Mental Health Conditions And The Risk Of Chronic Opioid Therapy Among Patients with Rheumatoid Arthritis: A Retrospective Veterans Affairs Cohort Study

Abstract

Objective:

Patients with rheumatoid arthritis (RA) often receive opioid analgesics for pain management. We examined the association between mental health conditions and the risk of chronic opioid therapy.

Methods:

A retrospective cohort of Veterans with RA initiating opioid use was assembled using Veterans Health Administration databases (2001–2012). Mental health conditions included anxiety (N=1,108, 12.9%), depression (N=1,912, 22.2%), bipolar disease (N=131, 1.5%), and post-traumatic stress disorder (N=768, 8.9%) and were identified by ICD coded diagnoses and use of specific medications. Cohort members were followed from opioid initiation through chronic opioid therapy, defined as the continuous availability of opioids for at least 90 days. Multivariable Cox proportional hazard regression models assessed the association between mental health conditions and chronic opioid therapy accounting for relevant covariates. Subgroup analyses examined whether the strength of the observed association varied by the duration of the initial opioid prescription.

Results:

We identified 14,767 patients with RA with 22,452 episodes of opioid use initiation. Mental health conditions were identified in 8,607 (38.3%) patients. Compared with patients without mental health conditions, patients with mental health conditions have a higher risk of developing chronic opioid therapy (469.3 vs 378.1 per 1000 person-years, adjusted hazard ratio [aHR]: 1.18, 95% CI 1.09, 1.29). The increased risk was highest for those with a history of opioid use disorder (aHR: 1.94, 95% CI 1.09, 3.46), and also elevated for those with other substance use disorders (aHR: 1.35, 95% CI 1.05, 1.73). Duration of the initial opioid prescription was independently associated with chronic opioid therapy, regardless of the estimated opioid daily dose.

Conclusions:

History of mental health conditions and duration of the initial opioid prescription were associated with an increased risk of chronic opioid therapy among patients with RA.

INTRODUCTION

Sales of prescription opioid analgesics in the US nearly quadrupled from 1999 to 2014.[1] In concordance with increased sales, deaths from prescription opioids have quadrupled since 1999.[2, 3] In 2017, healthcare providers wrote 191 million opioid prescriptions.[4] In 2016, rates of drug overdose deaths reached 4.4 per 100,000 standard population for natural and semisynthetic opioids and 6.2 per 100,000 standard population for synthetic opioids.[5] There have been recent declines in opioid prescribing, but use of these drugs remains common.[6]

Despite the high prevalence of opioid use in the US, our understanding of the effectiveness and safety of opioids is limited. Short-term opioid use is effective compared to placebo, for treating pain after surgical procedures[7], neuropathic pain[8] and for pain associated with acute medical emergencies.[9] In contrast, multiple studies on chronic opioid therapy demonstrate no benefit in reducing functional disability or improving functional ability compared to alternate therapies.[10–13] Furthermore, for treating patients with non-malignant pain, opioid use has been linked to mortality, increased hospital length of stay, readmission rates and risk of serious infections in a dose dependent fashion.[14–18]

Patients with chronic pain, including patients with rheumatoid arthritis (RA), are at increased risk of opioid exposure. Patients with RA report pain as a major factor for seeking medical care,[19] and opioid prescribing for pain associated with RA is common.[20] Furthermore, patients with mental health conditions, including depression and anxiety, are often prescribed acute opioid therapy, and have been reported to be more than twice as likely to use opioids for chronic pain.[21–23] Depression, anxiety and fibromyalgia are common co-morbidities in patients with RA, and could contribute to persistent opioid therapy.[24–26] Other factors associated with regular opioid use among patients with RA have included age 50–64 years, female gender, black race and concurrent glucocorticoid therapy.[20, 27] Nevertheless, despite the increased prevalence of opioid use among RA patients, few studies have evaluated the risk of chronic opioid therapy after opioid initiation, and the potential role of mental health conditions in this population. We conducted a nationwide retrospective cohort study of veterans with RA who initiated opioid use to examine the factors associated with chronic opioid therapy.

MATERIALS AND METHODS

Study Design and Data Source

We assembled a retrospective cohort of patients with RA in the Veterans Health Administration (VHA) using national VHA databases between 2001–2012. Data included demographic information, clinical inpatient and outpatient encounters, vital sign data and laboratory results. Pharmacy data included dispensed prescriptions, number of pills, date dispensed, and days of supply. For veterans enrolled in Medicare or Medicaid, we obtained enrollment, claims files, and pharmacy use data (Medicare Part D). The institutional review board of Vanderbilt University and the Tennessee Valley VA Healthcare System approved this study with a waiver of informed consent (VA IRB 310276–28).

