Medical marijuana access and prolonged opioid use among adolescents and young adults

Abstract

Background and Objectives

Laws liberalizing access to medical marijuana are associated with reduced opioid analgesic use among adults, but little is known about the impact of such policies on adolescents and young adults.

Methods

This retrospective cohort study used 2005 to 2014 claims from MarketScan® Commercial database, which covers all 50 states and Washington D.C. The sample included 195,204 adolescent and young adult patients (aged 12–25) who underwent one of 13 surgical procedures.

Results

Of the 195,204 patients, 4.8% had prolonged opioid use. Several factors were associated with a higher likelihood of prolonged opioid use, including being female (adjusted odds ratio [aOR], 1.27; 95% confidence interval [CI], 1.21–1.33), longer hospital stay (aOR, 1.04; 95% CI, 1.02–1.06), greater days of index opioid supply (8–14 days: aOR, 1.39, 95% CI, 1.33–1.45; greater than 14 days: aOR, 2.42, 95% CI, 2.26–2.59), rural residence (aOR, 1.07; 95% CI, 1.01–1.14), and cholecystectomy (aOR, 1.16; 95% CI, 1.08–1.25). There was not a significant association of medical marijuana dispensary laws on prolonged opioid use (aOR, 0.98; 95% CI, 0.81–1.18).

Conclusions and Scientific Significance

Medical marijuana has been suggested as a substitute for opioids, but our results focusing on adolescents and young adults provide new evidence that this particularly vulnerable population does not exhibit reductions in prolonged use of opioids after surgery when they have legal access to medical marijuana. These findings are the first to demonstrate potentially important age differences in sustained use of opioids, and point to the need for prescriber oversight and management with this vulnerable population.

INTRODUCTION

While opioid prescribing in the United States has been decreasing overall,¹ statistics on opioid prescribing to adolescents and young adults show varying trends. Data from Optum Clinformatics Data Mart suggest that any opioid prescribing to a child or adolescent has declined around 50% between 2012 and 2017 among those with commercial insurance, with much of the decline being driven by reductions in the number of hydrocodone prescriptions.² However, other work suggests that prescribing to adolescents and young adults in emergency and ambulatory settings has remained fairly constant.³

Adolescents and young adults are an at-risk population for opioid misuse and opioid harm,^4, 5 with adolescents and young adults who are prescribed opioids being at greater risk of long-term opioid use^6, 7 and more likely to progress to heroin use.⁸ When prescription opioids are taken for a short period of time following a doctor’s prescription, the opioids are safe to take and can reduce patients’ pain.⁹ However, like illicit opioids, prescription opioids can lead to dependence and addiction when they are used for prolonged periods of time. While rates of unintentional poisonings and overdoses from prescription opioids among children and adolescents have been declining since 2011,¹⁰ there has been a threefold increase in fatal opioid poisonings among children and adolescents between 1999 and 2016.⁵ In more recent years, between mid-2019 and the end of 2021, the number of drug overdose deaths among adolescents was 2231, and opioid-involved overdose deaths accounted for 91.3% of the total deaths.¹¹

State and federal agencies have implemented a wide range of strategies and targeted interventions to try to reduce overprescribing,¹² including implementation and mandated use of prescription drug monitoring programs (PDMPs),¹³ opioid prescribing guidelines,^14, 15 and pain clinic regulations.¹⁶ One policy that has garnered attention from researchers in recent years is medical marijuana laws. Randomized trials have documented that cannabinoids have a modest association with reduced levels of chronic pain,¹⁷ and studies examining the association between state medical marijuana law adoption and opioid prescribing have shown a reduced number of opioid prescriptions, daily opioid doses prescribed, spending on opioid prescriptions and opioid prescribing rates.^18–23 However, these effects have been found within adult populations or state-level analyses, and little attention has been given to the impact of medical marijuana laws on adolescent and young adult opioid use, particularly among postsurgical adolescents and young adults, populations likely to receive opioids for pain management.

