
==== Front
Subst Abuse Rehabil
Subst Abuse Rehabil
sar
Substance Abuse and Rehabilitation
1179-8467
Dove

482717
10.2147/SAR.S482717
Original Research
Incidence, Timing and Social Correlates of the Development of Opioid Use Disorder Among Clients Seeking Treatment for an Alcohol Use Problem: Changes Over the Three Waves of the Opioid Epidemic
Falls et al
Falls et al
http://orcid.org/0000-0003-4116-0441
Falls Zackary 1 *
Zhang Xueqing 2 *
http://orcid.org/0000-0001-9616-6811
Elkin Peter L 1
Jacobs David 3
http://orcid.org/0000-0001-7918-9317
Bednarczyk Edward M 4
Gibson Walter 4
Jette Gail P 5
http://orcid.org/0000-0001-6658-1264
Leonard Kenneth E 6 *
1 Department of Biomedical Informatics and Clinical and Research Institute on Addiction, University at Buffalo, The State University of New York, Buffalo, New York, USA
2 Department of Biostatistics and Clinical and Research Institute on Addiction, University at Buffalo, The State University of New York, Buffalo, New York, USA
3 School of Pharmacy and Department of Epidemiology and Environmental Health, University at Buffalo, the State University of New York, Buffalo, New York, USA
4 School of Pharmacy, University at Buffalo, The State University of New York, Buffalo, New York, USA
5 Division of Outcomes, Management, and Systems Information, Office of Addiction Services and Supports, Albany NY, United States
6 Department of Psychiatry and Clinical and Research Institute on Addiction, University at Buffalo, The State University of New York, Buffalo, New York, USA
Correspondence: Kenneth E Leonard, Department of Psychiatry and Clinical and Research Institute on Addictions, University at Buffalo, 1021 Main Street, Buffalo, NY, 14203, USA, Email Kleonard@buffalo.edu
* These authors contributed equally to this work

19 9 2024
2024
15 185195
14 6 2024
27 8 2024
© 2024 Falls et al.
2024
Falls et al.
https://creativecommons.org/licenses/by-nc/3.0/ This work is published and licensed by Dove Medical Press Limited. The full terms of this license are available at https://www.dovepress.com/terms.php and incorporate the Creative Commons Attribution – Non Commercial (unported, v3.0) License (http://creativecommons.org/licenses/by-nc/3.0/). By accessing the work you hereby accept the Terms. Non-commercial uses of the work are permitted without any further permission from Dove Medical Press Limited, provided the work is properly attributed. For permission for commercial use of this work, please see paragraphs 4.2 and 5 of our Terms (https://www.dovepress.com/terms.php).
Introduction

Opioid use disorder (OUD) and opioid overdose (OD) have shown to be strongly associated with alcohol use disorder (AUD). As a potential target population for secondary prevention, we examined the incidence and timing of OUD/OD among clients seeking treatment for alcohol problems and how this has changed over the three waves of the opioid epidemic corresponding to the primary opioid involved in fatal ODs, prescription painkillers (2007–2009), heroin (2010–2012), and fentanyl (2013–2016). We also examined social determinants of health as predictors of OUD/OD.

Methods

Clients (N = 59,186) presenting for a first treatment for alcohol use problems were extracted from the Client Data System (CDS) of the New York State Office of Addiction Services and Support (OASAS) and New York State (NYS) Medicaid Data Warehouse. Using this cohort, we employed the Kaplan–Meier method to determine the survival probabilities for patients admitted in each of the three waves of the epidemic.

Results

Patients in Cohort 3 (2013–2016) were diagnosed with OUD/OD more rapidly than patients in Cohort 1 (2007–2009) or Cohort 2 (2010–2012), although the overall estimated OUD/OD rate was comparable across the three cohorts.

Discussion

These findings provide a useful estimate of the incidence and the expected time frame of an opioid use disorder in clients with an alcohol use problem. Moreover, it suggests that as the opioid epidemic progressed, OUD/OD developed more rapidly but the overall prevalence did not increase.

Keywords

opioid use disorder
opioid overdose
substance use
alcohol
incidence
==== Body
pmcIntroduction

While deaths and emergency room admissions due to opioid ODs have dramatically increased over the past twenty years, there have also been substantial increases in ODs involving alcohol. Analysis of the Drug Abuse Warning Data found that emergency room visits for alcohol-involved drug OD visits doubled between 2005 and 2011.1 Nearly 60% of these cases included the use of central nervous system medications, with approximately half of these being opioid or benzodiazepine medications. Jones, Paulozzi, and Mack2 found that nearly 20% of opioid medication ODs and nearly 30% of benzodiazepine ODs involved alcohol. Similarly, 22% of the opiate OD deaths and 21% of the benzodiazepine deaths involved alcohol. More recently, research has shown that from 2000 through 2019, deaths from alcohol poisonings doubled, and deaths from combined alcohol and opioid ODs more than quadrupled.3 By 2017, alcohol was a contributing factor in approximately 15% of all opioid OD deaths.4 While these findings pertain to acute alcohol and opioid use, they suggest a potential relationship between alcohol use disorder (AUD) and opioid use disorder (OUD/OD).

