Abstract
Background
There has been an increase in over dose deaths and emergency department visits (EDVs) involving use of prescription opioids, but the association between opioid prescribing and adverse outcomes is unclear.
Methods
Data were obtained from administrative claim records from Arkansas Medicaid and HealthCore commercially insured enrollees, 18 years and older, who used prescription opioids for at least 90 continuous days within a 6-month period between 2000 and 2005 and had no cancer diagnoses. Regression analysis was used to examine risk factors for EDVs and alcohol- or drug-related encounters (ADEs) in the 12 months following 90 days or more of prescribed opioids.
Results
Headache, back pain, and preexisting substance use disorders were significantly associated with EDVs and ADEs. Mental health disorders were associated with EDVs in HealthCore enrollees and with ADEs in both samples. Opioid dose per day was not consistently associated with EDVs but doubled the risk of ADEs at morphine-equivalent doses over 120 mg/d. Use of short-acting Drug Enforcement Agency Schedule II opioids was associated with EDVs compared with use of non–Schedule II opioids alone (relative risk range, 1.09–1.74). Use of Schedule II long-acting opioids was strongly associated with ADEs (relative risk range, 1.64–4.00).
Conclusions
Use of Schedule II opioids, headache, back pain, and substance use disorders are associated with EDVs and ADEs among adults prescribed opioids for 90 days or more. It may be possible to increase the safety of chronic opioid therapy by minimizing the prescription of Schedule II opioids in these higher-risk recipients.
Chronic opioid therapy (COT) is now a common strategy for managing chronic noncancer pain (CNCP).1–3 Concurrent with an increase in COT for CNCP has been an increase in reported opioid abuse and deaths from prescription opioid overdose. In Washington state, an increase in overdose deaths among individuals receiving workers’ compensation benefits was observed between 1996 and 2002, concurrent with a shift toward the use of more potent Schedule II opioids and a 50% increase in average daily morphine dose.4 Similar increases in opioid-related fatal poisoning rates have been found in Massachusetts5 and in Oregon.6 Nonfatal adverse events resulting from prescription opioid use also appear to be increasing. The Centers for Disease Control7 estimates that the number of emergency department (ED) visits for nonmedical use of opioid analgesics increased 111% from 2004 to 2008 (from 144 600 to 305 900 visits) and increased 29% just during the 2007–2008 period. Oxycodone, hydrocodone, and methadone were most frequently noted, and the use of all of these showed statistically significant increases.
However, it is unclear whether specific aspects of COT prescribing are related to adverse events. Deaths related to prescription opioid use might be owing to abuse and diversion of these medications, or they might be purely accidental.8 Demographic and clinical factors associated with increased adverse medical outcomes among COT prescription recipients are unknown.
In this study, we investigated the relationship between COT prescription and subsequent ED utilization. We studied a state Medicaid population and a multistate commercially insured population, using administrative claims data to examine the prevalence and risk factors for ED use and visits for alcohol- and drug-related encounters (ADEs) among individuals who used opioids for at least 90 continuous days.
METHODS
DATA SOURCE
The TROUP (Trends and Risks of Opioid Use for Pain) study3 was designed to assess trends in and risks of COT for noncancer pain among insured individuals with 5 tracer conditions: arthritis and/or joint pain, back pain, neck pain, headaches, and human immunodeficiency virus and/or AIDS. Data were obtained from claims records from January 2000 through December 2005 from 2 sources: The HealthCore Integrated Research Database (HIRD) (HealthCore Inc, Wilmington, Delaware), containing data from large, commercial insurance plans in 14 states; and Arkansas Medicaid. These 2 diverse populations were chosen to describe the range of opioid use in different populations. Owing to the retrospective nature of the study and the use of a limited data set, a waiver of the requirements for informed consent was granted from the human subjects review committees. Additional details of the study population have been reported elsewhere.3,9
STUDY SAMPLE
The analytical sample consisted of adult enrollees (age ≥18 years) who used opioids for at least 90 continuous days over a 6-month period between January 1, 2001, and December 31, 2004. Continuous use was defined as opioid prescription claims without a gap of 32 or more days between the end of the days supplied of one opioid prescription and the fill date of the next opioid prescription. The first day of the opioid use episode was defined as the index date. Individuals were required to have 12 months of continuous eligibility prior to and following the index date. Nursing home residents, hospice patients, and patients with a cancer diagnosis in the year prior to the index date (other than nonmelanoma skin cancer) were excluded. From HealthCore and Arkansas Medicaid, 38 491 and 10 159 enrollees, respectively, met the inclusion criteria.
SOCIODEMOGRAPHIC AND CLINICAL VARIABLES
Data on sociodemographic and clinical characteristics came from claims records in the 12-month period prior to the index date of the opioid use episode. The Charlson comorbidity index10 was used as a measure of overall medical comorbidity. In addition, we also collected information on International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) pain diagnoses in the 12 months before the index date. Arthritis and/or joint pain, back pain, neck pain, and headache were selected as tracer pain diagnoses to be tracked individually because these were the most commonly reported pain sites in the World Health Organization Collaborative Study of Psychological Problems in General Health Care.11 We also collected information on the presence of the following other (non-tracer) pain diagnoses: extremity pain, abdominal pain, chest pain, kidney stones and/or gallstones, pelvic pain, rheumatoid arthritis, fractures, neuropathic pain, fibromyalgia, and temporomandibular joint pain. These conditions were summed to create the number of nontracer pain diagnoses. Mental health and substance use disorders were classified into the following groups based on ICD-9-CM diagnoses using validated grouping software developed by the Agency for Healthcare Research and Quality (Clinical Classifications Software [CCS], 2008 release; Agency for Healthcare Research and Quality, Rockville, Maryland): adjustment disorders, anxiety disorders, mood disorders, personality disorders, and substance use disorders. Mood disorders were further classified as unipolar depressive disorder or bipolar disorder; substance use disorders were further classified as alcohol-use disorder, nonopioid-use disorder, or opioid-use disorder. Adjustment, anxiety, mood, and personality disorders were also summed to create a variable identifying the number of mental health disorder types.
