Compounding risks of chronic health conditions and substance use disorder on healthcare burden in the USA: Analysis of NSDUH data (2021–2023)

Highlights

• •Severe substance use disorder increases emergency room healthcare use.
• •Black and Native/Alaska Native adults show higher emergency room use.
• •Mild and moderate drug use disorders raise ER visits across CHC levels.
• •Mental health conditions substantially increase ER use and substance risks.

Abstract

Background

Emergency room (ER) use reflects acute healthcare burden, but the roles of chronic health conditions (CHCs), substance use disorders (SUDs), drug use disorders (DUDs), and mental health conditions (MHCs) remain underexplored across populations.

Methods

Using nationally representative survey data (N = 226,838; weighted = 1,243,120,763), we applied survey-weighted logistic regression to examine predictors of ER visits. Covariates included CHCs, SUDs, DUDs, severity levels of these orders, MHCs, race/ethnicity, education, employment, residence, and body mass index.

Results

Adults with ≥ 1 CHC were more likely to visit the ER (OR = 1.72; 95 % CI: 1.60–1.85). DUD significantly increased ER use (OR = 1.70; 95 % CI: 1.54–1.88), while overall SUD was not significant after adjustment (OR = 1.05; 95 % CI: 0.98–1.12). Severe SUD elevated ER use even without CHCs (OR = 1.89; 95 % CI: 1.67–2.13). African Americans had higher odds of ER visits (OR = 1.28; 95 % CI: 1.21–1.36), and Native American/Alaska Natives were more likely to report DUD (OR = 1.55; 95 % CI: 1.31–1.82). Lower educational attainment (OR = 1.22; 95 % CI: 1.16–1.28) and unemployment (OR = 1.34; 95 % CI: 1.25–1.43) were linked to higher risks. MHCs predicted ER use (OR = 1.63; 95 % CI: 1.53–1.74) and substance-related disorders.

Conclusions

CHCs, DUD severity, and MHCs are strong predictors of ER utilization. Disparities among African Americans and Native American/Alaska Natives highlight the need for integrated care addressing chronic illness, behavioral health, and substance use—particularly for socioeconomically and racially marginalized groups.

1. Introduction

Healthcare utilization (HCU), especially for emergency services, offers valuable insight into population health needs and the burden on healthcare systems. Although hospitalization rates in the United States remain stable between 2014–2018, the age-adjusted rate hospitalization rate for individuals aged 1–64 increased from 5.1 % in 2018 to 5.9 % in 2019 (National Center for Health Statistics, 2024). This rising trend poses significant challenges for patients and health systems, increasing financial burdens and straining healthcare resources (Duffy et al., 2024, Hirani et al., 2025). Frequent hospital admissions can result in lost productivity, reduced earnings, and heightened out-of-pocket expenses, particularly for uninsured individuals (Dobkin et al., 2018).

While a substantial body of literature has examined the impact of socioeconomic factors and chronic health conditions (CHCs) on HCU (Bayer-Oglesby, L., Zumbrunn, A., Bachmann, N., & Team, on behalf of the S et al., 2022, Kim et al., 2024, Komaromy et al., 2018, Yong and Yang, 2021), relatively few studies have investigated the comorbidity of substance use disorders (SUD) and CHCs, particularly using nationally representative survey data. This gap is significant given the growing recognition that individuals with both SUD and CHCs often experience more complex health trajectories, higher HCU, and poorer outcomes (Stephens et al., 2020, Wu et al., 2018). Most existing research in this area relies on clinical or administrative data, which may not fully capture the broader population or the social determinants influencing these comorbidities. By leveraging nationally representative data, such as the National Survey on Drug Use and Health (NSDUH), this study aims to provide a more comprehensive understanding of how the intersection of behavioral health and chronic disease contributes to hospitalization risk across diverse demographic and socioeconomic groups.

According to the Centers for Disease Control and Prevention (CDC), more than 76 % of Americans live with at least one CHC, and 51.4 % have multiple CHCs, contributing to 86 % of total healthcare expenditures (Centers for Disease Control and Prevention, 2020, Watson, 2022, Watson, 2025). Conditions such as cancer, diabetes, obesity, cardiovascular disease, and stroke account for high morbidity and mortality rates. Simultaneously, SUD has emerged as a worsening public health crisis. Approximately 48.5 million Americans are affected by SUD, a chronic, relapsing condition involving compulsive use of substances such as alcohol, opioids, and stimulants American Addiction Center (2025). Drug use overdose is responsible for nearly 105,000 deaths annually (National Institute on Drug Abuse, 2024, Substance Abuse and Mental Health Services Administration., 2023) and SUD is a major cause of disability (Anderson & Sharp, 2025).

The coexistence of SUD with conditions such as diabetes, cardiovascular disease, liver disease, and cancer increases the complexity of care and the risk of hospitalization (Desai et al., 2021, Wu et al., 2018). Beal describe this dual burden as double jeopardy; a term used in health disparities research to refer to the compounded disadvantage faced by individuals who experience multiple intersecting risk factors—such as CHCs and SUD—that together amplify vulnerability and adverse health outcomes (Beal, 2008). This framing underscores the urgency of addressing this intersection. However, many existing studies rely on electronic health records (EHRs), which often lack comprehensive data on behavioral health and may underrepresent the broader population.