Research has also shown that both hazardous alcohol consumption and AUD are associated with the non-medical use of prescription opioids among college students.5,6 Similarly, Schepis and McCabe7 found that AUD was related to opioid misuse in a sample of adults aged 50 and above. Schepis and Hake8, using the National Epidemiological Survey of Alcohol and Related Conditions (NESARC), reported that participants with a previous AUD were more likely to have an OUD/OD than those without an AUD. Moreover, Weitzman and Ong9 found that the overlap between AUD and OUD/OD grew substantially between 2008 and 2015 and argued for research addressing the prospective relationship between these two disorders.

Given its predictive value, individuals with an AUD represent a potentially important population for preventing the development of OUD/OD. However, the value and nature of any intervention to reduce the likelihood of an OUD/OD may depend on the temporal aspects of the progression—that is, whether an individual identified with an AUD is likely to develop an OUD/OD rapidly or more gradually. The unfolding of the opioid epidemic has been characterized as three waves corresponding to the primary opioid involved in fatal ODs, prescription opioids (2007–2009), heroin (2010–2012), and fentanyl (2013–2016).10 From the year of our first full cohort, 2007, until 2009, the opioid overdose epidemic was largely an issue of prescription opioids. Beginning in 2010, there was a growth in heroin ODs which continued to increase until approximately 2016. However, beginning in 2013, there was a rapid increase in OD deaths due to fentanyl which continued in 2016 and increased dramatically after 2016. Unfortunately, there are few longitudinal studies of the development of OUD/OD that specifically examine the role of AUD nor examine how this relationship may have changed over the three waves of the epidemic.

One of the important changes in the development of OUD/OD over the three waves of the opioid epidemic has been the relationship between OUD/OD and the social determinants of health. For example, early in the epidemic, fatal ODs were more likely in the white non-Hispanic population and increased rapidly from 1999 through 2012 in this group.11 Beginning in 2013, the fatal OD rates among the African-American population began to increase, with large increases due to heroin and fentanyl and increases due to prescription opioids in 2016.11,12 Rapid increases were also observed among Hispanic individuals.13 In addition to racial/ethnic disparities, there were substantial disparities with respect to education14 and gender.12 Finally, although the early wave of prescription opioid ODs was more pronounced in rural areas, the subsequent heroin and fentanyl waves were stronger in urban areas.15

The purpose of this study was to estimate the prevalence and timing of OUD or OD among clients seeking treatment for an alcohol use problem over a ten-year period (2007–2016) and to determine whether the prevalence changed over the three waves of the opioid epidemic. In addition, we examine whether social determinants of health were related to the occurrence of an OUD/OD and whether these changed over the waves. This was accomplished by utilizing a large New York State (NYS) database of clients seeking help for the first time for an alcohol use problem and linking these data to Medicaid data for subsequent OUD/OD diagnoses.

Materials and Methods

Study Sample

Because this study involves de-identified secondary data analyses, it was ruled exempt by the University at Buffalo Institutional Review Board and complies with the Declaration of Helsinki. Data for this study was drawn from the Client Data System (CDS) of the NYS Office of Addiction Services and Supports (OASAS) and NYS Medicaid Data Warehouse (MDW). Clients with a first admission to an OASAS outpatient treatment program from 2007 through 2016 were included if they presented with a primary alcohol use problem (diagnoses are not recorded in the CDS) at the time of admission. Clients who reported any problem with heroin or other opioids at admission or at discharge were excluded, as were clients who were noted as receiving buprenorphine or methadone, or who were referred to an opioid treatment program after admission. OASAS staff matched and merged OASAS treatment data with the Medicaid data using a “unique client identifier” to anonymously track clients. Clients with no Medicaid records after the initial alcohol use admission were excluded as were clients with a diagnosis of OUD or OD in the Medicaid records preceding their initial alcohol admission. For this study, we only included clients who, according to the OASAS CDS, were seen in an outpatient clinic and who denied any previous treatment to provide an estimate of the risk of OUD/OD among patients seen for AUD in the earliest stages possible given these data. The final cohort selection is depicted in Figure 1. Figure 1 Inclusion/exclusion criteria for cohort selection.

There were 59,186 eligible cases with a primary alcohol use problem with matched Medicaid data. As seen in Table 1, the sample was predominantly male (68%). Approximately one-half (49%) of the sample was White Non-Hispanic, one-quarter (24%) were Black Non-Hispanic, and one-fifth (22%) were Hispanic. Nearly one-half (44%) were 30 years old or younger at the time of the first admission, and about one-third were between 31 and 45 years of age (34%). Only 45% of the sample were employed, and 16% were unemployed. Over 30% (32%) of the sample were not in the workforce (eg, student, in-training, taking care of children, retired). The majority had never been married (61%) or were separated or divorced (18%). Although there were changes in these variables across the three waves, these changes were small.Table 1 Distribution of Social Demographic Factors in Each of the Three Cohorts and Overall