In light of the evidence that adolescents and young adults are at substantial risk of prolonged opioid use after undergoing surgical and dental procedures^24, 25 and that medical marijuana legalization is associated with reductions in opioid prescribing in adults,^18, 22, 23 we investigate whether postsurgical adolescents and young adult patients in states with open medical marijuana dispensaries are at decreased risk of prolonged opioid use. We hypothesize that adolescents and young adults in states with open medical marijuana dispensaries are less likely to have prolonged opioid use in the 6 months after their surgical procedure.

METHODS

Data and sample

We used the IBM® MarketScan® Commercial Claims and Encounters database for the years 2005 through 2014 to examine whether open medical marijuana dispensaries are associated with reductions in prolonged opioid medication use. MarketScan collects patient-level data from self-insured employers and medium and large commercial health plans. The database covers all 50 US states and Washington DC and provides individual-level claims data on service utilization, expenditures, patient demographic information, and pharmacy claims.

Our study period ends in 2014 for two reasons. First, the transition from ICD-9 to ICD-10 codes in health claims data meant that transitions in coding could affect how our outcome measure was captured. Second, the Marketscan data experienced a decline in sample size from 2015 due to a few of their large data contributors (insurers) no longer providing their claims to the aggregator. The overall size of the database fell by about 40%, making it difficult to do longitudinal analysis over the jump from 2014 to 2015. Therefore, we used the data up to 2014 for our analysis.

We identified 512,223 adolescent (defined as 12–17 years old) and young adult (defined as 18–25 years old) patients who underwent one of 13 surgical procedures (appendectomy, tonsillectomy and/or adenoidectomy, cholecystectomy, colectomy, inguinal hernia, pectus repair, umbilical or epigastric hernia, supracondylar fracture fixation, epicondylar fracture fixation, posterior arthrodesis, arthroscopic knee repair, orchiopexy, and hypospadias repair) used in prior research between 2005 and 2014.²⁴ As we focused on examining prolonged opioid medication use after surgery, we only included patients with an index opioid prescription which was defined as filling at least one opioid prescription during the period from 30 days before admission to 2 weeks after discharge.^24, 26 In addition, we excluded individuals who were not continuously insured during the 6 months before their admission and 6 months after discharge, those who had additional surgeries or anesthesia during the 6 months after discharge, those diagnosed with opioid use disorder or an opioid-related overdose in the 6 months before their admission, and those whose census region was unknown. Study flow diagram can be found in Figure 1. As a sensitivity analysis, we repeated our analysis limiting the sample to the 2010–2014 study period to match the time frame of prior studies. The Pennsylvania State University IRB approved this study.

Measures

Our primary outcome measure was prolonged opioid use, defined as refilling at least one opioid prescription during the period 90–180 days after hospital discharge.^24, 26, 27 Opioid prescriptions were identified through pharmacy claims data using National Drug Codes (NDC). Among prescriptions containing hydrocodone or codeine, prescriptions for cold or cough treatment were excluded.²⁸

The main exposure variable was the state having a legally protected and open medical marijuana dispensary, similar to Powell et al.²⁹ A dichotomous medical marijuana dispensary variable was created that was equal to 1 if the state had medical marijuana dispensaries that were legally allowed and operational. States with open dispensaries without laws providing legal protections, and states with laws providing legal protections without open dispensaries were coded as 0. Because most states did not formalize their regulations specific to marijuana dispensaries until after the 2009 Department of Justice Issued Ogden memo,³⁰ we included a sensitivity analysis that replaced our main exposure variable with another dichotomous indicator of whether the state had passed any medical marijuana law. This sensitivity analysis also allowed us to directly compare our results to those of previous studies using adult populations. The information of the dates in which both policies became effective are from Williams et al.³¹ and are shown in Table 1.Table 1. Dates of medical marijuana law (MML) in effect and medical marijuana dispensaries legally protected and active.