HEALTH SERVICE USE VARIABLES
Data on ED visits and ADEs (alcohol intoxication, withdrawal, or overdose; drug intoxication or withdrawal; nonopioid drug overdose; or opioid drug overdose) were collected for the 12-month period following the start of the opioid use episode. Because the overall number was small, we included both outpatient and inpatient ADEs as well as those occurring in the ED. Of COT patients with ADEs, 65% from the HealthCore cohort (n=402) and 62% from Arkansas Medicaid (n=164) also had at least 1 ED visit during the 12-month period following the opioid use episode index date. Primary ICD-9-CM diagnoses for visits to the ED were tracked for the 12-month period following the index date of the opioid use episode. The ICD-9-CM diagnoses were classified into diagnostic categories using validated grouping software (CCS).
MEDICATION VARIABLES
Data on opioid use were collected for the 6-month period after the index date of the opioid use episode to match the time frame for the 90 days of opioid use required for inclusion in the sample. Opioids were categorized into 3 major groups: short-acting Drug Enforcement Agency (DEA) Schedule II opioids, long-acting DEA Schedule II opioids, and non–Schedule II opioids.3 The total morphine-equivalent dose (MED) for a single prescription was calculated by multiplying the quantity of each prescription by the strength of the prescription (milligrams of opioid per unit dispensed) and multiplying this total by a conversion factor.3 Opioid dose per day supplied was calculated by adding the total MEDs for the 3 major opioid groups and dividing by the sum of the total days’ supply (as entered by the dispensing pharmacist on each prescription claim) of the 3 major opioid groups. Opioid daily dose was further categorized into the following categories: less than the median (35 mg MED/d for the Arkansas sample and 32 mg MED/d for the HealthCore sample), from the median to 120 mg MED/d, and greater than 120 mg MED/d. The 120-mg threshold was chosen because Washington State Interagency Opioid Dosing Guidelines12 recommend pain specialty consultation for doses greater than this.
Types of opioid received were determined based on opioid class (as defined by DEA schedule and duration of action). Patients were coded as receiving a particular opioid class if they received at least 30 days’ supply of that class within a 6-month period. Seven mutually exclusive categories were thus derived: non–Schedule II, Schedule II short-acting, Schedule II long-acting, non–Schedule II plus Schedule II short-acting, Schedule II short-acting plus Schedule II long-acting, non–Schedule II plus Schedule II long-acting, and all 3 opioid types. Opioid days’ supply was divided empirically into 3 categories: 160 days or fewer, more than 160 to 185 days, and more than 185 days. Patients could receive more than 185 days’ supply if they took 2 or more different opioids simultaneously. We also collected data on receipt of sedative and/or hypnotic agents (benzodiazepines and nonbenzodiazepine sedative and/or hypnotic drugs).
STATISTICAL ANALYSIS
Variables were analyzed descriptively using frequencies and percentages for categorical variables and means with standard deviations for continuous variables.
The associations between health service use outcomes(ED visits and ADEs)and sociodemographic and clinical factors were examined using multiple regression analysis. For ED visits, Poisson regressions were used owing to the count nature of the data and skewed distributions. Negative binomial regressions were used for the number of ADEs owing to the high density of observations with 0 encounters. Regression coefficients and 95% confidence intervals were exponentiated to create estimates of relative risk.
Variables in the regression models were chosen based on factors previously shown in the literature to be associated with the outcomes of interest13–16 and on what was considered clinically relevant to the outcomes of interest. The regression models contained the following indicator variables: (1) sociodemographic characteristics (age and sex); (2) medical data (Charlson comorbidity index; dichotomous variables indicating the presence of arthritis and/or joint pain, back pain, neck pain, and headache; and number of nontracer pain conditions); (3) number of mental health disorders (binary indicator with no disorders as the reference group and 1 disorder and 2 or more disorders as the 2 categories of interest); (4) substance use disorders (alcohol use disorder, nonopioid drug use disorder, and opioid drug use disorder); and (5) medication use (days’ supply of opioids, opioid daily dose, and sedative and/or hypnotic agent use).
Dose is emerging as an issue of risk17 and policy.12 Opioid type (eg, Schedule II) has been a traditional focus of regulation and is one focus of planned new regulations.18 In addition, including type in the models also allowed us to address differing risks of long- vs short-acting preparations, which would not have been possible in a strictly dose-based analysis. Days’ supply of sedative and/or hypnotic agents was transformed by dividing by 30 for the purpose of easier interpretation (ie, the coefficient would now be interpreted as change in the model outcome for each change in 30-days’ supply of the medication).
Owing to the large sample sizes and the number of tests, P=.001 was chosen for statistical significance. Other P values are included only for descriptive purposes. STATA software, version 10.0 (StataCorp LLP, College Station, Texas), and SPSS, version 15.0 (IBM Company, Chicago, Illinois) were used for all analyses.
RESULTS
Table 1 lists the sociodemographic characteristics of adult enrollees in HealthCore and Arkansas Medicaid who used opioids for at least 90 continuous days within a 6-month period. On average, those undergoing COT were middle-aged and predominantly female. Approximately half of HealthCore COT patients and approximately three-quarters of those in the Arkansas Medicaid cohort had been diagnosed as having at least 1 noncancer tracer pain. The most prevalent diagnosis in both groups was back pain.
Table 1.