Defining the Population of patients with Rheumatoid Arthritis

The underlying population for this study consisted of veterans with RA age ≥ 18 years. Veterans with RA were identified by any prescription filled for Disease-Modifying Antirheumatic Drugs (DMARDs) used to treat RA and a coded inpatient or outpatient diagnosis for RA (ICD-9CM: 714.X) within the previous 180 days of the qualifying medication fill date. We excluded patients with recorded diagnoses of other autoimmune diseases including psoriasis or psoriatic arthritis (ICD9-CM: 696, 696.0, 696.1), juvenile rheumatoid arthritis (ICD9-CM: 714.3X), systemic lupus erythematosus (ICD9-CM: 710.0), Crohn’s disease (ICD9-CM: 555.X), ulcerative colitis (ICD9-CM: 556.X) or ankylosing spondylitis (ICD9-CM: 720.0) within the 365 days prior to cohort enrollment. The combination of DMARD therapy and coded diagnosis for RA has been shown to have a positive predictive value of 97% and a specificity of 97.1% in identifying patients with RA.[28]

Patients with Rheumatoid Arthritis initiating Opioid Use

Among patients with RA, we identified a cohort of patients who were new users of an opioid in the outpatient setting, defined as filling a prescription for a study opioid with no exposure to any opioid during the prior 180 days. Inpatient treatment initiation was not eligible; however, prescriptions filled on discharge were eligible for inclusion. The date of opioid initiation was the cohort enrollment date. To ensure that patients were users of healthcare services, patients were required to have at least one VA, Medicaid or Medicare healthcare encounter (medication fill, inpatient or outpatient medical encounter) in the baseline 365 days, and in the 366 – 730 days prior to cohort enrollment. Patients enrolled in Medicare Advantage programs during baseline were excluded, since their claims were not accessible. We also excluded patients with evidence of serious diseases (cancer, HIV, end stage renal disease or dialysis, liver failure, respiratory failure, organ transplant) identified during the baseline 365 days preceding cohort enrollment to exclude patients who may require special pain management and those receiving opioids for end of life care. Patients with an initial opioid prescription for more than 30 days of supply were also excluded, as those may represent intention for initiation of prolonged opioid therapy.

Study Opioids

The complete list of included study opioid analgesics were codeine, hydrocodone, hydromorphone, morphine, oxycodone, propoxyphene, and tramadol (Supplemental Content Table 1). Opioid duration was calculated based on the days of medication supply dispensed. Since use of some less traditional formulations may be difficult to monitor, opioids in powder or injectable form were excluded from our analysis. Additionally, we excluded opioid formulations used for cough and for diarrhea (e.g. opium tincture). We also excluded long-acting opioids, including methadone and transdermal fentanyl, as these are not commonly prescribed as initial treatment and likely represented less frequent and special situations.

Mental Health Conditions

We identified mental health conditions during the 365-day baseline period, defined as the presence of coded clinical diagnoses (ICD-9 codes), prescriptions for specific medications associated with mental health conditions or participation in psychotherapy.[29] Mental health diagnoses included anxiety, depression, bipolar disease, attention deficit hyperactivity disorder (ADHD), post-traumatic stress disorder (PTSD), history of suicide attempt or ideation, opioid use disorder, sleep disorders and schizophrenia. Medications associated with mental health conditions included: antipsychotics, mood stabilizers, benzodiazepines, non-benzodiazepine hypnotics, serotonin-norepinephrine reuptake inhibitor (SNRIs, excluding duloxetine hydrochloride due to use for treatment of neuropathic pain), and selective serotonin reuptake inhibitors (SSRIs). Psychotherapy use was measured using procedures codes.[29] A complete list of mental health conditions and associated medications is included in Supplemental Content Table 2.

Outcome

The study outcome was time to chronic opioid therapy. We defined chronic opioid therapy as having more than 90 cumulative days’ supply of opioids from at least two separate prescriptions dispensed within a 6-month window having no gaps in supply greater than 32 days.[29–32]

Follow-up

For the primary outcome assessment, follow-up continued from the date of cohort enrollment (opioid use initiation) through the date when the chronic opioid therapy definition was fulfilled, the end of the study (December 31^st 2012), date of death, the 180^th day without opioids, development of a serious disease, enrollment in a Medicare Advantage program or hospital admission for any cause, whichever came first. We chose to censor patients who developed a serious disease or were admitted to the hospital, as these events may alter opioid prescription trajectory for our patient population. As opioid use could be initiated multiple times during the study period, patients could contribute more than one episode of opioid use for the study, provided that all selection criteria were fulfilled and a completely new set of study covariates (below) was measured.

