Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2019 Aug 24.
Published in final edited form as: Subst Use Misuse. 2018 Jan 5;53(10):1591–1601. doi: 10.1080/10826084.2017.1416407

High-Risk Prescribing to Medicaid Enrollees Receiving Opioid Analgesics: Individual- and County-Level Factors

Sara E Heins a,b, Mark J Sorbero a, Christopher M Jones c, Andrew W Dick a, Bradley D Stein a,d
PMCID: PMC6033690  NIHMSID: NIHMS944788  PMID: 29303393

Abstract

Background

Prescription opioid overdoses have increased dramatically in recent years, with the highest rates among Medicaid enrollees. High-risk prescribing includes practices associated with overdoses and a range of additional opioid-related problems.

Objectives

To identify individual- and county-level factors associated with high-risk prescribing among Medicaid enrollees receiving opioids.

Methods

In a four-states, cross-sectional claims data study, Medicaid enrollees 18–64 years old with a new opioid analgesic treatment episode 2007–2009 were identified. Multivariate regression analyses were conducted to identify factors associated with high-risk prescribing, defined as high-dose opioid prescribing (morphine equivalent daily dose ≥100 mg for >6 days), opioid overlap, opioid-benzodiazepine overlap.

Results

High-risk prescribing occurred in 39.4% of episodes. Older age, rural county of residence, white race, and major depression diagnosis were associated with higher rates of all types of high-risk prescribing. Individuals with prior opioid, alcohol, and hypnotic/sedative use disorder diagnoses had lower odds of high-dose opioid prescribing but higher odds of opioid overlap and opioid-benzodiazepine overlap than individuals without such disorders. High-dose opioid prescribing in Massachusetts was less common than in California, Illinois, and New York, whereas the rate of benzodiazepine overlap in Massachusetts was more common than in other states.

Conclusions/Importance

High-risk prescribing was common and associated with several important demographic, clinical, and community factors. Findings can be used to inform targeted interventions designed to reduce such prescribing, and given state variation observed, further research is needed to better understand the effects of state policies on high-risk prescribing.

Keywords: Opioid analgesics, Medicaid, county characteristics, individual characteristics, depression, benzodiazepines

Introduction

Opioid addiction and overdoses are significant public health issues, with prescription opioid use and associated deaths increasing dramatically in recent decades (Rudd, Aleshire, Zibbell, & Gladden, 2016). At the federal level, multiple efforts to improve opioid prescribing practices include establishing prescribing guidelines for chronic noncancer pain (Dowell, Haegerich, & Chou, 2016), eliminating payments tied to patient survey questions about pain control that might unintentionally incentivize over-prescription of opioids (Centers for Medicaid and Medicare Services, n.d.), and supporting Prescription Drug Monitoring Programs (PDMPs), which by 2017 operate in every state but Missouri, and can assist in identifying patterns associated with higher risk for health harms, such as receiving opioids from multiple prescribers or pharmacies (Dowell, Zhang, Noonan, & Hockenberry, 2016; Katz et al., 2010; Logan, Liu, Paulozzi, Zhang, & Jones, 2013; Meara et al., 2016; Patrick, Fry, Jones, & Buntin, 2016; Paulozzi, Kilbourne, & Desai, 2011; Reifler et al., 2012; Reisman, Shenoy, Atherly, & Flowers, 2009;Wen, Schackman, Aden, & Bao, 2017).

Increased opioid analgesic use is associated with increased opioid-related morbidity and mortality, but specific high-risk prescribing practices such as high morphine equivalent daily doses (MEDD), overlapping opioid prescriptions (Liu, Logan, Paulozzi, Zhang, & Jones, 2013; Mack, Zhang, Paulozzi, & Jones, 2015; Paulozzi, Strickler, Kreiner, & Koris, 2015), and overlapping opioid and benzodiazepine prescriptions are associated with higher rates of negative outcomes, including opioid abuse and misuse (Hall et al., 2008), increased emergency department use and hospitalizations (Braden et al., 2010), and overdoses (Baumblatt et al., 2014; Dilokthornsakul et al., 2016; Ekstrom, Bornefalk-Hermansson, Abernethy, & Currow, 2014; Hall et al., 2008; Park, Saitz, Ganoczy, Ilgen, & Bohnert, 2015). Recent studies have examined the extent of high-risk opioid prescribing (Liu et al., 2013; Logan et al., 2013; Mack et al., 2015; Meara et al., 2016; Paulozzi et al., 2015), finding that almost a quarter of some populations receive overlapping opioid (Liu et al., 2013; Mack et al., 2015) and opioid/benzodiazepine prescriptions (Hackman et al., 2014; Hwang et al., 2016; Larochelle, Zhang, Ross-Degnan, & Wharam, 2015; Liu et al., 2013; Mack et al., 2015; Yang et al., 2015), and up to 16% receive high dose MEDD prescriptions (Logan et al., 2013; Mack et al., 2015; Meara et al., 2016; Paulozzi et al., 2015).

Compared to nonenrollees, Medicaid enrollees are at higher risk of prescription opioid-related mortality (Centers for Disease Control, 2009; Fernandes, Campana, Harwell, & Helgerson, 2015; Sharp & Melnik, 2015) and have higher rates of mental health and substance use disorders (Adelmann, 2003), yet, relatively few studies examine high-risk prescribing in this population (Yang et al., 2015). Medicaid enrollees with comorbid mental health, nonopioid substance use, and pain disorders appear to have higher rates of some types of high-risk prescribing than do enrollees without such characteristics (Stein et al., 2017).

Rates of high-risk opioid prescribing in non-Medicaid populations vary, with several studies finding higher rates among females, older patients (Hwang et al., 2016; Liu et al., 2013; Paulozzi et al., 2015; Yang et al., 2015), and individuals residing in rural and poorer counties (Stein et al., 2017). Community characteristics can affect the care residents receive (Dick et al., 2015; Stein et al., 2015; Stein, Gordon, et al., 2015), yet, the relative influence of individual- and county-level factors on high-risk opioid prescribing is relatively unexamined. To address this gap, our study examines individual- and county-level factors associated with multiple types of high-risk prescribing among Medicaid enrollees receiving opioid analgesics, hypothesizing that females, Caucasians, older patients, patients with depression or substance use disorders, and individuals residing in rural, low income counties will have higher rates of high-risk prescribing than Medicaid enrollees without these characteristics (Gu, Dillon, & Burt, 2010; Paulozzi & Xi, 2008; Tamayo-Sarver, Hinze, Cydulka, & Baker, 2003).

