FDA is requiring opioid pain medicine manufacturers to update prescribing information regarding long-term use

Using its authority under Section 505(o)(3) of the Federal Food, Drug, and Cosmetic Act, FDA required extended-release/long-acting (ER/LA) opioid analgesic (OA) new drug application holders to conduct epidemiologic studies to 1) quantify the serious risks of misuse, abuse, addiction, and fatal and non-fatal overdose in patients using OAs long term and 2) assess potential risk factors for these outcomes. The studies were conducted using prespecified protocols and statistical analysis plans that were reviewed by FDA and discussed in a public scientific workshop. Studies conducted under postmarket requirements (PMRs) 3033-1 and 3033-2 were large, multisite investigations that included patients enrolled in various health insurance plans and health systems across the United States. These studies were, by design, restricted to the relatively small proportion of patients receiving OAs who go on to use them long term and therefore do not inform quantitative questions of risk related to shorter-term use of OAs. 

PMR 3033-1 was a prospective, observational cohort study that estimated the risks of addiction, abuse, and misuse in adult patients initiating long-term use of Schedule II OAs. Patients were recruited and data were collected from 2017 through 2021. Study participants had been enrolled in selected health insurance plans or health systems for at least one year, and either 1) filled multiple ER/LA OA prescriptions during a 90-day period (“ER/LA cohort”); or 2) filled any Schedule II OA prescriptions covering at least 70 of 90 days (“long-term opioid therapy [LtOT] cohort”). Patients who received any of the qualifying opioid analgesics (i.e., ER/LA OAs or Schedule II OAs, depending on the study cohort) in the previous six months were excluded; however, patients were not excluded for use of other prescription OA therapy. In addition, patients with an existing diagnosis of a terminal illness or opioid use disorder (OUD) in the previous 12 months, receiving methadone or buprenorphine for the treatment of OUD, or receiving hospice care were excluded. Continued OA use was not required during follow-up. After meeting the study eligibility criteria, patients were required to be free of at least one outcome at baseline and to have completed a minimum number of follow-up assessments (at least two of the three-, six-, nine-, and 12-month assessments for misuse and abuse; the 12-month assessment for OUD) to be included in one or more analyses. This resulted in 978 and 1,244 patients being included in one or more analyses for the ER/LA and LtOT cohorts, respectively.

Opioid misuse was defined as the intentional use of a drug for a therapeutic purpose inappropriately outside labeling directions or in a way other than prescribed or directed by a health care professional. Opioid abuse was defined as the intentional use of a drug for a nontherapeutic purpose, repeatedly or sporadically, or for the purpose of achieving a positive psychological or physical effect. Misuse and abuse were measured using the Prescription Opioid Misuse and Abuse Questionnaire, a self-reported questionnaire validated for use in this population, which asks about symptoms in the past three months. Over 12 months, across the two cohorts, approximately 22% of included patients newly met criteria for prescription opioid misuse and approximately 9% of included patients newly met criteria for prescription opioid abuse, based on information reported at three-, six-, nine-, and 12-month assessments (Table 1).

Addiction was defined in this study as moderate-to-severe OUD, as assessed using a validated, semi-structured interview tool, the Psychiatric Research Interview for Substance and Mental Disorders, DSM-5, Opioid Version (PRISM-5-Op), which asks about symptoms in the past 12 months. OUD was based on a count of symptoms reported in the interview and defined in two ways: 1) using standard DSM-5 criteria, and 2) using modified criteria in which most DSM-5 symptoms were counted toward an OUD designation only when the patient indicated a non-pain reason for opioid use associated with that symptom, and the “persistent desire or attempts to quit or cut down on opioid use” criterion was counted only if multiple unsuccessful attempts were made. Over 12 months, across the two cohorts, the percentage of included patients who newly met criteria for moderate-to-severe OUD was approximately 3-6% using the standard DSM-5 criteria and approximately 1-2% using the modified DSM-5 criteria, based on information reported at the 12-month assessment (Table 1). Several factors (e.g., potential volunteer bias, predominance of managed care and integrated healthcare systems) may have limited the generalizability and interpretability of findings.

PMR 3033-2 was a retrospective, observational cohort study that estimated the risk of opioid-involved overdose or opioid overdose-related death (together, abbreviated as OOD) in adult patients with new long-term use of Schedule II OAs between 2006 and 2016 (n=220,249). The study included patients enrolled in either one of two commercial insurance programs, one managed care program, or one Medicaid program for at least nine months. New long-term use was defined as having Schedule II OA prescriptions¹ covering at least a 70-days’ supply over the three months before cohort entry and none during the preceding six months. Patients were excluded if they had an opioid-involved overdose in the nine months prior to cohort entry. The outcome was the first OOD event during the follow-up period, as measured using a validated medical-code-based algorithm with linkage to the National Death Index database. The cumulative incidence of OOD increased steadily throughout the five-year follow up period, resulting in a five-year cumulative incidence ranging from approximately 1.5% to approximately 4% (Table 2).  Over the entire study period (5-11 years, depending on study site), approximately 17% of first OOD events were fatal.