1.1. The current study

To address this gap, the present study utilizes nationally representative survey data to examine the relationship between SUD, CHCs, and HCU. Guided by Andersen’s Behavioral Model of Health Services Use (Andersen, 1968), we explore how individual characteristics and comorbidities influence the use of emergency and inpatient services. By applying this framework, we aim to enhance understanding of healthcare-seeking behaviors and inform strategies to reduce preventable hospitalizations. This population-level analysis advances current research by highlighting the impact of SUD, CHCs, and their co-occurrence on HCU, supporting efforts to promote more integrated and equitable healthcare delivery.

2. Conceptual Framework: Comorbidity of chronic health conditions and substance use disorder in healthcare utilization

This study is guided by Andersen’s Behavioral Model of Health Services Use (Andersen, 1968), which posits that HCU is influenced by a combination of predisposing, enabling, and need-based factors. In the context of HCU, these factors interact in complex ways, particularly when considering the comorbidity of CHC and SUD.

As represented in Fig. 1, HCU is often driven by socioeconomic determinants such as income, education, employment status, and neighborhood characteristics (Hatef et al., 2019, Taylor et al., 2006, Wallar et al., 2020). Lower income is associated with increased risk of HCU, in part, due to higher rates of CHC, limited access to preventive care, and delays in seeking treatment (Kim et al., 2024, Wallar et al., 2020). Financial barriers may also deter HCU among low-income populations, potentially exacerbating health conditions until emergency care becomes necessary. Education plays a protective role; individuals with higher educational attainment tend to experience fewer hospitalizations, likely due to improved health literacy and better self-management of chronic conditions (Magnani, J. W., Mujahid, M. S., Aronow, H. D., Cené, C. W., Dickson, V. V., Havranek, E., Morgenstern, L. B., Paasche-Orlow, M. K., Pollak, A., Willey, J. Z., On behalf of the American Heart Association Council on Epidemiology and Prevention; Council on Cardiovascular Disease in the Young; Council on Cardiovascular and Stroke Nursing; Council on Peripheral Vascular Disease; Council on Quality of Care and Outcomes Research; and Stroke Council., 2018, Yong and Yang, 2021). Employment status is similarly influential. Unemployed individuals face elevated risks of hospitalization due to multiple factors, including loss of health insurance, increased stress-related illness, and reduced access to routine care (Junna et al., 2022). Neighborhood disadvantage further compounds these risks, as residents in impoverished areas are more likely to suffer from CHCs and require frequent HCU (Hatef et al., 2019, Taylor et al., 2006).

Fig. 1.

Conceptual Model of the Interrelationships Among Socioeconomic Conditions, Comorbidity, and HCU.

Beyond socioeconomic factors, the presence of CHCs—such as diabetes, cardiovascular disease, and chronic obstructive pulmonary disease (COPD)—is a significant predictor of hospitalization (Ho, T.-W., Tsai, Y.-J., Ruan, S.-Y., Huang, C.-T., Lai, F., Yu, C.-J., Group, H. S et al., 2014, Yohannes et al., 2017). Individuals with multiple CHCs often require more frequent and intensive care, underscoring the importance of coordinated care models (Dantas et al., 2016, Deutschbein et al., 2020, Schneider et al., 2016). SUD also contributes substantially to hospitalization rates, with affected individuals experiencing longer, more frequent, and costlier hospital stays due to the physical and psychological complications of substance misuse (Pavarin et al., 2022, Peterson et al., 2021).

Recent research has increasingly focused on the combined impact of CHC and SUD, revealing that individuals with both conditions face significantly higher healthcare burden than those with either condition alone (Gardner et al., 2022). Comorbid SUD can complicate disease management, exacerbate symptoms, and reduce access to integrated care services (Volkow & Blanco, 2023). Fig. 1 shows the study’s focus on understanding how the intersection of behavioral health and CHCs—shaped by socioeconomic and environmental factors—drives disparities in HCU.

3. Methods

3.1. Data Source

This study utilized pooled public data from the National Survey on Drug Use and Health (NSDUH) for the years 2021 through 2023. The NSDUH is an annual, nationally representative cross-sectional survey administered by the Substance Abuse and Mental Health Services Administration (SAMHSA). It collects detailed information on substance use, mental health conditions, and HCU among the civilian, non-institutionalized population of the United States aged 12 years and older.

3.2. Sampling Method

The NSDUH employs a multistage probability sampling design to ensure national representativeness. The sampling process involves stratification and clustering at multiple levels, including geographic areas and households. Within selected households, individuals are randomly selected to participate in the survey. Data collection is conducted using computer-assisted interviewing (CAI) techniques, which include both computer-assisted personal interviewing (CAPI) and audio computer-assisted self-interviewing (ACASI). These methods are designed to enhance data quality and confidentiality, particularly for sensitive topics such as substance use and mental health. Sampling weights provided by NSDUH are applied in all analyses to account for the complex survey design and to produce unbiased population estimates.

