Abstract
Opioid overprescribing is a major driver of the current opioid overdose epidemic. However, annual opioid prescribing in the USA dropped from 782 to 640 morphine milligram equivalents per capita between 2010 and 2015, while opioid overdose deaths increased by 63%. To better understand the role of prescription opioids and health care utilization prior to opioid-related overdose, we analyzed the death records of decedents who died of an opioid overdose in Illinois in 2016 and linked to any existing controlled substance monitoring program (CSMP) and emergency department (ED) or hospital discharge records. We found that of the 1893 opioid-related overdoses, 573 (30.2%) decedents had not filled an opioid analgesic prescription within the 6 years prior to death. Decedents without an opioid prescription were more likely to be black (33.3% vs 20.2%, p < .001), Hispanic (16.3% vs 8.8%, p < .001), and Chicago residents (46.8% vs 25.6%, p < .001) than decedents with at least one filled opioid prescription. Decedents who did not fill an opioid prescription were less likely to die of an overdose involving prescribed opioids (7.3% vs 19.5%, p < .001) and more likely to fatally overdose on heroin (63% vs 50.4%, p < .001) or fentanyl/fentanyl analogues (50.3% vs 41.8%, p = .001). Between 2012 and the time of death, decedents without an opioid prescription had fewer emergency department admissions (2.5 ± 4.2 vs 10.6 ± 15.8, p < .001), were less likely to receive an opioid use disorder diagnosis (41.3% vs 47.5%, p = .052), and were less likely to be prescribed buprenorphine for opioid use disorder treatment (3.3% vs 8.6%, p < .001). Public health interventions have often focused on opioid prescribing and the use of CSMPs as the core preventive measures to address the opioid crisis. We identified a subset of individuals in Illinois who may not be impacted by such interventions. Additional research is needed to understand what strategies may be successful among high-risk populations that have limited opioid analgesic prescription history and low health care utilization.
Keywords: Opioid overdose, Drug overdose prescription opioids, Heroin, Fentanyl, Controlled substance monitoring programs
Introduction
Drug overdose, two thirds of which is opioid-related, is now the leading cause of accidental death in the USA [1]. The increasing prevalence of opioid misuse and opioid use disorder has been linked to a dramatic increase in opioid prescribing beginning in the late 1990s [2, 3]. A 2013 survey found that 79.5% of heroin users reported nonmedical use of prescription opioids prior to initiating heroin use [4]. This is a shift from the previous generations of patients with opioid use disorder (OUD), who predominantly initiated their opioid use with heroin. In a 2014 study, more than 80% of respondents who initiated opioid use in the 1960s used heroin as their first opioid, compared with fewer than 30% of respondents who initiated opioid use in the 2000s [3]. Driven by these findings, public health interventions have sought to curtail opioid prescribing through a variety of interventions including controlled substance monitoring programs (CSMPs), commonly referred to as prescription drug monitoring programs (PDMPs) [5].1 These statewide databases collect information about controlled substance prescriptions dispensed from retail pharmacies and are typically used to inform provider decisions and as a tool for law enforcement [6].
Evidence suggests CSMPs are an effective surveillance tool and may reduce opioid prescribing and diversion of medication [6–9]. Although some studies associate implementation of CSMPs with a reduction in mortality due to overdose involving prescription opioids [10, 11], decreased opioid prescribing has had a minimal effect on the number of overall opioid overdose–related deaths, which are increasingly driven by illicit opioids [12–15]. While opioid prescribing has begun to decrease nationally, the number of annual opioid overdose deaths has continued to climb. According to the Centers for Disease Control and Prevention, the total amount of opioids prescribed in the USA has dropped from its peak of 782 to 640 morphine milligram equivalents (MME) per capita between 2010 and 2015 [16]. Over the same time period, the number of deaths related to natural and semisynthetic opioids, a category that includes many prescription opioids, increased by 15%, and the total number of opioid overdose deaths increased by 57% [17]. Illinois follows the national trend, where the total number of opioid prescriptions dispensed decreased by 9.8% between 2013 and 2017 [18], and the concurrent opioid-related fatal overdoses have increased by 82% [18].
The divergence between prescribed opioid volume and opioid overdose deaths has led some authors to argue that too much emphasis has been placed on reducing prescribing, and not enough attention has been paid to structural factors that contribute to the development of substance use disorder, such as poverty and lack of opportunity [19, 20]. Additionally, little effort has been devoted to improving outcomes for patients with an existing OUD diagnosis who may have known risk factors for a fatal overdose, such as a previous nonfatal overdose, or who return to opioid use after a period of abstinence [21, 22]. Further data are needed to understand how the interaction of prescription opioid use, health care utilization, and social environment relates to the incidence of fatal overdose. In this study, we compare the demographics and health care utilization of opioid overdose decedents with and without opioid analgesic prescription histories in the 6 years preceding death by linking individual death records, Illinois Controlled Substance Monitoring program (IL CSMP) records, and emergency department (ED) and hospital discharge claims data.
Methods
Study Sample
Death records were obtained from the Illinois Department of Public Health Division of Vital Records. We extracted the records of all decedents who died between January 1, 2016 and December 31, 2016 where a contributing or underlying cause of death was drug overdose, designated by any one of the ICD-10 codes X40-X44, X60-X64, X85, and Y10-Y14.
Next, we retained records of overdoses only involving opioids by analyzing the ICD-10 codes and the death certificate text. We analyzed the text fields on the death certificate by splitting the text into single words and comparing each word with a list of drugs using a Jaro-Winkler string comparison algorithm [23]. If the Jaro-Winkler string distance exceeded 0.9, then the matches were manually inspected to ensure that each match represented an opioid drug or a misspelling of an opioid drug. An overdose was determined to be opioid involved if a specific opioid (listed in Table 1), the words “opioid,” “opiate,” or a misspelling thereof were found in the text analysis or if a contributing cause of death included any of the ICD-10 codes for opioid overdoses (T40.0, T40.1, T40.2, T40.3, T40.4). Details of this algorithm are reported separately [24].