Methods

Data and study population

Using 2007–2009 Medicaid claims from California, Illinois, Massachusetts, and New York, we identified enrollees ages 18–64 years who had filled two or more valid opioid analgesic prescriptions in a calendar year, allowing us to examine overlapping opioid analgesic prescribing rates. The data were obtained prior to federal redaction of substance use disorder treatment related claims and were a subset of the most populous states available from a larger project examining opioid use disorders among Medicaid enrollees. We excluded dually eligible individuals and individuals not consistently Medicaid-enrolled, defined as at least nine months enrollment in a calendar year. Consistent with other studies of high-risk prescribing, we excluded individuals with a cancer diagnosis (ICD-9 codes 140–172.9; 174–215.9, 217–229.10; 235–239.9, 338.3) (Liu et al., 2013; Mack et al., 2015; Paulozzi, Zhou, Jones, Xu, & Florence, 2016; Yang et al., 2015). We identified new opioid treatment episodes from February 1, 2007 through June 30, 2009 using National Drug Codes (NDC), quantity dispensed, and days’ supply in pharmacy claims. A new episode was defined as the period from the first observed fill date of any opioid prescription after a 30-day period with no filled opioid prescriptions, through the last day of medication in the last filled opioid prescription, with no more than a 29-day gap between opioid prescriptions. Multiple prescriptions from a single provider for different doses of the same drug filled on the same day were considered a single prescription. The Institution’s IRB approved the study.

Measures of high-risk opioid prescribing

Consistent with prior studies, we created three measures of high-risk prescribing. The first, high MEDD, converts opioid prescriptions to an equivalent morphine daily dose (Ripamonti et al., 1998), facilitating comparisons among different formulations and strengths. Various cutoffs to define high-dose prescribing have been proposed, notably Washington State and the Centers for Disease Control have advocated for cutoffs of 120 and 90 mg MEDD, respectively. To facilitate comparisons with other populations, for the present study, we used a commonly cited cutoff in the literature (Liu et al., 2013; Logan et al., 2013; Paulozzi et al., 2015) and defined high MEDD as MEDD of 100 mg or more for an episode of at least 7 days. Our second measure, overlapping opioid prescriptions, was defined as an episode of at least 14 days with a minimum of 7 days of opioid possession in which >25% of episode days had overlapping opioid prescriptions. This modification of definitions used in prior studies (Liu et al., 2013; Mack et al., 2015; Paulozzi et al., 2015) allowed us to identify extended periods when the daily opioid dose might possibly be greater than the prescriber intended due to early refills, doctor shopping, or poorly coordinated care. The final measure, overlapping opioid and benzodiazepine prescriptions, was defined as a benzodiazepine prescription of at least 5 days’ supply filled during an episode. This is a commonly used measure of high-risk prescribing (Hwang et al., 2016; Liu et al., 2013; Mack et al., 2015; Yang et al., 2015) given that concomitant use of benzodiazepines and opioids increases risk of respiratory distress and overdose (Ekstrom et al., 2014).

Individual-level factors

Individuals were categorized as having pre-existing alcohol use disorder, opioid use disorder, hypnotic or sedative use disorder, or major depressive disorder if they had a claim containing the respective International Classification of Disease version 9 (ICD-9) diagnosis code in the year prior to the opioid treatment episode. Consistent with prior research on high-risk opioid use (Mack et al., 2015), we used ICD-9 codes for pain and service utilization within 14 days of the start of the opioid treatment episode to categorize individuals with acute pain, chronic pain, both acute and chronic pain, or no pain. Information on patient age, gender, and race/ethnicity were obtained from Medicaid files.

County-level factors

We examined county-level factors for the patient’s county of residence that had previously been associated with potentially inappropriate prescribing or opioid misuse. County urbanicity was categorized based on Rural–Urban Continuum Codes (RUCC) from the Area Resource File (ARF) as (1) metropolitan (urban population of at least 50,000), (2) large nonmetropolitan (population of at least 20,000), (3) medium nonmetropolitan (population 2,500–20,000), and (4) small nonmetropolitan (population less than 2,500) (Service, 2013). Primary care provider (PCP) supply was defined as the number of nonfederal PCPs per 100,000 population, derived from the 2007–2010 Area Resource File. Median household income was calculated from the American Community Survey (ACS), and educational attainment was defined as the percentage of adults greater than 25 years of age with no high school degree. PCP supply, educational attainment, and median household income were categorized by quartile for the purpose of the analysis.

Analyses

We calculated descriptive statistics for the three indicators of high-risk prescribing separately. We also examined any high-risk prescribing practice, defined as the presence of any of the three indicators. We used multivariate logistic regression to identify the association of individual- and county-level factors with measures of high-risk prescribing, conducting separate analyses for any high-risk prescribing, high MEDD, overlapping opioids, and overlapping opioids and benzodiazepines. We used state level fixed effects as well as clustered standard errors at the level of the individual to adjust for individuals with multiple episodes. Episodes with missing covariate data (<5% of episodes) were excluded from regressions.

Results

We identified 801,413 Medicaid enrollees with 1,863,684 opioid treatment episodes, 2.33 episodes (SD = 1.59) per enrollee (Figure 1).

Figure 1.

Figure 1

Open in a new tab

Study exclusion criteria.

High-risk prescribing was common, with 39.4% of episodes involving at least one measure of such prescribing (Table 1). High-dose opioid prescribing was most common (29.0% of episodes), followed by overlapping opioid and benzodiazepine (9.6%), and overlapping opioid (7.3%) prescriptions. Approximately one-third of episodes (33.2%) involved one type of high-risk prescribing, with multiple types of potentially inappropriate prescribing occurring in 6.3% of episodes. As compared to episodes with no high-risk prescribing indicators, episodes with at least one high-risk prescribing indicator on average had higher MEDD (181.0 mg (SD = 144.8 mg) vs. 131.6 mg (SD = 197.8)), longer episode lengths (65.4 days (SD=103.4 days) vs. 25.7 days (SD = 59.5 days)), and a higher number of prescription fills (8.5 (SD = 10.2) vs. 4.9 (SD = 6.0)) (Table 2).

Table 1.

Frequency of episodes involving high-risk prescribing.

Outcome Number of
episodes		 Percentage of
episodes
Any high-risk prescribing 733,094 39.3%
>100 MEDD 540,036 29.0%
Overlapping opioids 135,784 7.3%
Overlapping benzodiazepines 178,974 9.6%
Number of high-risk indicators per episode
  0 1,130,590 60.7%
  1 618,298 33.2%
  2 107,892 5.8%
  3 6,904 0.4%
Open in a new tab

Abbreviation: MEDD, morphine equivalent daily dose.

Table 2.