Covariates

Study covariates (Table 1 and Supplemental Content Table 3) were obtained during the baseline 365 days preceding the enrollment date and included demographics (age, gender, race), body mass index [BMI], co-morbid pain conditions, comorbidities (acute myocardial infarction, arrhythmia, atrial flutter, cardiac valve abnormalities, congestive heart failure, hypertension, hyperlipidemia, pacemaker, chronic obstructive pulmonary disease (COPD), pneumonia, smoking history, carotid artery disease, peripheral artery disease, dementia, Parkinson’s disease, transient ischemic attack (TIA), stroke, diabetes, obesity, osteomyelitis, osteoporosis, urinary tract infection (UTI), sepsis, falls, fractures), and medications associated with disease. In addition, baseline pain scores were recorded as the highest documented score within 90 days prior to initiating opioid therapy. We also characterized the initial opioid prescription based on the dispensed days of supply: 1–7, 8–15, 16–29 and 30 days; and based on the opioid dose dispensed in estimated oral morphine equivalent (OME), using standard conversion factors (Supplemental Content Table 1).

Table 1:

Selected Characteristics of The Study

Patient Characteristics	Episodes WITH Mental Health Disease (N=8,607)	Episodes WITHOUT Mental Health Disease (N=13,845)
Age, Median (IQR)	62 [56, 72]	66 [58, 75]
Gender, N (%)
Male	7, 264 (84.4)	12,585 (90.9)
Race, N (%)
White	6,849 (79.6)	11,231 (81.1)
Black	1,165 (13.5)	1,665 (12.0)
Other	593 (6.9)	949 (6.9)
Pre-Existing Disease, N (%)
Cardiac
Arrhythmia	756 (8.8)	1,082 (7.8)
Atrial Flutter	518 (6.0)	786 (5.7)
CHF	574 (6.7)	717 (5.2)
Hypertension	4,246 (49.3)	6,158 (44.5)
Hyperlipidemia	3,026 (35.2)	4,105 (29.7)
Pulmonary
COPD	1,428 (16.6)	1,739 (12.6)
Smoking	1,103 (12.8)	1,218 (8.8)
Neurologic
Dementia	125 (1.5)	38 (0.3)
Stroke	160 (1.9)	119 (0.8)
Other
Diabetes	1,856 (21.6)	2,503 (18.1)
Obesity	796 (9.3)	615 (4.4)
Pain Conditions
Baseline Pain Scores, Median (IQR)*	7 (5, 8)	6 (4, 8)
Abdominal	407 (4.7)	634 (4.6)
Back	1,147 (13.3)	1,506 (10.9)
Musculoskeletal	4,316 (50.2)	7,738 (55.9)
Osteoarthritis	2,589 (30.1)	3,968 (28.7)
Medications, N (%)
Antihypertensives	2,515 (29.2)	3,481 (25.1)
Statin	4,661 (54.2)	6,808 (49.2)
Nitrates	1,080 (12.6)	1,522 (11.0)
NSAID	4,421 (51.4)	6,429 (46.4)
Oral Glucocorticoids	4,735 (55.0)	7,337 (53.0)
Healthcare Utilization, Median (IQR)
Outpatient Visits	6 (3, 9)	5 (3, 7)
ED Visits	0 (0, 1)	0 (0, 1)

^*Baseline pain scores were missing in 30% of the population.

Statistical Analysis

We used multivariable Cox proportional hazards regression models to examine the time to study outcomes between patients with and without mental health conditions, while accounting for relevant study covariates (Table 1 and Supplemental Content Table 3). Since the study encompassed several calendar years during which opioid use may have changed, we also accounted for the year of cohort entry as a covariate in our regression models. Adjusted hazard ratios and 95% confidence intervals were calculated from the regression models that included, study outcomes, exposures and all covariates listed above. Since a patient could contribute more than one episode of opioid use, we accounted for the clustering of observations at the patient level using the Huber-White Sandwich variance estimator, and computed robust standard errors for all estimates.[33] Since some covariates had missing values, we repeated our regression analyses using multivariate normal regressions for multiple imputation. For these, 10 imputed datasets were created and used for the estimations. In secondary analyses, we examined the role of individual mental health conditions and specific medications used on developing study outcomes. Furthermore, subgroup analyses were conducted to assess whether features of the first opioid prescription modified the association between mental health conditions and the risk of chronic opioid therapy. For these assessments, interaction terms in the multivariable regression models were examined. We verified the fulfilment of model assumptions using a log-log plot. All analyses were conducted using R (http://www.r-project.org) and Stata (StataCorp, College Station, TX).

