Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2023 Oct 2.
Published in final edited form as: J Subst Abuse Treat. 2022 May 20;142:108806. doi: 10.1016/j.jsat.2022.108806

Multimorbidity classes indicate differential patterns of health care engagement among people who inject drugs

Omeid Heidari a,b,*, Becky L Genberg c, Nancy Perrin d, Derek T Dangerfield II d,e, Jason E Farley b,d, Gregory Kirk c, Shruti H Mehta c
PMCID: PMC10544774  NIHMSID: NIHMS1884442  PMID: 35643587

Abstract

Background:

Aging people who inject drugs (PWID) have complex health needs. Health care management could be complicated by persistent substance use, multiple health challenges, and inconsistent access to care. However, we know little about the relationship between chronic multimorbidity and health care engagement in this population. The purpose of this study is to characterize patterns and correlates of chronic disease multimorbidity among PWID.

Methods:

We conducted a latent class analysis (LCA) using data from the AIDS Linked to the IntraVenous Experience (ALIVE) Study, a community-based observational cohort, to determine classes of multimorbid chronic diseases. We then conducted regressions to determine factors associated with class membership and the impact of each multimorbid class on health events and utilization.

Results:

Of 1387 individuals included, the majority were male (67%) and Black (81%), with a mean age of 53 years. We identified four classes of multimorbidity: Low Multimorbidity (54%), and Low Multimorbidity Including Psychiatric Comorbidity (26%), Multimorbidity (12%), and Multimorbidity Including Psychiatric Comorbidity (7%). Female sex, baseline age, and receipt of disability were factors significantly associated with membership in all three classes compared to the Low Multimorbidity class. Additionally, PWID in these three classes were significantly more likely to utilize emergency room and outpatient health care. Membership in both classes with psychiatric comorbidity was associated with significantly higher adjusted odds of receiving medication for opioid use disorder.

Discussion:

Holistic health care systems can best address the needs of aging PWID with integrated care that provides harm reduction, substance use and mental health treatment together, and wrap around services.

Keywords: Chronic disease, Opioid use, Healthcare utilization, Latent class analysis

1. Introduction

In the United States, aging people who inject drugs (PWID) have complex health needs. Treatment guidelines and health care services designed to provide care to PWID often limit the scope to infectious disease treatment and prevention, skin and soft tissue infections, and harm reduction strategies, with little attention paid to engagement in health care for treatment of chronic diseases that can be diagnosed and managed in primary care (Centers for Disease Control and Prevention, 2012; Low et al., 2016; Visconti et al., 2019). Increased primary care utilization generally leads to improved health outcomes for persons with substance use disorder and results in reduced cost to health systems (Chi et al., 2011; Park et al., 2015; Parthasarathy et al., 2012). However, PWID utilize primary care at low rates. Consequently, they may overutilize emergency and inpatient care, and accordingly are often seen in later stages of disease (Artenie et al., 2015; Larney et al., 2017; Mizuno et al., 2015; Okeke et al., 2015; Thompson et al., 2018; Westergaard et al., 2013). Therefore, we must understand the holistic health care needs of PWID to target services and optimize health care engagement.

According to the Behavioral Model for Vulnerable Populations, adapted from the Andersen Model of Healthcare Utilization, predisposing, enabling, need, and health behavior factors influence primary care utilization and engagement (Andersen, 1968; Gelberg et al., 2000). Predisposing factors are individual characteristics such as age, race, and sex; enabling characteristics include access, social, and structural barriers within the health care infrastructure; need characteristics refer to both evaluated and perceived health conditions; health behaviors for PWID are actions taken by individuals that relate to management of multimorbid diseases and substance use. The impact of predisposing, enabling, and health behavior characteristics on health care engagement for PWID has been well studied. Enabling characteristics such as insurance status, low clinician continuity, and stigma decrease primary care and outpatient service engagement (Hartzler et al., 2018; Westergaard et al., 2011, 2013). Predisposing characteristics of Black race, Hispanic ethnicity, and younger age have all been linked with decreased utilization and engagement (Hoots et al., 2017; Lesko et al., 2019). Conversely, a research has demonstrated that a number of enabling factors facilitate health care access and engagement, such as medication for opioid use disorder (MOUD), integrated care services, and focused retention interventions (Islam et al., 2013; Mehta et al., 2015; Miller et al., 2018; Morozova et al., 2017). However, how primary care engagement may be further influenced by “need”-based characteristics from the Behavioral Model, particularly multimorbid chronic diseases, remains unclear.