3.3. Sample Size

The original dataset contained 290,911 respondents. Participants younger than 18 years (n = 56,996) were excluded, leaving 233,915 adults eligible for analysis. To ensure valid estimates, cases with missing data on key study variables or failing analytic restrictions were removed, resulting in a final analytic sample of 226,838. This approach ensured complete-case analysis for regression models while preserving representativeness through survey weighting.

3.4. Ethical Considerations

The NSDUH dataset is de-identified and made accessible by SAMHSA for research purposes. As such, the use of this dataset does not require institutional review board (IRB) approval, in accordance with federal regulations governing research with publicly available, anonymized data. All analyses were conducted in compliance with ethical standards for secondary data use. No identifiable personal information was accessed or used, and all data handling procedures adhered to the guidelines set forth by SAMHSA to ensure confidentiality and data integrity.

3.5. Constructs and measures

Following Andersen’s Behavioral Model of Health Services Use (Andersen, 1968), this study incorporated a comprehensive set of variables that reflect the key domains influencing HCU. These included predisposing factors such as demographic characteristics (e.g., age, sex, race/ethnicity), enabling factors like socioeconomic status (e.g., education, employment, geographic location), and need-based factors including behavioral health indicators, CHC, and mental health status. By aligning with Andersen’s framework, the study aimed to explore how these interrelated components contribute to patterns of HCU among U.S. adults.

3.5.1. Dependent variables

Healthcare utilization (HCU) outcome in the study is the frequency of Emergency Room (ER) visits. ER visit variable was analyzed in both count and binary formats. For binary analysis, the frequency of healthcare visits was dichotomized: 0 indicated no visit, and 1 indicated one or more visits in the past year. This binary classification was used in binomial regression models, while the actual counts of visits were used in Poisson regression models to assess utilization patterns.

3.5.2. Predictors

CHCs were measured via self-report of ever being diagnosed with any of the following: heart condition, diabetes, chronic obstructive pulmonary disease (COPD), cirrhosis, hepatitis B/C, kidney disease, asthma, HIV/AIDS, cancer, or high blood pressure. Each CHC variable was originally coded as 1 (Yes), 0 (No), with additional codes for bad data (85), don’t know (94), refused (97), and legitimate skip (99). A composite CHC burden score was created by summing binary indicators, ranging from 0 to 9. Two categorical variables were derived: one with four levels (0 CHC, 1 CHC, 2–4 CHC, and 5–9 CHC) to examine disparities, and another binary variable indicating disease burden (1 = at least one CHC, 0 = none). Body Mass Index (BMI), which was categorized using CDC thresholds as normal (18.5–24.9), overweight (25.0–29.9), and obese (≥30.0). Mental health condition (MHC) was assessed through self-report of a major depressive episode in the past 12 months, coded as yes (1) or no (0).

SUD and DUD were defined using DSM-5 criteria. SUD included misuse of alcohol, cocaine, heroin, marijuana, methamphetamine, inhalants, hallucinogens, sedatives, stimulants, tranquilizers, and pain relievers. DUD included the same substances as SUD, excluding alcohol. Both disorders were coded into severity levels: mild (1), moderate (2), severe (3), or none (4). Additionally, binary variables were created for each disorder, coded as 0 (no disorder) or 1 (any disorder).

3.5.3. Covariates

Age was categorized into five groups after dropping those less than 18 years (i.e., 18–25, 26–34, 35–49, 50–64, and 65 years and older). Sex was coded as male (1) or female (2). Race and ethnicity classifications included Non-Hispanic White, Non-Hispanic Black/African American, Non-Hispanic Native American/Alaska Native, Non-Hispanic Native Hawaiian/Other Pacific Islander, Non-Hispanic Asian, Non-Hispanic individuals identifying with more than one race, and Hispanic. Two SES variables were included. Employment status was recorded as either employed or not employed, while education level was grouped into four categories: less than high school, high school graduate, some college or associate degree, and college graduate. Geographic location was defined as metro or non-metro based on residence classification in the dataset.

3.6. Statistical analysis

We conducted descriptive and inferential analyses to examine the relationship between CHCs, SUD, DUD, and HCU. First, we generated frequency distributions to describe the study variables. Rao-Scott Chi-Square was used to assess the prevalence of ER visits by 2-l2v2l variables (CHC, SUD, DUD) and 4-level severity of SUD and DUD.

Second, logistic regression model was applied to assess the association between the predictors (i.e., CHCs, SUD/DUD) and the binary ER visits (coded as 0 = no use, 1 = at least one visit). Fig. 2 shows that ER visits follow a heavily right-skewed distribution. The variance of 1.89 exceeds the mean of 0.42 (range 0–31), indicating overdispersion, therefore violates the Poisson regression assumption of equal mean and variance. To account for over-dispersion in the count data (Fig. 2), negative binomial regression (NBR) was employed for the frequency (count) of ER visits. Both logistic regression and NBR models adjusted for key covariates described in ‘subsection 3.5.2′ above.

Fig. 2.

Histogram of the Frequency of Emergency Room Visits.