RESULTS

There were 65,682 patients initially identified using DMARD therapy in the VA pharmacy databases. We applied selection criteria (Figure 1) and excluded those with data errors or inconsistencies on key demographic variables (N=161), diagnosis of serious illness (N=4,375), no active use of VA healthcare system in the prior year (N=6,371), absence of RA diagnosis (N=5,946), DMARD therapy for non-RA diagnosis (N=3,451), and Medicare advantage plan enrollment within the baseline year (N=3,858).

Figure 1:

Study Inclusion Flow Chart

Among 41,520 patients with RA meeting study selection criteria, 24,538 (59%) used opioids during the baseline study period and 15,324 (36.9%) had at least one episode of new opioid use, which required no use for the preceding 180 days. Among the new opioid users, 557 (3.6%) were further excluded due to initiation of non-study opioids (e.g. fentanyl, buprenorphine, methadone) or because they had an initial opioid prescription with more than 30 days of supply. Thus, we identified 14,767 new opioid users who contributed 22,452 episodes of new opioid use.

Patient Characteristics

Patients with RA and mental health conditions accounted for 38.3% (N=8,607) of observations and were predominantly white (80%), males (84%), with a median age of 62 (Interquartile range (IQR): 56, 72) years. Among episodes of new opioid use from patients with mental health conditions, 22% had diagnosis of depression (N=1,912), 13% (N=1,108) anxiety, 12% (N=1,048) were diagnosed with a sleep disorder, 9% (N=768) PTSD and 24% (N=2,041) had engaged in psychotherapy-related encounters prior to cohort enrollment. About 50% (N=4,328) of episodes from patients with mental health conditions had evidence of SSRI use, 11% (N=944) had antipsychotic medication user, 9% (N=743) had non-benzodiazepine hypnotic use, and 6% (N=544) had SNRI use (Supplemental Content Table 2). The most commonly reported pain conditions in episodes from patients with mental health conditions were musculoskeletal (50%, N=4,316), osteoarthritis (30%, N=2,589), and back pain (13%, N=1,147).

Patients with RA and without mental health conditions accounted for 61.7% (N=13,845) of the study population, were predominantly white (81%), males (91%) with a median age of 66 years (IQR: 58, 75). The distribution of relevant characteristics and pain conditions was generally similar between patients with and without mental health conditions (Table 1). However, episodes of new opioid use from patients with mental health conditions were more likely to have a diagnosis of dementia (1.5% vs 0.3%, p < 0.001), diabetes (22% vs 18%, p < 0.001), obesity (9% vs 4%, p < 0.001), prescriptions for proton pump inhibitors (54% vs. 43%, p < 0.001), and prescriptions for NSAIDs (51% vs 46%, p < 0.001). Patients contributing episodes with mental health conditions were also less likely to be male (84% vs 91%, p < 0.001) and less likely to have a diagnosis of musculoskeletal pain than patients without history of mental health conditions (50% vs 56%, p < 0.001; Table 1 and Supplemental Table 3).

Factors associated with Chronic opioid therapy

Patients were followed from initiation of opioid use through the earliest of date of death (N=135), enrollment into Medicare Advantage program (N=207), hospital admission (N=2,911), development of a serious illness (N=453), opioid discontinuation for more than 180 days (N=14,263), reaching end of study (N=793) or meeting the definition of chronic opioid therapy (N=3,690).

Among patients with mental health conditions who initiated opioid use, the rate of chronic opioid therapy during follow-up was 469.3 per 1000 person-years, compared with 378.1 per 1000 person-years among patients without mental health conditions. In the adjusted multivariable analysis, history of mental health conditions was significantly associated with chronic opioid therapy (aHR: 1.18; 95% CI 1.09 to 1.29, Table 2 and Figure 2).

Table 2.

