Associations of Mental Health and Family Background With Opioid Analgesic Therapy

A Nationwide Swedish Register-Based Study

Patrick D. Quinn; Martin E. Rickert; Johan Franck; Amir Sariaslan; Katja Boersma; Paul Lichtenstein; Henrik Larsson; Brian M. D'Onofrio

Disclosures

Pain. 2019;160(11):2464-2472. 

In This Article

Results

Opioid Initiation

In this cohort of 5,071,193 adolescents and adults who were naive to prescription opioid analgesics, the estimated cumulative incidence of initiating an episode of opioid prescription within 3 years was 11.4% (pointwise 95% confidence interval [CI], 11.3%-11.4%). As shown in eTable 4, supplemental digital content (available at http://links.lww.com/PAIN/A838), older individuals were more likely to initiate opioids than were younger individuals.

From 2007 onward, the cohort was dispensed 5,399,576 opioid prescriptions (eTable 5, supplemental digital content, details the specific opioid drugs, available at http://links.lww.com/PAIN/A838). The majority were for weak opioids, including tramadol (32.9% of dispensed prescriptions) and codeine (32.4%). Among opioid recipients, 52.3% were dispensed at least 1 prescription for a codeine-containing product, 40.1% tramadol, 20.8% oxycodone, 10.4% morphine, 9.5% dextropropoxyphene (removed from the market in 2011), 2.2% buprenorphine or methadone, and 2.2% other opioids.

Individuals with most preexisting mental health conditions had greater rates of opioid initiation than did individuals without those conditions. These HRs, adjusted for demographics, ranged from 1.24 (1.20–1.27) for bipolar disorder to 2.12 (2.04–2.21) for OUD (Table 1). Only schizophrenia spectrum disorders were inversely associated with opioid initiation (HR, 0.84; 0.83–0.86). Moreover, relative to no conditions, the number of mental health conditions was also associated with greater initiation rates (Figure 1). As shown in eTable 6 and eTable 7, supplemental digital content (available at http://links.lww.com/PAIN/A838), we found comparable associations between prior psychoactive medications and initiation, as well as in sensitivity analyses examining past-6-year mental health diagnoses and excluding buprenorphine and methadone recipients.

Figure 1.

Preexisting mental health conditions and opioid initiation, long-term opioid therapy, and concurrent benzodiazepine-opioid therapy. Kaplan–Meier estimates of cumulative incidence (%) of opioid initiation since 2007 in 5,071,193 individuals who were naive to prescription opioid analgesics (A) and, among 1,298,083 individuals initiating opioids, long-term opioid therapy (B), and concurrent benzodiazepine-opioid therapy (C) by number of preexisting mental health conditions. Shaded areas are pointwise 95% confidence intervals.

Long-term Opioid Therapy

Among the 1,298,083 individuals who received opioids before censoring, the estimated cumulative incidence of LTOT was 7.6% (7.6%-7.7%) within 3 years of opioid initiation and was greater among older opioid recipients (eTable 4, supplemental digital content, available at http://links.lww.com/PAIN/A838). For each mental health condition, individuals with preexisting diagnoses had greater rates of transitioning from opioid initiation to LTOT relative to those without (Table 2). Adjusted HRs ranged from 1.66 (1.56–1.77) for bipolar disorder to 2.20 (2.15–2.25) for SUD and 3.82 (3.51–4.15) for OUD. The number of mental health conditions was again associated with greater rates of LTOT (Figure 1), and the relative magnitudes of all LTOT associations were greater than their respective magnitudes for opioid initiation.

Similarly, across each class of psychoactive medications, prior recipients had greater LTOT rates than did nonrecipients (eTable 6, supplemental digital content, available at http://links.lww.com/PAIN/A838). As shown in eTable 8, supplemental digital content, (available at http://links.lww.com/PAIN/A838), sensitivity analyses examining past-6-year mental health diagnoses and excluding buprenorphine and methadone recipients also produced comparable associations. In a test of the sensitivity of the results to the LTOT definition, we found comparable but stronger associations when we defined LTOT as strong opioids in 3 of 4 consecutive quarters. Although the cumulative incidence was lower (1.7% [1.7%-1.7%] within 3 years of opioid initiation), adjusted HRs for this definition ranged from 2.46 (2.21–2.75) for bipolar disorder to 3.54 (3.40–3.69) for SUD and 8.43 (7.44–9.55) for OUD(eTable 8, supplemental digital content, available at http://links.lww.com/PAIN/A838).

Concurrent Benzodiazepine Therapy

The estimated cumulative incidence of concurrent benzodiazepine-opioid therapy was 5.7% (5.7%-5.8%) within 3 years of opioid initiation, and opioid recipients with each preexisting mental health condition had greater rates than did recipients without the respective condition (Table 2 and Figure 1). Adjusted HRs ranged from 3.56 (3.48–3.65) for suicide attempt or other self-injury to 6.70 (6.23–7.20) for OUD.

Family Background

Among the subcohort of 1,482,462 individuals with familial data (aged 13–29 years in 2007), parental mental health history was associated with differences in rates of dispensed opioid prescriptions (Table 3). Specifically, adolescents and young adults with parents who had been hospitalized for anxiety or depressive disorder, opioid or nonopioid SUD, or suicide attempt or other self-injury had greater rates of opioid initiation as well as subsequent LTOT. Adjusted HRs for opioid initiation ranged from 1.27 (1.25–1.29) for anxiety or depressive disorder to 1.41 (1.39–1.44) for suicide attempt or other self-injury. Among opioid recipients, adjusted HRs for LTOT ranged from 1.34 (1.27–1.41) for anxiety or depressive disorder to 1.44 (1.36–1.53) for SUD. As shown in eTable 9, supplemental digital content (available at http://links.lww.com/PAIN/A838), these associations persisted after adjustment for offspring mental health conditions and family SES.

We additionally found variability in opioid prescription as a function of family-of-origin SES. Individuals from lower-SES families, broadly, had greater rates of opioid therapy initiation than did those from higher-SES families. Moreover, among opioid recipients, those from lower-SES families had greater rates of transitioning to LTOT. These associations held across multiple indices of SES, including parental education, family income, and neighborhood deprivation. Although children of immigrants and those from metropolitan areas had modestly greater opioid initiation rates, their rates of subsequent LTOT did not differ from those with Nordic-born parents or those from nonmetropolitan areas, respectively.

processing....