Multimorbidity (having two or more co-occurring diseases) has been previously reported to be common and under-addressed for PWID (Johnston et al., 2019), with prior studies demonstrating prevalence as high as 58% for diabetes, obstructive lung disease, liver disease, kidney dysfunction, and hypertension among PWID (Cullen et al., 2009; Krupski et al., 2015; Monroe et al., 2011; Salter et al., 2011). Moreover, multimorbid diseases among PWID are widely undiagnosed or under-treated. As many PWID in the United States age, the prevalence of multimorbidity is likely to increase; thus understanding the impact of having multiple and sometimes undiagnosed chronic conditions will be important for optimizing care (Piggott et al., 2013, 2016; Salter et al., 2011).

The purpose of this study is to define the phenotypes of multimorbidity using latent class analysis (LCA) for a population of community based PWID and associations between multimorbid disease states and health care engagement, accounting for predisposing, enabling, and health behavior factors. Taken together, the findings of this study classify all domains of health care engagement from the Behavioral Model specific to PWID and highlight areas for targeted and holistic healthcare.

2. Methods

2.1. Study sample

The study derived data from the AIDS Linked to the Intravenous Experience (ALIVE), a community-based cohort study with current and former people who inject drugs in Baltimore, MD. The ALIVE Study was first established in 1988 to characterize the incidence and natural history of HIV among PWID (Vlahov et al., 1991). Subsequent recruitment waves of new participants occurred during 1994–1995, 1998, 2005–2008, and 2015–2018. Eligibility criteria for the ALIVE Study included being 18 years or older, negative for HIV or free of clinical AIDS, and reporting injection drug use in the 6 months prior to enrollment. Biannual study visits with participants included an interview, clinical examination, and biospecimen collection, which included HIV antibody tests for HIV negative participants, HIV RNA and CD4 cell count for participants living with HIV, complete blood count, comprehensive metabolic panel, and hemoglobin A1c. All participants provided written informed consent prior to participation in the study. The Johns Hopkins Bloomberg School of Public Health Institutional Review Board approved study procedures.

The sample for the current study included participants with data collected in 2015–2019 (N = 1646) who had at least two study visits by the end of 2019 (N = 1387). From January 2015 through the end of December 2019, participants included in the analysis had a median of 8 visits (25th–75th percentile: 6–10). The first study visit for each participant in the study period represents their baseline visit for this analysis. Compared to those included in this analysis, those who were excluded because they did not have at least two study visits were significantly more likely to be male (p = 0.03), younger (p < 0.001), non-Black race (p < 0.001), report injection drug use in the last 6 months (p < 0.001), emergency room utilization in the last 6 months (p < 0.001), and less likely to be living with HIV (p = 0.001), but were otherwise similar on baseline disability status, outpatient health care utilization, insurance status, and baseline prevalence of all other diseases included as latent class indicators.

2.2. Measures

2.2.1. Latent class indicators

The study examined data on the presence or absence of chronic diseases commonly treated in the outpatient clinical setting salient to a population of PWID: Anxiety/Depression, Major psychiatric disease (Bipolar disorder and/or Schizophrenia), Diabetes, Hypertension, Arthritis, Obstructive airway disease, HIV, Liver fibrosis/Cirrhosis, and Renal dysfunction. Individuals were classified as having or not ever having each disease based on time-updated clinical, laboratory, and self-reported data collected during follow-up visits using the established criteria we describe in Table 1.

Table 1.

Definitions for chronic disease indicators, covariates associated with class membership, and health events and healthcare utilization.