Association of Mental Health Disease with Chronic Opioid Therapy

	Patients with Mental Health Disease	Patients without Mental Health Disease
Total Events	1,622	2,068
Person-Years*	3,456.2	5,468.9
Unadjusted rate/1000 person-years†	469.3	378.1
Adjusted Hazard Ratio (95% confidence interval)‡	1.18 (1.09 to 1.29)	Reference

^*Person-Years: Cumulative person-time at risk that participants contributed to the study

^†Unadjusted rate/1000 person-years: Crude incidence rate of chronic opioid use expressed per 1000 persons-years of observation

^‡Hazard ratio from multivariable Cox proportional hazards model adjusting for study covariates (Supplemental Content Table 3)

Figure 2: Forest Plot of the Adjusted Hazard Ratio* of Mental Health Conditions, Medications and Their Association with Chronic Opioid Therapy.

*Both the model for the overall effect and the model for independent individual effects are adjusted for study covariates (Supplemental Content Table 3)

The initial opioid prescription characteristics were also strongly associated with chronic opioid therapy. In multivariable analyses and compared to patients prescribed 0–7 days of opioid medication, patients prescribed 8–15 days of opioid medication had an aHR: 1.12 (95% CI 0.93 to 1.35), 16–29 days had an aHR: 1.52 (95% CI 1.16 to 2.01), and a 30 days’ supply had an aHR: 1.78 (95% CI 1.53 to 2.08), (Figure 3 and Supplemental Content Table 4).

Figure 3: Forest Plot of Adjusted Hazard Ratios* of Race, Gender, Day Supply of opioids and Their Association with Chronic Opioid Therapy.

*Both the model for the overall effect and the model for independent individual effects are adjusted for study covariates (Supplemental Content Table 3), including gender (Reference: Female), race (Reference: Black) and opioid day supply (Reference: 0–7 days)

In the assessment of individual mental health conditions and medications, non-opioid substance use disorder (aHR: 1.35; 95% CI 1.05 to 1.73), benzodiazepines (aHR: 1.13; 95%CI 1.01 to 1.27), non-benzodiazepine sedative hypnotics (aHR: 1.26; 95% CI 1.01 to 1.56), previous opioid use disorder (aHR: 1.94; 95%CI 1.09 to 3.46), SSRIs (aHR: 1.11; 95% CI 1.00 to 1.24) and antipsychotics (aHR: 1.37; 95% CI 1.13 to 1.67) were significantly associated with an increased risk of chronic opioid therapy (Figure 2). Results from models with multiple imputation were very similar to results from the main analyses. Multiple imputation results, which accounted for missing covariates especially pain scores (missing in about 30% of study episodes), were consistent with our primary analysis of the risk of developing chronic opioid therapy (aHR: 1.21; 95% CI 1.12 to 1.30).

Subgroup analysis

The association between history of mental health conditions and risk of chronic opioid therapy did not vary by the duration of the initial opioid prescription (p for interaction term: 0.404).

DISCUSSION

This national retrospective cohort study of veterans with RA found that history of mental health conditions, in particular both history of opioid and non-opioid substance use disorder, and use of medications associated with mental health conditions such as use of benzodiazepines and SSRIs, were independently associated with a high risk of chronic opioid therapy. In addition, the characteristics of the initial opioid prescription were strongly and independently associated with the study outcomes. These findings complement previous observations from cross-sectional studies that explored factors associated with prescription opioid use.[34, 35]

Our observations suggest that careful consideration of the history of mental health conditions, specifically prior substance use disorders, and the characteristics of initial opioid prescriptions are warranted when planning initiation of opioid therapies among patients with RA. Previous studies have documented the association between history of substance use disorder and opioid use.[36, 37] Our study also found a significant association between medications associated with mental health conditions, specifically benzodiazepines and SSRIs, and development of chronic opioid therapy. The use of benzodiazepines and SSRIs has been associated with prolonged opioid therapy in non-RA populations.[38, 39] Furthermore, a report by the Centers for Disease Control and Prevention (CDC) National Violent Death Reporting System described the association between anti-depressants, opioid use and suicide-related drug overdoses. According to the report, 23% of deceased people tested positive for anti-depressants and 20.8% tested positive for opioids.[40]

There is also a growing body of evidence focusing on the association between the characteristics of the initial opioid prescriptions and the risk of subsequent persistence on opioid therapies.[41, 42] Although most previous evidence comes from studies of patients without RA, our findings from patients with RA are consistent with previous observations, and suggest that consideration of patient history and evaluation of initial opioid prescribing patterns could be an actionable target for future preventative initiatives. Persistence of opioid therapy has been associated with worsening of chronic pain and diagnosis of depression. In a retrospective cohort study of 49,770 patients receiving care in the Department of Veterans Affairs Healthcare system without a previous history of depression, opioid prescriptions for 90–180 days was associated with a 25% increased risk of depression and those using opioids for more than 180 days had a more than 50% increase in depression.[43]

Our study findings also extend our understanding of the role of history of mental health conditions and initial opioid prescription patterns among a population of primarily male adults with RA.[22, 44] Due to well-recognized gender differences in disease prevalence, males affected with RA usually represented a small fraction of participants in previous studies. Thus, our study provides an examination on this understudied population.