Table 1.
Classification of drugs found on postmortem toxicology.
| Class | Compounds |
|---|---|
| Heroin | Heroin |
| Fentanyl and fentanyl analogues | Fentanyl Fentanyl analogues: Acetyl fentanyl Acryl fentanyl 3-Methyl fentanyl 4-ANPP Norfentanyl Carfentanyl U-47700 Furanyl fentanyl |
| Prescription opioids | Hydrocodone Oxycodone Hydromorphone Oxymorphone Tramadol Buprenorphine |
Overdoses were further categorized by specific substances listed on the death certificate (Table 1). Each overdose was allowed to contribute to the counts for each drug class that contributed to the death. Fentanyl was categorized separately from other opioid analgesics because it is commonly obtained in an illicitly manufactured form [25]. An analysis of IL CSMP records supported this classification as only 5.5% of decedents who overdosed on fentanyl filled a fentanyl prescription in the last 30 days before death.
Because morphine is a metabolite of heroin and codeine, the mention of morphine on the death certificate does not imply that the decedent ingested exogenous morphine. Therefore, if morphine intoxication alone was listed as the cause of death, then overdoses were not categorized into the scheme above unless they occurred in Cook County where a team worked with the medical examiner to manually reclassify each overdose based on external circumstances, such as the presence of drugs or drug paraphernalia at the scene and/or known history of heroin use per history provided in the report [26]. For overdoses outside of Cook County, such an analysis was not available, and overdoses involving morphine only (N = 28) were not categorized into the prescription, fentanyl, or heroin categories. We performed a sensitivity analysis classifying all morphine-only overdoses outside of Cook County first as prescription opioid overdoses and then as heroin overdoses. The direction or significance of the differences in Table 1 was not significantly changed by either of these categorizations.
Demographic information for each decedent was obtained from the death certificate. Decedents were classified as urban or rural, based on the classification of their residence ZIP code by the US Health Resources and Services Administration [27]. The median income in decedents’ residence ZIP code was obtained from the 2016 American Community Survey, downloaded from the census bureau website [28].
IL CSMP Records
Prescription records for the decedents from 2010 to 2016 were obtained from the IL CSMP, which is formally known as the Illinois Prescription Monitoring Program. The IL CSMP records schedule II–V prescriptions dispensed by all retail pharmacies in Illinois as well as states that share CSMP records with the Illinois Program.2 The IL CSMP does not record methadone dispensed for by opioid treatment programs for the treatment of opioid use disorder.
Decedents’ IL CSMP profiles were retrieved by linking on first name, last name, date of birth, and sex. We matched 1473/1893 (77.8%) of decedents in this study to a record in the IL CSMP. Decedents were stratified by whether they filled an opioid analgesic prescription between 2010 and the time of death. Decedents were classified as having received an opioid analgesic if they filled one or more prescriptions for buprenorphine (transdermal patch only), butorphanol, codeine, fentanyl, hydrocodone, hydromorphone, meperidine, methadone (only captured in IL CSMP when prescribed for outpatient pain treatment), morphine, oxycodone, oxymorphone, propoxyphene, tapentadol, and tramadol. Decedents were classified as having received a benzodiazepine prescription if they filled one or more prescriptions for alprazolam, clonazepam, lorazepam, temazepam, diazepam, chlordiazepoxide, oxazepam, triazolam, estazolam, flurazepam, or clorazepate.
MME was calculated from the IL CSMP with the formula MME/day = strength per unit × (number of units/days supply) × MME conversion factor. We used MME conversion factors compiled by the CDC for research purposes [29].
ED and Hospitalization Records
ED and hospital discharge records were obtained for decedents from 2012 to the time of death from the Illinois Department of Public Health Division of Patient Safety and Quality with a fuzzy match on first name, last name, date of birth, and sex. We matched 1616/1893 (85.4%) decedents in this study to a hospitalization/ED discharge record. Diagnosis codes were in the International Statistical Classification of Diseases and Related Health Problems 9th edition (ICD-9) or 10th edition (ICD-10) format. ICD-9 diagnosis codes were converted to ICD-10 using the general equivalence mapping provided by the Centers for Medicare and Medicaid Services [30]. We identified decedents with a diagnosis of opioid use disorder by searching for ICD-10 code F11 and all sub-codes. We identified decedents with previous overdose by searching for ICD-10 codes T40.0, T40.1, T40.2, T40.3, T40.4, and T40.6. To count only diagnoses that were made prior to the fatal overdose, we excluded visits where the date of discharge was on the date of death.
For a given diagnosis code, we calculated the per-visit probability that this code was used by dividing the number of hospitalizations with the given code by the total number of hospitalizations in each study group. For each ICD-10 code, we compared the per-visit probability that this code was used between the two study groups. The difference in per-visit probability was considered significant if the Bonferroni-corrected p value of Fisher’s exact test was less than .05. We excluded all codes that were used fewer than ten times due to data protection requirements.
Each decedent’s insurance status was derived from the discharge data by assigning each decedent the primary payer type that was associated with their hospitalization records. When different payer types were used for successive visits, we assigned the most commonly listed payer type.
Statistical Analysis
Unless otherwise specified, statistical comparisons between groups were conducted using chi-squared tests for categorical variables and t tests for continuous variables. All analyses were done in R Version 3.