Disease Definition
Arthritis Self-report arthritis diagnosis or treatment in the last 6 months
Anxiety/depression Self-report anxiety or depression diagnosis/ treatment in the last 6 months, or Centers for Epidemiologic Studies Depression Scale (CES-D) scores of 23+ (Carleton et al., 2013).
Diabetes Self-reported diabetes medication taken in past 6 months, or Hemoglobin A1c (HbA1c) concentration of >6.5
Hypertension Self-reported hypertension medication in the last 6 months, or 2+ study visits with blood pressure readings of systolic ≥140 or diastolic ≥90
HIV Self-reported antiretroviral therapy (ART) in the last 6 months or confirmed HIV antibody test.
Liver fibrosis Transient elastography (FibroScan) measurements of median liver stiffness, classified by cutoffs for F2 fibrosis (kPa ≥ 9.3) (Kirk et al., 2009)
Major psychiatric disorders Self-reported medications for/diagnosis of bipolar disorder, schizophrenia, or schizoaffective disorder.
Obstructive airway Forced expiratory volume in 1 s to forced vital capacity of ≤70%
Renal dysfunction Estimated glomerular filtration rate < 60 mL/min/1.73m2 using MDRD equation.
Covariate Definition (at first study visit)
Age Measured as a continuous variable at the first study visit and transformed into a dichotomous variable near the median age of 55 years old.
Sex Dichotomous variable comparing self-report male vs. female sex.
Race Classified as Black compared to non-Black race.
Income A dichotomous variable comparing individuals with an income less than $5000 dollars per year to income greater than $5000 per year
Disability Measured as a dichotomous variable based on the following question: “Currently, are you receiving SSI (Supplemental Security Income) or SSDI (Social Security Disability Income), or any other type of disability?”
Insurance Self-report health insurance in the last six months
Continuity of care Self-report of healthcare visits with the same healthcare provider for the last 2 years at the first visit
Drug use Self-report of heroin, cocaine, crack, and/or speedball use and route in the last 6 months classified the following categories: Injection only, non-injection only, both injection and non-injection and no substance use (ref)
Alcohol use Self-report of any alcohol use in the last 6 months.
Cigarette use Binary variable for self-report of any cigarette use in the last 6 months.
Health events Definition
Infectious Self-report of diagnosis or treatment for sepsis, pneumonia, endocarditis or skin infection in the last 6 months at a study visit between 2015 and 2019
Cardiovascular Self-report of diagnosis or treatment for myocardial infarction, heart failure, heart disease, or stroke in the last 6 months at a study visit between 2015 and 2019
Healthcare utilization Definition (at last study visit)
Emergency room Self-report utilization of emergency room services in the last six months
Outpatient visit Self-report utilization of outpatient healthcare services in the last 6 months
Medication for opioid use disorder Self-report prescription of buprenorphine, naltrexone, or methadone in the last 6 months
Open in a new tab

2.2.2. Covariates associated with class membership

The study team selected covariates based on characteristics described by the Behavioral Model for Vulnerable Population as related to health care utilization and engagement in three categories: predisposing, enabling and health behaviors, with definitions listed in Table 1.

2.2.3. Health events and health care utilization

Health events were dichotomous variables in two major groups: infectious or cardiovascular. This study classified individuals as having or not ever having each event from each study visit with criteria described in Table 1. We used health care utilization measures from participant’s last visit with definitions described in Table 1.

2.3. Statistical analysis

The research team conducted latent class analysis (LCA) to identify profiles of multimorbidity in Mplus Version 8.1 (Muthen & Muthen, 2011). The Akaike Information Criteria (AIC), Bayesian Information Criteria (BIC), sample size adjusted Bayesian Information Criteria (aBIC), Lo-Mendell-Rubin adjusted likelihood ratio test (LMR-LRT), and Bootstrapped likelihood ratio test identified the optimal number of multimorbid disease classes (Nylund et al., 2007). The study also considered interpretability of the classes based on the conditional probabilities of disease indicators and entropy of class membership in selecting the final class distributions (Celeux & Soromenho, 1996).

Following the identification of the optimal number of classes, the study included predisposing, enabling, and health behavior characteristics as covariates with unadjusted regressions using the R3STEP function in MPlus. This function examines that covariate’s correlation with a multimorbidity class, relative to a reference class. The reference for all models was the class with the lowest percent of multimorbid diseases. Guided by the Behavioral Model, the study selected a priori variables of age, sex, race, insurance, and drug use (defined in Table 1) for the final adjusted model regardless of statistical significance (Gelberg et al., 2000). We selected other enabling and health behavior variables with statistically significant (p < 0.05) unadjusted associations with any class for the final adjusted model. The team applied controls for number of study visits and number of days between first and last visit in the adjusted model.

Finally, we examined the association of multimorbid class on dichotomous outcomes of health care utilization and health events using logistic regressions, controlling for predisposing, enabling, and health behavior characteristics using Stata (StataCorp, 2013).