Episode characteristics by presence of high-risk prescribing indicators.

Characteristic No high-risk
indicators		 At least one
high-risk indicator
MEDD, mean (SD) 131.6 (197.8) 181.0 (144.8)
Episode duration in days, mean (SD) 25.7 (59.5) 65.4 (103.4)
Number of prescription fills, mean (SD) 4.9 (6.0) 8.5 (10.2)
Open in a new tab

Abbreviations: MEDD, morphine equivalent daily dose; SD, standard deviation.

Enrollees in the sample were aged 18–64 years with a relatively even distribution of episodes among age groups and the majority of episodes were attributed to females (73.5%) (Table 3). Several factors were consistently associated with both any high-risk prescribing (Table 3) and the various types of high-risk prescribing (Table 4). For example, the relationship between older age cohorts and all types of high-risk prescribing were relatively consistent and robust, with individuals aged 55 to 64 significantly more likely to experience any high-risk prescribing than individuals age 18 to 25 (52% vs. 24%, OR 3.35, 95% CI 3.30 to 3.40). Similarly, individuals living in nonmetropolitan counties were more likely to have any high-risk prescribing overall and across all indicators of high-risk prescribing, and blacks were significantly less likely to have high-risk prescribing than whites (37% vs. 41%; OR 0.73, 95% CI 0.73 to 0.74). Individuals with a recent prior diagnosis of major depression were also consistently more likely to have all types of high-risk prescribing, and significantly higher rates of any high-risk prescribing than individuals without major depression (51% vs. 38%; OR 1.42, 95% CI 1.40 to 1.44).

Table 3.

Individual- and county-level characteristics of Medicaid enrollees receiving opioids overall and by high-risk prescribing and county-level characteristics by state.a

Variable N (%) of opioid
analgesic episodes		 % of opioid
analgesic episodes with any high-risk prescribing		 Multivariate regression
OR (95% CI)
Total 1,863,684 39.4%
Individual-level characteristics
  Age
  18–25 304,910 (16.4%) 24.1% Reference
  26–35 425,823 (22.8%) 31.6% 1.43 (1.41, 1.45)*
  36–45 434,142 (23.3%) 41.0% 2.14 (2.12, 2.17)*
  46–55 436,054 (23.4%) 48.1% 2.86 (2.82, 2.90)*
  56–64 262,755 (14.1%) 52.2% 3.35 (3.30, 3.40)*
Gender
  Male 493,511 (26.5%) 40.4% Reference
  Female 1,370,173 (73.5%) 39.0% 1.04 (1.03, 1.05)*
Race
  White 804,532 (43.2%) 40.8% Reference
  Black 423,609 (22.7%) 36.6% 0.73 (0.73, 0.74)*
  Hispanic 440,294 (23.6%) 38.5% 0.84 (0.83, 0.84)*
  Other 195,249 (10.5%) 41.1% 0.85 (0.84, 0.87)*
Opioid dependencyb
  No 1,781,300 (95.6%) 39.2% Reference
  Yes 82,384 (4.4%) 43.2% 1.04 (1.02, 1.06)*
Alcohol dependencyb
  No 1,775,975 (95.3%) 39.1% Reference
  Yes 87,709 (4.7%) 43.7% 1.04 (1.02, 1.06)*
Hypnotic or sedative dependencyb
  No 1,856,888 (99.6%) 39.3% Reference
  Yes 6,796 (0.4%) 51.7% 1.44 (1.35, 1.53)*
Major depressionb
  No 1,716,878 (92.1%) 38.3% Reference
  Yes 146,806 (7.9%) 51.4% 1.42 (1.40, 1.44)*
Pain diagnoses
  No pain diagnosesc 167,673 (9.0%) 31.2% Reference
  Acute pain diagnoses onlyc 451,203 (24.2%) 49.1% 1.72 (1.70, 1.74)*
  Non-acute pain diagnoses onlyc 156,966 (8.4%) 39.1% 1.29 (1.27, 1.31)*
  Both acute and non-acute pain diagnosesc 1,087,842 (58.4%) 36.6% 1.16 (1.15, 1.18)*
County-level characteristics RUCCd
  Metropolitan 1,679,205 (90.1%) 39.1% Reference
  Large nonmetropolitan 107,362 (5.8%) 40.8% 1.05 (1.03, 1.07)*
  Medium nonmetropolitan 72,631 (3.9%) 43.0% 1.18 (1.15, 1.20)*
  Small nonmetropolitan 4,486 (0.2%) 45.3% 1.23 (1.14, 1.33)*
PCPs per 100,000 population,d mean (SD)
  Q1 (lowest concentration) 458,464 (24.6%) 40.2% Reference
  Q2 483,775 (26.0%) 38.6% 0.98 (0.97, 1.00)*
  Q3 445,644 (23.9%) 39.8% 1.04 (1.02, 1.05)*
  Q4 (highest concentration) 475,801 (25.5%) 38.9% 0.99 (0.97, 1.01)
% No high school degreed
  Q1 (lowest %) 464,838 (24.9%) 36.7% Reference
  Q2 609,008 (32.7%) 39.4% 0.99 (0.98, 1.00)
  Q3 342,205 (18.4%) 39.7% 1.02 (1.01, 1.04)*
  Q4 (highest %) 447,633 (24.0%) 41.8% 1.21 (1.19, 1.23)*
Median household income,d mean (SD)
  Q1 (lowest median) 458,841 (24.6%) 40.2% Reference
  Q2 537,372 (28.8%) 40.7% 1.00 (0.98, 1.01)
  Q3 285,365 (15.3%) 38.5% 1.04 (1.02, 1.05)*
  Q4 (highest median) 582,106 (31.2%) 37.8% 0.87 (0.86, 0.88)*
State
  CA 751,592 (40.3%) 39.9% Reference
  IL 352,773 (18.9%) 39.0% 1.19 (1.17, 1.21)*
  MA 178,689 (9.6%) 32.5% 0.80 (0.79, 0.82)*
  NY 580,630 (31.2%) 40.9% 1.25 (1.23, 1.27)*
Open in a new tab

Abbreviations: PCP, primary care physician; RUCC, Rural–Urban Continuum Code.

a

Numbers are % and calculations are made at the episode level.

b

At least one diagnosis present in calendar year prior to the episode start date.

c

Within 14 days of start of episode, pain definitions used as described by Mack et al., 2015.

d

County attribute of patient’s place of residence at the start of the episode.

*

p < 0.05.

Table 4.