State	MML effective date	Medical marijuana dispensaries legally protected and active
California	11/6/1996	1/1/2004
Oregon	12/3/1998	3/24/2014
Washington	12/3/1998	7/22/2011
Alaska	3/4/1999
Maine	12/23/1999	3/31/2011
Hawaii	6/14/2000	8/8/2017*
Colorado	12/28/2000	6/7/2010
Nevada	10/1/2001	7/31/2015*
Maryland	10/1/2003	7/6/2017*
Vermont	7/1/2004	6/21/2013
Montana	11/2/2004	12/7/2016*
Rhode Island	1/3/2006	4/19/2013
New Mexico	7/1/2007	7/1/2009
Michigan	12/4/2008
New Jersey	10/1/2010	12/6/2012
Washington, DC	7/27/2010	7/29/2013
Arizona	12/14/2010	12/6/2012
Delaware	7/1/2011	6/26/2015*
Connecticut	10/1/2012	8/20/2014
Massachusetts	1/1/2013	6/24/2015*
New Hampshire	7/23/2013	4/30/2016*
Illinois	1/1/2014	11/9/2015*
Minnesota	5/30/2014	7/1/2015*
New York	7/5/2014	1/7/2016*
Pennsylvania	5/17/2016*
Ohio	9/8/2016*
Arkansas	11/9/2016*
North Dakota	12/8/2016*
Florida	1/3/2017*

• Note: The asterisk indicates policies were effective after our study period (2005–2014). So, these were not included in the analysis.
• Source: Williams et al.³¹

Patient characteristics included sex, age, geographic region, urban versus rural residence, length of hospital stay, days supply of index opioid prescription, form of opioids for their index opioid prescription, inpatient surgery, type of surgical procedure, pre-existing behavioral health diagnoses, and previous opioid prescription, which was defined as filling at least one opioid prescription during the period 30–180 days before admission. We identified individuals with a behavioral health diagnosis claim in the 6 months before their surgical procedure, including depression, anxiety, attention deficit disorder/attention deficit hyperactivity disorder (ADD/ADHD), substance use disorder, and other mental health disorders not specified earlier, including bipolar disorder and conduct disorder, using International Classification of Diseases, 9th Revision (ICD-9) codes.

Statistical analyses

Bivariate analyses were conducted to identify the characteristics of adolescent and young adult patients with prolonged opioid medication use, and used a comparative interrupted time series (CITS) design to compare the likelihood of prolonged opioid use before and after having access to medical marijuana dispensaries among adolescents and young adults. While the difference-in-differences approach compares pre-post differences in means between treatment and comparison groups and thus requires a parallel trends assumption, the CITS model explicitly admits differences in pre-treatment trends between treatment and comparison groups.³² By extending the pre-trend into the post-period, CITS models effectively change the counterfactual against which treatment effects are defined.³³ Logistic regression models were used to estimate odds ratios of prolonged opioid use along with 95% confidence intervals (CIs). We controlled for state-fixed effects, year-fixed effects, and seasonal effects, and calculated Huber-White standard errors clustered at the state level. We also estimated the model with a reduced study period (2010–2014) that was used by previous studies^18, 19, 22 to investigate the robustness of our estimates and to compare them with the literature. We followed the REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) guidelines for this study.³⁴

RESULTS

In our 10-year sample of commercially insured adolescent and young adult patient procedures, we identified 512,223 adolescent and young adult patients who underwent one of 13 surgical procedures, and 195,204 patients were prescribed opioids in the first place and met inclusion criteria (Figure 1). Of the 195,204 patients residing in all 50 states and Washington DC, 4.8% had prolonged opioid use (Table 2). Compared to patients without prolonged opioid use, patients with prolonged opioid use were more likely to be female (61.7% vs. 53.6%; p < .01), live in rural areas (19.6% vs. 17.8%; p < .01) and be aged 18–25 years (63.9% vs. 52.7%; p < .01). Patients with prolonged opioid use were also statistically more likely to have a pre-existing mental health diagnosis (including depression, anxiety disorder, substance use disorder, and other mental health disorder), have a previous opioid prescription (29.1% vs. 13.7%; p < .01), have a longer length of hospital stay (1.39 ± 1.51 days vs. 1.31 ± 1.17 days; p < .01), and be prescribed more days of the index opioid prescription (11.35 ± 12.72 days vs. 8.03 ± 8.28 days; p < .01) than patients without prolonged opioid use.Table 2. Sample summary characteristics of people aged 12–25 who had one of 13 surgical procedures (n = 195,204).