Results
There were 1893 Illinois residents who died of an opioid-related overdose in 2016 (Table 2). Of these decedents, 72% were male, 24% were black, and 11% were Hispanic; the mean age decedents was 42. Overdoses most commonly involved heroin (57%) and fentanyl or fentanyl analogues (44%). Among the 840 decedents who died of an overdose involving fentanyl or fentanyl analogues, 44 decedents (5.2%) filled a fentanyl prescription within the last 30 days of life. In contrast, among the 309 (16.3%) decedents who died of an overdose involving prescription opioids, 157 (52%) filled a prescription for one of the opioids involved within 30 days of the fatal overdose. There were 280 (15.6%) decedents who died of an overdose involving benzodiazepines, and of those, 120 (42.4%) filled a prescription for a benzodiazepine within the last 30 days of life. Among the 1893 decedents, 1616 (85%) were seen in a hospital or ED between 2012 and 2016. At these visits, 729 decedents (38.5%) were diagnosed with OUD, and 410 (21.7%) were diagnosed with a nonfatal overdose.
Table 2.
Study group characteristics
| Overall | |
|---|---|
| n | 1893 |
| Demographics | |
| Male (%) | 1367 (72.2) |
| Black (%) | 457 (24.1) |
| Hispanic (%) | 203 (11.0) |
| Age (mean (SD)) | 41.66 (12.82) |
| Rural (%) | 179 (9.5) |
| Chicago resident (%) | 606 (32.0) |
| Mean ZIP code income 2016 in USD (mean (SD)) | 56,079 (22712) |
| Drugs involved in overdose | |
| Prescription (%) | 309 (16.3) |
| Prescription alone (%) | 188 (9.9) |
| Filled prescription for ≥ 1 opioid involved in overdose in last 30 days of life (%) | 157 (8.3) |
| Heroin (%) | 1085 (57.3) |
| Fentanyl or fentanyl analogues (%) | 840 (44.4) |
| Filled prescription for fentanyl within 30 days of fatal overdose involving fentanyl (%) | 44 (2.3) |
| Cocaine (%) | 416 (22.0) |
| Alcohol (%) | 352 (18.6) |
| Benzodiazepine (%) | 284 (15.0) |
| Filled prescription for benzodiazepine within 30 days of fatal overdose involving benzodiazepine (%) | 120 (6.3) |
| Health care utilization (2012–2016) | |
| Any opioid analgesic prescription (%) | 1320 (69.7) |
| Any > 100 MME/day (%) | 371 (19.6) |
| Any opioid analgesic in last 12 months (%) | 828 (43.7) |
| Any > 100 MME/day (%) | 139 (7.3) |
| Buprenorphine in the last 12 months (%) | 133 (7.0) |
| Any Benzodiazepine prescription (%) | 988 (52.2) |
| Benzodiazepine prescription in last 12 months (%) | 698 (36.9) |
| Any ED or hospital admission (%) | 1616 (85.4) |
| Hospital admissions (mean (SD)) | 2.37 (5.66) |
| ED admissions (mean (SD)) | 8.14 (13.90) |
| Total hospital stay (mean (SD)) | 11.41 (34.55) |
| OUD diagnosis = 1 (%) | 729 (38.5) |
| Previous OD = 1 (%) | 410 (21.7) |
| Primary insurance (%) | |
| Commercial | 393 (24.3) |
| Medicaid | 713 (44.1) |
| Medicare | 207 (12.8) |
| Other | 77 (4.8) |
| Self-pay | 226 (14.0) |
ED emergency department, MME morphine milligram equivalents, SD standard deviation, OD overdose, OUD opioid use disorder
Of the 1893 opioid-related overdose decedents in this study, 573 (30.2%) did not fill an opioid analgesic prescription recorded in the IL CSMP in the 6 years preceding death (Table 3). Decedents without an opioid prescription filled were more likely to be black (33.3% vs 20.2%, p < .001), Hispanic (16.3% vs 8.8%, p < .001), Chicago residents (46.8% vs 25.6%, p < .001), and residents in a ZIP code with a lower median income ($53,230 vs $57,307, p < .001) than decedents with at least one filled opioid prescription between 2010 and 2016. Decedents without an opioid prescription history were less likely to overdose on prescription opioids (7.3% vs 20.2%, p < .001) or benzodiazepines (11.2% vs. 16.7%, p = 0.003) but more likely to overdose on heroin (66.1% vs 53.5%, p < .001) or fentanyl/fentanyl analogues (50.3% vs 41.8%, p = .001).
Table 3.