Variable	Value	Percent in
Cohort 1	Percent in
Cohort 2	Percent in
Cohort 3	Total
percent	
		n=19,293	n=18,076	n=21,817	N=59,186	
Gender	F	30%	32%	33%	32%	
M	70%	68%	67%	68%	
Age	18–21	14%	12%	7%	11%	
22–25	16%	17%	16%	16%	
26–30	16%	17%	19%	17%	
31–35	12%	13%	14%	13%	
36–40	12%	11%	11%	11%	
41–45	11%	10%	10%	10%	
46–50	9%	9%	9%	9%	
51–55	5%	6%	6%	5%	
55+	6%	7%	9%	7%	
Race/ethnicity	White Non-Hispanic	52%	49%	45%	49%	
Black Non-Hispanic	22%	24%	24%	24%	
Other Non-Hispanic	5%	6%	7%	6%	
Hispanic	20%	21%	24%	22%	
Education	Less than 8th Grade	6%	6%	5%	6%	
9–11 grade/No Diploma	23%	22%	20%	22%	
HS/GED/Voc with Diploma	39%	38%	37%	38%	
Some College - No Degree	20%	20%	20%	20%	
Associates Degree	5%	6%	7%	6%	
Bachelors Degree	6%	7%	9%	7%	
Graduate Degree	2%	2%	3%	2%	
Employment	Employed	46%	41%	48%	45%	
Unemployed	9%	9%	10%	9%	
Unemployed, In-Treatment	8%	8%	6%	7%	
Disabled	6%	6%	6%	6%	
Not in Labor Force	30%	35%	30%	32%	
Income source	Wages/Salary	44%	39%	46%	43%	
Public Assistance/Family	41%	45%	42%	43%	
None	15%	16%	12%	14%	
Residence	Independent	90%	90%	91%	90%	
Dependent	4%	4%	3%	4%	
Homeless	4%	4%	4%	4%	
Other	2%	2%	2%	2%	
Marital status	Never Married	60%	61%	62%	61%	
Married	16%	15%	16%	16%	
Living as Married	4%	4%	4%	4%	
Widowed	1%	1%	2%	2%	
Separated	7%	7%	6%	7%	
Divorced	11%	11%	11%	11%	
Notes: HS/GED/Voc with Diploma: High school graduate, General Educational Development (High school equivalency tests), Vocational education with a diploma.

Measures

OASAS Client Data System

The CDS collects admissions and discharges for NYS residents seeking addiction services provided through OASAS certified treatment programs. The CDS data are client self-reports and treatment providers usually provide the information through batch submission. To ensure reliability and validity, business rules that cross-check data items are utilized to minimize inconsistencies and are placed throughout the CDS to minimize data entry errors. Treatment providers are routinely audited to ensure they meet OASAS quality standards. The CDS provides substance use data elements for the Substance Abuse and Mental Health Services Administration (SAMHSA) Treatment Episode Data Set (TEDS) and is also used to support National Outcome Measures (NOMS).

At admission, the CDS collects information about demographic variables such as age, gender, and social determinants of health including racial/ethnic identity, employment, primary income source, current residence (independent, dependent, homeless and other) and marital status.

Medicaid Claims

New York’s Medicaid program is one of the largest insurance programs in the nation, providing health coverage to over seven million people, approximately 4.4 million of whom receive their health care through enrollment in a managed care plan. The NYS MDW includes individual-level NYS Medicaid enrollment and Medicaid healthcare utilization claims. The MDW is a comprehensive and secure healthcare information system environment. Data are stored in an Oracle database and include 15 years (March 2005 –2020) of Medicaid claims, which are updated weekly. In the present study, we used the MDW to identify the first occurrence and date of a diagnosis of an OUD or of an opioid OD.16 The specific codes for these can be found in Table S1.

Data Analysis

We conducted a series of preliminary analyses to examine the prevalence of and relationship between OD and OUD across the entire cohort. Kaplan–Meier survival curves were plotted for the three opioid epidemic cohorts, and the Log rank test was performed. To analyze the associations between the diagnosis of OUD/OD and sociodemographic factors, we first employed models that included the cohort variable and each sociodemographic factor individually. Covariates that showed a statistically significant association with the diagnosis of OUD/OD were then included in a multivariable logistic regression model, from which odds ratios were derived. The main effects were examined in a multivariate model including all social determinants as well as cohort. Additionally, interaction effects were assessed in a multivariate model comprising the cohort, all the social determinant variables, and their interactions with the cohort to ascertain potential relationship changes over time. All analyses were performed using R 3.6, with a two-sided p-value of ≤0.05 considered statistically significant.

Results

From the preliminary analyses, the prevalence of OD and OUD were 1.4% and 17.0% over the entire observation period. In addition, nearly everyone who experienced an OD also had a diagnosis of OUD (89.5%), although there were many clients with an OUD who did not experience an OD (92.5%). This led us to focus our analyses on the development of an OUD or an opioid OD, which we will refer to as OUD/OD.

Given the escalation in opioid ODs over the years of this study, we examined the prevalence and time to develop OUD/ODs over the time period. From the year of our first full cohort, 2007, until 2009, the opioid OD epidemic was largely an issue of prescription opioids. Our second cohort was defined as beginning in 2010 when there was a growth in heroin ODs until 2012, although deaths from heroin continued to increase after 2012. The final cohort was defined as beginning in 2013 because of the rapid increase in OD deaths due to fentanyl which began then and continued throughout 2016.