Demographic and Clinical Characteristics of Adult Enrollees Who Used Opioids Continuously for 90 Days or Longer During a 6-Month Period From 2000 to 2005a
| Characteristic | HealthCore (n=38 491) | Arkansas Medicaid (n=10 159) |
|---|---|---|
| Age, mean (SD), y | 50.4 (12.9) | 52.9 (16.2) |
| Female sex | 22 772 (59.2) | 7279 (71.7) |
| Charlson comorbidity index, mean (SD) | 0.4 (0.9) | 1.1 (1.4) |
| Any noncancer tracer pain diagnosis | 21 202 (55.1) | 7555 (74.4) |
| Arthritis and/or joint pain | 6776 (17.6) | 3733 (36.8) |
| Back pain | 14 181 (36.8) | 5104 (50.2) |
| Neck pain | 3944 (18.2) | 1709 (16.8) |
| Headache | 6237 (16.2) | 2159 (21.3) |
| Tracer pain types, mean (SD), No. | 0.9 (0.9) | 1.2 (1.0) |
| Any noncancer nontracer pain diagnosis | 22 243 (57.8) | 7645 (75.3) |
| Extremity pain | 12 985 (33.7) | 4655 (45.8) |
| Abdominal pain | 7474 (19.4) | 3255 (32.0) |
| Chest pain | 4963 (12.9) | 2828 (27.8) |
| Kidney/gallstones | 1145 (3.0) | 389 (3.8) |
| Pelvic pain | 3618 (9.4) | 875 (8.6) |
| Rheumatoid arthritis | 1307 (3.4) | 555 (5.5) |
| Fractures | 4042 (10.5) | 2209 (21.7) |
| Neuropathic pain | 2164 (5.6) | 500 (4.9) |
| Fibromyalgia | 3280 (8.5) | 1027 (10.1) |
| Temporomandibular joint | 285 (0.7) | 73 (0.7) |
| Nontracer pain types, mean (SD), No. | 1.1 (1.2) | 1.6 (1.4) |
| Any diagnosed mental health disorder | 6411 (16.7) | 3098 (30.5) |
| Adjustment disorder | 747 (1.9) | 200 (2.0) |
| Anxiety disorder | 2755 (7.2) | 1713 (16.9) |
| Unipolar depression | 2732 (7.1) | 2030 (20.0) |
| Bipolar disorder | 2220 (5.8) | 281 (2.8) |
| Personality disorder | 77 (0.2) | 186 (1.8) |
| Mental health disorder diagnoses, mean (SD), No. | 0.2 (0.5) | 0.4 (0.7) |
| Any diagnosed substance use disorder | 960 (2.5) | 632 (6.2) |
| Alcohol abuse or dependence | 472 (1.2) | 296 (2.9) |
| Nonopioid drug abuse or dependence | 464 (1.2) | 378 (3.7) |
| Opioid abuse or dependence | 270 (0.7) | 64 (0.6) |
Unless otherwise indicated, data are reported as number (percentage) of patients.
Table 2 lists the ED visits and ADEs among adult enrollees in HealthCore and Arkansas Medicaid who used opioids for at least 90 continuous days within a 6-month period. One-quarter to one-third of enrollees in both insurers had an ED visit in the 12 months following the index date of COT use, and fewer than 3% had an ADE.
Table 2.
Twelve-Month Health Service Utilization by Adult Enrollees Who Used Opioids Continuously for 90 Days or Longer During a 6-Month Period From 2000 to 2005a
| Characteristic | HealthCore (n=38 491) | Arkansas Medicaid (n=10 159) |
|---|---|---|
| Any ED visits | 9297 (24.2) | 2863 (28.2) |
| ED visits, mean (SD), No. | 0.5 (1.5) | 1.1 (3.2) |
| Any alcohol- or drug-related encounters | 622 (1.6) | 264 (2.6) |
| Alcohol withdrawal | 94 (0.2) | 22 (0.2) |
| Alcohol intoxication | 4 (0.0) | 28 (0.3) |
| Alcohol overdose | 12 (0.0) | 9 (0.1) |
| Drug withdrawal | 271 (0.7) | 86 (0.8) |
| Drug intoxication | 68 (0.2) | 25 (0.2) |
| Nonopioid drug overdose | 179 (0.5) | 104 (1.0) |
| Opioid overdose | 96 (0.2) | 36 (0.4) |
| Alcohol- or drug-related encounters, mean (SD), No. | 0.02 (0.20) | 0.04 (0.60) |
Abbreviation: ED, emergency department.
Unless otherwise indicated, data are reported as number (percentage) of patients. All ED visits and alcohol and/or drug encounters are for the time period 12 months after the index date of initiation of long-term opioid use.
PRIMARY DIAGNOSES FOR VISITS TO THE ED
The most common primary diagnoses given for the first ED visit in the HealthCore cohort were headache (10.0% of visits), back problems (9.9%), abdominal pain (6.8%), sprains and strains (6.6%), and diseases of the heart (6.3%). The most common primary diagnoses given for the first ED visit in the Arkansas Medicaid cohort were back problems (10.4% of visits), diseases of the heart (7.7%), headache (5.3%), respiratory infections (5.3%), and sprains and strains (5.2%).
The 2 samples had similar opioid usage (Table 3). About 55% to 60% of the samples had less than a 160-day supply, and three-quarters of the samples used non–Schedule II opioids only. Forty-three percent of the samples had at least some use of a sedative and/or hypnotic agent for an average of about 2 months.
Table 3.
Opioid and Sedative and/or Hypnotic Use by Adult Enrollees Who Used Opioids Continuously for 90 Days or Longer During a 6-Month Period From 2000 to 2005a
| Characteristic | HealthCore (n=38 491) | Arkansas Medicaid (n=10 159) |
|---|---|---|
| Opioid daily dose | ||
| <Medianb | 19 190 (49.9) | 5152 (50.7) |
| Median to 120 mg MED | 16 552 (43.0) | 4385 (43.2) |
| >120 mg MED | 2749 (7.1) | 622 (6.1) |
| Opioid days’ supply, d | ||
| ≤160 | 23 020 (59.8) | 5497 (54.1) |
| >160 to 185 | 5740 (14.9) | 1804 (17.8) |
| >185 | 9731 (25.3) | 2858 (28.1) |
| Opioid type | ||
| Non–Schedule II only | 29 778 (77.4) | 7969 (78.4) |
| Schedule II, short-acting only | 794 (2.1) | 257 (2.5) |
| Schedule II, long-acting only | 2056 (5.3) | 534 (5.3) |
| Non–Schedule II and Schedule II short-acting | 1138 (3.0) | 311 (3.1) |
| Non–Schedule II and Schedule II long-acting | 3015 (7.8) | 669 (6.6) |
| Schedule II, both long- and short-acting | 1037 (2.7) | 283 (2.8) |
| Non–Schedule II and Schedule II, both long- and short-acting | 673 (1.7) | 136 (1.3) |
| Any use of sedative and/or hypnotic | 16 617 (43.2) | 4329 (42.6) |
| ≥30 d of benzodiazepine | 11 489 (29.8) | 3629 (35.7) |
| ≥30 d of nonbenzodiazepine sedative and/or hypnotic | 4408 (11.5) | 674 (6.6) |
| Duration of sedative and/or hypnotic use, mean (SD), mo | 1.7 (2.8) | 1.8 (2.6) |
Abbreviations: MED, morphine-equivalent dose.