Demographics, postmortem toxicology, and health care utilization among decedents in 2016 stratified by any opioid analgesic prescription from 2010 to 2016 in the IL CSMP
| Any opioid analgesic prescription | |||
|---|---|---|---|
| Yes | No | Test statistic (p) | |
| n | 1320 | 573 | |
| Demographics | |||
| Male (%) | 921 (69.8) | 446 (77.8) | 12.5*** |
| Black (%) | 266 (20.2) | 191 (33.3) | 37.2*** |
| Hispanic (%) | 113 (8.8) | 90 (16.3) | 21.9*** |
| Age (mean (SD)) | 42.16 (12.85) | 40.51 (12.69) | 6.8** |
| Rural (%) | 144 (10.9) | 35 (6.1) | 10.2** |
| Chicago resident (%) | 338 (25.6) | 268 (46.8) | 81.3*** |
| ZIP median income, 2016 USD (mean (SD)) | 57,307 (21907) | 53,230 (24258) | 11.9*** |
| Drugs involved in overdose | |||
| Prescription (%) | 267 (20.2) | 42 (7.3) | 47.7*** |
| Prescription alone (%) | 166 (12.6) | 22 (3.8) | 33.1*** |
| Heroin (%) | 706 (53.5) | 379 (66.1) | 25.7*** |
| Fentanyl and fentanyl analogue (%) | 552 (41.8) | 288 (50.3) | 11.2** |
| Cocaine (%) | 269 (20.4) | 147 (25.7) | 6.2* |
| Alcohol (%) | 228 (17.3) | 124 (21.6) | 4.8* |
| Benzodiazepine (%) | 220 (16.7) | 64 (11.2) | 9.0** |
| Health care utilization (2012–2016) | |||
| Buprenorphine in last 12 months (%) | 114 (8.6) | 19 (3.3) | 16.5*** |
| Any Benzodiazepine prescription (%) | 886 (67.1) | 102 (17.8) | 387.5*** |
| Benzodiazepine prescription in the last 12 months (%) | 637 (48.3) | 61 (10.6) | 241.2*** |
| Any ED or hospital admission (%) | 1261 (95.5) | 355 (62.0) | 357.9*** |
| Hospital admissions (mean (SD)) | 3.06 (6.47) | 0.77 (2.39) | 126.7*** |
| ED admissions (mean (SD)) | 10.58 (15.80) | 2.52 (4.21) | 295.5*** |
| Total hospital stay (mean (SD)) | 14.78 (39.97) | 3.65 (13.35) | 81.4*** |
| OUD diagnosis (%) | 591 (44.8) | 138 (24.1) | 71.4*** |
| Previous nonfatal overdose (%) | 333 (25.2) | 77 (13.4) | 32*** |
| Primary insurance (%) | 73.8*** | ||
| Commercial | 310 (24.6) | 83 (23.4) | |
| Medicaid | 562 (44.6) | 151 (42.5) | |
| Medicare | 196 (15.5) | 11 (3.1) | |
| Other | 55 (4.4) | 22 (6.2) | |
| Self-pay | 138 (10.9) | 88 (24.8) | |
Note. The test statistic is t statistic or χ-squared statistic for continuous and categorical variables, respectively
*p < .05; **p < .01; ***p < .001
ED emergency department, SD standard deviation, OUD opioid use disorder
Between 2012 and the time of death, decedents without an opioid prescription had fewer ED admissions (2.5 ± 4.2 vs 10.6 ± 15.8, p < .001) and lower number of days spent as an inpatient (3.7 ± 13.4 vs 14.8 ± 40.0, p < .001). They were less likely to receive an OUD diagnosis (24.1% vs 44.8%, p < .001) or present with a previous, nonfatal overdose (13.4% vs 25.2%, p < .001). However, among decedents with at least one hospital admission, the rates of OUD diagnosis were similar (41.3% vs 47.5%, p = .05), as were rates of nonfatal overdose (23.1% vs 26.7%, p = .20). Decedents without an opioid prescription were less likely to fill prescriptions for formulations of buprenorphine approved for OUD treatment (sublingual, buccal) in the last year of life (3.3% vs 8.6%, p < .001).
Insurance status differed significantly between the two study groups (p < .001). Decedents with no opioid prescription history were more than twice as likely to be self-pay (24.8% vs 10.9%) and less likely to be on Medicare (3.1% vs 15.5%) than decedents with an opioid prescription history, whereas the rates of Medicaid were similar (42.5% vs 44.6%).
Table 4 lists ICD-10 diagnosis codes that were significantly more likely to be applied to hospital/ED visits of decedents with an opioid prescription history. These diagnosis codes include codes for low back pain (M545 5.5% vs 1.2% of visits, p < .001) or other chronic pain (G8929 7.2% vs 0.9% of visits, p < .001). Additionally, hospital/ED visits by decedents with an opioid prescription were more likely to be associated with chronic health conditions such as primary hypertension (I10 19.1% vs 12% of visits, p < .001), hyperlipidemia (E785 4.2% vs 1.3% of visits, p < .001), and uncomplicated type 2 diabetes mellitus (E119 4.7% vs 1.9% of visits, p < .001), as well as for sequelae of these conditions such as atherosclerotic heart disease (I2510 2.9% vs 0.5% of visits) and heart failure (I509 2.2% vs 0.6% of visits, p < .001).
Table 4.
Diagnosis codes that are more likely during ED/hospital visits of decedents with opioid prescription history than of decedents with no opioid prescription history
| ICD-10 code | Description | Percentage of visits with diagnosis | Difference (%) | |
|---|---|---|---|---|
| Any opioid analgesic prescription | ||||
| Yes | No | |||
| I10 | Essential (primary) hypertension | 19.1 | 12 | − 7*** |
| G8929 | Other chronic pain | 7.2 | 0.9 | − 6.3*** |
| M545 | Low back pain | 5.5 | 1.2 | − 4.2*** |
| M549 | Dorsalgia, unspecified | 4.8 | 0.8 | − 4*** |
| F419 | Anxiety disorder, unspecified | 11.3 | 7.6 | − 3.7** |
| M5489 | Other dorsalgia | 3.8 | 0.6 | − 3.2*** |
| K219 | Gastro-esophageal reflux disease without esophagitis | 4.7 | 1.7 | − 3*** |
| Z79891 | Long-term (current) use of opiate analgesic | 6.8 | 3.8 | − 3*** |
| E785 | Hyperlipidemia, unspecified | 4.2 | 1.3 | − 2.9*** |
| E119 | Type 2 diabetes mellitus without complications | 4.7 | 1.9 | − 2.8*** |
| F319 | Bipolar disorder, unspecified | 5.5 | 2.7 | − 2.8*** |
| J449 | Chronic obstructive pulmonary disease, unspecified | 4.6 | 1.8 | − 2.8*** |
| I2510 | Atherosclerotic heart disease of the native coronary artery without angina pectoris | 2.9 | 0.5 | − 2.4*** |
| E784 | Other hyperlipidemia | 2.8 | 0.7 | − 2.1*** |
| I509 | Heart failure, unspecified | 2.2 | 0.6 | − 1.6*** |
| Z8673 | Personal history of transient ischemic attack (TIA) and cerebral infarction without residual deficits | 2.1 | 0.6 | − 1.5** |
| Z7982 | Long-term (current) use of aspirin | 1.9 | 0.5 | − 1.4* |
Note. Columns “Yes” and “No” contain the percentage of all visits in each group that carried each ICD-10 code. The p values are Bonferroni-corrected p values calculated using Fisher’s exact test
*p < .05; **p < .01; ***p < .001
Table 5 lists ICD-10 diagnosis codes that were significantly more likely to be applied to hospital/ED visits of decedents without an opioid prescription history. This group of decedents was more likely to receive visit diagnoses related to psychiatric illness such as suicidal ideation (R45851 8.1% vs 4.4% of visits, p < .001), schizophrenia (F209 2.3% vs 0.8% of visits, p < .001), and schizoaffective disorder (F259 3% vs 0.9% of visits, p < .001). In addition, these decedents were more likely to receive visit diagnoses related to substance use disorder such as opioid use disorder (F1110 11.9% vs 6.4% of visits, p < .001) and cannabis use disorder (F1210 4.7% vs 2.6% of visits, p = .005) and with diagnoses related to opioid overdose such as accidental heroin poisoning (T40.1X1A 6.4% vs 2.2% of visits, p < .001).