3. Results

3.1. Demographic characteristics

Of the 1387 individuals included, the majority were male (67%) and Black (81%). At baseline, the mean age was 53 years old (25th–75th percentile: 24–74). Less than one-third reported an annual income greater than $5000 (31%), and most reported receiving disability in the prior 6 months (80%). Three percent of the study sample reported being uninsured in the last 6 months at baseline, with 85% reporting coverage with Medicare or Medicaid. At the first visit in this analysis, 56% of this study population reported substance use in the previous 6 months, with 14% reporting noninjection use only, 37% reporting both injection and noninjection use, and 5% reporting injection drug use exclusively (5%). Finally, at baseline, almost a third of the participants reported using emergency room services in the previous 6 months and three quarters utilizing outpatient health care services (Table 2).

Table 2.

Study sample baselinea characteristics, 1387 ALIVE participants with a study visit in 2015 and 2019.

Total
n = 1387
n %
Predisposing Age, median (range) 53 (27–77)
Female sex 459 33
Black race 1119 81
Enabling Annual income <$5000 959 69
Disability 1105 80
Insurance last 6 months 1349 97
Same provider in the last 2 years 775 56
Health behaviors (last 6 months) Non-injection drug use only 193 14
Injection drug use only 70 5
Both injection and non-injection drug use 512 37
Healthcare utilization (last 6 months) Emergency Department Visit 439 32
Outpatient Clinic Visit 1046 75
Medication for Opioid Use 734 53
Disorderb
Need characteristics HIV 413 30
Diabetes 124 9
Hypertension 826 60
Renal Disease 110 8
Arthritis 465 34
Major Psychiatric 393 28
Comorbidityc
Anxiety/Depression 629 45
Fibrosis (≥ 9.3 kPa)d 182 24
Obstructive Airway Diseasee 152 23
Open in a new tab
a

First study visit in 2015–2019 represents the baseline visit.

b

Methadone, Buprenorphine, or Naltrexone.

c

Bipolar or Schizophrenia.

d

Fibroscan available for only 753 participants at first visit.

e

FEV/FCV available for only 658 participants at first visit.

3.2. Model selection

A comparison of model fit indices indicated that a four-class solution was preferred (Supplemental Table 1). HIV was the only disease indicator that the team dropped from analysis, as its conditional probabilities were similar in each class; model fit and interpretability improved with its exclusion. Additionally, the study identified HIV and viral load status as covariates of class membership, but neither was associated with class membership. Information criteria showed a split between a 3 and 4 class solution, with the BIC lower for the former and AIC lower for the latter. Additionally, the bootstrapped LMR/LTR indicated a four-class solution significantly improved fit compared to three classes (Nylund et al., 2007). A four-class solution and its interpretability highlighted the following patterns of multimorbidity: Low multimorbidity (54%), Low multimorbidity including psychiatric comorbidity (26%), Multimorbidity (12%), and Multimorbidity including psychiatric comorbidity (7%).

3.3. Multimorbidity classes

Table 3 lists the conditional probabilities of each disease in the four multimorbid classes, the probability of being diagnosed with each disease given inclusion in that class, converted to percentages. The four classes had similar conditional percentages of obstructive airway disease (24–38%) and hypertension (96–100%). Within classes, Low multimorbidity and Low multimorbidity including psychiatric comorbidity had similar lower conditional percentages of diabetes, liver fibrosis, renal dysfunction, and arthritis compared to the Multimorbidity and Multimorbidity including psychiatric comorbidity classes, which had high conditional percentage of these nonpsychiatric diseases. Between classes, psychiatric comorbidity distinguished these classes, with high conditional percentage of Anxiety/Depression and Major Psychiatric Disorders in the Low multimorbidity including psychiatric comorbidity and Multimorbidity including psychiatric comorbidity classes compared to their counterparts.

Table 3.

Latent classes of multimorbidity classified by individual disease indicators, n = 1387 ALIVE participants from 2015 to 2019.

Class 1 Class 2 Class 3 Class 4
n = 745 (54%) n = 368 (26%) n = 172 (12%) n = 102 (7%)
Low multimorbidity Low multimorbidity including psychiatric comorbidity Multimorbidity Multimorbidity including psychiatric comorbidity
Arthritis 21% 31% 49% 48%
Major psychiatric disorders (bipolar & schizophrenia) 2% 100% 0% 100%
Anxiety/depression 58% 100% 61% 100%
Diabetes 7% 0% 75% 100%
Liver fibrosis 12% 14% 38% 28%
Hypertension 96% 97% 98% 100%
Obstructive airway 38% 34% 24% 29%
Renal dysfunction 16% 17% 29% 32%
Open in a new tab

Bolded are conditional prevalences that are noteworthy for that class.