	Total	Patients with prolonged opioid use	Patients without prolonged opioid use
	n	Row %	Row %	p Value
Total	195,204	4.8	95.2
Geographic regions				<.01
Northeast	23,713	3.2	96.8
North Central	54,946	4.5	95.5
South	79,862	5.3	94.7
West	36,683	5.3	94.7

	n	Column %	Column %	p Value
Female	105,443	61.7	53.6	<.01
Age
12–17	91,230	36.1	47.3	<.01
18–25	103,974	63.9	52.7	<.01
Rural	34,621	19.6	17.8	<.01
Pre-existing behavioral health diagnoses
Depression	7,016	6.3	3.5	<.01
Anxiety	5,406	5.0	2.7	<.01
ADD/ADHD	6,470	3.6	3.3	.18
Other mental health disorder	9,978	8.6	4.9	<.01
Substance use disorder	1,227	1.1	0.6	<.01
Previous opioid prescriptiona	28,197	29.1	13.7	<.01
Medical marijuana dispensary (legal and active)b	23,285	12.3	11.9	.25

	Mean ± SD	Mean ± SD	p Value
Average length of hospital stay	1.39 ± 1.51	1.31 ± 1.17	<.01
Average days supply of index opioid prescription	11.35 ± 12.72	8.03 ± 8.28	<.01

• Note: Statistical significance was set at p < .05. Bold indicates a p < .05.
• ^a Previous opioid prescription was defined as filling at least one opioid prescription during the period 30–180 days before admission.
• ^b Medical marijuana dispensary (legal and active) means that medical marijuana dispensaries are legally allowed to operate and known to be operating in the states.

Table 3 shows the estimated adjusted odds ratios (aOR) from our CITS model predicting prolonged opioid medication use. We used two models, one is the long period which is our main model (2005–2014) and the other is a restricted study period (2010–2014). We did not find a statistically significant protective effect of medical marijuana dispensaries on prolonged opioid use in 2005–2014 (aOR, 0.98; 95% confidence interval [CI], 0.81–1.18) and 2010–2014 (aOR, 0.84; 95% CI, 0.62–1.14) after adjusting for demographics, type of surgery, pre-existing behavioral health diagnoses, year and state fixed effects. Living in a state with medical marijuana dispensaries had a larger negative association with prolonged opioid use in the later part of the study period (2010–2014), but it remained not statistically significant. Table 3. Risk factors for prolonged opioid medication use.