We utilized the Kaplan–Meier method to generate a cumulative hazard plot illustrating the time (in months) to an incident diagnosis of OUD/OD across different OD waves, as displayed in Figure 2. As can be seen in this figure, patients in the most recent cohort received an OUD/OD diagnosis much more rapidly than clients in the earlier cohorts (Chi-Square for Log Rank Test = 375, df = 2, p < 0.001). Of note, even though clients in the most recent cohort had been observed for a much shorter period, the overall prevalence rates were very similar with 18.2%, 17.3% and 16.1% for Cohorts 1, 2 and 3, respectively. Figure 2 Cumulative proportion of three cohorts receiving OUD/OD diagnosis after AUD treatment.

As can be seen in Figure 2, there was a substantial difference in the rates by year 4. Given this substantial difference and the fact that clients seen in 2016 had approximately 4 years of follow-up information, we examined the predictors of receiving an OUD/OD diagnosis within 4 years of initial admission using logistic regression. The main effects were examined in a multivariate model including all social determinants as well as cohort. Subsequently, interaction effects were assessed in a multivariate model comprising the cohort, all the social determinant variables, and their interactions with the cohort to ascertain potential relationship changes over time.

The results of the main effect analysis appear in Table 2. Every social determinant was associated with OUD/OD in the multivariate analysis. Males were at a lower risk of developing OUD/OD than females with an odds ratio of 0.78 compared to females, indicating that the odds of developing OUD/OD are 22% lower in males than in females. Notably, individuals who were 22–25 years old when admitted were at a greater risk for OUD/OD than those who were 18–21 years old, although the magnitude of this increased risk was only about 11%. Beginning at age 36, clients who were older at their first admission were at a much reduced risk of OUD/OD, approximately 20% lower with an even lower risk for those age 55 and older (47%) when they were initially admitted. White non-Hispanic individuals had the highest risk of developing OUD/OD, significantly higher than Black non-Hispanic, Hispanic and other non-Hispanic groups who had a 27–35% lower risk. There was not a significant difference in risk between those with less than an 8th-grade education and those with more education, except for those with a bachelor’s or graduate degree. Individuals holding bachelor’s or graduate degrees exhibit odds ratios of 0.63 and 0.68, respectively, implying a significant 37% and 32% lower odds of OUD/OD compared to those with less than an 8th-grade education.Table 2 Main Effect Predictors of OUD/OD Using Multivariate Logistic Regression Model

Variable	Comparison	Odds Ratio	
Gender	Male vs Female	0.78 ***	
Age	22–25 vs 18–21	1.11*	
26–30 vs 18–21	1.06	
31–35 vs 18–21	1.01	
36–40 vs 18–21	0.85**	
41–45 vs 18–21	0.80***	
46–50 vs 18–21	0.79***	
51–55 vs 18–21	0.79**	
55+ vs 18–21	0.53***	
Race	Black Non-Hispanic vs White Non-Hispanic	0.73***	
Other Non-Hispanic vs White Non-Hispanic	0.65***	
Hispanic vs White Non-Hispanic	0.72***	
Education	9–11/No Diploma vs Less than 8th Grade	1.07	
HS/GED/Vocational with Dipl vs Less than 8th Grade	0.98	
Some College-No degree vs Less than 8th Grade	0.93	
Associates Degree vs Less than 8th Grade	0.86	
Bachelors Degree vs Less than 8th Grade	0.63***	
Graduate Degree vs Less than 8th Grade	0.68**	
Employment	Unemployed vs Employed	1.68***	
Unemployed, In-Treatment vs Employed	1.64***	
Disabled vs Employed	1.45***	
Not in Labor Force vs Employed	1.32***	
Income source	Public Assistance/Family vs Wages/Salary	1.23***	
None vs Wages/Salary	1.19*	
Region	Eastern vs Central	1.12	
Hudson Valley vs Central	1.17*	
New York City vs Central	1.38***	
North Country vs Central	1.03	
Western vs Central	1.18**	
Residence	Dependent vs Independent	1.26***	
Homeless vs Independent	1.66***	
Other vs Independent	1.28**	
Notes: Significance: ***p < 0.001, **p < 0.01, *p < 0.05.

For indicators of economic difficulties, all non-wage/salary employment statuses, including unemployment, disability, and not being in the labor force, show a substantial increase in odds of OUD/OD, ranging from 32% to 68%, compared to employed individuals. Likewise, individuals dependent on public assistance/family support or without any income source face odds ratios of 1.23 and 1.19, respectively, suggesting a higher risk of OUD/OD by 23% and 19%, respectively. Finally, individuals living independently demonstrated odds ratios ranging from 1.26 to 1.66, indicating that those living dependently, homeless or in another living arrangement were 26% to 66% more likely to have an OUD/OD diagnosis. Those who indicated they were homeless at admission had a substantially higher risk (OR = 1.66) than those who were living in a dependent living situation or any other situation.

Geographically, residents of New York City, the Western region, and the Hudson Valley exhibited odds ratios of 1.38, 1.18 and 1.17, respectively, indicating 38%, 18% and 17% higher risk OUD/OD compared to the Central NYS region.