3.4. Baseline correlates of class membership

Table 4 shows the correlates of class membership relative to class 1, Low multimorbidity. From the predisposing domain, sex and age were significant covariates. Female sex was significantly associated with membership in all three classes, with highest adjusted odds of membership in the Multimorbidity including psychiatric comorbidity class (aOR 4.2; 95% CI 2.7–6.6). Participants older than 55 years of age had 70% lower adjusted odds (aOR 0.3; 95% CI 0.2–0.5) of being in the Low multimorbidity including psychiatric comorbidity class than the Low multimorbidity class. Baseline disability, income, and insurance, all enabling characteristics, were associated with class membership. Baseline disability was associated with all three classes but remained statistically significant only for the two psychiatric comorbidity classes in the adjusted model, and strongest for the Multimorbidity including psychiatric comorbidity class (aOR 2.2; 95% CI 1.4–3.6). Participants reporting an annual income less than $5000 at baseline had 60% lower adjusted odds of membership in the Multimorbidity class (aOR 0.4; 95% CI 0.3–0.8). Baseline insurance was associated with nearly 4 times higher adjusted odds of membership in the Low multimorbidity including psychiatric comorbidity class (aOR 3.8; 95% CI 1.1–13.3). Health behaviors of substance use and its association with class membership varied. Reporting baseline noninjection and injection- and noninjection drug use was significantly associated with over two-fold higher adjusted odds of membership in the Low multimorbidity including psychiatric comorbidity class. Additionally, reporting injection- and noninjection drug use was significantly associated with 60% lower odds of membership in the Multimorbidity class in adjusted analysis (aOR 0.4; 95% CI 0.2–0.9). Baseline alcohol use was associated with 50% lower adjusted odds of membership in the Multimorbidity with psychiatric comorbidity class (aOR 0.5; 95% CI 0.3–0.8) while baseline cigarette use was associated with 50% lower odds of membership in the Multimorbidity class in adjusted analysis (aOR 0.5; 95% CI 0.2–0.9).

Table 4.

Estimates of class membership with baseline characteristics (relative to low multimorbidity class), n = 1387 ALIVE participants from 2015 to 2019.

Class Low multimorbidity with psychiatric comorbidity Multimorbidity Multimorbidity including psychiatric comorbidity
Covariate OR (95% CI) aORa (95% CI) OR (95% CI) aORa (95% CI) OR (95% CI) aORa (95% CI)
Predisposing characteristics
Age > 55 years 0.5 (0.3–0.7) 0.3 (0.2–0.5) 2.3 (1.3–4.1) 1.1 (0.6–2.1) 0.9 (0.6–1.4) 0.8 (0.5–1.3)
Female 6.2 (3.8–10.7) 2.6 (1.9–3.5) 2.1 (1.1–4.1) 2.7 (1.5–4.7) 2.5 (1.6–3.7) 4.2 (2.7–6.6)
Black race 0.7 (0.5–1.0) 0.8 (0.5–1.1) 20.5 (0.5–851.5) 4.0 (0.9–16.5) 1.9 (0.9–3.5) 1.5 (0.7–3.1)
Baseline enabling variables
Income <$5000/year 0.9 (0.7–1.3) 0.8 (0.6–1.2) 0.4 (0.2–0.6) 0.4 (0.3–0.8) 0.9 (0.5–1.4) 0.9 (0.5–1.5)
Disability 1.6 (1.2–2.1) 1.9 (1.4–2.6) 1.8 (1.1–2.9) 1.4 (0.8–2.6) 2.4 (1.5–3.8) 2.2 (1.4–3.6)
Insurance 4.5 (1.2–17.0) 3.8 (1.1–13.3) 0.7 (0.2–2.5) 0.5 (0.1–2.1) 3.5 (0.4–27.1) 3.2 (0.5–23.0)
Continuity of care 1.3 (0.9–1.7) 1.3 (0.9–1.7) 1.9 (1.1–3.1) 1.0 (0.5–1.8) 1.6 (1.0–2.6) 1.1 (0.6–1.7)
Baseline health behaviors: substance use in the last 6 months
Injection only 0.7 (0.4–1.4) 1.6 (0.8–3.1) 0.7 (0.2–2.2) 0.7 (0.2–2.5) 0.2 (0.02–1.2) 0.2 (0.03–2.0)
Non-injection only 1.7 (1.2–2.4) 2.5 (1.6–3.9) 0.5 (0.2–1.3) 0.4 (0.2–1.2) 1.4 (0.8–2.5) 1.5 (0.8–3.0)
Injection and non-injection 1.6 (1.2–2.1) 2.3 (1.6–3.3) 0.2 (0.1–0.5) 0.4 (0.2–0.9) 0.9 (0.6–1.4) 1.6 (0.9–2.9)
Alcohol use 1.1 (0.8–1.4) 0.9 (0.6–1.2) 0.6 (0.3–0.9) 0.9 (0.5–1.7) 0.5 (0.4–0.8) 0.5 (0.3–0.8)
Cigarette use 1.4 (0.9–2.1) 1.1 (0.7–1.7) 0.3 (0.2–0.5) 0.5 (0.2–0.9) 1.0 (0.5–1.7) 1.0 (0.5–1.9)
Open in a new tab

