Date of Award


Degree Type


Degree Name

Doctor of Philosophy in Pharmaceutical Sciences


Biomedical and Pharmaceutical Sciences

First Advisor

Stephen J. Kogut


Approximately 50 million adults in the United States (U.S.) suffered from chronic pain in 2016, with almost 20 million of whom had high impact pain that interfered with work or daily life. In 2017, there were more than 75 million dispensed prescriptions of opioid medications to the Medicare Part D population alone. Fatal overdoses involving opioids, including heroin and illicit fentanyl, have been increasing annually in the U.S., from an estimated 25,052 in 2013 to over 50,000 in 2019; however, prescription opioids make up about one-third of these overdose fatalities. While opioid prescribing has received increased attention in recent years, all prescription controlled substances have potential adverse risks and require judicious prescribing to improve patient outcomes. Concurrent use of prescription opioids and other prescription controlled substances significantly increases risks of adverse outcomes such as respiratory depression and death. Benzodiazepines have an increased risk of falls leading to fractures and hospitalizations. In addition, all prescription controlled substances carry an inherent risk of addiction, which is evidenced in their classification as controlled substances by the U.S. Drug Enforcement Agency (DEA).

Prescription controlled substance utilization is an important and challenging concept for clinicians to consider when treating patients. Controlled substance prescriptions have indications for treating several disease states such as pain, anxiety, epilepsy, attention deficit disorder, and insomnia, among many other conditions. While these medications can have benefits for patients when used as prescribed, they also have important safety risks. By creation of the U.S DEA regulations, all controlled substance prescriptions inherently carry a risk of addiction, with risk of addiction increasing as the controlled substance classification increases from schedule-V to schedule-II. In addition, there are a myriad of additional risks specific to each medication class. Assessment of prescription controlled substance utilization on a population level can help to identify areas of opportunity where interventions can be made in the prescribing and utilization of controlled substances, leading to an improvement in patient health outcomes.

Benzodiazepines are one of the most prescribed medication classes with three benzodiazepines being in the top 55 prescribed drugs in 2017 in the U.S. This class of medications have important safety considerations including an increased risk of dependence and addiction, falls, and death from opioid overdose. Although benzodiazepine safety and prescribing encompasses a rich and important research area, there is a lack of pharmacoepidemiologic literature addressing benzodiazepine dosing intensity in real-world settings. Opioid prescribing is often assessed by dose intensity via standardized morphine milligram equivalents (MME) for medication safety to improve health outcomes and reduce the risk of opioid overdose. Measuring a standardized benzodiazepine dose with a diazepam milligram equivalents approach could be of similar importance to help reduce adverse outcomes and promote safer prescribing and utilization.

In August of 2014, tramadol was classified as a Schedule-IV controlled substance by the U.S. Drug Enforcement Administration on a national level (although it was classified as a Schedule-IV controlled substance by 12 states prior to this). Even in light of it becoming a controlled substance on the national level, tramadol’s prescribing has significantly increased in recent years while Schedule-II opioid prescribing has decreased. Tramadol has numerous drug-drug interactions as well as warnings associated with its prescribing.

The purpose of the studies in this dissertation was to better understand prescription controlled substance utilization and highlight areas that pose the greatest opportunity for improvement. We examined 2 data sources: (1): the Rhode Island Prescription Drug Monitoring Program (PDMP) data set for the calendar year 2018 for manuscripts 1 and 2, and (2) a large commercial claims database in the U.S. for the calendar years 2016 and 2017 for manuscript 3.

Manuscript 1: The objective of this study was to develop and apply a standardized benzodiazepine milligram equivalency conversion algorithm and assess the dosing intensity of benzodiazepine prescriptions dispensed in Rhode Island (RI) in 2018. We created a benzodiazepine dosing equivalency algorithm taking into account drug half-life and dosing recommendations listed in the FDA prescribing information to create a standardized diazepam milligram equivalency (DME). We then assessed prescribed benzodiazepine dose intensity in RI by applying the algorithm to data from the RI Prescription Drug Monitoring Program (PDMP) for the calendar year of 2018. The RI PDMP includes controlled substances dispensed by all pharmacies with a retail license in RI as well as a controlled substance registration with the state. In addition, information on prescription controlled substances dispensed to RI-residents by pharmacies in neighboring states also is included. We then assessed the impact of different demographic and clinical covariates on benzodiazepine dose intensity. We performed an ordinary least squares (OLS) multiple regression analysis to determine which independent variables were most predictive of DME, with a log transformation used to address skewness of the dependent variable. These analyses determined that compared to patients who did not utilize opioids, those concurrently utilizing opioids had a statistically significantly higher mean daily DME, and patients with a longer duration of concurrent utilization had significantly higher mean DME per day. We also found that patients concurrently taking other prescription controlled medications such as buprenorphine or stimulants had statistically significantly higher mean daily DME compared to those without concurrent utilization of those medications.

Manuscript 2: The objective of this study was to develop a population-based controlled medication utilization measurement framework that uses pharmacy data solely; and subsequently to pilot test its application using “real-world” data. We performed an environmental scan, reviewed the literature, and consulted with clinical experts to locate existing measures and/or develop new measures related to prescription controlled medication utilization. In total, the measurement framework contained 16 measures that represented four domains: opioid utilization (five measures), benzodiazepine utilization (five measures), care coordination (two measures), and buprenorphine (four measures). We then applied the framework we created to data from the RI PDMP for the calendar year 2018, which was the same dataset used for manuscript 1. Measure rates were calculated for the overall population as well as for different covariates. Multivariate logistic regression models were developed for each dichotomous quality measure to assess differences across covariates. We hypothesized that utilization of prescribed controlled substances in RI differs from optimal use as defined by the quality measures evaluated in the framework, which is what we observed for most quality measures.

Manuscript 3: The objective of this study was to assess prescribing of tramadol among patients with contraindications and higher risks of adverse events in a large commercially insured population in the U.S. The data source used for this study was Optum’s de-identified Clinformatics® Data Mart Database for the calendar years 2016 through 2017. This data source is a commercial claims database representing approximately 35 million unique patients in the U.S. Patients who received at least one paid claim for tramadol during this time period were included in the analysis unless they had a diagnosis for either cancer (n=158,884) or sickle cell disease (n=586), or were missing demographic information (n=3,087). We assessed several categories of contraindications or higher risk utilization: children under the age of 18; patients with epilepsy or seizure disorders; and concurrent use with interacting medications, including other high risk serotonergic medications; medications that are strong or moderate inhibitors of CYP 2D6 inhibitors and/or CYP 3A4 inhibitors or inducers; concurrent use with other non-tramadol opioids; concurrent use with benzodiazepines; concurrent use with non-benzodiazepine sedative hypnotics; and a composite that included any of these high risk prescribing circumstances. Measure rates were calculated for the overall population as well as for different covariates. Multivariate logistic regression models were developed for each measure of higher risk utilization to assess differences across covariates. We hypothesized that tramadol is often prescribed to patients who have a higher risk of adverse outcomes, which is what we observed with 31.17% (99% Confidence Intervals: 31.06-31.27) of patients with at least one tramadol prescription included in the composite outcome of any higher risk utilization.



To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.