The analyses of interactions indicated that the relationship between OUD/OD with race/ethnicity, education, income source and residence did not differ across the three cohorts. However, there were significant changes for the relationship of OUD/OD with gender, age, employment and region across the three cohorts (see Table 3). The rate of OUD/OD for each of these variables for each level of the social factors is presented in Table 4. As can be seen in the table, the rate of OUD/OD among men increased slightly more than the rate among women across cohorts. The rate of OUD/OD nearly tripled among those aged 51 to 55, while the rates for other groups roughly doubled from Cohort 1 to Cohort 3. The most notable aspect of the age by cohort effect was that there were notable differences among the age groups among Cohort 1 with differences between the younger and older cohorts ranging from 2% to 4%. In contrast, the differences between the younger and older age groups in cohort 3 were about 1% to 2%.Table 3 Odds Ratios for Interactions of Cohort and Social Determinants Predicting OUD/OD

Variable	Comparison	Odds Ratio	
Gender	Male vs Female within cohort 2 compared to cohort 1	1.07	
Male vs Female within cohort 3 compared to cohort 1	1.18*	
Age	22–25 vs 18–21 within cohort 2 compared to cohort 1	1.03	
26–30 vs 18–21 within cohort 2 compared to cohort 1	1.26.	
31–35 vs 18–21 within cohort 2 compared to cohort 1	1.18	
36–40 vs 18–21 within cohort 2 compared to cohort 1	1.02	
41–45 vs 18–21 within cohort 2 compared to cohort 1	1.18	
46–50 vs 18–21 within cohort 2 compared to cohort 1	1.03	
51–55 vs 18–21 within cohort 2 compared to cohort 1	1.36	
55+ vs 18–21 within cohort 2 compared to cohort 1	1.18	
22–25 vs 18–21 within cohort 3 compared to cohort 1	1.09	
26–30 vs 18–21 within cohort 3 compared to cohort 1	1.32*	
31–35 vs 18–21 within cohort 3 compared to cohort 1	1.41*	
36–40 vs 18–21 within cohort 3 compared to cohort 1	1.39*	
41–45 vs 18–21 within cohort 3 compared to cohort 1	1.28.	
46–50 vs 18–21 within cohort 3 compared to cohort 1	1.38*	
51–55 vs 18–21 within cohort 3 compared to cohort 1	1.75**	
55+ vs 18–21 within cohort 3 compared to cohort 1	1.67**	
Employment	Unemployed vs Employed within cohort 2 compared to cohort 1	0.88	
Unemployed, In-Treatment vs Employed within cohort 2 compared to cohort 1	0.81	
Disabled vs Employed within cohort 2 compared to cohort 1	0.69	
Not in Labor Force vs Employed within cohort 2 compared to cohort 1	0.81	
Unemployed vs Employed within cohort 3 compared to cohort 1	0.87	
Unemployed, In-Treatment vs Employed within cohort 3 compared to cohort 1	0.80	
Disabled vs Employed within cohort 3 compared to cohort 1	0.66*	
Not in Labor Force vs Employed within cohort 3 compared to cohort 1	0.91	
Region	Eastern vs Central within cohort 2 compared to cohort 1	1.13	
Hudson Valley vs Central within cohort 2 compared to cohort 1	0.80	
New York City vs Central within cohort 2 compared to cohort 1	0.74*	
North Country vs Central within cohort 2 compared to cohort 1	1.35	
Western vs Central within cohort 2 compared to cohort 1	0.97	
Eastern vs Central within cohort 3 compared to cohort 1	1.65**	
Hudson Valley vs Central within cohort 3 compared to cohort 1	1.52**	
New York City vs Central within cohort 3 compared to cohort 1	1.03	
North Country vs Central within cohort 3 compared to cohort 1	1.42.	
Western vs Central within cohort 3 compared to cohort 1	1.04	
Notes: BOLD indicates significant comparisons. Significance: **p < 0.01, *p < 0.05, p < 0.10.

Table 4 Patient Counts (Percentage) of Indicators of OUD/OD Within 4-Year Follow-Up in Each Level of Control Variables for Each Cohort Category