	Year 2005–2014		Year 2010–2014
	aOR	95% CI	aOR	95% CI
Medical marijuana dispensary (legal and active)a
No medical marijuana dispensary (not legal or not active)	1.00	Ref	1.00	Ref
Having medical marijuana dispensary (legal and active)a	0.98	0.81–1.18	0.84	0.62–1.14
Pre-policy time trend	1.01	0.98–1.05	0.98	0.86–1.10
Sex
Male	1.00	Ref	1.00	Ref
Female	1.27***	1.21–1.33	1.24***	1.16–1.32
Length of hospital stay	1.04***	1.02–1.06	1.06***	1.03–1.09
Number of days supply of index opioid prescriptionb
1–7 days	1.00	Ref	1.00	Ref
8–14 days	1.39***	1.33–1.45	1.45***	1.36–1.55
Greater than 14 days	2.42***	2.26–2.59	2.56***	2.35–2.79
Type of opioids (form of index opioid)
All other forms of opioid (other than elixir or solution) (index opioid prescription)	1.00	Ref	1.00	Ref
Elixir or solution (index opioid prescription)c	0.81***	0.74–0.88	0.77***	0.69–0.87
Admissions
Outpatient surgery	1.00	Ref	1.00	Ref
Inpatient surgery	0.91**	0.84–0.99	0.89**	0.80–0.98
Procedure
Appendectomy	1.00	Ref	1.00	Ref
Tonsillectomy and/or adenoidectomy	0.79***	0.71–0.88	0.75***	0.66–0.85
Cholecystectomy	1.16***	1.08–1.25	1.14**	1.03–1.26
Colectomy	0.97	0.51–1.84	1.10	0.37–3.25
Inguinal hernia	0.97	0.88–1.07	0.91	0.78–1.07
Umbilical or epigastric hernia	1.10	0.93–1.30	0.90	0.70–1.15
Pectus repair	0.64***	0.48–0.86	0.61***	0.44–0.84
Supracondylar fracture fixation	1.34	0.86–2.09	1.11	0.65–1.89
Epicondylar fracture fixation	0.57**	0.34–0.97	0.54	0.25–1.16
Posterior arthrodesis	0.72***	0.58–0.91	0.65***	0.48–0.89
Arthroscopic knee repair	0.71***	0.64–0.79	0.66***	0.59–0.75
Orchiopexy	0.64***	0.47–0.87	0.61**	0.42–0.89
Hypospadias repair	0.74	0.22–2.50	1.00	1.00–1.00
Residence
Urban	1.00	Ref	1.00	Ref
Rural	1.07**	1.01–1.14	1.08	0.98–1.19
Pre-existing behavioral health diagnosesd
Depression	1.34***	1.23–1.46	1.16**	1.01–1.33
Anxiety	1.44***	1.30–1.59	1.60***	1.38–1.85
ADD/ADHD	1.00	0.88–1.14	0.96	0.81–1.14
Other mental health disorder	1.40***	1.31–1.49	1.40***	1.26–1.56
Substance use disorder	1.38***	1.11–1.71	1.57***	1.24–1.98
Previous opioid prescriptione
No previous opioid prescription	1.00	Ref	1.00	Ref
Having previous opioid prescriptione	2.18***	2.07–2.30	2.19***	2.05–2.34

• Note: The models also include state-fixed effects, year-fixed effects, and quarter dummy variables to account for seasonal effects, but these results are not shown in the table. Asterisks indicate statistical significance (*p < .10, **p < .05, ***p < .01).
• ^a Medical marijuana dispensary (legal and active) means that medical marijuana dispensaries are legally allowed to operate and known to be operating in the states.
• ^b Index opioid prescription was defined as filling at least one opioid prescription during the time 30 days before admission and 2 weeks after discharge.
• ^c The form of index opioid is either elixir or solution.
• ^d We identified pre-existing behavioral health conditions by using International Classification of Diseases, 9th Revision (ICD-9) codes and included the conditions that were diagnosed during the 6 months preceding the hospital admission.
• ^e Previous opioid prescription was defined as filling at least one opioid prescription during the period 30–180 days before admission.

Being female (aOR, 1.27; 95% CI, 1.21–1.33), having a longer hospital stay (measured in days) (aOR, 1.04; 95% CI, 1.02–1.06) and greater days of index opioid supply (8–14 days: aOR, 1.39, 95% CI, 1.33–1.45; greater than 14 days: aOR, 2.42, 95% CI, 2.26–2.59) were associated with a higher likelihood of a prolonged opioid use. Patients who had elixir or solution forms of opioids for their index opioid prescription (aOR, 0.81; 95% CI, 0.74–0.88) or received inpatient surgery (aOR, 0.91; 95% CI, 0.84–0.99) were less likely to have prolonged opioid use (Table 3).