Unless otherwise indicated, data are reported as number (percentage) of patients. Medication use occurred in the time period 6 months after the index date of initiation of long-term opioid use.
Median dose was 32 mg MED for HealthCore and 35 mg MED for Arkansas Medicaid.
Table 4 lists the mean daily opioid dose for each of the 7 categories of opioid type. For both samples, the lowest mean opioid dose was among those receiving only non–Schedule II drugs, and the highest mean opioid dose was among those receiving both Schedule II long-acting and Schedule II short-acting drugs.
Table 4.
Daily Opioid Dose by Opioid Type Among Adult Enrollees Who Used Opioids Continuously for 90 Days or Longer During a 6-Month Period From 2000 to 2005a
| Opioid Type | HealthCore | Arkansas Medicaid |
|---|---|---|
| Non–Schedule II only | 35.8 (29.0) | 40.9 (29.1) |
| Schedule II, short-acting only | 64.7 (115.8) | 42.0 (44.2) |
| Schedule II, long-acting only | 156.9 (223.5) | 115.7 (108.0) |
| Non–Schedule II and Schedule II short-acting | 47.8 (33.5) | 43.5 (27.9) |
| Non–Schedule II and Schedule II long-acting | 89.0 (115.2) | 83.9 (72.3) |
| Schedule II, both long- and short-acting | 216.7 (313.5) | 192.1 (198.5) |
| Non–Schedule II and Schedule II, both long- and short-acting | 146.1 (246.6) | 52.9 (63.1) |
Data are reported as mean (SD) daily morphine-equivalent dose in milligrams.
We also examined opioid dose by frequency of ED use. Mean (SD) daily opioid dose increased from 53.5 (98.6) mg MED for the HealthCore COT users with 0 to 2 ED visits to 71.6 (141.5) mg MED among those with 3 or more ED visits in the year following the initiation of COT (P<.001). For Arkansas Medicaid, the mean daily dose did not differ significantly between the 2 groups (52.8 and 53.6 mg MED).
REGRESSION MODELS
Table 5 and Table 6 list the relative risks for variables associated with ED visits and ADEs among adult Health-Core and Arkansas Medicaid COT users in the first 12 months following the index date of COT use, unadjusted and adjusted for all other variables in the models. For both samples, most of the unadjusted relative risks were statistically significant (P <.001). Results reported herein are for the adjusted models.
Table 5.
Variables Associated With Emergency Department Visits Among Adult Enrollees Who Used Opioids Continuously for 90 Days or Longer During a 6-Month Period From 2000 to 2005 in the First Year After Initiation of Continuous Opioid Usea
| Characteristic | HealthCore (n=38 491)
|
Arkansas Medicaid (n=10 159)
|
||
|---|---|---|---|---|
| Unadjusted | Adjustedb | Unadjusted | Adjustedb | |
| Age, y | 0.97 (0.97–0.98)c | 0.975 (0.97-0.97)c | 0.95 (0.94–0.95)c | 0.95 (0.95-0.95)c |
| Female sex | 1.31 (1.27–1.35)c | 1.13 (1.09–1.16)c | 1.21 (1.16–1.26)c | 1.20 (1.14–1.26)c |
| Charlson comorbidity index | 1.14 (1.12–1.15)c | 1.08 (1.07–1.10)c | 1.05 (1.04–1.07)c | 1.08 (1.06–1.09)c |
| Tracer pain type | ||||
| Arthritis and/or joint pain | 0.96 (0.92–0.99)d | 0.87 (0.83–0.90)c | 0.60 (0.58–0.63)c | 0.84 (0.81–0.88)c |
| Back pain | 1.54 (1.49–1.58)c | 1.11 (1.07–1.14)c | 1.91 (1.83–1.98)c | 1.32 (1.26–1.38)c |
| Neck pain | 1.55 (0.50–0.61)c | 0.93 (0.90–0.97)c | 1.47 (1.41–1.54)c | 0.89 (0.85–0.94)c |
| Headache | 2.55 (2.46–2.62)c | 1.68 (1.63–1.74)c | 2.46 (2.36–2.55)c | 1.30 (1.25–1.36)c |
| Number of nontracer pain conditions | 1.38 (1.36–1.39)c | 1.23 (1.21–1.24)c | 1.44 (1.43–1.46)c | 1.28 (1.27–1.30)c |
| Mental health disorders, No. | ||||
| 0 | 1 [Reference] | 1 [Reference] | 1 [Reference] | 1 [Reference] |
| 1 | 1.73 (1.66–1.80)d | 1.12 (1.08–1.16)c | 1.60 (1.52–1.67)d | 0.94 (0.90–0.99)d |
| ≥2 | 2.68 (2.56–2.81)c | 1.26 (1.20–1.33)c | 2.55 (1.43–2.67)c | 1.03 (0.98–1.09) |
| Substance abuse or dependence | ||||
| Alcohol | 2.37 (2.18–2.59)c | 1.31 (1.19–1.43)c | 1.74 (1.60–1.90)c | 1.18 (1.08–1.30)c |
| Nonopioid drug | 4.48 (4.19–4.78)c | 1.73 (1.60–1.88)c | 3.30 (3.10–3.50)c | 1.65 (1.55–1.76)c |
| Opioid drug | 4.66 (4.29–5.06)c | 1.51 (1.37–1.67)c | 3.03 (2.64–3.48)c | 1.22 (1.06–1.41)d |
| Supply of sedative and/or hypnotic medication per 30 d | 1.06 (1.05–1.06)c | 1.03 (1.02–1.04)c | 1.04 (1.03–1.05)c | 1.02 (1.01–1.02)c |
| Opioid daily dose, mg MED | ||||
| <Median | 1 [Reference] | 1 [Reference] | 1 [Reference] | 1 [Reference] |
| Median to 120 | 1.61 (1.56–1.66)c | 1.30 (1.26–1.34)c | 1.00 (0.96–1.04) | 1.03 (0.99–1.07) |
| >120 | 1.58 (1.49–1.67)c | 1.08 (1.02–1.15)d | 1.16 (1.07–1.25)c | 0.97 (0.89–1.06) |
| Days’ supply of opioids, d | ||||