Logistic regression of high-risk prescribing outcomes on individual- and county-level variables.a

Variable >100 MEDD Opioid overlap Opioid-benzodiazepine overlap
Individual level characteristics % OR (95% CI) % OR (95% CI) % OR (95% CI)
Age
  18–25 18.5 Reference 4.7 Reference 3.9 Reference
  26–35 22.8 1.29 (1.28, 1.31)* 6.4 1.33 (1.30, 1.36)* 7.4 1.95 (1.89, 20.1)*
  36–45 29.1 1.83 (1.80, 1.85)* 8.2 1.59 (1.55, 1.62)* 10.8 3.08 (2.99, 3.18)*
  46–55 35.2 2.43 (2.40, 2.47)* 8.7 1.63 (1.59, 1.67)* 12.9 3.98 (3.86, 4.10)*
  56–65 40.7 3.01 (2.96, 3.06)* 7.9 1.46 (1.42, 1.50)* 12.4 3.95 (3.81, 4.08)*
Gender
  Male 29.0 Reference 9.2 Reference 8.8 Reference
  Female 29.0 1.04 (1.03, 1.05)* 6.6 0.76 (0.75, 0.77)* 9.9 1.37 (1.34, 1.39)*
Race
  White 27.6 Reference 8.3 Reference 13.0 Reference
  Black 28.6 0.95 (0.93, 0.96)* 6.8 0.79 (0.78, 0.81)* 6.4 0.39 (0.38, 0.40)*
  Hispanic 30.4 1.08 (1.07, 1.10)* 6.3 0.67 (0.66, 0.69)* 7.3 0.52 (0.51, 0.54)*
  Other 32.4 1.13 (1.11, 1.15)* 6.6 0.69 (0.67, 0.70)* 8.1 0.49 (0.48, 0.51)*
Opioid dependencyb
  No 29.4 Reference 7.2 Reference 9.0 Reference
  Yes 20.5 0.59 (0.57, 0.60)* 10.0 1.21 (1.18, 1.24)* 21.8 2.52 (2.45, 2.60)*
Alcohol dependencyb
  No 29.1 Reference 7.1 Reference 9.4 Reference
  Yes 27.6 0.96 (0.94, 0.97)* 10.4 1.20 (1.17, 1.23)* 14.5 1.10 (1.07, 1.13)*
Hypnotic or sedative dependencyb
  No 29.0 Reference 7.3 Reference 9.5 Reference
  Yes 21.1 0.91 (0.84, 0.97)* 13.8 1.38 (1.28, 1.49)* 30.7 2.00 (1.85, 2.17)*
Major depressionb
  No 28.7 Reference 7.2 Reference 8.5 Reference
  Yes 32.8 1.05 (1.03, 1.06)* 8.8 1.12 (1.10, 1.14)* 22.1 2.57 (2.52, 2.63)*
Pain diagnoses
  No pain diagnosesc 21.2 Reference 6.5 Reference 8.3 Reference
  Acute pain diagnoses onlyc 36.2 1.73 (1.71, 1.76)* 10.5 1.56 (1.53, 1.60)* 11.4 1.21 (1.18, 1.23)*
  Non-acute pain diagnoses onlyc 25.2 1.18 (1.16, 1.20)* 10.5 1.63 (1.58, 1.67)* 11.2 1.19 (1.16, 1.22)*
  Both acute and non-acute pain diagnosesc 27.7 1.29 (1.27, 1.31)* 5.6 0.83 (0.81, 0.85)* 8.8 1.04 (1.02, 1.07)*
County characteristics RUCCd
  Metropolitan 28.6 Reference 7.3 Reference 9.6 Reference
  Large nonmetropolitan 31.5 1.07 (1.04, 1.09)* 7.8 1.05 (1.02, 1.08)* 8.6 1.01 (0.97, 1.05)
  Medium nonmetropolitan 33.0 1.15 (1.12, 1.17)* 7.0 1.13 (1.08, 1.17)* 11.5 1.27 (1.22, 1.32)*
  Small nonmetropolitan 33.3 1.12 (1.03, 1.22)* 7.0 1.15 (1.01, 1.31)* 14.0 1.40 (1.22, 1.61)*
PCPs per 100,000 population,d mean (SD)
  Q1 30.8 Reference 7.4 Reference 9.1 Reference
  Q2 27.9 0.96 (0.94, 0.97)* 7.3 0.99 (0.97, 1.01) 9.9 1.07 (1.05, 1.10)*
  Q3 28.5 0.97 (0.95, 0.98)* 7.3 1.00 (0.98, 1.02) 10.9 1.15 (1.13, 1.18)*
  Q4 28.8 1.01 (0.99, 1.03) 7.2 0.95 (0.92, 0.97)* 8.6 0.95 (0.92, 0.98)*
% No high school degreed
  Q1 25.1 Reference 7.2 Reference 10.9 Reference
  Q2 29.7 1.03 (1.02, 1.04)* 7.7 0.91 (0.90, 0.93)* 7.9 0.94 (0.92, 0.97)*
  Q3 28.4 1.11 (1.09, 1.13)* 6.4 1.01 (0.99, 1.03) 12.0 1.06 (1.04, 1.09)*
  Q4 32.5 1.31 (1.28, 1.33)* 7.5 0.88 (0.85, 0.90)* 8.8 1.08 (1.04, 1.11)*
Median household income,d mean (SD)
  Q1 31.4 Reference 7.2 Reference 8.3 Reference
  Q2 30.4 0.93 (0.91, 0.94)* 6.4 1.00 (0.98, 1.02) 11.0 1.30 (1.27, 1.34)*
  Q3 26.2 0.93 (0.92, 0.95)* 8.6 1.16 (1.13, 1.19)* 10.6 1.27 (1.23, 1.31)*
  Q4 27.1 0.73 (0.72, 0.74)* 7.5 1.04 (1.01, 1.06)* 8.9 1.34 (1.30, 1.39)*
State
  CA 29.4 Reference 8.3 Reference 9.1 Reference
  IL 30.1 1.37 (1.34, 1.39)* 4.5 0.47 (0.46, 0.49)* 11.5 1.46 (1.41, 1.50)*
  MA 16.7 0.61 (0.59, 0.62)* 5.7 0.51 (0.50, 0.53)* 15.9 1.82 (1.76, 1.88)*
  NY 31.6 1.46 (1.43, 1.48)* 8.2 0.92 (0.90, 0.94)* 7.2 0.82 (0.80, 0.84)*
Open in a new tab

Abbreviations: PCP, primary care physician; RUCC, Rural–Urban Continuum Code.

a

Regressions are conducted at the episode level with clustered robust standard errors at the individual level.

b

At least one diagnosis present in calendar year prior to the episode start date.

c

Within 14 days of start of episode, pain definitions used as described by Mack et al., 2015.

d

County attribute of patient’s place of residence at the start of the episode.