In addition, NBR models were stratified by the number of co-occurring CHCs—categorized as none, 1 CHC, 2–4 CHCs, and ≥ 5 CHCs—to further assess whether the effect of SUD severity on HCU varied by comorbidity burden. Survey weights, strata, and primary sampling units were accounted for using Stata’s mi svyset command to produce nationally representative estimates. All models report adjusted odds ratios (ORs) for logistic regression and incidence rate ratios (IRRs) for NBR, along with their corresponding 95 % confidence intervals (CIs). Analyses were conducted using Stata Statistical Software version 18 (StataCorp, 2023).

4. Results

4.1. Frequency distribution

Of the 233,915 respondents meeting inclusion criteria, 226,838 had complete data and were included in the analytic models. Table 1 summarizes the demographic, health, and substance use characteristics of the study sample. The sample was relatively young, with nearly half aged 18–34 years (50.36 %) and another 26.78 % aged 35–49 years. Women represented a slightly higher proportion than men (55.68 % vs. 44.32 %). Educational attainment was moderately high, with 35.21 % holding a college degree, while 10.05 % had not completed high school. Non-Hispanic Whites accounted for the majority of participants (60.54 %), followed by Hispanics (17.10 %) and Non-Hispanic Blacks (11.35 %), while other racial/ethnic groups were less represented. Most participants were employed (64.11 %), resided in metropolitan areas (84.00 %), and were overweight or obese (65.46 %).

Table 1.

Frequency Distribution of Demographic, Health, and Substance Use Variables (N = 233,915).

Variable	Category	Frequency	Percent
Age Category	18–25 years	70,159	29.99
	26–34 years	47,637	20.37
	35–49 years	62,653	26.78
	50–64 years	27,148	11.61
	≥65 years	26,318	11.25
Sex at Birth	Male	103,679	44.32
	Female	130,236	55.68
Education Level	Less than high school	23,509	10.05
	High school graduate	58,582	25.04
	Some college/Assoc. degree	69,459	29.69
	College graduate	82,365	35.21
Race/Ethnicity	Non-Hispanic White	141,604	60.54
	Non-Hispanic Black/Afr. Am	26,548	11.35
	Native Am./Alaska Native	2,933	1.25
	Native HI/Other Pac. Islander	1,001	0.43
	Non-Hispanic Asian	12,732	5.44
	More than one race, NH	9,109	3.89
	Hispanic	39,988	17.10
Employment Status	Employed	149,972	64.11
	Not employed	83,943	35.89
Geographic Location	Metro	196,492	84.00
	Nonmetro	37,423	16.00
BMI Category	Non-obese	80,800	34.54
	Overweight	66,577	28.46
	Obese	86,538	37.00
MHC	Yes	45,071	19.95
	No	180,900	80.05
CHC Category (2-level)	None	156,062	66.72
	≥1 condition	77,853	33.28
CHC Category (4-level)	0 conditions	156,062	66.72
	1 condition	53,221	22.75
	2–4 conditions	23,789	10.17
	5–9 conditions	843	0.36
ER Visits (binary)	No visit	178,306	78.60
	Had visit	48,532	21.40
DUD (binary)	No	206,687	88.36
	Yes	27,228	11.64
SUD (binary)	No	185,561	79.33
	Yes	48,354	20.67
DUD Severity	Mild	14,253	6.09
	Moderate	6,808	2.91
	Severe	6,167	2.64
	None	206,687	88.36
SUD Severity	Mild	25,795	11.03
	Moderate	11,379	4.86
	Severe	11,180	4.78
	None	185,561	79.33
Continuous variables
		Mean (SD)	Min-Max
CHC c (count)		0.49	0–9
BMI c		28.33 (6.87)	9.37–68.56
ER Visit c (count)		0.42 (1.37)	0–31

^ccontinuous variable.

Note: SUD = substance use disorder (alcohol and/or drugs); DUD = Drug Use Disorder (excluding alcohol); CHC = Chronic Health Conditions; MHC = Mental Health Condition; BMI Body Mass Index; ER Emergency Room; SD Standard Deviation; Min-Max Minimum-Maximum.

Regarding health and healthcare use, approximately one-third (33.28 %) reported at least one chronic health condition, but only 0.36 % had ≥ 5 conditions. Mental health concerns were reported by 19.95 % of participants. About one in five respondents (21.40 %) had at least one emergency room visit. SUD was more prevalent than DUD (20.67 % vs. 11.64 %), and most substance users reported mild severity (11.03 %). Continuous measures showed a mean BMI of 28.33 (SD = 6.87) and the average of ER visit of 0.42.

4.2. Prevalence of ER visit by CHC, DUD, SUD, and substance Use-related disorder severity

The results of Rao-Scott Chi-Square analysis show significant variation between ER visit and CHC (p < 0.0001), SUD (p < 0.0001), DUD (p < 0.0001), and their severity (p < 0.0001). The prevalence of ER visits was 54.8 % among individuals with at least one CHC, 23.2 % among those with a SUD and 14.65 % among those with a DUD. When examining the severity of substance-related disorders, ER visit prevalence was highest among individuals with mild SUD (11.66 %) and mild DUD (7.82 %; result not presented).