There are a number of considerations when interpreting the findings from PMR 3033-2. OOD events may not have involved the prescribed OAs; rather, they could have occurred after a patient discontinued prescribed OA use and could have involved illicit opioids like heroin or fentanyl. Since the outcome for this study included only the first OOD event, a patient could have experienced subsequent events, including fatal overdose, that would not have been included in the OOD incidence estimates. Additionally, the intentionality of the OOD event (i.e., suicide vs. accidental) could not be adequately confirmed. Several factors could have contributed to bias. There was substantial attrition over the first five years of follow-up, and although this study used an incidence measure designed to account for loss to follow-up, if those who left the cohort (e.g., due to insurance disenrollment) systematically had a different risk of OOD than those remaining in the study, incidence estimates could have been biased. In addition, limiting the cohort to patients with no recent documented opioid-involved overdose likely selected patients at lower risk of OOD during follow-up. Finally, opioid overdoses that were reversed by a bystander or that otherwise did not result in either a medical claim or death were not captured.  

Both studies collected patient- and drug-related characteristics at baseline and conducted exploratory analyses of potential risk factors for misuse, abuse, addiction, and overdose. Many factors were associated with one or more outcomes, with one of the strongest and most consistent risk factors being a personal history of a substance use disorder.² In addition, higher OA dose during the three months before cohort entry was a strong and consistent risk factor for OOD. Neither study was designed to assess associations between changes in OA dose or discontinuation of opioids and adverse opioid-related outcomes. 

¹ Hydrocodone fixed combinations were reclassified from Schedule III to Schedule II in October 2014. Products containing hydrocodone were treated as a Schedule II opioid throughout PMR 3033-2.
² A notable proportion of patients starting long-term OA therapy in these studies had a personal history of SUD, whether in the past year (in PMR 3033-1, 6.5% to 8% had a past-year non-opioid, non-nicotine substance use disorder, based on interview measures; in PMR 3033-2, approximately 4-6% each had OUD, alcohol use disorder, or another substance use disorder, based on diagnostic codes). PMR 3033-1 also assessed SUDs prior to the past year (29% to 34.1% had a prior-to-past-year non-opioid, non-nicotine substance use disorder, based on interview measures).

Table 1. Incidence of Prescription Opioid Misuse, Prescription Opioid Abuse, and OUD in Prospective PMR 3033-1

Source: Adapted from Final Study Report for Prospective PMR 3033-1. 

¹ Moderate-to-severe pain-adjusted DSM-5-OUD was defined as having four or more pain-adjusted DSM-5 criteria for OUD related to prescription opioid use or two or more DSM-5 criteria related to heroin use, as measured by the PRISM-5-Op. 
² Moderate-to-severe DSM-5 OUD was defined as having four or more standard DSM-5 criteria for OUD related to prescription opioid use or two or more DSM-5 criteria related to heroin use, as measured by the PRISM-5-Op. 
³ Includes patients who initiated an ER/LA OA that included at least 28 days’ supply of an ER/LA OA within a 60-day window followed by a subsequent ER/LA OA prescription within a 7-day period, all within a 90-day period prior to the patient’s baseline interview. Patients could not have used an ER/LA OA in the 6 months before the initial 28 days’ supply of an ER/LA OA, but patients on IR/SA OAs during the same 6 months were still eligible for this cohort.
⁴ Includes patients who initiated either an ER/LA OA or a Schedule II IR/SA OA for at least 70 of the past 90 days. Patients could not have used an ER/LA OA or a Schedule II IR/SA OA in the 6 months before the initial ER/LA OA or Schedule II IR/SA OA prescription contributing to at least 70 days of use, but other prescription OA therapy would not exclude them (e.g., tramadol use).
Abbreviations: CI, confidence interval; DSM-5, Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition; ER/LA, extended-release/long-acting; LtOT, long-term opioid therapy; N, number; OA, opioid analgesic; OUD, opioid use disorder; PMR, postmarketing requirement; PRISM-5-Op, Psychiatric Research Interview for Substance and Mental Disorders, DMS-5, Opioid Version
 

Table 2. Cumulative Incidence of OOD in PMR 3033-2

Source: Adapted from the Final Study Report for PMR 3033-2. 

¹ Five-year cumulative incidence is the complement of the Kaplan-Meier OOD-free survival preceding 5 years measured in percent (%) scale
Abbreviations: CI, confidence interval; KPNW, Kaiser Permanente Northwest; N, number; OOD, opioid-involved overdose or opioid overdose-related death; PMR, postmarketing requirement; VUMC, Vanderbilt University Medical Center (Medicaid)