*

p < 0.05.

However, a number of other factors varied by prescribing indicator. For example, compared to individuals without such disorders, individuals with prior opioid, alcohol, or hypnotic/sedative use disorder diagnoses had lower odds of high MEDD but higher odds of overlapping opioid prescriptions and overlapping opioid and benzodiazepine prescriptions (Table 4). Individuals with only acute pain or only chronic pain were both more likely to have potentially inappropriate prescribing across all indicators than were individuals with no pain. However, this pattern did not hold for individuals with both acute and chronic pain. Such individuals were more likely than individuals with no pain to have high MEDD and overlapping opioids and benzodiazepines but were less likely to have overlapping opioids. Overlapping opioid use rates were also significantly lower in females compared to males (6.6% vs. 9.2%, OR 0.76, 95% CI 0.75 to 0.77), but overlapping opioids and benzodiazepines and high MEDD were significantly higher for females as compared to males.

With respect to county characteristics such as PCP supply, household income, and county population without a high school diploma, there were some significant differences across quartiles from the referent quartile, but there did not appear to be a clinically meaningful pattern either within an indicator of high-risk prescribing or across indicators. However, we found very different patterns across states in measures of high-risk prescribing. For example, the rate of high MEDD in Massachusetts (16.7%) was slightly more than half that seen in California, Illinois, and New York, whereas the rate of opioid and benzodiazepine overlap in Massachusetts was substantially higher than the other states and more than double than that seen in New York (Table 4).

Discussion

In examining potentially inappropriate opioid prescribing among Medicaid enrollees with at least two opioid prescriptions in four populous states, we found relatively high rates of high-risk prescribing, with approximately 4 in 10 opioid treatment episodes involving at least one type. Our findings from a sample of over 800,000 Medicaid enrollees from four states with over 2 million opioid treatment episodes over a 3-year period complement recent examinations of opioid prescribing among a national sample of disabled Medicare beneficiaries (Meara et al., 2016; Paulozzi et al., 2015), with both finding high rates of high-risk prescribing. The higher rate of high dose opioid episodes we observe in Medicaid enrollees filling two or more opioid analgesic prescriptions in a year compared to rates seen in studies of populations receiving at least one prescription (Meara et al., 2016; Paulozzi et al., 2015) suggest that initiatives targeting such populations may increase efficiency while decreasing unintended consequences. The frequency with which opioid treatment episodes involve high-risk prescribing also reinforces the need for further research examining the relationship of such prescribing with negative clinical outcomes, the motivations among prescribers and patients involved in these behavior patterns, as well as efforts to identify interventions that are effective in aligning prescribing behavior more consistently with practice guidelines and safe prescribing practices.

In 2012, the 259 million opioid prescriptions written were enough for every adult in the United States to receive a prescription (CDC, 2014), and the United States is now the top consumer of opioids per capita in the world (Board, 1968). The growth in opioid analgesic prescriptions has been associated with an increase in opioid-related overdoses (CDC, 2014). Given the consistently documented increased risk for opioid-related overdoses among Medicaid enrollees, it is concerning that we found that approximately one-third of new opioid treatment episodes in our study population involved high dosages of opioids and that rates of overlapping opioids and overlapping opioids and benzodiazepines, while less frequent, were also high. Given the high rates of high-risk prescribing observed, we are encouraged by recent efforts such as the dissemination of recent guidelines from the Centers for Disease Control which explicitly mentions the importance of not prescribing benzodiazepines concurrently with opioids and avoiding high dose opioids (Dowell et al., 2016).

Prescribing potentially addictive medications such as opioids and benzodiazepines to individuals previously treated for substance use disorders must be done thoughtfully, with clinicians weighing the benefits of treatment with the increased risk of relapse or precipitating a new substance use disorder. We found individuals with previously diagnosed opioid use, alcohol use, and hypnotic/sedative use disorders were all significantly less likely to receive high-dose opioids but were all more likely to receive overlapping opioid and overlapping opioid and benzodiazepine prescriptions, consistent with previous research (Paulozzi et al., 2016). We are uncertain why such differences between indicators of high-risk prescribing may exist. Overlapping prescriptions may involve multiple prescribers, some of whom may be unaware of prior diagnoses or concurrent prescribers. It may also be that when prescribing to individuals with previously identified substance use disorders, prescribers are more sensitive to prescribing a higher dose on a single prescription than they are to potential issues associated with writing multiple prescriptions. Electronic health records and PDMPs have the potential to decrease such overlapping prescribing by providing timely information to prescribers and pharmacists (Brandeis Prescription Drug Monitoring Program Center of Excellence, 2016), and additional research can help provide a better understanding of the magnitude of risk such prescribing practices present to individuals previously identified with substance use disorders. Medicaid Patient Review and Restriction (PRR) programs are another potential tool to prevent opioid abuse. In these programs, Medicaid enrollees who meet certain criteria, such as having visited multiple prescribers, received a high volume of opioid prescriptions, or a previous substance dependency diagnosis, are “locked-in” to a restricted network of prescribers and pharmacists to better coordinate care and prevent doctor shopping (The Pew Charitable Trusts, 2016). None of the states in our analysis used prior substance dependence diagnosis as a criteria for enrollment in a PRR program (The Pew Charitable Trusts, 2016), but given the significant associations between prior substance dependence diagnosis and overlapping opioid and overlapping opioid and benzodiazepine prescriptions in our data, states may wish to consider adopting PRR programs which use these criteria.

Rural counties have higher rates of opioid-related overdoses (Centers for Disease Control, 2009; Hall et al., 2008) and nonmedical opioid use (Fiscella, Franks, Gold, & Clancy, 2000), yet, have far fewer resources to provide effective treatment for opioid use disorders (Bohnert et al., 2011; Fiscella et al., 2000). Even after controlling for county income, education, and physicians per capita—factors that have been associated with poor quality care in other areas of healthcare (Fiscella et al., 2000)—we found that rural counties had higher rates of high-risk prescribing, potentially contributing to their higher rates of opioid-related morbidity and mortality. Research has also found that whites and individuals with major depression have higher rates of clinically unsupported and nonmedical opioid use as well as higher opioid overdose rates than other individuals (Bohnert et al., 2011; Dunn et al., 2010; Sullivan, Edlund, Zhang, Unützer, & Wells, 2006). In our study, both groups had higher rates of high-risk prescribing than individuals without those characteristics. Understanding the characteristics of patients and communities at increased risk for high-risk prescribing may be a first step toward targeted patient or community focused educational efforts designed to complement many current prescriber focused initiatives (Dowell et al., 2016).