Boldface indicates statistical significance (p<0.05).

a

Controlling for number of visits and number of days between first and last visit.

3.5. Differences in health care utilization and health events

Table 5 displays odds ratios of associations between class membership and health care utilization and health event outcomes, using the Low multimorbidity class as the reference and controlling for baseline age, sex, race, income, disability, care continuity, substance use, number of study visit, and number of days between first and last visit in the adjusted logistic regression model. Compared to those in the Low multimorbidity class, persons in the Multimorbidity and Multimorbidity including psychiatric comorbidity classes had significantly higher odds of an infectious disease health event, but only the later class remained statistically significant in the adjusted model (aOR 2.1; 95% CI 1.4–3.3). Both Multimorbidity (OR 2.0; 95% CI 1.2–3.5) and Multimorbidity including psychiatric comorbidity (OR 2.1; 95% CI 1.1–4.1) had significantly higher odds of a cardiovascular health event, but neither were statistically significant in the adjusted model. Both of these classes had approximately 2 times significantly higher adjusted odds of inpatient hospitalization in the last 6 months compared to the Low multimorbidity group. All three classes had significantly higher odds of emergency room and outpatient health care visits in the last 6 months compared to the Low multimorbidity class, with the Multimorbidity having the highest adjusted odds of outpatient (aOR 2.0; 95% CI 1.2–3.3) and emergency room (aOR 2.0; 95% CI 1.3–2.8) utilization. Finally, both classes with psychiatric comorbidity had significantly higher odds of MOUD prescription in the last 6 months, with the Multimorbidity including psychiatric comorbidity having 2 times higher adjusted odds and the Low multimorbidity including psychiatric comorbidity having 1.4 times higher adjusted odds compared to the Low multimorbidity class.

Table 5.

Association between class membership on healthcare utilization and health events relative to Low Multimorbidity class, n = 1384 ALIVE participants from 2015 to 2019.

Healthcare utilization
 Health events
Medication for opioid use disordera Outpatient visit Emergency room visit Inpatient stay Cardiovascularb Infectiousc
OR (95% CI) aORd (95% CI) OR (95% CI) aORd (95% CI) OR (95% CI) aORd (95% CI) OR (95% CI) aORd (95% CI) OR (95% CI) aORd (95% CI) OR (95% CI) aORd (95% CI)
Low multimorbidity Psychiatric comorbidity 1.9 (1.5–2.4) 1.4 (1.1–1.9) 1.4 (1.1–1.9) 1.6 (1.2–2.2) 1.6 (1.2–2.1) 1.5 (1.1–2.1) 1.4 (0.9–1.9) 1.3 (0.9–1.9) 0.8 (0.4–1.4) 0.9 (0.5–1.7) 1.6 (1.2–2.0) 1.3 (0.9–1.7)
Multimorbidity 0.7 (0.5–1.1) 1.0 (0.7–1.4) 2.6 (1.6–4.3) 2.0 (1.2–3.3) 1.8 (1.3–2.6) 2.0 (1.3–2.8) 1.8 (1.2–2.8) 1.9 (1.2–2.9) 2.0 (1.2–3.5) 1.7 (0.9–2.9) 1.1 (0.7–1.5) 1.2 (0.8–1.7)
Multimorbidity Psychiatric comorbidity 2.1 (1.3–3.3) 2.1 (1.3–3.5) 2.2 (1.2–3.9) 1.8 (1.0–3.4) 1.8 (1.3–2.7) 1.7 (1.1–2.8) 2.3 (1.4–3.8) 2.2 (1.3–3.7) 2.1 (1.1–4.1) 1.9 (0.9–3.8) 2.5 (1.6–3.8) 2.1 (1.4–3.3)
Open in a new tab
a