| ≤160 | 1 [Reference] | 1 [Reference] | 1 [Reference] | 1 [Reference] |
| >160 to ≤185 | 1.03 (0.98–1.07) | 1.01 (0.97–1.06) | 0.83 (0.79–0.88)c | 0.88 (0.83–0.92)c |
| >185 | 1.36 (1.32–1.40)c | 1.02 (0.99–1.06) | 1.03 (0.99–1.08) | 0.94 (0.89–0.98)d |
| Opioid type | ||||
| Non–Schedule II only | 1 [Reference] | 1 [Reference] | 1 [Reference] | 1 [Reference] |
| Schedule II, short-acting only | 1.62 (1.49–1.77)c | 1.31 (1.20–1.44)c | 2.30 (2.11–2.51)c | 1.64 (1.50–1.79)c |
| Schedule II, long-acting only | 1.34 (1.26–1.42)c | 1.00 (0.93–1.07) | 1.27 (1.17–1.38)c | 0.97 (0.89–1.06) |
| Non–Schedule II and Schedule II short-acting | 2.41 (2.26–2.56)c | 1.74 (1.64–1.86)c | 2.24 (2.07–2.43)c | 1.55 (1.43–1.69)c |
| Non–Schedule II and Schedule II long-acting | 1.88 (1.80–1.97)c | 1.39 (1.32–1.46)c | 1.54 (1.45–1.66)c | 1.06 (0.99–1.14) |
| Schedule II, both long- and short-acting | 1.65 (1.53–1.78)c | 1.16 (1.07–1.27)c | 1.58 (1.43–1.75)c | 1.09 (0.97–1.22) |
| Non–Schedule II and Schedule II, both long- and short-acting | 2.82 (2.62–3.03)c | 1.78 (1.65–1.93)c | 2.04 (1.80–2.32)c | 1.24 (1.09–1.42)c |
Abbreviation: MED, morphine-equivalent dose.
All data are reported as relative risk (95% confidence interval). All covariates are from the 12-month preindex date of initiation of continuous opioid use except for pharmacologic variables, which are from the 6 months after the index date.
Adjusted for all other model terms.
P<.001.
P<.01.
Table 6.
Variables Associated With Alcohol- or Drug-Related Encounters Among Adult Enrollees Who Used Opioids Continuously for 90 Days or Longer During a 6-Month Period From 2000 to 2005 in the First Year After Initiation of Continuous Opioid Usea
| Characteristic | HealthCore (n=38 491)
|
Arkansas Medicaid (n=10 159)
|
||
|---|---|---|---|---|
| Unadjusted | Adjustedb | Unadjusted | Adjustedb | |
| Age, y | 0.96 (0.95–0.97)c | 0.97 (0.96–0.98)c | 0.97 (0.96–0.98)c | 0.99 (0.98–1.00) |
| Female sex | 0.92 (0.78–1.10) | 0.88 (0.74–1.05) | 0.52 (0.38–0.72)c | 0.68 (0.50–0.93)e |
| Charlson comorbidity index | 1.09 (1.01–1.19)e | 1.04 (0.95–1.13) | 1.03 (0.92–1.14) | 1.03 (0.93–1.13) |
| Tracer pain type | ||||
| Arthritis and/or joint pain | 0.94 (0.75–1.19) | 0.99 (0.78–1.26) | 0.43 (0.31–0.60)c | 0.64 (0.46–0.89)d |
| Back pain | 1.79 (1.51–2.12)c | 1.22 (1.02–1.46)e | 1.96 (1.45–2.26)c | 1.08 (0.79–1.46) |
| Neck pain | 1.52 (1.22–1.89)c | 0.92 (0.73–1.16) | 2.35 (1.63–3.40)c | 1.19 (0.82–1.71) |
| Headache | 1.82 (1.48–2.24)c | 1.15 (0.93–1.42) | 3.09 (2.22–4.29)c | 1.73 (1.25–2.40)c |
| Number of nontracer pain conditions | 1.25 (1.18–1.34)c | 1.01 (0.93–1.08) | 1.36 (1.24–1.50)c | 1.03 (0.92–1.16) |
| Mental health disorders, No. | ||||
| 0 | 1 [Reference] | 1 [Reference] | 1 [Reference] | 1 [Reference] |
| 1 | 3.22 (2.63–3.96)c | 1.90 (1.54–2.35)c | 1.97 (1.38–2.82)c | 1.31 (0.92–1.87) |
| ≥2 | 5.61 (4.43–7.61)c | 2.02 (1.51–2.70)c | 5.54 (3.70–8.82)c | 2.29 (1.52–3.44)c |
| Substance abuse or dependence | ||||
| Alcohol | 9.80 (6.34–15.16)c | 4.85 (3.29–7.14)c | 7.73 (3.87–15.45)c | 5.12 (2.97–8.82)c |
| Nonopioid drug | 14.55 (9.88–21.44)c | 3.13 (2.11–4.64)c | 10.13 (5.72–17.94)c | 2.95 (1.73–5.01)c |
| Opioid drug | 15.59 (9.16–26.52)c | 3.29 (2.02–5.37)c | 8.64 (8.64–40.39)d | 2.73 (0.89–8.40) |
| Supply of sedative and/or hypnotic medication per 30 d | 1.15 (1.12–1.18)c | 1.11 (1.08–1.14)c | 1.06 (1.01–1.13)e | 1.08 (1.02–1.14)d |
| Opioid daily dose, mg MED | ||||
| <Median | 1 [Reference] | 1 [Reference] | 1 [Reference] | 1 [Reference] |
| Median to 120 | 2.51 (2.07–3.04)c | 1.66 (1.35–2.03)c | 1.67 (1.22–2.28)c | 1.17 (0.85–1.61) |
| >120 | 5.11 (3.89–6.71)c | 2.18 (1.58–3.00)c | 4.01 (2.29–7.03)c | 2.06 (1.15–3.67)e |
| Days’ supply of opioids, d | ||||
| ≤160 | 1 [Reference] | 1 [Reference] | 1 [Reference] | 1 [Reference] |
| >160 to ≤185 | 1.10 (0.84–1.43) | 0.98 (0.76–1.27) | 0.70 (0.45–1.08) | 0.86 (0.56–1.30) |
| >185 | 2.20 (1.83–2.65)c | 1.07 (0.86–1.32) | 1.34 (0.95–1.88) | 0.93 (0.64–1.34) |
| Opioid type | ||||
| Non–Schedule II only | 1 [Reference] | 1 [Reference] | 1 [Reference] | 1 [Reference] |
| Schedule II, short-acting only | 2.18 (1.30–3.63)d | 1.34 (0.80–2.26) | 1.30 (0.51–3.30) | 1.41 (0.60–3.32) |
| Schedule II, long-acting only | 3.06 (2.27–4.12)c | 1.64 (1.18–2.26)d | 6.52 (3.88–10.97)c | 2.47 (1.42–4.31)c |
| Non–Schedule II and Schedule II short-acting | 2.21 (1.44–3.40)c | 1.56 (1.02–2.40)e | 2.50 (1.19–5.27)e | 1.27 (0.60–2.66) |
| Non–Schedule II and Schedule II long-acting | 3.49 (2.73–4.47)c | 2.22 (1.70–2.90)c | 3.43 (2.08–5.67)c | 2.63 (1.58–4.38)c |