Consistent with prior studies (Paulozzi et al., 2015), we found indicators of high-risk prescribing varied substantially across states. State policies and regulations that may influence high-risk prescribing practices, such as regulations related to the use of PDMPs, medication quantity limits, requirements associated with physician exams or verification by a pharmacist, and use of tamper-resistant prescription forms, vary greatly (Adelmann, 2003; Clark, Eadie, Kreiner, & Strickler, 2012; Meara et al., 2016). However, relatively few studies have examined in what way such policies may influence high-risk prescribing practices (Haegerich, Paulozzi, Manns, & Jones, 2014), and to date, studies examining the effects of state policies are inconsistent (Haegerich et al., 2014; Meara et al., 2016). Notably, all four states in our dataset had well-established PDMPs as of 2007, the start of the study period. However, our examination of information from four states over a three-year period is not well suited to examine the effectiveness of these types of policies. As states seek to identify and implement policies designed to address high-risk opioid prescribing and associated negative sequelae, a better understanding of the effects of policies is needed to inform policymaking.

Our findings must be interpreted within the context of the study’s limitations. We used claims data from Medicaid enrollees from four states with fee-for-service or high quality managed care encounter data appropriate for research, and do not know to what extent our findings would generalize to non-Medicaid populations, dually eligible enrollees, or Medicaid-enrollees from other states. We also do not know to what extent Medicaid eligibility criteria or opioid prescribing policies in the states may explain between-state variation or affect generalizability with other states, and further research with more states and more years of data is needed to examine those issues. Furthermore, we do not observe opioid analgesics not paid for by Medicaid, prescriptions which are written and not filled, or nonprescription opioids, such as heroin. Our data are dated, having been obtained prior to federal redaction of information related to substance use disorders under 42 CFR. We are unable to examine how high-risk prescribing may have increased or decreased in more recent years, but in the absence of more recent data, our findings contribute to our understanding of patterns of high-risk prescribing, and help to establish a baseline through which the impact of policy changes using more recent data can be interpreted. By requiring a 30-day period with no prescribed opioids prior to the beginning of an opioid treatment episode and two filled opioid prescriptions, we systematically excluded individuals who were already receiving opioids and individuals who filled a single prescription. We are unable to consistently and confidently identify prescribers, dispensing pharmacies, and treatment setting of the initial opioid prescription in our data, and we are therefore unable to examine the characteristics of prescribers of high-risk opioids nor the treatment settings in which they were first prescribed. Because of this limitation, we are unable to determine the relative contribution of patient (demand) and provider (supply) characteristics to county-level variation in prescribing practices. We are also unable to determine to what extent multiple prescribers may have been involved in an episode with high-risk prescribing.

Most importantly, we have no clinical information on patient history, clinical severity, patient pain ratings, planned titration of medications, reactions to medications, and other circumstances of care that inform a prescriber’s use of opioid analgesics. As a result, we are unable to judge the clinical appropriateness of the prescriptions filled by any given patient and instead rely on patterns of prescribing at a population level that appear to be inconsistent with prescribing guidelines.

Despite these limitations, the present study contributes valuable information for patients, providers, policymakers, and researchers on the frequency of high-risk prescribing in an at-risk population of individuals. The study population included a large number of individuals from four populous, diverse states representing different geographic regions, policy environments, and cultural norms, and the analysis examined a number of individual- and county-level factors exceeding those examined in prior studies. At a time of wide ranging efforts to address the public health crisis related to opioids, our findings suggest targets for patient and prescriber focused efforts designed to address some of the most questionable prescribing behaviors as well as the need for continued efforts to better understand some of the substantial state level variation that occurs with such prescribing.

Acknowledgments

Funding and support

The National Institute on Drug Abuse of the National Institutes of Health (NIH) (award R01DA032881-01A1) supported this study. NIDA was not directly involved in the conduct of the study, collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or the decision to submit the manuscript for publication.

The authors are indebted to Hilary Peterson, BA, of the RAND Corporation, for assistance in manuscript preparation, and to participants in the RAND Summer Associate Program and the RAND Drug Policy Research Center for prior feedback on the research approach and findings.

National Institutes of Health, National Institute on Drug Abuse grant DA032881-01A1 (Stein, PI).

Footnotes

Declaration of interest

The authors declare that they have no conflict of interest. The authors alone are responsible for the content and writing of the article.