4.3. Logistic regression Assessing the Association between Demographics, health Characteristics, and healthcare use

Table 2 shows demographic predictors of ER visits, DUD, SUD and CHCs. Compared to adults aged 65 or older, young adults (18–25) had dramatically higher odds of DUD (OR = 5.25, 95 % CI = 4.18–6.58), and SUD (OR = 3.69, 95 % CI = 3.19–4.26), with elevated risks persisting but declining across older age groups. In contrast, increasing age was significantly associated with the prevalence of CHCs but ER visit odds did not vary greatly by age, with only a slight reduction among young adults (OR = 0.92, 95 % CI = 0.84–1.01). Female participants were more likely to have more ER visits but reduced odds of DUD and SUD than male participants. Education strongly influenced outcomes: individuals with less than a high school education were more than twice as likely to have ER visits (OR = 2.16, 95 % CI = 1.94–2.40) or DUD (OR = 2.42, 95 % CI = 2.12–2.76) compared to college graduates. However, less than a high school education was associated with a lower prevalence of CHCs (OR = 83, 95 % CI = 0.75––0.92) while some college/associate degrees was associated with increased odds of CHCs. Employment was protective, with employed adults showing reduced odds of ER visits (OR = 0.73), DUD (OR = 0.69), SUD (OR = 0.92) and CHCs (OR = 0.72, 95 % CI 0.68–0.77).

Table 2.

Association between Demographic Variables and Key Variables: Emergency Room Visits, Overnight Hospitalization, Drug Use and Substance Use Disorders.

	Emergency Room Visit		Drug Use Disorder		Substance Use Disorder		Chronic Health Conditions
	OR	95 % CIs	OR	95 % CIs	OR	95 % CIs	OR	95 % CIs
Age Group (+65 ref)
18–25	0.92**	0.84, 1.01	5.25***	4.18, 6.58	3.69***	3.19, 4.26	0.09***	0.08, 0.10
26–34	0.98	0.89, 1.08	5.69***	4.57, 7.09	4.15***	3.58, 4.80	0.11***	0.10, 0.12
35–49	0.96	0.88, 1.05	3.81***	3.05, 4.76	3.09***	2.69, 3.57	0.18***	0.17, 0.20
50–64	0.96	0.87, 1.07	2.39***	1.82, 3.12	2.17***	1.85, 2.54	0.43***	0.39, 0.47
Sex at Birth (Male ref)
Female	1.14***	1.08, 1.21	0.63***	0.58, 0.68	0.61***	0.57, 0.65	0.95	0.90, 1.00
Education (College graduate ref)
less than high school	2.16***	1.94, 2.40	2.42***	2.12, 2.76	1.43***	1.27, 1.61	0.83***	0.75, 0.92
High school graduate	1.73***	1.61, 1.87	2.03***	1.85, 2.24	1.27***	1.18, 1.38	0.95	0.89, 1.01
Some college/associate degree	1.47***	1.38, 1.57	1.93***	1.724, 2.15	1.31***	1.21, 1.42	1.07*	1.01, 1.14
Employment (Not employed ref)
Employed	0.73***	0.69, 0.76	0.69***	0.65, 0.75	0.92**	0.86, 0.99	0.72***	0.68, 0.77
Race/Ethnicity (Whites ref.)
Black- NH	1.40***	1.29, 1.52	1.09	0.98, 1.20	0.98	0.90, 1.06	1.08	0.98, 1.18
Native America/Alaska Native	1.43**	1.05, 1.96	2.03**	1.61, 2.56	1.73***	1.35, 2.22	1.01	0.78, 1.31
Native Hawaii/Other Pacific Islanders	0.99	0.65, 1.53	1.28	0.71, 2.31	0.99	0.63, 1.57	0.80	0.51, 1.23
Asian, NH	0.57***	0.48, 0.67	0.44***	0.35, 0.56	0.40***	0.34, 0.48	0.86*	0.74, 0.99
Multiple races, NH	1.12	0.97, 1.30	1.39***	1.17, 1.65	1.29**	1.09, 1.53	1.16*	1.01, 1.33
Hispanics	0.88**	0.79, 0.96	0.76***	0.68, 0.85	0.78***	0.71, 0.85	0.80***	0.73, 0.86
Nonmetro (Metro ref)	1.09**	1.00, 1.17	0.92	0.81, 1.03	0.87**	0.80, 0.95	1.09*	1.02, 1.16
BMI (Normal ref.)
Overweight	1.07	0.99, 1.16	0.78***	0.71, 0.85	0.89**	0.83, 0.96	1.40	1.31, 1.47
Obese	1.29***	1.21, 1.38	0.68***	0.62, 0.75	0.73***	0.68, 0.78	2.18	2.09, 2.27
Adult Mental Health (No ref.)	1.61***	1.52, 1.71	3.35***	3.07, 3.66	2.72***	2.51, 2.95	1.80***	1.67, 1.94

Note: BMI body mass index; NH non-Hispanics.

* p < 0.05, ** p < 0.001, *** p < 0.0001.