Variable	Value	Cohort 1	Cohort 2	Cohort 3	
No OUD/OD	OUD/OD	No OUD/OD	OUD/OD	No OUD/OD	OUD/OD	
Gender	F	5204 (90%)	570 (10%)	5112 (87%)	764 (13%)	5975 (83%)	1212 (17%)	
M	12595 (93%)	924 (7%)	11,021 (90%)	1178 (10%)	12,516 (86%)	2039 (14%)	
Age	18–21	2360 (91%)	242 (9%)	1869 (88%)	251 (12%)	1362 (85%)	238 (15%)	
22–25	2857 (91%)	296 (9%)	2690 (88%)	378 (12%)	2953 (84%)	562 (16%)	
26–30	2858 (92%)	247 (8%)	2630 (88%)	376 (12%)	3501 (84%)	662 (16%)	
31–35	2110 (92%)	174 (8%)	2014 (89%)	257 (11%)	2528 (84%)	493 (16%)	
36–40	2075 (93%)	157 (7%)	1743 (91%)	173 (9%)	1994 (85%)	342 (15%)	
41–45	1968 (93%)	145 (7%)	1580 (90%)	173 (10%)	1822 (87%)	281 (13%)	
46–50	1619 (93%)	125 (7%)	1485 (91%)	148 (9%)	1580 (85%)	274 (15%)	
51–55	895 (94%)	54 (6%)	935 (90%)	103 (10%)	1026 (85%)	186 (15%)	
55+	1057 (95%)	54 (5%)	1188 (94%)	83 (6%)	1784 (89%)	229 (11%)	
Race/ethnicity	White Non-Hispanic	9295 (92%)	817 (8%)	7786 (89%)	998 (11%)	8350 (85%)	1509 (15%)	
Black Non-Hispanic	3978 (92%)	343 (8%)	3904 (89%)	479 (11%)	4366 (84%)	845 (16%)	
Other Non-Hispanic	957 (94%)	59 (6%)	1052 (93%)	83 (7%)	1418 (88%)	192 (12%)	
Hispanic	3569 (93%)	275 (7%)	3392 (90%)	382 (10%)	4416 (86%)	721 (14%)	
Education	Less than 8th Grade	1090 (94%)	75 (6%)	917 (90%)	108 (10%)	987 (86%)	157 (14%)	
9–11/No Dipl	4070 (91%)	421 (9%)	3507 (88%)	492 (12%)	3521 (83%)	740 (17%)	
HS/GED/Voc with Dipl	6854 (92%)	579 (8%)	6069 (89%)	737 (11%)	6879 (85%)	1257 (15%)	
Some College - No Degree	3463 (92%)	298 (8%)	3238 (90%)	379 (10%)	3712 (85%)	641 (15%)	
Associates Degree	951 (95%)	50 (5%)	934 (89%)	112 (11%)	1290 (86%)	213 (14%)	
Bachelors Degree	1095 (95%)	55 (5%)	1173 (93%)	91 (7%)	1667 (89%)	198 (11%)	
Graduate Degree	276 (95%)	16 (5%)	296 (93%)	23 (7%)	494 (89%)	61 (11%)	
Employment	Employed	8414 (95%)	479 (5%)	6828 (92%)	582 (8%)	9224 (88%)	1222 (12%)	
Unemployed	1570 (89%)	199 (11%)	1419 (84%)	267 (16%)	1693 (79%)	439 (21%)	
Unemployed, In-Treatment	1359 (89%)	177 (11%)	1249 (85%)	215 (15%)	1009 (80%)	249 (20%)	
Disabled	1118 (90%)	126 (10%)	1017 (88%)	139 (12%)	1166 (84%)	218 (16%)	
Not in Labor Force	5338 (91%)	513 (9%)	5621 (88%)	739 (12%)	5458 (83%)	1139 (17%)	
Income source	Wages/Salary	8011 (95%)	452 (5%)	6546 (92%)	547 (8%)	8858 (88%)	1174 (12%)	
Public Assistance/Family	7179 (90%)	778 (10%)	7143 (87%)	1035 (13%)	7528 (82%)	1631 (18%)	
None	2609 (91%)	264 (9%)	2445 (87%)	360 (13%)	2164 (82%)	462 (18%)	
Region	Central	1988 (93%)	153 (7%)	1685 (89%)	209 (11%)	1961 (88%)	275 (12%)	
Eastern	1279 (94%)	83 (6%)	1064 (89%)	126 (11%)	1204 (84%)	228 (16%)	
Hudson Valley	1884 (93%)	135 (7%)	1670 (91%)	157 (9%)	1939 (84%)	370 (16%)	
New York City	7989 (92%)	736 (8%)	7745 (89%)	917 (11%)	9016 (84%)	1662 (16%)	
North Country	849 (94%)	54 (6%)	738 (88%)	98 (12%)	768 (86%)	121 (14%)	
Western	3773 (92%)	333 (8%)	3196 (88%)	435 (12%)	3615 (86%)	604 (14%)	
Residence	Independent	16156 (93%)	1272 (7%)	14,580 (90%)	1659 (10%)	16,949 (86%)	2834 (14%)	
Dependent	763 (90%)	85 (10%)	583 (86%)	93 (14%)	608 (81%)	142 (19%)	
Homeless	615 (85%)	106 (15%)	620 (82%)	139 (18%)	722 (76%)	226 (24%)	
Other	265 (90%)	31 (10%)	351 (87%)	51 (13%)	271 (81%)	65 (19%)	

Discussion

The present findings demonstrate that clients presenting for treatment of an alcohol use problem at the time of the fentanyl wave of the opioid epidemic had a more rapid progression to an OUD/OD than clients initially seen during the prescription pain and heroin waves of the epidemic. The rapid progression is obvious within the first year after treatment admission with nearly three times as many Cohort 3 clients (9.3%) being diagnosed with an OUD/OD compared to 3.5% and 5.9% of Cohort 1 and 2 clients. The more rapid development of OUD/OD continued with the largest difference observed at four years after admission.

The more rapid development of OUD/OD in Cohort 3 may not be unexpected given the well-documented increase in OD deaths17 and hospitalizations18 occurring during the follow-up observation period of Cohort 3, 2016 to 2020. However, it is important to note that while the development of OUD/OD was more rapid among those in Cohort 3, the data suggests that all three cohorts appeared to reach a comparable asymptote at approximately 18%. This may provide a useful estimate of the percentage of Medicaid clients with an alcohol use problem who are at risk for progression to an OUD/OD. Moreover, estimates of OUD/OD based on survey data suggest that the prevalence of OUD/OD was stable and declined from 2015 to 2019.19 This suggests that the overall prevalence of OUD/OD in this sample may not be increasing, although the lethality of OD continues to increase. Further follow-up of these cohorts would be necessary to establish this. In addition, the extent to which this would generalize to other alcohol treatment samples is unknown.