| Schedule II, both long- and short-acting | 3.79 (2.58–5.56)c | 1.89 (1.26–2.84)d | 4.45 (2.17–9.15)c | 1.87 (0.91–3.86) |
| Non–Schedule II and Schedule II, both long- and short-acting | 6.89 (4.58–10.36)c | 3.20 (2.11–4.83)c | 7.36 (2.77–19.53)c | 4.00 (1.68–9.53)d |
Abbreviation: MED, morphine-equivalent dose.
All data are reported as relative risk (95% confidence interval). All covariates are from the 12-month preindex date of initiation of continuous opioid use except for pharmacologic variables, which are from the 6 months after the index date.
Adjusted for all other model terms.
P<.001.
P<.01.
P<.05.
Younger age, female sex, more medical comorbidities, presence of back pain, presence of headaches, and greater number of nontracer pain conditions were associated with more ED visits. Alcohol, nonopioid drug use, and opioid abuse or dependence each showed strong association with ED utilization for both insurers. Presence of mental health disorders was significantly associated with ED utilization only in the HealthCore sample.
Opioid dose over the median but below 120 mg/d was significantly associated with ED visits in the Health-Core sample but not in Arkansas Medicaid. Dose greater than 120 mg was not significantly associated with ED visits in either sample. Days’ supply of opioids was not significant in the HealthCore model. In the Arkansas Medicaid sample, a greater than 160 but less than 185 days’ supply was protective compared with a supply of less than 160 days. Use of Schedule II short-acting opioids and use of sedative and/or hypnotic agents each was associated with increased ED visits in both samples.
Older age and presence of arthritis and/or joint pain were negatively related to the number of ADEs. Higher number of mental health disorders was positively associated with ADEs in the HealthCore sample, and 2 or more mental health disorders increased the risk of encounters over 2-fold in the Arkansas Medicaid sample. Alcohol and/or nonopioid drug abuse or dependence were very strongly related to ADEs. Opioid abuse or dependence was strongly positively related to ADEs in the Health-Core sample but not in the Arkansas Medicaid model.
Opioid dose of more than 120 mg/d was associated with an increased number of ADEs, but the increase was statistically significant only in the HealthCore sample (P<.001). In general, days’ supply of opioids was not related to ADEs. The results for opioid type were mixed. In the HealthCore model, only use of non–Schedule II plus Schedule II long-acting drugs or use of all 3 types reached statistical significance (P≤.001). In the Arkansas Medicaid sample, use of Schedule II long-acting drugs only or use of non–Schedule II plus Schedule II long-acting drugs was significantly associated with more ADEs. Use of sedative and/or hypnotic medication was related to ADEs in the HealthCore sample but not in the Arkansas Medicaid model.
COMMENT
Among COT users, previously diagnosed substance use disorders and opioid types received were the variables most strongly associated with ED visits and ADEs in the subsequent year. Any use of short-acting Schedule II opioids was associated with a higher number of ED visits, while any use of long-acting Schedule II opioids was associated with more ADEs. Opioid days’ supply was not a predictor of these outcomes. Higher opioid daily dose was inconsistently associated with these outcomes. A dose between the median and 120 mg MED/d was associated with ED visits in the HealthCore population only. Doses higher than 120 mg MED/d were not associated with ED visits in either population but were associated with a doubling of risk of ADEs in both populations. Thus, contrary to our hypothesis, we did not observe a clear dose response effect for the health service outcomes we examined.
Daily opioid dose has lately received more scientific and regulatory attention as a risk factor for poor outcomes, but to our knowledge it has not previously been examined in models that also control for opioid type and days’ dose supplied; our data suggest that receipt of any high-potency Schedule II opioids (vs non–Schedule II drugs) may be a more important determinant of medical risk associated with COT. This may be owing to the greater difficulty of safely adjusting the dose for higher-potency opioids. The use of Schedule II opioids may be a sign of more-difficult-to-manage pain and possibly more frequent dose adjustments. Visits to the ED could be prompted by adverse effects occurring in the context of dose adjustment.