Medication for Opioid Use Disorder (MOUD) includes prescription of buprenorphine, naltrexone, or methadone in the last 6 months of the participant’s last visit.

b

Cardiovascular events include self-report myocardial infraction, heart failure, heart disease, or stroke between 2015 and 2019.

c

Infectious events include self-report sepsis, pneumonia, endocarditis, or skin infection between 2015 and 2019.

d

Controlling for baseline age, sex, race, income, disability, insurance, same provider>90% of visits, substance use, number of visits, and number of days between first and last visit.

4. Discussion

This study provides new evidence regarding health care engagement for PWID. Using the Behavioral Model for Vulnerable populations, we identified the need characteristics with unique classes of multimorbidity; the predisposing, enabling, and health behavior characteristics associated with each class; and impact of these need-based multimorbidity classes on health care engagement. From predisposing characteristics, we observed higher levels of multimorbidity among women and from the enabling characteristics, those receiving disability were significantly associated with a psychiatric comorbidity class. Moreover, the data suggest that high levels of multimorbidity are associated with greater health care utilization and adverse health events, including some driven by psychiatric comorbidity. Our findings reinforce the complex nature of health care engagement for PWID, and through the lens of the need characteristics, identify avenues for clinical interventions based on multimorbidity class, especially given the high burden of psychiatric comorbidity observed.

Multiple studies have examined classes of multimorbidity in the general aging population using latent class analysis (Barile et al., 2015; Cigolle et al., 2011; Gonsoulin et al., 2017; Hall et al., 2018; Islam et al., 2014; Juul-Larsen et al., 2020; Khorrami et al., 2020; Kuwornu et al., 2014; Larsen et al., 2017; Olaya et al., 2017; Park et al., 2019; Sibley et al., 2014; Whitson et al., 2016). While those studies differ in their selection and measurement of diseases, their findings, like ours, demonstrated heterogeneity in multimorbidity classes. One study in particular (Barile et al., 2015) found a similar four class structure, but with more cardiometabolic, gastrointestinal, and musculoskeletal diagnoses considered. A systematic review of multimorbidity profiles using different clustering approaches found in its synthesis of latent class analysis studies two interpretable clusters, one of mental health disorders and another of asthma and COPD; the review also saw an additional discernable axis of cardiometabolic diseases (Busija et al., 2019). Our findings reinforce heterogeneity in multimorbidity clustering, particularly as it relates to clustering around mental health disorders. Our findings expand on this work by examining multimorbidity in an understudied population of PWID, and by examining the relationship between multimorbidity and health care engagement and utilization.

We found a high burden of psychiatric comorbidity, with two named classes based on this finding. While measured inconsistently across studies, seven studies had a distinct class with a psychiatric comorbidity (Barile et al., 2015; Islam et al., 2014; Juul-Larsen et al., 2020; Kuwornu et al., 2014; Larsen et al., 2017; Olaya et al., 2017; Park et al., 2019). Five of these studies found depression and/or anxiety in a distinct class, or comorbid with noncardiometabolic disease (e.g., allergies, headache, asthma) similar to the Low multimorbidity including psychiatric comorbidity class (Islam et al., 2014; Juul-Larsen et al., 2020; Kuwornu et al., 2014; Larsen et al., 2017; Park et al., 2019). Olaya et al. (2017) identified a class similar to the Multimorbidity including psychiatric comorbidity class with high conditional percentage of depression with cardiorespiratory and arthritis multimorbidity. Only one study (Barile et al., 2015) found two classes with psychiatric comorbidity, both of which were similar to the psychiatric comorbidity classes in this study. Both Low multimorbidity including psychiatric comorbidity and Multimorbidity including psychiatric comorbidity classes also included high conditional prevalence of major psychic disorder (bipolar disorder and/or schizophrenia), which were not included as indicators in any other studies. These two classes also had much higher conditional prevalence of anxiety/depression compared to similar classes in other studies. Conditional percentage for anxiety/depression indicators in other studies ranged from 17 to 70% (Islam et al., 2014; Juul-Larsen et al., 2020; Kuwornu et al., 2014; Larsen et al., 2017; Park et al., 2019), which was similar to the conditional percentages seen in the Low multimorbidity and Multimorbidity classes, indicating a high burden of psychiatric comorbidity for PWID. Additionally, this identification of distinct classes of psychiatric comorbidity had implications for health care utilization, as well as for areas for clinicians to target when engaging this population in outpatient care.

