hospital mortality [the relative mediation effect estimate (95% CI): 0.17 (0.14, 0.20), 0.14 (0.12, 0.17), 0.02 (0.017, 0.03) and 0.05 (0.047, 0.06), respectively] suggesting that if smoking, alcohol abused, depression and anxiety had the same distributions between POAD and non-POAD patients, the disparity of in-hospital mortality would have been reduced by 17%, 14%, 2% and 5% among the respective groups. These findings highlight potential avenues for enhancing clinical outcomes in TBI patients with POAD and reinforce the importance of employing statistical approaches to unveil potential causal relationships when randomized clinical trials are not feasible.