Among the 13 procedures, cholecystectomy (aOR, 1.16; 95% CI, 1.08–1.25) was associated with higher odds of prolonged opioid use. Tonsillectomy and/or adenoidectomy (aOR, 0.79; 95% CI, 0.71–0.88), pectus repair (aOR, 0.64; 95% CI, 0.48–0.86), epicondylar fracture fixation (aOR, 0.57; 95% CI, 0.34–0.97), posterior arthrodesis (aOR, 0.72; 95% CI, 0.58–0.91), arthroscopic knee repair (aOR, 0.71; 95% CI, 0.64–0.79), and orchiopexy (aOR, 0.64; 95% CI, 0.47–0.87) were associated with a lower likelihood of prolonged opioid use (Table 3). Rural residence and pre-existing diagnosis of depressive disorder, anxiety disorder, substance use disorder, and other mental health disorders were all associated with higher odds of prolonged opioid use. Furthermore, adolescents and young adults with a previous opioid prescription 30–180 days before admission were more likely to have prolonged opioid use (aOR, 2.18; 95% CI, 2.07–2.30) (Table 3).

We reran the model several ways to assess the robustness of our initial null finding regarding living in a state that allows medical marijuana dispensaries. The top row of Table 4 reports our original findings from just the medical marijuana dispensary variable from the two separate periods reported in Table 3. In the second row of Table 4, we show the results from rerunning the model excluding the pre-existing behavioral health diagnoses to see if they might generate a spurious correlation with this policy variable. We found there was very little impact on the medical marijuana dispensary variable when these variables were excluded. We then examined the extent to which the pre-policy time interaction affected the results by estimating the model by excluding the pre-policy time interaction from the specification. The adjusted odds ratios on the medical marijuana dispensary variable in row 3 of Table 4 are similar to our original findings. In row 4, we removed both the pre-policy time interaction as well as the behavioral health diagnoses from the model. Again, we found the coefficient on the medical marijuana dispensary variable remained similar to the results of our original model. Lastly, we tested the effects of any medical marijuana laws rather than the medical marijuana dispensaries. The last row of Table 4 shows that medical marijuana laws had no significant association with prolonged opioid use in both time periods (year 2005–2014: aOR, 1.00, 95% CI, 0.91–1.10; year 2010–2014: aOR, 1.00, 95% CI, 0.82–1.21).Table 4. Sensitivity analyses to check robustness of null finding for legally protected and operational medical marijuana dispensary.

		Year 2005–2014		Year 2010–2014
		aOR	95% CI	aOR	95% CI
Original model	Medical Marijuana Dispensary (MMD)a	0.98	0.81–1.18	0.84	0.62–1.14
Exclude Pre-existing Behavioral Health	MMD	0.99	0.82–1.19	0.86	0.63–1.18
Exclude Pre-policy Time Trend	MMD	0.93	0.82–1.07	0.86	0.66–1.12
Exclude Pre-existing Behavioral Health Diagnoses & Pre-policy Time Trend	MMD	0.94	0.82–1.07	0.89	0.67–1.17
Effects of Any Medical Marijuana Laws rather than MMD	Medical Marijuana Laws	1.00	0.91–1.10	1.00	0.82–1.21

• Abbreviations: aOR, adjusted odds ratio; CI, confidence interval.
• ^a Medical Marijuana Dispensary (legal and active) means that medical marijuana dispensaries are legally allowed to operate and known to be operating in the states.

Our study found no compelling evidence that medical marijuana dispensaries reduced prolonged opioid use, and these results were robust to a variety of empirical specifications and study periods. This is consistent with the results of a study by Segura et al.³⁵ who found no significant association of medical marijuana laws with nonmedical prescription opioid use and prescription opioid use disorder in adolescents (12–17 years old) and young adults (18–25 years old) using National Survey on Drug Use and Health (NSDUH) data from 2004 to 2014, a similar study period. On the other hand, our finding stands in contrast to prior works suggesting that liberalization of marijuana laws is effective at reducing opioid prescribing.^18–23

