Reviewing the Impact of Informatics on Substance Abuse Disorders and the Opioid Epidemic
Drug overdoses are now the top reason for unintentional mortality in the United States, and prescription opioid abuse is a major contributor to the public health crisis (Sun et al., 2018). The review explores the contributions of informatics in combating substance abuse disorders and the opioid epidemic. Substance abuse informatics incorporates availability and implementation of educational and preventative resources, analyzing associations and trends, identifying predictors, treatment outcomes, and prescription drug monitoring programs.
The review was conducted on articles identified by PubMed and Medline Ovid. Articles were chosen based on quality and relevance.
Educating both healthcare providers and the public has increased awareness of corrective drugs like Narcan® and awareness of available educational programs that may reduce the substance abuse related deaths disorders. The identification of key predictors of successful treatment and predictors of potential abusers is still being researched, but many studies have identified relevant factors associated with successful outcomes such as decreased overdose deaths, decreased SADs, and a decrease in appropriate prescription patterns. Through the use of available patient data, geographic systems, and informatics tools, associations and trends were analyzed and used to identify those with the highest risk of abuse and the greatest need for assistance. A review of treatment programs and outcomes revealed the need for greater participation in educational programs, improved patient management, and increased use of residential treatment facilities. Lastly, prescription drug monitoring programs require a higher rate of participation by healthcare providers and patients in order to be effective.
Informatics is constantly improving and positively contributing to research in efforts to reduce substance abuse. Despite the advances in informatics in recent years, there remains a need for discovery and incorporation of more effective uses of informatics to combat the substance abuse epidemic.
The review investigates recently published papers that assess preventative measures, treatment, and public health issues surrounding the substance abuse crisis. Current studies are producing promising results regarding where and to whom epidemics are occurring, educating providers and the public, identifying abusers before the problem becomes too severe to manage, and determining which treatments are the most effective. The review summarizes current findings and recommends future research in specific areas such as research on a national level, implementation of PDMP standards, mandatory healthcare provider participation, identifying predictors, and treatment outcomes to combat substance abuse disorders.
Substance abuse has become a significant issue affecting economic and social aspects of society. Due to the exponential increase in individuals with substance abuse disorders (SADs), researchers and healthcare professionals are working to prevent, manage, and successfully treat these disorders. According to the National Institute on Drug Abuse (NIDA), abuse of tobacco, alcohol, and illicit drugs is costly to our nation, exacting more than $740 billion annually in costs related to crime, lost work, productivity, and healthcare costs. Although tobacco use accounts for the majority of the economic burden, illicit drugs and opioids contribute to an overall cost of $271.5 billion as of 2013. Another alarming statistic published by NIDA estimated that 24.6 million Americans aged 12 or older (9.4% of the population) have used an illicit drug in the past month. Drug abuse is highest among people in their late teens and twenties but abuse is rapidly increasing among people in their fifties and early sixties. NIDA continues to address the large “”treatment gap in the United States, with an estimated 8.6% of Americans needing treatment for a problem related to drug or alcohol, but only 0.9% receiving help.
Biomedical informatics, a useful tool for understandings SADs, deals with storage, retrieval and optimal use of data, knowledge for discovery, problem solving, and decision-making. Informatics aids in the analysis of patient data (e.g., overdose rates, frequently used drugs, comorbidities, user demographics, and treatment outcomes). Clinical informatics created prescription drug monitoring programs that collect, monitor, and analyze the prescribing and dispensing of controlled substances. A basic literature search reveals a large number of recent patient data analyses that have allowed physicians and healthcare professionals to identify demographic predictors of future substance abusers. Frequently, healthcare providers are unable to intervene until the substance abuse has become so severe that a patient overdoses or serious social and financial repercussions occur. If healthcare providers identify risk factors earlier, the quality of patient care may improve. The funding for and amount of substance abuse research has significantly increased in the last few decades and has contributed to more effective treatment program characteristics and improved patient care quality. Effectively managing substance abuse disorders through the proper use of informatics, can improve patient care by offering effective treatment programs and preventative measures, decreasing the financial burden, and benefiting physicians.
The purpose of the review is to investigate and critique current research and the role that informatics plays regarding the education, prevention, management, and treatment outcomes of programs and studies focused on substance abuse disorders with an emphasis on prescription opioid abuse.
The review included articles selected from PubMed and Medline Ovid. PubMed is a database found on the National Center for Biotechnology Information, NCBI. PubMed was chosen because of its popularity, relevance, and its availability of a large number of recent articles and relevance to topic. Medline Ovid is the Medical Literature Analysis and Retrieval System Online. This database is compiled by the United States National Library of Medicine and promotes an availability of biomedical information.
