The Epidemiology of Androgen Toxicity : A Six-year Retrospective Cohort Study of the Risk of Primary Health Outcomes among Inpatients with Androgen Toxicity in the United States / Scottie L. Howell. - Ann Arbor : ProQuest, 2019. - (ProQuest 27543963).
- Dissertation Manuscript presented to the Faculty of the College of Health Sciences of Trident University International). 17 October, 2019)
Abstract
Background:
Androgen prescriptions have increased substantially from 2000 to 2010. In response to the rise in prescriptions and a questionably credible cardiovascular risk study in 2013, the FDA issued mandates on testosterone labeling requirements to include consumer risk warnings. Despite the FDA mandates and well-established side effects, the reported risk estimates of health outcomes associated with androgen therapies remain equivocal across study populations with little evidence to characterize risk in an inpatient population. The epidemiological research detailed herein addressed this knowledge gap by identifying androgen toxicity types and classifying androgen-induced disease states from the literature into testable primary outcomes. As of 2019, no large-scale nationally representative studies have examined the risk of primary health outcomes with androgen toxicity among inpatients.
Objective:
The objective of this study was to characterize the epidemiology of androgen toxicity by examining the risk of primary health outcomes, incidence of inpatient variables, and mean healthcare costs in a nationally representative population of inpatients with and without androgen toxicity exposure.
Design:
A quantitative population-based retrospective cohort design was employed using National Inpatient Sample data spanning a period of six years (2010-2015) to identify an index cohort of 488 androgen toxicity exposures and a reference cohort of nearly 34 million nonexposures for the analysis.
Results:
In log-binomial GLM regression analysis adjusting for demographic characteristics, androgen toxicity exposure was associated with an increased risk of polycythemia (RR = 152.49, 95% CI [120.97, 192.22], p < .001), hypercoagulability (RR = 6.28, 95% CI [3.59, 10.98], p < .001), drug-induced liver injury (RR = 47.27, 95% CI [27.65, 80.81], p < .001), and venous thromboembolism (RR = 6.42, 95% CI [4.77, 8.63], p < .001), but not mortality (RR = 0.50, 95% CI [0.20, 1.19], p = .119). In negative binomial GLM regression analysis adjusting for demographic characteristics, exposure was associated with an increased incidence of chronic conditions (IRR = 1.20, 95% CI [1.14, 1.27], p < .001), diagnoses (IRR = 1.18, 95% CI [1.12, 1.23], p < .001), external causes of injury (IRR = 4.88, 95% CI [4.08, 5.83], p < .001), but not length of stay (IRR = 1.03, 95% CI [0.96, 1.12], p = .324). Exposure was also associated with a decreased incidence of procedures (IRR = 0.86, 95% CI [0.77, 0.96], p = .009). In log-gamma GLM regression analysis adjusting for demographic characteristics, exposure increased mean healthcare costs (Δ = 4,178.53, p = .206), although the effect was not statistically significant. An odds weighted, inverse probability of treatment weighted, and propensity score weighted analysis replicated that exposure increased the risk of polycythemia, hypercoagulability, druginduced liver injury, and venous thromboembolism, but not mortality. The weighted analysis also replicated that exposure increased the incidence of chronic conditions, diagnoses, and external causes of injury, but not length of stay. The decreased incidence of procedures found in the main adjusted negative binomial GLM analysis failed replication in the weighted analysis. Graphical trend analysis showed consistent year-to-year trends in relative risk estimates of mortality, hypercoagulability, drug-induced liver injury, and venous thromboembolism but not polycythemia. The incidence rate ratio estimates of length of stay, diagnoses, chronic conditions, external causes of injury, and procedures trended consistently from year-to-year. Mean healthcare costs dropped 14.24% from 2010 to 2011, trended upwards 52.55% from 2011 to 2014, and dropped 24.01% from 2014 to 2015.
Conclusion:
Androgen toxicity increased the risk of several primary health outcomes including polycythemia, hypercoagulability, drug-induced liver injury, and venous thromboembolism in this epidemiologic study. Given the importance of reducing inpatient risk factors and adverse health outcomes in U.S. hospital admissions, further exploration of androgen toxicity has extensive clinical and public health relevance.