Preexisting substance abuse has been repeatedly identified as a risk factor for abuse of prescribed opioids, but our data suggest that it also poses significant risks for ED visits and ADEs. This might be owing to the association of substance abuse diagnoses with opioid misuse.19–22 Our results suggest that preexisting substance abuse diagnoses should be included in any risk stratification scheme for COT and likely indicate the need for closer follow-up of these individuals.23
In our sample of COT recipients, fewer than 3% had any ADEs in the year following 90 days or longer opioid use. However, those with a substance use disorder diagnosis in the preceding year had from 18% to 73% more ED utilization and were 3 to 5 times more likely to have an ADE. According to 2006 DAWN data,7 approximately 1.5% of ED visits were associated with drug misuse or abuse.
In prior studies, ongoing substance abuse or a history of such abuse has been shown to be associated with an increased risk of prescription drug misuse.19,21,22,24 There has been an increase in the reported number of overdoses from prescription opioids, but studies are only now revealing the connection between the likelihood of overdose and the prescribed opioid dose or type.4–6 In our study, combining sedative and/or hypnotic drugs with prescription opioids increased the likelihood of ED visits and ADEs. In a recent study of overdose deaths in West Virginia,8 79% of the patients who died of overdose also tested positive for alcohol and other drugs. Because our study identified opioid abuse or dependence based on diagnoses provided in claims records, we were unable to clinically validate the diagnoses or characterize the particular opioids of abuse. Mental health disorder diagnoses were also associated with the outcomes examined in our study.15,16 However, the associations were generally not as strong or as consistent as for substance use disorder diagnoses.
Strengths of this study include the availability of data from 2 large samples representing different payer types, regions, and socioeconomic status, thus increasing the generalizability of the findings. We also examined a broad range of medical, psychiatric, and medication use variables potentially associated with the health service utilization outcomes examined.
Limitations of this study are those associated with the use of administrative data. Diagnoses were based on ICD-9-CM codes recorded in clinical encounters. These tend to be specific but not sensitive,25 which means that all cases are not detected, but few are misclassified. We had no information on the effects of opioid therapy on pain intensity or pain interference with daily activities, so we are unable to address these potential benefits of opioid therapy. Data on opioid and other medication use were limited to prescriptions dispensed and may not reflect how patients actually used the medications. We lacked information on the severity of different conditions, particularly mental health and pain disorders, that might influence health service utilization. Sociodemographic data available for both samples were limited to age and sex; therefore, we were unable to control for race or income, additional characteristics that may influence health service use. We controlled for the most common pain conditions, but it is possible that there may be other pain and nonpain conditions that may influence the outcomes examined and that were not adequately controlled for in our models. It is possible that there are other factors important in the association between opioid use and ED visits and ADEs that were not accounted for in our models. While the results of our study demonstrate a strong association between opioid type and ED visits and ADEs, we do not have the information necessary to claim that these visits were caused by prescription opioid use. In addition, everyone in the study sample had at least 90 days of opioid use, and thus our findings may not apply to populations with more short-term or intermittent opioid use.
In summary, this report describes clinical and demographic characteristics associated with ED visits and ADEs among adult enrollees in a state Medicaid and commercially insured population who used prescription opioids for at least 90 continuous days Receipt of any Schedule II opioids and substance use disorder diagnoses were the variables most strongly associated with the health service utilization outcomes examined. Short-acting Schedule II opioids were associated with increased numbers of ED visits, while long-acting Schedule II opioids were associated with increased numbers of ADEs. It may be possible to increase the safety of COT by minimizing the prescription of Schedule II opioids in higher-risk recipients. To increase the safety of COT, further research is warranted to investigate in more detail the relationship observed between receipt of Schedule II opioids and ED use and the interaction between opioid type received and other clinical risk factors identified.
Acknowledgments
Funding/Support: This research was supported by grant DA022560 from the National Institute on Drug Abuse (Dr Sullivan). Dr Braden was supported by Ruth L. Kirschstein National Research Service Award Institutional Research Training Grant T32 MH20021.
Footnotes
Financial Disclosure: None reported.
Author Contributions: Study concept and design: Braden, Fan, Edlund, Martin, DeVries, and Sullivan. Acquisition of data: Martin and DeVries. Analysis and interpretation of data: Braden, Russo, Fan, Edlund, Martin, and Sullivan. Drafting of the manuscript: Braden and Russo. Critical revision of the manuscript for important intellectual content: Braden, Fan, Edlund, Martin, DeVries, and Sullivan. Statistical analysis: Russo and Fan. Obtained funding: Edlund, DeVries, and Sullivan. Administrative, technical, and material support: Martin, DeVries, and Sullivan. Study supervision: Sullivan.
Additional Contributions: Gary Moore, MS, and Tom Puenpatom, MPH, provided invaluable programming support in preparing the analytic files (Mr Puenpatom is employed by HealthCore). The Arkansas Department of Human Services, Division of Medical Services, provided no-cost access to the Arkansas Medicaid claims files.
References
- 1.Caudill-Slosberg MA, Schwartz LM, Woloshin S. Office visits and analgesic prescriptions for musculoskeletal pain in US: 1980 vs. 2000. Pain. 2004;109(3):514–519. doi: 10.1016/j.pain.2004.03.006. [DOI] [PubMed] [Google Scholar]
- 2.Olsen Y, Daumit GL. Opioid prescribing for chronic nonmalignant pain in primary care: challenges and solutions. Adv Psychosom Med. 2004;25:138–150. doi: 10.1159/000079063. [DOI] [PubMed] [Google Scholar]
- 3.Sullivan MD, Edlund MJ, Fan MY, Devries A, Brennan Braden J, Martin BC. Trends in use of opioids for non-cancer pain conditions 2000–2005 in commercial and Medicaid insurance plans: the TROUP study. Pain. 2008;138(2):440–449. doi: 10.1016/j.pain.2008.04.027. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Franklin GM, Mai J, Wickizer T, Turner JA, Fulton-Kehoe D, Grant L. Opioid dosing trends and mortality in Washington State workers’ compensation, 1996–2002. Am J Ind Med. 2005;48(2):91–99. doi: 10.1002/ajim.20191. [DOI] [PubMed] [Google Scholar]
- 5.Fernandez W, Hackman H, McKeown L, Anderson T, Hume B. Trends in opioid-related fatal overdoses in Massachusetts, 1990–2003. J Subst Abuse Treat. 2006;31(2):151–156. doi: 10.1016/j.jsat.2006.04.008. [DOI] [PubMed] [Google Scholar]
- 6.Oregon Department of Health Services. [Accessed January 5, 2010];Methadone deaths (and distribution) on the rise. http://www.oregon.gov/DHS/ph/cdsummary/2003/ohd5214.pdf.