Search results were narrowed to peer reviewed articles published in the last ten years (2008-current). Both clinical trials and reviews were included. Articles were grouped based on to relevance to education, identifying problem usage, treatment outcomes, and prescription drug monitoring programs. The driving question of this research was, “”What is the role of informatics in substance abuse disorders?. Articles were selected based on research methods, validity of results, and populations studied. Results were restricted to articles published in the English language and studies that included human subjects exclusively.
1. Substance Abuse Education and Naloxone Distribution
A geographic information system (GIS) was developed to help pharmacists effectively distribute naloxone (Narcan®), a lifesaving opioid antagonist that rapidly reverses opioid overdose. The study, conducted in Pennsylvania, used overdose death data and ZIP Code Tabulation Areas, (ZCTAs). The article stated that analysis will continue over the next five years to better understand the association between pharmacies that carry naloxone and opioid related deaths. The researchers reasoned that the number of overdose deaths in ZCTAs with naloxone-distributing pharmacies was significantly higher than the average number of deaths in all ZCTAs in Allegheny County: 7.38 deaths versus 4.84 deaths, respectively (P = 0.021) (Burrell et al., 2017). The article recommends further consideration of naloxone in opioid dependent populations. To prevent opioid related mortality, Olivia et al. (2017) summarized the development of a national opioid overdose education and Naloxone distribution program. The program focused on the facilitation of a health care system-based approach of distributing naloxone, developing patient and provider education resources, and implementation and evaluation resources. One hundred seventy-two overdose reversals were completed through this program. Through the use of informatics, a primarily community-based public health approach into a health care system-based approach to combat SADs, was the contributing factor in the success of the program.
In 2011, Kothari et al. used Medline to review studies of substance abuse disorders directed towards a population of future healthcare professionals: undergraduate medical students. Kothari et al. (2011) suggested improved methods of curriculum evaluation and publication guidelines. The Substance Abuse Research Education and Training Program (SARET), developed in 2012 by Truncali et al. aimed to stimulate current healthcare professionals to become more involved with and contribute to substance abuse research. A significant result of SARET was an interactive web-based module series on different research substance abuse topics. Results of focus groups and online surveys were analyzed and demonstrated that an online program enhanced interest in substance abuse professionals by 35-38%. In 2014, Bruckner et al. reviewed educational standards for substance abuse prevention programs in public schools. Prior to the publishing of this article, no research that systematically coded the educational standards across grades and states. The researchers found that the majority of states fell well below the recommended amount of instruction and overall, states vary widely in their content and standard coverage of substance abuse. The authors promote research that examines the relation among state alcohol, tobacco, and other drug use standards, classroom instruction, and adolescent drug use.
Informatics research continues to investigate many of the drugs involved in SADs. Informatics has significantly improved the education of substance abuse disorders and increased the distribution of naloxone through educational standard reviews, data analysis, and program implementation. In 2017, Cepeda et al. assessed the impact of a risk evaluation and mitigation strategy (REMS), for extended release and long-acting opioids. REMS evaluates prescribers, training courses, patient counseling and education, drug utilization, and surveillance monitoring. REMS consists of education, monitoring, proper medication disposal, and enforcement. The authors found that the implementation of REMS resulted in the leveling off and even decrease in opioid abuse. Led by the American Academy of Addiction Psychiatry, the Provider’s Clinical Support System for Medication Assisted Treatment initiative focused on training and mentoring health professionals in the treatment of opioid use disorders (Levin et al., 2016). The researchers aimed to increase evidence-based practices with medications for opioid use disorders. Levin et al. (2017) reviewed current initiatives that included training and mentoring for primary care physicians, outreach to multidisciplinary professional organizations, and creation of a resource portal for families, patients, and communities. The article recommends working with health care providers to offer medication assisted therapy (MAT).
2. Identifying Predictors of Problem Usage
Recognizing key predictors of successful treatment can serve to identify disparities, strengths, and weaknesses in service delivery, leading to treatment success and reducing unmet treatment needs (Arndt, 3013). In 2018, Acion et al. used a machine learning framework with multiple prediction models to determine SAD treatment success. Super learning (SL) is a methodology that facilitates this decision by combining all identified prediction algorithms pertinent for a particular prediction problem. SL generates a final model that is at least as good as any of the other models considered for predicting the outcome. The overarching aim of this work is to introduce SL to analysts and practitioners. Using the prediction model suggested by SL and TMLE, one could answer questions such as “”What is the treatment success rate difference between Hispanics with comorbid psychiatric disorders and those without comorbid disorders? and “”Is this difference different from zero?. Acion et al. (2018) found the most successful model was super learning, which combined all identified prediction algorithms pertinent for a specific problem. The authors compared the performance of logistic regression, penalized regression, random forests, deep learning neural networks, and super learning to predict treatment outcomes. The study used common statistical methods in a realistic setting with valid and applicable results. The data was limited to the population studied (Hispanics) but the results are a useful framework for future prediction model research. Future directions of this work include determining the benefits of applying targeted learning for different effect estimations and inference in the addiction field (Acion et al., 2018).