Table 5.
Diagnosis codes that are more likely during ED/hospital visits of decedents with no opioid prescription history than of decedents with opioid prescription history
| ICD-10 code | Description | Percentage of visits with diagnosis | Difference | |
|---|---|---|---|---|
| Any opioid analgesic prescription | ||||
| Yes | No | |||
| F1110 | Opioid use disorder, uncomplicated | 6.4 | 11.9 | 5.5*** |
| J45901 | Unspecified asthma with (acute) exacerbation | 2.2 | 6.4 | 4.2*** |
| T401X1A | Poisoning by heroin, accidental (unintentional), initial encounter | 2.3 | 6.1 | 3.8*** |
| R45851 | Suicidal ideations | 4.4 | 8.1 | 3.7*** |
| I469 | Cardiac arrest, cause unspecified | 1.6 | 4.4 | 2.7*** |
| F1210 | Cannabis use disorder, uncomplicated | 2.6 | 4.7 | 2.1** |
| F259 | Schizoaffective disorder, unspecified | 0.9 | 3 | 2.1*** |
| F209 | Schizophrenia, unspecified | 0.8 | 2.3 | 1.5*** |
| F29 | Unspecified psychosis not due to a substance or known physiological condition | 0.4 | 1.9 | 1.5*** |
| T401X4A | Poisoning by heroin, undetermined, initial encounter | 1.3 | 2.8 | 1.5* |
| T401X2A | Poisoning by heroin, intentional self-harm, initial encounter | 1.1 | 2.5 | 1.4* |
| G931 | Anoxic brain damage, not elsewhere classified | 0.6 | 1.9 | 1.3*** |
| I468 | Cardiac arrest due to other underlying condition | 0.2 | 1 | 0.8** |
Note. Columns “Yes” and “No” contain the percentage of all visits in each group that carried each ICD-10 code. The p values are Bonferroni-corrected p values calculated using Fisher’s exact test
*p < .05; **p < .01; ***p < .001
Discussion
We found that more than 30% of those who died of an opioid-related overdose in the state of Illinois in 2016 had no opioid analgesic prescription history during the prior 6 years. This finding is consistent with recent surveys of OUD patients, which show that between 20% and 35% of individuals who report nonmedical opioid use initiated their opioid use with heroin [3, 4]. The early public understanding of the opioid epidemic was shaped by accounts that highlight the role of prescribed opioids [31] and subsequent interventions focused on the role of opioid prescribing as a driver of the opioid crisis [19, 32]. Those interventions are unlikely to benefit a population of overdose decedents who have filled any opioid prescriptions in the 6 years before fatal overdose. Our study adds to the growing body of evidence that demonstrates the importance of a multimodal approach in the public health response to the opioid overdose epidemic [19].
Fewer than 20% of all decedents in our study cohort had benzodiazepines listed as a contributing factor to the cause of death. However, medical examiners in Illinois were not routinely testing for benzodiazepines in 2016. While Cook County, the most populous jurisdiction in the state, began routinely testing all overdose decedents for benzodiazepines in 2017, these practices still vary across the state and the country. This presents an important gap in the national overdose surveillance, because the combined use of opioids and benzodiazepines greatly increases the risk of fatal overdose, and the two drug classes are frequently prescribed in combination [33, 34]. Among decedents in our study who filled at least one opioid prescription in the last 6 years of life, two thirds also filled a benzodiazepine prescription during this period.
We observed significant differences in health care utilization between our study groups. Decedents without an opioid prescription history spent an average of 3.7 nights in the hospital between 2012 and 2016 compared to 14.8 nights for decedents with an opioid prescription history. High hospital utilization may cause the stratification into the two groups because hospitalized patients are frequently discharged with opioid prescriptions and may subsequently develop chronic opioid use [35, 36]. However, the difference in health care utilization may also be related to disparities in health care access; decedents without an opioid prescription history were more likely to reside in ZIP codes with lower median incomes and were more likely to be self-pay than decedents with an opioid prescription history.
Decedents without an opioid prescription history were more likely to be black or Hispanic than decedents with an opioid prescription history (49.6% vs 29.0%). Blacks and Hispanics may be less likely to receive opioid prescriptions as they continue to have lower rates of health care access than whites [37]. Racial differences between our study groups may also be driven by provider bias, as previous studies have shown that black patients are less likely than white patients to receive prescription opioids when presenting to the ED with certain pain conditions [38, 39].