- 7.Centers for Disease Control and Prevention (CDC) Emergency department visits involving nonmedical use of selected prescription drugs - United States, 2004–2008. MMWR Morb Mortal Wkly Rep. 2010;59(23):705–709. [PubMed] [Google Scholar]
- 8.Hall AJ, Logan JE, Toblin RL, et al. Patterns of abuse among unintentional pharmaceutical overdose fatalities. JAMA. 2008;300(22):2613–2620. doi: 10.1001/jama.2008.802. [DOI] [PubMed] [Google Scholar]
- 9.Braden JB, Fan MY, Edlund MJ, Martin BC, DeVries A, Sullivan MD. Trends in use of opioids by noncancer pain type 2000–2005 among Arkansas Medicaid and HealthCore enrollees: results from the TROUP study. J Pain. 2008;9(11):1026–1035. doi: 10.1016/j.jpain.2008.06.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40(5):373–383. doi: 10.1016/0021-9681(87)90171-8. [DOI] [PubMed] [Google Scholar]
- 11.Gureje O, Von Korff M, Simon GE, Gater R. Persistent pain and well-being: a World Health Organization Study in Primary Care. JAMA. 1998;280(2):147–151. doi: 10.1001/jama.280.2.147. [DOI] [PubMed] [Google Scholar]
- 12.Washington State Agency Medical Directors’ Group. [Accessed January 5, 2010];Interagency guideline on opioid dosing for chronic non-cancer pain: an educational pilot to improve care and safety with opioid treatment. http://www.agencymeddirectors.wa.gov/opioiddosing.asp.
- 13.DeFrances CJ, Lucas CA, Buie VC, Golosinskiy A. 2006 National Hospital Discharge Survey. Natl Health Stat Report. 2008 Jul;30(5):1–20. [PubMed] [Google Scholar]
- 14.Pitts SR, Niska RW, Xu J, Burt CW. National Hospital Ambulatory Medical Care Survey: 2006 emergency department summary. Natl Health Stat Report. 2008 Aug;6(7):1–38. [PubMed] [Google Scholar]
- 15.de Boer AG, Wijker W, de Haes HC. Predictors of health care utilization in the chronically ill: a review of the literature. Health Policy. 1997;42(2):101–115. doi: 10.1016/s0168-8510(97)00062-6. [DOI] [PubMed] [Google Scholar]
- 16.Fogarty CT, Sharma S, Chetty VK, Culpepper L. Mental health conditions are associated with increased health care utilization among urban family medicine patients. J Am Board Fam Med. 2008;21(5):398–407. doi: 10.3122/jabfm.2008.05.070082. [DOI] [PubMed] [Google Scholar]
- 17.Dunn KM, Saunders KW, Rutter CM, et al. Opioid prescriptions for chronic pain and overdose: a cohort study. Ann Intern Med. 2010;152(2):85–92. doi: 10.1059/0003-4819-152-2-201001190-00006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Woodcock J. A difficult balance--pain management, drug safety, and the FDA. N Engl J Med. 2009;361(22):2105–2107. doi: 10.1056/NEJMp0908913. [DOI] [PubMed] [Google Scholar]
- 19.Dowling K, Storr CL, Chilcoat HD. Potential influences on initiation and persistence of extramedical prescription pain reliever use in the US population. Clin J Pain. 2006;22(9):776–783. doi: 10.1097/01.ajp.0000210926.41406.2c. [DOI] [PubMed] [Google Scholar]
- 20.Edlund MJ, Sullivan M, Steffick D, Harris KM, Wells KB. Do users of regularly prescribed opioids have higher rates of substance use problems than nonusers? Pain Med. 2007;8(8):647–656. doi: 10.1111/j.1526-4637.2006.00200.x. [DOI] [PubMed] [Google Scholar]
- 21.Reid MC, Engles-Horton LL, Weber MB, Kerns RD, Rogers EL, O’Connor PG. Use of opioid medications for chronic noncancer pain syndromes in primary care. J Gen Intern Med. 2002;17(3):173–179. doi: 10.1046/j.1525-1497.2002.10435.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Tetrault JM, Desai RA, Becker WC, Fiellin DA, Concato J, Sullivan LE. Gender and non-medical use of prescription opioids: results from a national US survey. Addiction. 2008;103(2):258–268. doi: 10.1111/j.1360-0443.2007.02056.x. [DOI] [PubMed] [Google Scholar]
- 23.Chou R, Fanciullo GJ, Fine PG, et al. American Pain Society-American Academy of Pain Medicine Opioids Guidelines Panel. Clinical guidelines for the use of chronic opioid therapy in chronic noncancer pain. J Pain. 2009;10(2):113–130. doi: 10.1016/j.jpain.2008.10.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Edlund MJ, Steffick D, Hudson T, Harris KM, Sullivan M. Risk factors for clinically recognized opioid abuse and dependence among veterans using opioids for chronic non-cancer pain. Pain. 2007;129(3):355–362. doi: 10.1016/j.pain.2007.02.014. [DOI] [PubMed] [Google Scholar]
- 25.Quan H, Li B, Saunders LD, et al. IMECCHI Investigators. Assessing validity of ICD-9-CM and ICD-10 administrative data in recording clinical conditions in a unique dually coded database. Health Serv Res. 2008;43(4):1424–1441. doi: 10.1111/j.1475-6773.2007.00822.x. [DOI] [PMC free article] [PubMed] [Google Scholar]