Our study used administrative and clinical data derived from electronic medical records to allow the identification of patients with RA and enable longitudinal follow-up and identification of study outcomes. To minimize misclassification in the identification of our study population, we applied definitions that combined coded diagnoses and medication use and have been previously demonstrated to have high positive predictive values in identification of patients with RA.[28, 45] Our assessment of medication use over time was based on pharmacy dispensing records, which are free of recall issues and have a strong concordance with self-reported medication use.[46, 47] Although to our knowledge, there is no standardized definition for chronic opioid therapy, we adopted a definition that has been applied in previous studies. [1, 12, 20, 48] Furthermore, we identified an extensive set of relevant covariates, and conducted multivariable regression analyses to allow the identification of factors independently associated with our study outcomes.

Nevertheless, our findings must be interpreted in light of several limitations. First, veterans may not receive all of their care or medications at facilities captured in our study, which may lead to underreporting outcomes or medication data. We addressed this concern by supplementing VA data with Medicare and Medicaid data. Second, although we relied on recorded pharmacy prescription fills to ascertain exposure, we were unable to verify the reason for opioid prescription or that patients were taking the medications as prescribed which may lead to misclassification. Third, we did not have direct measurements of RA disease severity or disease duration for our study, and although our analyses accounted for a number of conditions and medications use, we cannot rule out the possibility of residual confounding. Forth, diagnosis of some mental health conditions can be challenging, and some conditions may be underrecognized. The retrospective identification of mental health conditions applied in our study may introduce misclassification, which would lead to an underestimation of the real association between mental health conditions and study outcomes. We also acknowledge that although we identified and accounted for a number of common pain related conditions, it can be challenging to identify some less common conditions in a retrospective manner. For example, during the study years fibromyalgia did not have a specific diagnosis code and could only be identified through unspecified diagnoses of symptoms, and thus, it was not included in our study to reduce concerns about misclassification. Although opioid use by itself could increase the risk sensitivity to pain (opioid-induced hyperalgesia) and the need for additional pain management in some patients, the contribution of this phenomenon to the study outcomes could not be examined with the available data. Last, although RA is more common among women, our population is largely composed of veteran men, which may limit the generalizability of our findings. However, our study provides a valuable insight on male patients with RA, who are underrepresented in the existing literature.

Our findings indicate that history of mental health conditions, especially history of substance use disorders, and duration of the initial opioid prescription are associated with chronic opioid use among patients with RA. An increased emphasis on the examination of the past medical history of patients with RA could reduce unwarranted opioid exposure and minimize the risk of chronic opioid therapy and its consequences. Further understanding on how initial prescriptions of opioid medication for acute pain episodes can shape patient behaviors is imperative to curb the national opioid epidemic.

Table 3:

Frequency and Initial Day Supply of Opioid Medications

Patient Characteristics	Episodes WITH Mental Health Disease (N=8,607)	Episodes WITHOUT Mental Health Disease (N=13,845)
Opioid, N (% of episodes)
Hydrocodone	3,642 (42.3)	5,586 (40.4)
Tramadol	1,721 (20.0)	3,043 (22.0)
Codeine	1,243 (14.4)	2,226 (16.4)
Oxycodone	1,140 (13.3)	1,709 (12.3)
Propoxyphene	712 (8.3)	1,126 (8.1)
Morphine	136 (1.6)	89 (0.6)
Hydromorphone	12 (0.1)	26 (0.2)
Days’ Supply, N (% of episodes)
0 – 7	1,846 (21.5)	2,704 (19.5)
8 – 15	1,457 (16.9)	2,434 (17.6)
16 – 29	282 (3.3)	472 (3.4)
30	5,022 (58.4)	8,235 (59.5)

Key Points.

• Approximately a third of patients with RA are exposed to opioid analgesics.

• Patients with RA and history of mental health disease, especially substance use disorders, who initiate opioid use have an increased risk of chronic opioid therapy.

• This study provides insight in an underrepresented population of mainly male patients with RA.