Prescription Patterns of Family Members After Discontinued Opioid or Benzodiazepine Therapy of Users

Abstract

This study uses data from a national database of privately insured individuals from 2007 to 2016 to examine first-prescription filling patterns exhibited by family members of high-volume users of opioids or benzodiazepine whose drug treatment has been discontinued.

The epidemic of opioid use disorder has led to increased scrutiny of patients prescribed large amounts of controlled substances, such as opioids and benzodiazepines. States and payers, such as commercial insurers and Medicaid programs, are imposing stricter limits on prescribing opioids¹ raising concerns that these policies may pressure physicians to discontinue opioid therapy.^2,3 Because of the physical dependence inherent with long-term opioid or benzodiazepine use, individuals whose therapy is discontinued may resort to extreme measures to avoid withdrawal symptoms.⁴ The objective of this study was to examine the prescription patterns of family members of high-volume opioid or benzodiazepine users whose therapy is stopped. We hypothesized that when clinicians discontinue therapy for high-volume users, their family members may be more likely to obtain prescriptions for these medications to potentially divert those prescriptions to the high-volume user.

Methods

We used a national database of privately insured individuals from 2007-2016 to identify families (including spouses, children, or other dependents) with 2 or more members enrolled for 3 months or longer, and identified all prescription fills for a 30-day or more supply of opioids or benzodiazepines. Among these individuals with a 30-day or more supply, based on the total number of opioid or benzodiazepine prescription fills during this period, we classified individuals in the top 10% as “high-volume users” and those in the bottom 50% as “low-volume users.” For each individual, we also identified the day when the prescribed medications were expected to end for the rest of the study period based on the last day supplied for the final opioid or benzodiazepine prescription (“discontinuation date”). The study was classified as nonhuman subjects research and thus determined to be exempt from review by the institutional review board at Harvard Medical School. Our primary objective was to determine the rate of the first opioid or benzodiazepine prescriptions filled by family members of individuals with discontinued therapy. Among these individuals, we compared those whose family members had a first-time opioid or benzodiazepine fill with those whose family members did not while focusing on the 2 days before and after the discontinuation date to capture the highest-risk period around that date. We estimated the relative risk of family members’ first-time prescription fill (binary variable) using beneficiary-day–level logistic regression with indicator variables for each day relative to the discontinuation date (which was defined as the last day’s supply for the last prescription filled by the index user). We also adjusted for user age, sex, comorbidities,⁵ and indicators for the calendar year. We used robust SEs clustered at the individual level for all estimates. The significance threshold was P < .05 and the significance testing was 2-sided.

Results

A total of 463 637 high-volume and 4.9 million low-volume opioid users, and 357 632 high-volume and 2.1 million low-volume benzodiazepine users with discontinued therapy were identified. Among the high-volume users, 788 (0.2%) of the opioid users and 683 (0.2%) of the benzodiazepine users had family members who filled first-time opioid or benzodiazepine prescriptions within 2 days of the discontinuation date, and 2397 (0.5%) of the opioid users and 1921 (0.5%) of the benzodiazepine users had family members who did so within 14 days of that date. Among the low-volume users, 3064 (0.06%) of the opioid prescription fills and 1328 (0.06%) of the benzodiazepine prescription fills were observed within 2 days of the discontinuation date, and 21 938 (0.4%) of the opioid prescription fills and 7068 (0.3%) of the benzodiazepine prescription fills within 14 days of that date. High-volume opioid users whose family members filled prescriptions were more likely than those whose family members did not to have higher daily doses (406 vs 353 morphine equivalents, P < .001) (Table) and reside in rural areas (4.8% vs 1.8%, P < .001). Similarly, high-volume benzodiazepine users whose family members filled prescriptions were more likely to have higher health care spending during the previous 12 months ($17 076 vs $12 452, P < .001) and reside in rural areas (6.0% vs 1.9%, P < .001) compared with those who did not have family members fill prescriptions.

Table. Characteristics of High-Volume Opioid and Benzodiazepine Users by Whether Their Family Member Filled the First Prescription Within 2 Days of Discontinuation Date^a.