References

  1. Adelmann PK. Mental and substance use disorders among Medicaid recipients: Prevalence estimates from two national surveys. Administration and Policy in Mental Health and Mental Health Services Research. 2003;31(2):111–129. doi: 10.1023/B:APIH.0000003017.78877.56. [DOI] [PubMed] [Google Scholar]
  2. Baumblatt JAG, Wiedeman C, Dunn JR, Schaffner W, Paulozzi LJ, Jones TF. High-risk use by patients prescribed opioids for pain and its role in overdose deaths. JAMA Internal Medicine. 2014;174(5):796–801. doi: 10.1001/jamainternmed.2013.12711. [DOI] [PubMed] [Google Scholar]
  3. Board, I.N.C. Report of the International Narcotics Control Board for …. United Nations Publications; 1968. [Google Scholar]
  4. Bohnert ASB, Valenstein M, Bair MJ, Ganoczy D, McCarthy JF, Ilgen MA, Blow FC. Association between opioid prescribing patterns and opioid overdose-related deaths. Journal of the American Medical Association. 2011;305(13):1315–1321. doi: 10.1001/jama.2011.370. [DOI] [PubMed] [Google Scholar]
  5. Braden JB, Russo J, Fan M-Y, Edlund MJ, Martin BC, DeVries A, Sullivan MD. Emergency department visits among recipients of chronic opioid therapy. Archives of Internal Medicine. 2010;170(16):1425–1432. doi: 10.1001/archinternmed.2010.273. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Brandeis Prescription Drug Monitoring Program Center of Excellence. Options for Unsolicited Reporting. 2016 Retrieved from: http://www.pdmpassist.org/pdf/COE_documents/Add_to_TTAC/Update%20to%20Brandeis%20COE%20Guidance%20on%20Unsolicited%20Reporting%20final.pdf.
  7. CDC. [Retrieved May 30, 2017];CDC Vital signs—Opioid painkiller prescribing. 2014 Jul 1; from http://www.cdc.gov/vitalsigns/opioid-prescribing/index.html.
  8. Clark T, Eadie J, Kreiner P, Strickler G. Prescription Drug Monitoring Programs. 2012 Retrieved from http://www.pewtrusts.org~/media/assets/0001/pdmp_update_1312013.pdf.
  9. Centers for Medicaid and Medicare Services-1656-FC/IFC Medicare Program: Hospital Outpatient Prospective Payment and Ambulatory Surgical Center Payment Systems and Quality Reporting Programs; Organ Procurement Organization Reporting and Communication; Transplant Outcome Measures and Documentation Requirements; Electronic Health Record (EHR) Incentive Programs; Payment to Nonexcepted Off-Campus Provider-Based Department of a Hospital; Hospital Value-Based Purchasing (VBP) Program; Establishment of Payment Rates under the Medicare Physician Fee Schedule for Nonexcepted Items and Services Furnished by an Off-Campus Provider-Based Department of a Hospital. 42 C.F.R. Stat. 2016 (n.d.) [PubMed] [Google Scholar]
  10. Centers for Disease Control. Overdose deaths involving prescription opioids among Medicaid enrollees – Washington, 2004–2007. Morbidity and Mortality Weekly Report. 2009;58(42):1171–1175. https://doi.org/mm5842a1 [pii] [PubMed] [Google Scholar]
  11. Dick AW, Pacula RL, Gordon AJ, Sorbero M, Burns RM, Leslie D, Stein BD. Growth in Buprenorphine waivers for physicians increased potential access to opioid agonist treatment, 2002–11. Health Affairs (Project Hope) 2015;34(6):1028–1034. doi: 10.1377/hlthaff.2014.1205. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Dilokthornsakul P, Moore G, Campbell JD, Lodge R, Traugott C, Zerzan J, Page RL. Risk factors of prescription opioid overdose among Colorado Medicaid beneficiaries. The Journal of Pain. 2016;17(4):436–443. doi: 10.1016/j.jpain.2015.12.006. [DOI] [PubMed] [Google Scholar]
  13. Dowell D, Haegerich TM, Chou R. CDC guideline for prescribing opioids for chronic pain—United States, 2016. Journal of the American Medical Association. 2016;315(15):1624–1645. doi: 10.1001/jama.2016.1464. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Dowell D, Zhang K, Noonan RK, Hockenberry JM. Mandatory provider review and pain clinic laws reduce the amounts of opioids prescribed and overdose death rates. Health Affairs. 2016;35(10):1876–1883. doi: 10.1377/hlthaff.2016.0448. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Dunn KM, Saunders KW, Rutter CM, Banta-Green CJ, Merrill JO, Sullivan MD, Psaty BM. Opioid prescriptions for chronic pain and overdose: A cohort study. Annals of Internal Medicine. 2010;152(2):85–92. doi: 10.7326/0003-4819-152-2-201001190-00006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Ekstrom MP, Bornefalk-Hermansson A, Abernethy AP, Currow DC. Safety of benzodiazepines and opioids in very severe respiratory disease: National prospective study. BMJ (Clinical Research Ed.) 2014;348:g445. doi: 10.1136/bmj.g445. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Fernandes JC, Campana D, Harwell TS, Helgerson SD. High mortality rate of unintentional poisoning due to prescription opioids in adults enrolled in Medicaid compared to those not enrolled in Medicaid in Montana. Drug and Alcohol Dependence. 2015;153:346–349. doi: 10.1016/j.drugalcdep.2015.05.032. [DOI] [PubMed] [Google Scholar]
  18. Fiscella K, Franks P, Gold MR, Clancy CM. Inequality in quality. Journal of the American Medical Association. 2000;283(19):2579–2584. doi: 10.1001/jama.283.19.2579. [DOI] [PubMed] [Google Scholar]
  19. Gu Q, Dillon CF, Burt VL. Prescription drug use continues to increase: U.S. prescription drug data for 2007–2008. NCHS Data Brief. 2010;42(42):1–8. [PubMed] [Google Scholar]
  20. Hackman DT, Greene MS, Fernandes TJ, Brown AM, Wright ER, Chambers RA. Prescription drug monitoring program inquiry in psychiatric assessment: Detection of high rates of opioid prescribing to a dual diagnosis population. The Journal of Clinical Psychiatry. 2014;75(7):750–756. doi: 10.4088/JCP.14m09020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Haegerich TM, Paulozzi LJ, Manns BJ, Jones CM. What we know, and don’t know, about the impact of state policy and systems-level interventions on prescription drug overdose. Drug and Alcohol Dependence. 2014;145:34–47. doi: 10.1016/j.drugalcdep.2014.10.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Hall AJ, Logan JE, Toblin RL, Kaplan JA, Kraner JC, Bixler D, Paulozzi LJ. Patterns of abuse among unintentional pharmaceutical overdose fatalities. Journal of the American Medical Association. 2008;300(22):2613–2620. doi: 10.1001/jama.2008.802. [DOI] [PubMed] [Google Scholar]
  23. Hwang CS, Kang EM, Kornegay CJ, Staffa JA, Jones CM, McAninch JK. Trends in the concomitant prescribing of opioids and benzodiazepines, 2002–2014. American Journal of Preventive Medicine. 2016;51(21):151–160. doi: 10.1016/j.amepre.2016.02.014. [DOI] [PubMed] [Google Scholar]
  24. Katz N, Panas L, Kim M, Audet AD, Bilansky A, Eadie J, Carrow G. Usefulness of prescription monitoring programs for surveillance—Analysis of Schedule II opioid prescription data in Massachusetts, 1996–2006. Pharmacoepidemiology and Drug Safety. 2010;19(2):115–123. doi: 10.1002/pds.1878. [DOI] [PubMed] [Google Scholar]