Racial and geographic patterns were also evident. Non-Hispanic Black adults (OR = 1.40) and Native American/Alaska Natives (OR = 1.43) had greater odds of ER visits, and Native American/Alaska Natives faced particularly high DUD (OR = 2.03) and SUD (OR = 1.73) risks. Asians had significantly lower odds across all outcomes, while Hispanics showed modestly reduced risks for ER visits, substance-related disorders, and CHCs (OR = 0.86, 95 % 0.74–0.99). Living in nonmetro areas slightly increased the odds of CHCs and ER visits but reduced the odds of SUD (Table 2). Obesity was associated with higher ER use but lower odds of DUD and SUD. Notably, adults with a lifetime mental health diagnosis had substantially higher odds of all outcomes, highlighting mental health as a critical factor in both HCU, substance-related risks, and CHCs.

4.4. Interaction models of chronic health Condition, substance Use, Drug Use, and mental health

Table 3 presents the results of the logistic regression examining main and interaction effects of CHC, SUD, DUD, and MHC on the odds of ER visits. Having at least one CHC was strongly associated with increased odds of ER visits (OR = 1.72, 95 % CI = 1.56–1.87), indicating that individuals with chronic conditions are 72 % more likely to use ER services compared to those without. DUD was also a significant predictor (OR = 1.70, 95 % CI = 1.47–1.95), suggesting a 70 % higher likelihood of ER use among individuals with drug use disorders. MHC showed a moderate but significant effect (OR = 1.41, 95 % CI = 1.30–1.53). In contrast, SUD was not significantly associated with ER visits. None of the interaction terms reached statistical significance, indicating that the combined effects of CHC with SUD or DUD, and MHC with SUD or DUD, did not meaningfully modify the odds of ER visits beyond their independent contributions.

Table 3.

Main and Interaction Effect Model for Chronic Health Conditions, Substance Disorder and Healthcare Utilization.

	Emergency Room Visit
	OR	95 % CIs
Main effects
CHC	1.72***	1.56, 1.87
SUD	1.11	0.87, 1.27
DUD	1.70***	1.47, 1.95
MHC	1.41***	1.30, 1.53
Interaction effects
CHC *SUD	1.14	0.97,1.34
CHC *DUD	0.89	0.97, 1.33
MHC*DUD	0.86	0.66, 1.11
MHC*SUD	1.03	0.82, 1.29

Note: All analyses adjusted for demographic (i.e., age, sex, race/ethnicity), employment status, location (i.e., metro status), and body mass index.

CHC= Chronic Health Condition; SUD = substance use disorder; DUD = Drug Use Disorder; MHC = Mental Health Conditions.

* p < 0.05, ** p < 0.001, *** p < 0.0001.

4.5. Negative binomial regression of the association between disorder severity and healthcare use stratified by level of CHC

The result of the negative binomial poison regression is shown in Table 4. Overall, severe SUD was significantly associated with increased ER utilization (IRR = 1.97, 95 % CI = 1.22–3.17), while mild or moderate SUD showed no significant effect (Table 4). Stratification by CHC burden revealed that severe SUD remained a significant predictor of ER visits for individuals with no CHCs (IRR = 1.39, 95 % CI = 1.03–4.81), 1 CHC (IRR = 1.89, 95 % CI = 1.41–2.54), 2–4 CHCs (IRR = 1.53, 95 % CI = 1.15–2.02), and notably for those with > 5 CHCs (IRR = 4.19, 95 % CI = 1.33–13.12), indicating a strong dose–response relationship between SUD severity and ER use.

Table 4.

Association of Drug and Substance Use Disorder Severity and Healthcare Utilization Stratified by the Number of Chronic Health Conditions.

	Chronic Health Conditions
	Overall		None		1-CHC		2–4 CHCs		> 5 CHCs
	IRR	95 % CIs	IRR	95 % CIs	IRR	95 % CIs	IRR	95 % CIs	IRR	95 % CIs
SUD (None ref)
Mild	1.04	0.94, 1.16	0.98	0.85, 1.14	1.06	0.86, 1.30	1.34*	1.05, 1. 69	1.06	0.35, 3.21
Moderate	1.05	0.91, 1.20	1.11	0.95, 1.30	0.84	0.63, 1.11	1.30	0.89,1.90	1.24	0.43, 3.61
Severe	1.97**	1.22, 3.17	1.39*	1.03, 4.81	1.89***	1.41, 2.54	1.53**	1.15, 2.02	4.19*	1.33, 13.12
DUD (none ref)
Mild	1.69***	1.47, 1.94	1.83***	1.48, 2.28	1.55**	1.22, 1.97	1.28	0.95, 1.73	0.74	0.18, 3.12
Moderate	1.76***	1.43, 2.15	1.77***	1.35, 2.31	2.01***	1.43, 1.97	1.25	0.83, 1.86	0.34	0.09, 1.23
Severe	1.14	0.70, 1.75	1.09	0.49, 2.38	1.09	0.76. 1.57	1.33	0.86, 2.06	0.22*	0.06, 0.81

Note: All analyses adjusted for demographic (i.e., age, sex, race/ethnicity), employment status, location (i.e., metro status), mental health, and body mass index. * p < 0.05, ** p < 0.001, *** p < 0.0001.