Characteristic	High-Volume Opioid Users			High-Volume Benzodiazepine Users
	Family Member With Prescription Fill (n = 788)	No Family Member With Prescription Fill (n = 462 849)	P Valueb	Family Member With Prescription Fill (n = 683)	No Family Member With Prescription Fill (n = 356 949)	P Valueb
Prescription use
No. of opioid or benzodiazepine prescriptions, mean (SD)	49.7 (42.4)	44.5 (40.6)	<.001	38.6 (21.0)	35.7 (19.0)	<.001
6 mo Before discontinuation, mean (SD)
Daily morphine equivalents	405.8 (408.0)	353.3 (410.4)	<.001	NA	NA	NA
Days supplied	232.1 (104.9)	209.6 (110.3)	<.001	191.1 (63.4)	169.8 (72.0)	<.001
Concurrent use of opioid or benzodiazepine for respective groups, %	60.7	56.5	.02	72.3	70.5	.29
Demographics
Age, mean (SD)	51.0 (10.3)	49.8 (10.1)	<.001	49.9 (11.3)	49.0 (11.5)	.04
Female, %	52.5	52.4	.93	62.7	64.7	.28
Male, %	47.5	47.6	.93	37.3	35.3	.28
Rural residence, %	4.8	1.8	<.001	6.0	1.9	<.001
Relationship of high-volume user to insurance policyholder, %
Policyholder	28.2	44.7	<.001	26.5	42.9	<.001
Spouse	67.0	53.4	<.001	65.7	50.9	<.001
Child or other dependent	4.8	2.0	<.001	7.8	6.2	.10
Census region, %
Northeast	11.4	12.6	.32	12.6	16.2	.01
Midwest	20.3	22.3	.17	20.1	23.3	.046
South	51.4	45.3	<.001	52.0	42.2	<.001
West	12.1	18.0	<.001	9.2	16.3	<.001
Health care spending and utilization
In 12 mo before discontinuation
Total health care spending, mean (SD), $	18 916 (47 983.6)	21 389 (57 480.7)	.23	17 076 (51 687.1)	12 452 (36 711.7)	<.001
Hospitalization, %	21.1	18.5	.07	17.3	14.0	.01
Total comorbidities and mental health diagnoses
No. of Elixhauser comorbidities, mean (SD)c	4.1 (3.2)	4.0 (3.3)	.51	4.0 (11.5)	4.0 (11.3)	.76
Alcohol use disorder, %	5.2	4.9	.70	7.6	7.5	.92
Substance use disorder, %	15.4	15.5	.92	12.6	12.6	.98
Psychosis, %	4.2	4.7	.48	8.5	8.0	.63
Depression, %	43.7	42.6	.55	56.8	61.8	.01

Abbreviation: NA, not applicable.

^aThe columns show characteristics of high-volume users whose family members did or did not have a first opioid or benzodiazepine prescription fill within 2 days of the therapy discontinuation date.

^bP values were estimated using 2-sample t tests (2-tailed) or z tests for proportions, as appropriate.

^cThe number of Elixhauser comorbidities was calculated as the number of 29 Elixhauser comorbidities present during a hospital admission.⁵

A large relative increase was observed in the probability of a first-time fill by family members of high-volume opioid users in the days surrounding the discontinuation date (relative risk, 3.53 [95% CI, 2.45-4.60] on day −2 and 4.16 [95% CI, 2.91-5.40] on day 0 vs day −14) (Figure). Family members of high-volume benzodiazepine users exhibited similar increases (relative risk, 4.46 [95% CI, 2.85-6.07] on day −2 and 4.23 [95% CI, 2.70-5.77] on day 0 vs day −14). These prescription filling patterns were not observed for family members of low-volume opioid and benzodiazepine users.

Figure. Adjusted Relative Risk of Family Members Filling First Prescription for Opioids or Benzodiazepine on Days Around the Therapy Discontinuation Date for High- and Low-Volume Users.

The y-axis shows the adjusted relative risk of family members obtaining a prescription of the respective medication for the first time 14 days before through 14 days after the discontinuation date of the therapy for the index user. Calculation of relative risk and the use of robust SEs are explained in the Methods section. Day −14 before the discontinuation date for the index user is the reference day (Ref) for relative risks. The top panels depict the adjusted estimates for opioid users’ family members, whereas the bottom panels show estimates for benzodiazepine users’ family members. We compare the patterns for family members of high-volume users (top 10%) vs those of low-volume users (bottom 50%). The low-volume users’ families serve as a control under the assumption that discontinuation of opioid or benzodiazepine use among low-volume users is less likely to lead to diversion of these medications by another family member. Limit lines indicate 95% CIs.

Discussion

We observed a clear increase in first-time opioid or benzodiazepine prescriptions filled by family members around the therapy discontinuation date for high-volume users, whereas this pattern was not observed for the family members of low-volume users. Although family member fills were very uncommon, they represent a potentially concerning scenario: some patients who were unprepared for therapy discontinuation had family members who obtained medication for them. One key limitation of this study is that our analysis covered only privately insured patients and may not be generalizable to other populations.

Given elevated concerns about the potential adverse effects of long-term use of opioids and benzodiazepines and the risk of diversion, the pressure on clinicians to discontinue therapy may increase. Our results suggest one possible mechanism that patients with discontinued therapy may use to avoid withdrawal or transition to a new prescriber.