The observed disparities in health care utilization may adversely affect care and outcomes for patients with OUD. In our setting, decedents who lacked an opioid prescription history were more likely to be urban, black, and Hispanic than those who had filled at least one opioid prescription. Previous studies have shown that blacks and Hispanics are more likely to receive no care or delayed care for substance use disorders [40]. Similarly, this analysis showed that individuals with no opioid prescription history were less likely to receive any hospital or emergency room care (62% vs 95.5%, p < .001), less likely to receive a diagnosis of OUD (24.1% vs 44.8%, p < .001) and less likely to be prescribed buprenorphine formulations that are indicated for OUD treatment (3.3% vs 8.6%, p < .001). These results are consistent with the previous studies which document that blacks are less likely to initiate opioid dependence treatment with an opioid agonist [41].
An analysis of diagnosis codes showed substantial disparities between decedents with and without an opioid analgesic prescription history. Decedents with an opioid prescription history were more likely to receive hospital/ED care for back pain or chronic pain, which is not surprising given that pain is a common indication for prescription opioid analgesics. However, visits by this group of decedents were also more likely to be related to common chronic health conditions such as diabetes, hypertension, and hyperlipidemia. This finding is surprising because decedents with an opioid prescription history were more likely to be white, a demographic that has lower rates of hypertension [42] and diabetes [43] than Hispanics and blacks. One reason for the differences in inpatient diagnoses may be the age difference between the two groups. Decedents with an opioid prescription history are on average 2 years older than decedents without an opioid analgesic prescription history. Alternatively, the differences in inpatient diagnoses between the groups may be accounted for by disparities in health care access. Decedents with an opioid prescription history, who have higher overall health care utilization, may be more likely to receive outpatient care for chronic health conditions, despite lower prevalence of these conditions.
Decedents without an opioid prescription history were much less likely to be seen in the hospital or ED between 2012 and 2016. However, when they were seen, visits by these decedents were often related to drug use, overdose, asthma exacerbations, and psychiatric conditions like schizophrenia, psychosis, and suicidal ideation. This group of decedents may have had limited access to longitudinal care for chronic health conditions. For example, we have shown that this group of decedents was less likely to receive buprenorphine for OUD treatment. Therefore, this group of decedents presented to the hospital and ED predominantly in times of crisis, during exacerbations of untreated chronic conditions.
This analysis has several limitations. The data described are limited to Illinois residents, where the opioid epidemic has disproportionately affected black communities [44]. This finding may not be generalizable to other states. IL CSMP data on opioid prescribing before 2010 was not captured, so only 6 years of controlled substance prescription history could be reviewed. Additionally, decedents may have had access to diverted opioid medications not prescribed to them; this information would not be available in the IL CSMP. Among decedents who died of an overdose involving prescription opioids, 51% filled an opioid prescription in the last 30 days of life. Similarly, among decedents who died of an overdose involving benzodiazepines, only 42% filled a benzodiazepine prescription in the last 30 days of life. This indicates that many decedents had access to diverted drugs or filled prescriptions outside of the 36 states that share data with the IL CSMP. Notably, Missouri, a border state, only recently initiated its CSMP in 2017. Additionally, IL CSMP does not record methadone dispensed in opioid treatment programs. Our ED and hospitalization data does not include outpatient interactions like outpatient addiction treatment, which may account for a larger proportion of care related to OUD than hospitalization and ED discharge records. Finally, in the analysis of diagnostic codes associated with decedent types, the use of single-level modeling does not account for the possible interdependence of diagnosis codes. Analysis of this data using multilevel models is currently underway.
Conclusion
We showed that a substantial proportion of opioid overdose decedents in Illinois did not fill an opioid prescription in the 6 years preceding fatal overdose. A narrowed focus on opioid prescribing may not adequately address the continued rise of overdose, particularly in urban settings and among black and Hispanic populations. Additional research is needed to understand which public health approaches may help reduce morbidity and mortality among urban, minority communities where opioids are prescribed at low rates, heath care utilization is low, and OUD treatment with buprenorphine is uncommon.
Acknowledgments
The authors were funded by the University of Chicago Institute of Politics (ABA), NIDA UG3DA044829 (MTP), and AHRQ R00HS022433 (MTP). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Footnotes
We are using the terminology CSMP to more accurately describe these databases, which house information about receipt of controlled substance prescriptions (as opposed to all prescriptions) that are obtained in retail pharmacies.
AL, AR, AZ, CA, CO, FL, GA, IA, ID, IL, IN, KS, KY, LA, MA, MD, MI, MN, MO, MS, NC, ND, NE, NH, NV, NY, OH, OK, OR, PA, TN, TX, UT, VA, WI.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Kochanek KD, Murphy S, Xu J, Arias E. Mortality in the United States, 2016. NCHS Data Brief. 2017;293:1–8. [PubMed] [Google Scholar]
- 2.Morone NE, Weiner DK. Pain as the fifth vital sign: exposing the vital need for pain education. Clin Ther. 2013;35(11):1728–1732. doi: 10.1016/j.clinthera.2013.10.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Cicero TJ, Ellis MS, Surratt HL, Kurtz SP. The changing face of heroin use in the United States: a retrospective analysis of the past 50 years. JAMA Psychiat. 2014;71(7):821–826. doi: 10.1001/jamapsychiatry.2014.366. [DOI] [PubMed] [Google Scholar]
- 4.Muhuri P, Gfroerer J, Davies M. Associations of nonmedical pain reliever use and initiation of heroin use in the United States. CBHSQ Data Review. 2013. Available at https://www.samhsa.gov/data/sites/default/files/DR006/DR006/nonmedical-pain-reliever-use-2013.htm. Accessed 1 March 2018.