For DUD, both mild and moderate severity were consistently associated with higher ER utilization overall (IRRs = 1.69 and 1.76, respectively; p < 0.001) and across most CHC categories. Mild DUD showed particularly high effects among those with no CHCs (IRR = 1.83, 95 % CI = 1.48–2.28) and 1 CHC (IRR = 1.55, 95 % CI = 1.22–1.97). Severe DUD, however, did not show consistent significant effects, except for a reduced ER risk among individuals with > 5 CHCs (IRR = 0.22, 95 % CI = 0.06–0.81).

5. Discussion

5.1. Contextualizing the findings

This study investigated the prevalence and predictors of HCU among individuals with health and behavioral comorbidity (i.e., CHCs and SUD) in a large, nationally representative adult sample. Several important findings emerged.

First, CHCs and DUD were the strongest predictors of ER use. Adults with at least one CHC were 72 % more likely to visit the ER than those without, consistent with prior research linking chronic disease burden to high-cost acute care (Han et al., 2018, Magnusson et al., 2020, McPhail, 2016, Stephens et al., 2020). DUD also significantly elevated ER use, indicating that acute episodes stemming from drug use are a significant driver of emergency care utilization. In contrast, the overall substance use disorder was not a strong independent predictor once other factors were considered, although severe SUD was associated with increased ER utilization across all levels of CHC burden, reinforcing evidence that high-severity substance use places considerable strain on emergency services (Wu, Zhu, et al., 2018).

The dose–response relationship observed between SUD or DUD severity and ER visits further underscores the importance of early intervention. Mild and moderate DUD showed consistent associations with higher ER use even among individuals without chronic conditions, suggesting that even less severe substance-related problems can disrupt health stability and precipitate acute care use. Similar patterns have been reported in national surveys and hospital data, where repeated ER visits often occur before individuals engage in formal addiction treatment (Ashford et al., 2019, D’Onofrio et al., 2015, Fleury et al., 2022).

Racial and ethnic disparities in ER utilization and substance use disorders were notable. African Americans and Native American/Alaska Natives had higher odds of ER visits, and Native American/Alaska Natives had particularly elevated risks for DUD and SUD. These findings align with prior work linking structural racism, historical trauma, and inequitable access to preventive care to higher rates of behavioral and substance-related disorders among these groups (Farahmand et al., 2020, Soto et al., 2022). Access to culturally appropriate care remains limited due to structural barriers, including underfunded health systems, geographic isolation, and a shortage of culturally competent providers (Allen et al., 2022, O’Keefe et al., 2021). Initiatives like the Native Collective Research Effort to Enhance Wellness (N-CREW) highlight the value of community-led, culturally responsive interventions that integrate traditional healing practices and Indigenous knowledge systems (National Institute of Health, 2024). In contrast, Asians showed consistently lower odds across outcomes, while Hispanics had modestly reduced risks, echoing previous studies that suggest protective cultural or community factors may mitigate some substance-related risks (Rogers-LaVanne et al., 2023).

Socioeconomic factors also shaped the use of healthcare. Lower educational attainment was associated with significantly higher odds of ER visits and substance-related disorders, a pattern often attributed to barriers in accessing primary care and health literacy (Armoon et al., 2023). Employment was protective against all outcomes, highlighting the role of economic stability in reducing both substance-related risks and reliance on emergency care. Nonmetro residence slightly increased ER visits, possibly reflecting reduced access to outpatient and specialty services in rural areas.

The strong associations between mental health conditions and ER visits, DUD, and SUD echo evidence that untreated mental illness often co-occurs with substance use and contributes to preventable acute care use (Avery & Barnhill, 2017). Integrating behavioral health screening and treatment into primary care or community-based settings may reduce both substance-related harm and unnecessary ER utilization.

Notably, interaction effects between CHCs, SUD, DUD, and MHC did not significantly modify the odds of ER visit, suggesting that their contributions are largely additive rather than synergistic. This finding supports prior observations that while comorbid conditions individually increase acute care use, their combined effects may not always exceed the sum of their parts when demographic and socioeconomic factors are accounted for.

While studies such as Wakeman et al. (2019) and Carruthers and Sutton-Inoncencio (2023) have shown that SUD frequently go untreated in primary care settings, a more critical issue is that individuals with SUD may not disclose their substance use to primary care providers. This lack of disclosure can lead to gaps in the comprehensive management of co-occurring CHCs, resulting in fragmented care and increased reliance on acute services. The finding that moderate and severe SUD significantly predicted emergency room use across all CHC categories reinforces the argument that SUDs function as independent contributors to acute care needs, and that their under-recognition in primary care settings may exacerbate the burden of chronic disease management.

Our study contributes to this discourse by employing a nationally representative dataset, which contributes to the general picture of the issues and addresses limitations tied to localized electronic health record data. Moreover, we stratified our analysis by CHC burden, allowing us to demonstrate that the impact of SUD on HCU remains significant even in the absence of multiple chronic diseases.

5.2. Implications for Practice and policy

The findings from this study underscore the urgent need to address the structural separation between behavioral health and general medical care. Individuals with SUD, particularly those with moderate to severe severity levels, exhibit significantly higher rates of emergency room visits—even in the absence of diagnosed CHCs. This pattern highlights the role of SUD as an independent driver of acute HCU, which is often reactive, costly, and poorly integrated with long-term treatment planning.