In 2015, Carrell et al. used natural language processing to identify prescription opioid usage. The researchers combined natural language processing and computer-assisted manual review of clinical notes to identify problem usage in current electronic health records. The study used a large number of patient records (22,142), and produced valid results that demonstrated that natural language processing efficiently and accurately identified evidence of opioid abuse. The methods may increase estimates by as much as one-third when compared to traditional methods.
The Center for Disease Control (CDC), recently published a study (Faul, 2017) on Methadone prescribing, overdose, and the association with Medicaid preferred drug list policies in the U.S.. Methadone is a prescribed controlled substance used in SAD treatment for those with an opioid addiction. The study aimed to compare the percentage of deaths involving Methadone with the rate of prescribing Methadone for pain, characterize variation in Methadone prescribing among payers and states, and assess whether an association existed between state Medicaid reimbursement preferred drug list (PDL) policies and Methadone overdose rates. Faul found that Methadone accounted for approximately 1% of all opioids prescribed for pain but accounted for approximately 23% of all prescription opioid deaths in 2014. State drug management practices and reimbursement policies can affect methadone prescribing practices and, in turn, might reduce methadone overdose rates within a state. Drug utilization management policies that reduce the use of risky opioids such as methadone might reduce opioid-related morbidity and mortality. This evidence of decreases in methadone overdoses and use of preferred drug list policies could serve as a model for future decreases in other specific opioid drug-related mortality. The article concludes that Methadone should not be the first choice for an extended release/long acting opioid which contradicts current belief and practice.
Hume et al. (2017) aimed to create accurate definitions and measurable factors to identify drug overdose poisonings in four different states. Four drug poisoning indicators were used (acute or chronic, drug or opioid) and were compared to principal diagnosis or all medical history. These authors found that studying the entire diagnosis and medical history was needed to fully capture the state burden of opioid poisonings.
Another study focused on creating a model for identifying life-threatening admissions and their associations with drug dependence (Nguyen et al., 2017). The model demonstrates an accurate prediction of death or ICU admission in hospitalized drug users. This is the first model created that allows for early identification of life-threatening cases in drug users admitted to acute care hospitals. The model used patients’ age, sex, entrance model, and diagnosis as predictors of an in-hospital death or ICU admission. Based on their validation cohort, the researchers discovered a method of identifying life-threatening admissions, which may support improved patient care management. The model can also serve as a routine tool to improve clinical decision-making. This model will ultimately improve the early management of critically ill patients, who increasingly contribute to the global financial burden of healthcare. The study incorporated a user-friendly method that is accessible to most healthcare providers and may benefit life-threatening drug abuse cases.
In addition to age, sex, entrance model, and diagnosis, psychosocial factors are also significant predictors in substance abuse. Psychosocial risk factors, including social determinants of health, mental health disorders, and history of substance abuse disorders, are less amenable to rapid and systematic data analyses. Psychosocial risk factors are not often collected or stored as formatted data. Due to US Health Insurance Portability and Accountability Act (HIPAA) regulations, this data is not available as claims data. Psychosocial factors in Electronic Health Record screening need to be integrated into healthcare and care delivery (Oreskovic et al., 2017). Medicaid and non-Medicaid patient data was used to identify 22 psychosocial electronic health record terms and compare them to patient outcomes. This amount of search terms is a significantly higher number of terms than other studies have used. The researchers concluded that selected informatics tools such as word recognition software could improve healthcare delivery through accurate identification of psychosocial risk factors in substance abuse.
Bioinformatics researchers, Way et al. (2017) evaluated alcohol dependence in populations with the genetic variant, ALDH1B1. While no significant allelic association was observed, individuals with this gene variant may be more likely to abuse drugs. The study focused on a biomedical factor while many other studies only considered social and demographic associations.
Lastly, Shah et al. (2017) reviewed rates and identified risk factors for opioid dependence and overdose after urological surgery. Through analyzing patient risk factors, these researchers found five patient risk factors associated with opioid dependence or overdose. Risk factors included younger age, inpatient surgery and increasing hospitalization duration, baseline depression, tobacco use and chronic obstructive pulmonary disease, insurance provider, including Medicaid, Medicare (age less than 65 years) and noninsured status.