- 5.Compton WM, Jones CM, Baldwin GT. Relationship between nonmedical prescription-opioid use and heroin use. N Engl J Med. 2016;374(2):154–163. doi: 10.1056/NEJMra1508490. [DOI] [PubMed] [Google Scholar]
- 6.Morgan L, Weaver M, Sayeed Z, Orr R. The use of prescription monitoring programs to reduce opioid diversion and improve patient safety. J Pain Palliat Care Pharmacother. 2013;27:4–9. doi: 10.3109/15360288.2012.738288. [DOI] [PubMed] [Google Scholar]
- 7.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 Aff (Millwood) 2016;35(10):1876–1883. doi: 10.1377/hlthaff.2016.0448. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Patrick SW, Fry CE, Jones TF, Buntin MB. Implementation of prescription drug monitoring programs associated with reductions in opioid-related death. Health Aff (Millwood) 2016;35(7):1324–1332. doi: 10.1377/hlthaff.2015.1496. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Reifler LM, Droz D, Bailey JE, Schnoll SH, Fant R, Dart RC, Bucher Bartelson B. Do prescription monitoring programs impact state trends in opioid abuse/misuse? Pain Med. 2012;13:434–442. doi: 10.1111/j.1526-4637.2012.01327.x. [DOI] [PubMed] [Google Scholar]
- 10.Patrick SW, Carrie EF, Jones TF, Buntin MB. Implementation of prescription drug monitoring programs associated with reductions in opioid-related death rates. Health Aff. 2016;35(7):1324–1332. doi: 10.1377/hlthaff.2015.1496. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Bryce P. Do more robust prescription drug monitoring programs reduce prescription opioid overdose? Addiction. 2017;112(10):1773–1783. doi: 10.1111/add.13741. [DOI] [PubMed] [Google Scholar]
- 12.Paulozzi LJ, Kilbourne EM, Desai HM. Prescription drug monitoring programs and death rates from drug overdose. Pain Med. 2011;12:747–754. doi: 10.1111/j.1526-4637.2011.01062.x. [DOI] [PubMed] [Google Scholar]
- 13.Li G, Brady JE, Lang BH, Giglio J, Wunsch H, DiMaggio C. Prescription drug monitoring and drug overdose mortality. Injury Epidemiol. 2014;1(9):9. doi: 10.1186/2197-1714-1-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Brady JE, Wunsch H, Dimaggio C, Lang BH, Giglio J, Li G. Prescription drug monitoring and dispensing of prescription opioids. Public Health Rep. 2014;129(2):139–147. doi: 10.1177/003335491412900207. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Green TC, Bowman S, Davis C, Los C, McHugh K, Friedmann PD. Discrepancies in addressing overdose prevention through prescription monitoring programs. Drug Alcohol Depend. 2015;153:355–358. doi: 10.1016/j.drugalcdep.2015.05.009. [DOI] [PubMed] [Google Scholar]
- 16.Guy GP Jr, Zhang K, Bohm MK, et al. vital signs: changes in opioid prescribing in the United States, 2006–2015. MMWR Morb Mortal Wkly Rep. 2017;66:697–704. 10.15585/mmwr.mm6626a4. [DOI] [PMC free article] [PubMed]
- 17.Overdose Death Rates. https://www.drugabuse.gov/related-topics/trends-statistics/overdose-death-rates. Accessed 4/1/2018; 2018.
- 18.2017 State of Illinois Comprehensive Opioid Data Report. Springfield, IL: Illinois Department of Public Health; 2017. Available at http://www.dph.illinois.gov/sites/default/files/publications/publicationsdoil-opioid-data-report.pdf. Accessed 13 Feb 2018.
- 19.Dasgupta N, Beletsky L, Ciccarone D. Opioid crisis: no easy fix to its social and economic determinants. Am J Public Health. 2018;108(2):182–186. doi: 10.2105/AJPH.2017.304187. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Pain management and the opioid epidemic: balancing societal and individual benefits and risks of prescription opioid use. In: Phillips JK, Ford MA, Bonnie RJ, editors. Committee on Pain Management and Regulatory Strategies to Address Prescription Opioid Abuse. Washington, DC: National Academies of Sciences, Engineering, and Medicine; Health and Medicine Division; Board on Health Sciences Policy; 2017. Available at https://www.ncbi.nlm.nih.gov/pubmed/29023083. Accessed 23 Feb 2018. [PubMed]
- 21.Davoli M, Bargagli AM, Perucci CA, Schifano P, Belleudi V, Hickman M, Salamina G, Diecidue R, Vigna-Taglianti F, Faggiano F, for the VEdeTTE Study Group Risk of fatal overdose during and after specialist drug treatment: the VEdeTTE study, a national multi-site prospective cohort study. Addiction. 2007;102(12):1954–1959. doi: 10.1111/j.1360-0443.2007.02025.x. [DOI] [PubMed] [Google Scholar]
- 22.Caudarella A, Dong H, Milloy MJ, Kerr T, Wood E, Hayashi K. Non-fatal overdose as a risk factor for subsequent fatal overdose among people who inject drugs. Drug Alcohol Depend. 2016;162:51–55. doi: 10.1016/j.drugalcdep.2016.02.024. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Winkler WE. String comparator metrics and enhanced decision rules in the Fellegi Sunter model of record linkage. In: Proceedings of the Section on Survey Research Methods. Alexandria, VA: American Statistical Association; 1990.
- 24.Abbasi AB, Clark H, Agbodo N. Learning algorithm to identify drugs involved in drug overdose deaths from literal text on death certificates. West Palm Beach, FL: Council of State and Territorial Epidemiologists; 2018. [Google Scholar]
- 25.Fentanyl: a briefing guide for first responders. Washington, DC: U.S. Department of Justice, Drug enforcement administration; 2017. Archived at: https://www.nvfc.org/wp-content/uploads/2018/03/Fentanyl-Briefing-Guide-for-First-Responders.pdf. Accessed 25 Dec 2017.