  25. Larochelle MR, Zhang F, Ross-Degnan D, Wharam JF. Trends in opioid prescribing and co-prescribing of sedative hypnotics for acute and chronic musculoskeletal pain: 2001–2010. Pharmacoepidemiology and Drug Safety. 2015;24(8):885–892. doi: 10.1002/pds.3776. [DOI] [PubMed] [Google Scholar]
  26. Liu Y, Logan JE, Paulozzi LJ, Zhang K, Jones CM. Potential misuse and inappropriate prescription practices involving opioid analgesics. The American Journal of Managed Care. 2013;19(8):648–665. doi: 85159 [pii] [PubMed] [Google Scholar]
  27. Logan J, Liu Y, Paulozzi L, Zhang K, Jones C. Opioid prescribing in emergency departments: The prevalence of potentially inappropriate prescribing and misuse. Medical Care. 2013;51(8):646–653. doi: 10.1097/MLR.0b013e318293c2c0. [DOI] [PubMed] [Google Scholar]
  28. Mack KA, Zhang K, Paulozzi L, Jones C. Prescription practices involving opioid analgesics among Americans with Medicaid, 2010. Journal of Health Care for the Poor and Underserved. 2015;26(1):182–198. doi: 10.1353/hpu.2015.0009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Meara E, Horwitz JR, Powell W, McClelland L, Zhou W, O’Malley AJ, Morden NE. State legal restrictions and prescription-opioid use among disabled adults. New England Journal of Medicine. 2016;375(1):44–53. doi: 10.1056/NEJMsa1514387. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Park TW, Saitz R, Ganoczy D, Ilgen MA, Bohnert ASB. Benzodiazepine prescribing patterns and deaths from drug overdose among U.S. veterans receiving opioid analgesics: Case-cohort study. BMJ (Clinical Research Ed.) 2015;350:h2698. doi: 10.1136/bmj.h2698. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Patrick SW, Fry CE, Jones TF, Buntin MB. Implementation Of Prescription Drug Monitoring Programs Associated With Reductions In Opioid-Related Death Rates. Health Affairs. 2016;35(7):1324–1332. doi: 10.1377/hlthaff.2015.1496. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Paulozzi LJ, Kilbourne EM, Desai HA. Prescription drug monitoring programs and death rates from drug overdose. Pain Medicine. 2011;12(5):747–754. doi: 10.1111/j.1526-4637.2011.01062.x. [DOI] [PubMed] [Google Scholar]
  33. Paulozzi LJ, Strickler GK, Kreiner PW, Koris CM. Controlled substance prescribing patterns—prescription behavior surveillance system, eight states, 2013. MMWR Surveillance Summaries. 2015;64(9):1–14. doi: 10.15585/mmwr.ss6409a1. [DOI] [PubMed] [Google Scholar]
  34. Paulozzi LJ, Xi Y. Recent changes in drug poisoning mortality in the United States by urban–rural status and by drug type. Pharmacoepidemiology and Drug Safety. 2008;17(10):997–1005. doi: 10.1002/pds.1626. [DOI] [PubMed] [Google Scholar]
  35. Paulozzi LJ, Zhou C, Jones CM, Xu L, Florence CS. Changes in the medical management of patients on opioid analgesics following a diagnosis of substance abuse. Pharmacoepidemiology and Drug Safety. 2016;25(5):545–552. doi: 10.1002/pds.3980. [DOI] [PubMed] [Google Scholar]
  36. Pew Charitable Trusts. [Retrieved September 6, 2017];Curbing Prescription Drug Abuse With Patient Review and Restriction Programs. 2016 from http://pew.org/1q69pHC.
  37. Reifler LM, Droz D, Bailey JE, Schnoll SH, Fant R, Dart RC, Bartelson BB. Do prescription monitoring programs impact state trends in opioid abuse/misuse? Pain Medicine. 2012;13(3):434–442. doi: 10.1111/j.1526-4637.2012.01327.x. [DOI] [PubMed] [Google Scholar]
  38. Reisman RM, Shenoy PJ, Atherly AJ, Flowers CR. Prescription opioid usage and abuse relationships: An evaluation of state prescription drug monitoring program efficacy. Substance Abuse: Research and Treatment. 2009;3:41. doi: 10.4137/sart.s2345. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Ripamonti C, Groff L, Brunelli C, Polastri D, Stavrakis A, Conno FD. Switching from morphine to oral methadone in treating cancer pain: What is the equianalgesic dose ratio? Journal of Clinical Oncology : Official Journal of the American Society of Clinical Oncology. 1998;16(10):3216–3221. doi: 10.1200/JCO.1998.16.10.3216. [DOI] [PubMed] [Google Scholar]
  40. Rudd RA, Aleshire N, Zibbell JE, Gladden RM. Increases in drug and opioid overdose deaths–United States, 2000–2014. Morbidity and Mortality Weekly Report. 2016;64(50–51):1378–1382. doi: 10.15585/mmwr.mm6450a3. [DOI] [PubMed] [Google Scholar]
  41. Service, U. D. of A. E. R. 2003 Rural-Urban Continuum Codes. 2013;2016 [Google Scholar]
  42. Sharp MJ, Melnik TA. Control, C. for D., & (CDC), P. Poisoning deaths involving opioid analgesics – New York State, 2003–2012. Morbidity and Mortality Weekly Report. 2015;64(14):377–380. doi.org/mm6414a2 [pii] [PMC free article] [PubMed] [Google Scholar]
  43. Stein BD, Gordon AJ, Dick AW, Burns RM, Pacula RL, Farmer CM, Sorbero M. Supply of buprenorphine waivered physicians: The influence of state policies. Journal of Substance Abuse Treatment. 2015;48(1):104–111. doi: 10.1016/j.jsat.2014.07.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Stein BD, Mendelsohn J, Gordon AJ, Dick AW, Burns RM, Sorbero MJ, Pacula RL. Opioid analgesic and benzodiazepine prescribing among Medicaid enrollees with opioid use disorders. Journal of Addictive Diseases. 2017;36(1):14–22. doi: 10.1080/10550887.2016.1211784. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Stein BD, Pacula RL, Gordon AJ, Burns RM, Leslie DL, Sorbero MJ, Dick AW. Where Is Buprenorphine Dispensed to Treat Opioid Use Disorders? The Role of Private Offices, Opioid Treatment Programs, and Substance Abuse Treatment Facilities in Urban and Rural Counties. Milbank Quarterly. 2015;93(3):561–583. doi: 10.1111/1468-0009.12137. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Sullivan MD, Edlund MJ, Zhang L, Unützer J, Wells KB. Association between mental health disorders, problem drug use, and regular prescription opioid use. Archives of Internal Medicine. 2006;166(19):2087–2093. doi: 10.1001/archinte.166.19.2087. [DOI] [PubMed] [Google Scholar]
  47. Tamayo-Sarver JH, Hinze SW, Cydulka RK, Baker DW. Racial and ethnic disparities in emergency department analgesic prescription. American Journal of Public Health. 2003;93(12):2067–2073. doi: 10.2105/AJPH.93.12.2067. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Wen H, Schackman BR, Aden B, Bao Y. States with prescription drug monitoring mandates saw a reduction in opioids prescribed to medicaid enrollees. Health Affairs. 2017;36(4):733–741. doi: 10.1377/hlthaff.2016.1141. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Yang Z, Wilsey B, Bohm M, Weyrich M, Roy K, Ritley D, Melnikow J. Defining risk of prescription opioid overdose: Pharmacy shopping and overlapping prescriptions among long-term opioid users in medicaid. The Journal of Pain. 2015;16(5):445–453. doi: 10.1016/j.jpain.2015.1.475. [DOI] [PubMed] [Google Scholar]

RESOURCES