DISCUSSION

There are several differences between our study and previous studies. In contrast to prior studies of Medicaid or Medicare enrollees or adult populations,^18–23 our sample consists of commercially insured adolescents and young adults who underwent one of 13 surgical procedures. We also examined prolonged opioid medication use, in contrast to previous works which looked at overall opioid prescribing, such as number of opioid prescriptions, daily opioid doses prescribed, and spending on opioid prescriptions.^18–23 Furthermore, the outcome measures of the previous studies are mostly state-level measures, such as average number of opioid prescriptions in states with medical marijuana laws versus without.^18–22 However, we used individual-level claims data, so our analysis was conducted at the individual level. Third, our work has a longer study period than existing studies, which have study periods of 2010–2013, 2010–2015, or 2011–2016.^18, 19, 22 Many of the medical marijuana dispensaries opened around 2010 and 2011, which provided previous studies little opportunity to identify any pre-period trends among many of the states that opened medical marijuana dispensaries during the study period. Of the 10 states that opened medical marijuana dispensaries in 2010–2014, three of them did so in 2010 or 2011. Therefore, an important strength of our work is the long study period, allowing us to determine pre-period trends that are fundamental to the determination of the counterfactual. It also provides an opportunity to examine the parallel trends assumption that is required by the difference-in-differences model.

Regarding the medical marijuana dispensary variable, our results call into question some of the results found previously by other researchers who analyzed the association between medical marijuana legalization and prescription opioids using data from 2010.^18, 19, 22 They found medical marijuana legalization was associated with reductions in opioid prescriptions, including daily opioid doses prescribed and opioid prescribing rates.^18, 19, 21 We re-estimated our model limiting the study period to 2010–2014 and produced results similar to those of the literature, showing the negative association between having access to a medical marijuana dispensary and prolonged opioid medication use (aOR, 0.84) (Table 3). When using the full study period (2005–2014), not only did we find no evidence of a protective medical marijuana dispensary result, but as shown in Table 3, we also found that most of the other results remained unchanged. Our findings reinforce the importance of studies using difference-in-differences approaches having a pre-intervention period of sufficient duration to correctly identify pre-intervention trends.

Limitations

The study does suffer from relevant limitations. First, we defined prolonged opioid use based on information available in claims data. It is possible that some patients obtained opioids outside of the medical system, and this would not be captured. The exclusion of opioids from other sources could lead to over-estimating the likelihood of dispensaries reducing opioid use. In addition, it is unknown whether patients actually took the prescribed opioids. Second, the study is focused on commercially insured adolescents and young adults who received opioids in response to a surgical procedure, and we do not know if our findings generalize to populations receiving opioids unrelated to surgery or noncommercially insured adolescents and young adults. Finally, difference-in-differences and CITS models rely heavily on assumptions about the treatment group counterfactual. Interpretation of these results should be made in the context of these assumptions.

CONCLUSION

This study provides novel evidence from a sample of adolescents and young adults who underwent one of 13 surgical procedures that legal access to medical cannabis did not lead to a reduction in prolonged opioid use within this population, contrary to what has been found for some adult populations.³⁶ While recent evidence suggests that youth and young adults (ages 16–25) are less likely to report use of cannabis for medicinal purposes than adults between the ages of 26–55, 17% of youth and young adults living in states that prohibit any use report using cannabis for medicinal reasons while 19% of those living in states allowing medical use report such use, suggesting this is not that uncommon among this age group.³⁷

These findings are the first to demonstrate potentially important age differences in sustained use of opioids. Prolonged opioid use after surgery might lead to prescription opioid misuse, so it would be especially important to prevent long-term exposure to opioids and prolonged opioid use after surgeries in this population. Therefore, opioid prescribers should be more cautious when prescribing opioids to adolescents and young adults, and prescriber oversight and management with this vulnerable population would be necessary.

ACKNOWLEDGMENTS

This study was supported by a grant from the National Institute on Drug Abuse (NIDA), R01DA047396 (PI: Leslie).

Research Article –

Kyungha Kim DrPH, MHA, Rosalie L. Pacula PhD, Andrew W. Dick PhD, Bradley D. Stein MD, PhD, Benjamin G. Druss MD, MPH, Edeanya Agbese MPH, Austin C. Cohrs MPH, Douglas L. Leslie PhD

First published: 08 June 2023

https://doi.org/10.1111/ajad.13440