Importantly, SUD and DUD should not be viewed solely through the lens of behavioral health. These conditions are also chronic, relapsing medical disorders that require sustained, multidisciplinary care approaches (Volkow & Blanco, 2023). Framing SUD/DUD exclusively as behavioral issues risks reinforcing stigma and may hinder the development of comprehensive treatment strategies. Therefore, embedding screening and treatment for SUD/DUD within both primary care and emergency care settings is essential—not only to improve continuity of care but also to recognize and treat these conditions as part of the broader spectrum of chronic disease management.

Several actionable implications emerge from these findings. First, healthcare systems should prioritize the implementation of integrated care models, such as the CoCM, which bring together primary care and behavioral health services to support patients with complex needs. Second, the high utilization among populations with comorbid conditions—including racial and ethnic minority groups such as Native Americans and Alaska Natives—calls for culturally responsive, community-based interventions that address both clinical and structural determinants of health.

At the policy level, the results support ongoing efforts to expand insurance coverage for SUD treatment, increase behavioral health workforce capacity in underserved regions, and promote payment models that incentivize coordinated care. Public health strategies should also focus on early intervention for mild and moderate SUD, which, as this study shows, are associated with meaningful increases in healthcare use and represent opportunities for upstream prevention. Future research should explore how housing stability, social support, and access to outpatient services mediate the relationship between SUD and HCU.

5.3. Study limitations and Strengths

Several limitations should be considered when interpreting the findings of this study. First, although the NSDUH provides a large and nationally representative sample of the U.S. civilian, non-institutionalized population, it excludes individuals in institutional settings such as hospitals, prisons, nursing homes, and long-term care facilities—populations that may have higher rates of both CHC and SUD. This exclusion limits the generalizability of the findings to the broader U.S. population, particularly those with severe illness or socioeconomic disadvantage.

Second, all health and behavioral variables in the NSDUH are based on self-reported data, which may be subject to recall bias or underreporting, especially for stigmatized conditions such as drug use, HIV/AIDS, and mental illness. The potential for misclassification of SUD severity or underreporting of HCU could attenuate the observed associations.

Third, the NSDUH dataset codes sex as a binary variable (male/female), which reflects the survey’s data collection methodology. While this approach aligns with the structure of the dataset, it does not account for the full spectrum of gender diversity recognized in current public health research. This limitation may affect the inclusivity and generalizability of findings, particularly for individuals whose gender identity does not align with binary classifications. Future research should consider more inclusive measures of gender to better capture disparities in HCU.

Fourth, the CHC burden score treats all chronic conditions equally, despite substantial differences in severity (e.g., asthma vs. cancer). This simplification, common in population-level studies, may obscure differences in condition-specific impacts. Future research could apply weighted scoring or severity-adjusted indices to better capture differential disease burden.

Lastly, the cross-sectional design of the study limits causal inference. While we observed strong associations between SUD, chronic health burden, and HCU, we cannot determine the temporal order of these relationships. Future longitudinal research is needed to explore causality and identify mechanisms linking SUD and comorbidity to healthcare use over time.

Despite these limitations, this study offers important population-level insights into how comorbid behavioral and physical health conditions drive healthcare demand. It complements prior HER-based studies by providing population-based estimates and emphasizes the value of integrated care models for addressing complex health needs in the community.

In conclusion, this study emphasizes the growing evidence that the intersection of SUD and CHC significantly increases the burden on acute healthcare systems in the United States. Using nationally representative survey data from the 2021–2023 NSDUH, we demonstrated that individuals with moderate to severe SUD are disproportionately more likely to utilize emergency room services, even when they do not have co-occurring chronic physical health conditions. These findings remained robust after adjusting for sociodemographic, mental health, and geographic factors. Importantly, this study highlights the need for healthcare systems to adopt an integrated care model that bridges the gap between behavioral and physical health. The elevated HCU among individuals with SUD, particularly those with multiple CHC, signals a missed opportunity for early intervention and community-based behavioral health care. Without addressing these intersections, reliance on costly ER visits will likely persist. As such, public health strategies should prioritize the early identification and treatment of SUD, improve access to mental health services, and enhance coordination between primary care and specialty providers. Addressing these challenges holistically is vital to reducing preventable healthcare use and advancing health equity for vulnerable populations.

CRediT authorship contribution statement

Ayodeji Iyanda: Writing – review & editing, Writing – original draft, Visualization, Validation, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. Richard Adeleke: Writing – review & editing, Writing – original draft. Omowunmi Iyanda: Writing – review & editing, Writing – original draft.

Funding

This study received funding support from PVAMU RISE research grants during the preparation of this manuscript.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

The authors declare that there are no conflicts of interest related to the research, authorship, or publication of this study. No financial, personal, or professional affiliations or relationships have influenced the conduct or outcomes of this research.

All research activities were conducted in accordance with ethical standards, and the findings presented are solely the result of the authors’ independent analysis and interpretation.

Data availability

Data will be made available on request.