The social determinants of health were strongly related to the development of OUD/OD. Most of these determinants operated similarly to findings with respect to the opioid epidemic overall. For example, white non-Hispanic clients were at a higher risk than the other racial/ethnic categories, notwithstanding the more rapid increase of mortality among Hispanics.14 Similarly, clients with less education were at higher risk than clients with more education. Clients who were employed at the time of admission were at lower risk than all the other employment categories and those with an independent living status were at lower risk than those in a dependent (living with family) or homeless status. However, other factors were not associated with OUD/OD in a manner similar to the OD data. For example, female clients and younger clients at admission (18–35) were at higher risk for OUD/OD than were male clients and older clients (36 and older), respectively. However, opioid mortality is higher among men than women and those aged 35 to 44.20 As noted above, this may reflect the fact that the factors associated with OUD/OD and opioid mortality may be different. For example, women are more likely to use opioid analgesics, while men are more likely to use heroin, which could lead to higher prevalence of OUD/OD in women and higher mortality in men.21

Although there were several significant interactions that suggested that certain subgroups (males, older) increased more rapidly than their comparators, these differences were not large. Moreover, this was usually due to a group with a lower prevalence of OUD/OD among Cohort 1 having an increased prevalence among Cohort 3, which was numerically comparable to the comparators, but proportionally higher. As a result, the effects of the social determinants did not change appreciably from Cohort 1 to Cohort 3.

There are several limitations that should be considered. The identification of OUD/OD was reliant on the presence of an OUD/OD diagnosis in the Medicaid billing. Clients may have developed an OUD/OD but were not identified medically, or they may have transitioned from Medicaid. Although both probably occurred to some extent, the overall rate of OUD/OD was quite substantial, reaching approximately 18%, so it seems likely that most cases were captured by this method. It also seems unlikely that this would have substantially altered the findings with respect to cohort differences, although it might have influenced the relationships with the social factors. Second, while we have focused specifically on the progression to OUD/OD, it is likely that some in the sample developed polysubstance use problems, including the use of benzodiazepine. Because we were interested primarily in the relationship with the unfolding opioid epidemic, the consideration of these substances was beyond the scope of this paper. Third, this sample is an important one for the purposes of secondary prevention. Although the CDS did not provide AUD diagnoses, nearly 80% were diagnosed with AUD in the Medicaid data at a subsequent point. Given that they were treatment seeking, the estimates of OUD/OD may be different for individuals with an AUD who do not present for treatment, as well as for individuals who are not on Medicaid. Finally, there is evidence that we are now in a “fourth wave” insofar as fentanyl is being combined with cocaine, leading to deaths among individuals who perhaps had not used opioids. This may lead to the development of OUD/OD among an additional and new group of individuals who had not intended to use opioids. Moreover, the continued increase of fentanyl may result in a substantial increase in both ODs and OUDs. Further follow-up of this sample may be able to address this issue.

Conclusion

Clients presenting for treatment for alcohol use problems in the third wave of the opioid epidemic were identified with an OUD/OD much more rapidly than clients in the first and second waves. At one and four years after treatment, those in Cohort 3 had twice the prevalence of those in Cohort 1. Despite the more rapid onset of OUD/OD in Cohort 3, the overall prevalence for all three cohorts appears to converge on a prevalence of approximately 18%. Social determinants were associated with the development of OUD/OD in ways that reflected OUD/OD prevalence in other populations, but in ways that differed from opioid OD mortality. Despite minor changes in the relationships over time, the relationship between the social determinants and OUD/OD was basically comparable across cohorts.

Acknowledgments

This study was supported by grants from NIH NLM [T15LM012495]; NIAAA [R21AA026954 and R33AA0226954]; NIDA [K01DA056690], and NCATS [UL1TR001412]. This study was funded in part by the Department of Veterans Affairs. The authors acknowledge the New York State Office of Addiction Services and Supports for providing data used in this study.

Disclosure

Dr Zackary Falls reports grants from the National Institute on Drug Abuse, during the conduct of the study; is a Co-Founder and Board Member for Meditati, Inc., Mansarover, Inc., Mansarover Therapeutics, Inc., and AmritX, Inc., outside the submitted work. The authors report no conflicts of interest in this work.
==== Refs
References