- 26.Epidemiology report: increase in overdose deaths involving opioids – Chicago 2015–2016. Chicago, IL: Chicago Department of Public Health; 2017. Available athttps://www.cityofchicago.org/content/dam/city/depts/cdph/tobacco_alchohol_and_drug_abuse/2016ChicagoOpioidReport.pdf. Accessed Accessed 14 March 2018.
- 27.Health Resources & Services Administration. Defining rural population. https://www.hrsa.gov/ruralhealth/about-us/definition/index.html. Accessed 1 April 2018.
- 28.Cicero TJ, Kasper ZA, Ellis MS. Increased use of heroin as an initiating opioid of abuse: further considerations and policy implications. Addict Behav. 2018;87:267–271. doi: 10.1016/j.addbeh.2018.05.030. [DOI] [PubMed] [Google Scholar]
- 29.CDC. Data resources: analyzing prescription data and morphine milligram equivalents (MME). https://www.cdc.gov/drugoverdose/resources/data.html. Accessed 8/22/2017; 2017.
- 30.Surratt HL, Kurtz SP, Buttram M, Levi-Minzi MA, Pagano ME, Cicero TJ. Heroin use onset among nonmedical prescription opioid users in the club scene. Drug Alcohol Depend. 2017;179:131–138. doi: 10.1016/j.drugalcdep.2017.06.034. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Quinones S. Dreamland: the true tale of America’s opiate epidemic. New York, NY: Bloomsbury Press; 2015. [Google Scholar]
- 32.Kolodny A, Courtwright DT, Hwang CS, Kreiner P, Eadie JL, Clark TW, Alexander GC. The prescription opioid and heroin crisis: a public health approach to an epidemic of addiction. Annu Rev Public Health. 2015;36:559–574. doi: 10.1146/annurev-publhealth-031914-122957. [DOI] [PubMed] [Google Scholar]
- 33.Park TW, Saitz R, Ganoczy D, Ilgen MA, Bohnert AS. Benzodiazepine prescribing patterns and deaths from drug overdose among US veterans receiving opioid analgesics: case-cohort study. BMJ. 2015;350:h2698. doi: 10.1136/bmj.h2698. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Jones JD, Mogali S, Comer SD. Polydrug abuse: a review of opioid and benzodiazepine combination use. Drug Alcohol Depend. 2012;125(1–2):8–18. doi: 10.1016/j.drugalcdep.2012.07.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Calcaterra SL, Yamashita TE, Min SJ, Keniston A, Frank JW, Binswanger IA. Opioid prescribing at hospital discharge contributes to chronic opioid use. J Gen Intern Med. 2016;31(5):478–485. doi: 10.1007/s11606-015-3539-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Hill MV, Stucke RS, Billmeier SE, Kelly JL, Barth RJ., Jr Guideline for discharge opioid prescriptions after inpatient general surgical procedures. J Am Coll Surg. 2018;226(6):996–1003. doi: 10.1016/j.jamcollsurg.2017.10.012. [DOI] [PubMed] [Google Scholar]
- 37.Chen J, Vargas-Bustamante A, Mortensen K, Ortega AN. Racial and ethnic disparities in health care access and utilization under the affordable care act. Med Care. 2016;54(2):140–146. doi: 10.1097/MLR.0000000000000467. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Singhal A, Tien YY, Hsia RY. Racial-ethnic disparities in opioid prescriptions at emergency department visits for conditions commonly associated with prescription drug abuse. PLoS One. 2016;11(8):e0159224. doi: 10.1371/journal.pone.0159224. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Pletcher MJ, Kertesz SG, Kohn MA, Gonzales R. Trends in opioid prescribing by race/ethnicity for patients seeking care in US emergency departments. JAMA. 2008;299(1):70–78. doi: 10.1001/jama.2007.64. [DOI] [PubMed] [Google Scholar]
- 40.Wells K, Klap R, Koike A, Sherbourne C. Ethnic disparities in unmet need for alcoholism, drug abuse, and mental health care. Am J Psychiatr. 2001;158(12):2027–2032. doi: 10.1176/appi.ajp.158.12.2027. [DOI] [PubMed] [Google Scholar]
- 41.Stein BD, Gordon AJ, Sorbero M, Dick AW, Schuster J, Farmer C. The impact of buprenorphine on treatment of opioid dependence in a Medicaid population: recent service utilization trends in the use of buprenorphine and methadone. Drug Alcohol Depend. 2012;123(1–3):72–78. doi: 10.1016/j.drugalcdep.2011.10.016. [DOI] [PubMed] [Google Scholar]
- 42.Fryar CD, Ostchega Y, Hales CM, Zhang G, Kruszon-Moran D. Hypertension prevalence and control among adults: United States, 2015–2016. NCHS Data Brief. 2017;(289):1–8. [PubMed]
- 43.Centers for Disease Control and Prevention. National Diabetes Statistics Report, 2017. Atlanta, GA: Centers for Disease Control and Prevention, U.S. Dept of Health and Human Services; 2017. Available at https://www.cdc.gov/diabetes/data/statistics-report/index.html. Accessed 12 Feb 2018.
- 44.Cicero TJ, Ellis MS, Kasper ZA. Increases in self-reported fentanyl use among a population entering drug treatment: the need for systematic surveillance of illicitly manufactured opioids. Drug Alcohol Depend. 2017;177:101–103. doi: 10.1016/j.drugalcdep.2017.04.004. [DOI] [PubMed] [Google Scholar]