Objective: To compare the influence of various statistical analysis approaches while assessing the marginal effect of polypharmacy (prescription of multiple psychotropics including a first-line antidepressant) on all-cause hospital admission among veterans diagnosed with post-traumatic stress disorder.
Methods: Data were obtained on 398 Iraq/Afghanistan veterans being followed in a southwestern U.S. health care system from October 2005 through September 2009, diagnosed with post-traumatic stress disorder and receiving first-line antidepressants (serotonin selective or serotonin norepinephrine reuptake inhibitors). High-dimensional propensity score (hd-PS) approaches were considered, examining both covariate adjustment per PS deciles and propensity weighting, with results compared to those of standard multivariable logistic regression.
Results: Veterans prescribed polypharmacy did not appear to have increased odds of admission in either the decile-adjusted hd-PS model (odds ratio [OR] = 2.1; 95% confidence interval [CI]: 0.9-4.9, not significant [ns]) or traditional covariate-adjusted logistic model (OR = 2.1; 0.9-5.0, ns). Inverse probability of treatment (OR = 2.1; 1.3-3.3) and standardized-morbidity-ratio-weighted (OR = 2.2; 1.3-3.6) hd-PS models estimated similar odds of admission with narrower CIs.
Conclusions: Future research should incorporate alternate analytical methods for observational data and investigate their respective biases relative to clinician treatment decision-making. For several analytical approaches, greater risk of admission among patients prescribed antidepressant-related polypharmacy was observed despite recommended guidelines, suggesting the need to investigate why clinicians may deviate from guidelines.
Reprint & Copyright © 2016 Association of Military Surgeons of the U.S.