1. Castle IJ, Dong C, Haughwout SP, White AM. Emergency department visits for adverse drug reactions involving alcohol: United States, 2005 to 2011. Alcohol Clin Exp Res. 2016;40 (9 ):1913–1925. doi:10.1111/acer.13167 27488763
2. Jones CM, Paulozzi LJ, Mack KA. Centers for disease C, prevention. Alcohol involvement in opioid pain reliever and benzodiazepine drug abuse-related emergency department visits and drug-related deaths - United States, 2010. MMWR Morb Mortal Wkly Rep. 2014;63 (40 ):881–885.25299603
3. Buckley C, Ye Y, Kerr WC, et al. Trends in mortality from alcohol, opioid, and combined alcohol and opioid poisonings by sex, educational attainment, and race and ethnicity for the United States 2000–2019. BMC Med. 2022;20 (1 ):405. doi:10.1186/s12916-022-02590-z 36280833
4. Tori ME, Larochelle MR, Naimi TS. Alcohol or benzodiazepine co-involvement with opioid overdose deaths in the United States, 1999-2017. JAMA network open. 2020;3 (4 ):e202361. doi:10.1001/jamanetworkopen.2020.2361 32271389
5. McCabe SE, Teter CJ, Boyd CJ. Illicit use of prescription pain medication among college students. Drug Alcohol Depend. 2005;77 (1 ):37–47. doi:10.1016/j.drugalcdep.2004.07.005 15607840
6. McCabe SE, West BT, Wechsler H. Alcohol-use disorders and nonmedical use of prescription drugs among US college students. Journal of Studies on Alcohol and Drugs. 2007;68 (4 ):543–547. doi:10.15288/jsad.2007.68.543 17568959
7. Schepis TS, McCabe SE. Prescription opioid misuse in us older adults: Associated comorbidities and reduced quality of life in the national epidemiologic survey of alcohol and related conditions-III. J Clin Psychiatry. 2019;80 (6 ). doi:10.4088/JCP.19m12853
8. Schepis TS, Hakes JK. Age of initiation, psychopathology, and other substance use are associated with time to use disorder diagnosis in persons using opioids nonmedically. Substance Abuse. 2017;38 (4 ):407–413. doi:10.1080/08897077.2017.1356791 28723266
9. Weitzman ER, Ong MS. Rising prevalence of comorbid alcohol and opioid use disorders in adolescents and young adults in the United States. J Gen Intern Med. 2019;34 (10 ):1987–1989. doi:10.1007/s11606-019-05068-6 31152359
10. Understanding the opioid overdose epidemic- centers for disease control and prevention Available from: https://www.cdc.gov/opioids/basics/epidemic.html#three-waves. Accessed March 8, 2024.
11. Furr‐Holden D, Milam AJ, Wang L, Sadler R. African Americans now outpace whites in opioid‐involved overdose deaths: A comparison of temporal trends from 1999 to 2018. Addiction. 2021;116 (3 ):677–683. doi:10.1111/add.15233 32852864
12. Hoopsick RA, Homish GG, Leonard KE. Differences in opioid overdose mortality rates among middle-aged adults by race/ethnicity and sex, 1999-2018. Public Health Rep. 2021;136 (2 ):192–200. doi:10.1177/0033354920968806 33211981
13. Romero R, Friedman JR, Goodman-Meza D, Shover CL. US drug overdose mortality rose faster among Hispanics than non-Hispanics from 2010 to 2021. Drug Alcohol Depend. 2023;246 :109859. doi:10.1016/j.drugalcdep.2023.109859 37031488
14. Cano M, Mendoza N, Ignacio M, Rahman A, Daniulaityte R. Overdose deaths involving synthetic opioids: Racial/ethnic and educational disparities in the eastern and western US. Drug Alcohol Depend. 2023;251 :110955. doi:10.1016/j.drugalcdep.2023.110955 37699286
15. Peters DJ, Monnat SM, Hochstetler AL, Berg MT. The opioid hydra: Understanding overdose mortality epidemics and syndemics across the rural‐urban continuum. Rural Sociology. 2020;85 (3 ):589–622. doi:10.1111/ruso.12307
16. Lu CH, Jette G, Falls Z, et al. A cohort of patients in New York state with an alcohol use disorder and subsequent treatment information - A merging of two administrative data sources. J Biomed Inform. 2023;144 :104443. doi:10.1016/j.jbi.2023.104443 Epub 2023 Jul 16. PMID: 37455008.37455008
17. Hedegaard H, Miniño AM, Spencer MR, Warner M. Drug overdose deaths in the United States, 1999–2020. In: NCHS Data Brief, No 428. Hyattsville, MD: National Center for Health Statistics; 2021. doi:10.15620/cdc:112340.
18. Singh JA, Cleveland JD. National US time-trends in opioid use disorder hospitalizations and associated healthcare utilization and mortality. PLoS One. 2020;15 (2 ):e0229174. doi:10.1371/journal.pone.0229174 32069314
19. Keyes KM, Rutherford C, Hamilton A, et al. What is the prevalence of and trend in opioid use disorder in the United States from 2010 to 2019? using multiplier approaches to estimate prevalence for an unknown population size. Drug and Alcohol Dependence Reports. 2022;3 :100052. doi:10.1016/j.dadr.2022.100052 35783994
20. Spencer MR, Garnett MF, Miniño AM. Drug overdose deaths in the United States, 2002–2022. NCHS Data Brief, No 491. Hyattsville, MD:National Center for Health Statistics;2024. 10.15620/cdc:135849.
21. McHugh RK, Nguyen MD, Chartoff EH, Sugarman DE, Greenfield SF. Gender differences in the prevalence of heroin and opioid analgesic misuse in the United States, 2015–2019. Drug Alcohol Depend. 2021;227 :108978. doi:10.1016/j.drugalcdep.2021.108978 34488078