While all three classes had significantly higher odds of health care utilization compared to the Low multimorbidity class, both psychiatric comorbidity classes also had high significant adjusted odds of receipt of MOUD. Other studies with PWID have demonstrated co-location of MOUD with primary and outpatient care as a factor correlated with engagement (Islam et al., 2013; McNeil et al., 2014; Mehta et al., 2015; Morozova et al., 2017). Recent evidence from the ALIVE Study also demonstrated greater use of MOUD among those seeking treatment for mental health conditions (Genberg et al., 2019). Finally, these psychiatric comorbidity classes also had high odds of emergency room utilization, and the Multimorbidity including psychiatric comorbidity class had high significant adjusted odds of an inpatient hospitalization in the last 6 months compared to the Low multimorbidity class. While some of these utilization measures were also true of the Multimorbidity class, the psychiatric diagnoses may play a salient role as one study determined severity of mental health disorders significantly associated with increased utilization of emergency room services (Niedzwiecki et al., 2018) with another systematic review showing that attempted suicide or self-harm were potential drivers of this utilization (Barratt et al., 2016). Research should prioritize closer follow-up of individuals in this class with integrated mental health and other outpatient health care services, including wrap-around case management.

The Low multimorbidity class was the largest of the four described here. Multiple studies have demonstrated a class with low conditional percentage of diseases compared to the other classes (Islam et al., 2014; Larsen et al., 2017; Olaya et al., 2017; Park et al., 2019; Whitson et al., 2016). However, our Low multimorbidity class also had a high conditional percentage of hypertension (96%) compared to similar Low multimorbidity classes in these studies. Three studies (Kuwornu et al., 2014; Larsen et al., 2017; Sibley et al., 2014) found an additional class characterized by a high conditional percentage of hypertension alone, indicating a higher burden of hypertension for PWID, highlighting an unmet need that could be addressed with greater primary care engagement. The Multimorbidity class was characterized by high probabilities of all nonpsychiatric diseases. Though studies with geriatric populations had similar classes with high conditional percentage of all diseases studied (Hall et al., 2018; Juul-Larsen et al., 2020; Whitson et al., 2016), others had classes that were characterized by high conditional percentage of cardio-metabolic disease (i.e. hypertension, diabetes, hyperlipidemia, vascular disease, etc.) in a similar manner to our Multimorbidity class (Gonsoulin et al., 2017; Islam et al., 2014; Khorrami et al., 2020; Kuwornu et al., 2014; Larsen et al., 2017; Olaya et al., 2017; Park et al., 2019; Sibley et al., 2014). Future work with this population should consider additional comorbidities, particularly cardio-metabolic diseases, to determine if identifiable classes existed based on disease system for targeted clinical interventions.

Our findings also highlighted other Behavioral Model characteristics that impact health care engagement. Receipt of disability was an important enabling characteristic significantly associated with both classes with psychiatric comorbidity. This association has been well documented. One study found that individuals with mental health disorders represented a large component of disability services, and that more than 50% of individuals require multiple applications to Supplemental Security Income (SSI) and Social Security Disability Insurance (SSDI) to access these essential social services (Goldman et al., 2018). Some programs have assisted those who have mental health and other co-occurring disorders with these applications (Dennis et al., 2011). These outreach services can be initial touchpoints and can also perform linkage to essential health care services and other enabling factors like health insurance. From health behaviors, individuals reporting injection and noninjection drug use at the baseline visit in our study were significantly less likely to be in the Multimorbidity class. Previous studies within the ALIVE cohort also demonstrated the decreased odds of injection drug use with increased multimorbidity (Salter et al., 2011). Additionally, the decreased odds of injecting for this group could be related to the increased utilization of outpatient care or may reflect the impact of comorbid disease, though prior studies have only found associations with active injection drug use and lower health care utilization (Maragh-Bass et al., 2017). From the predisposing domain, while female sex was associated with membership in all three classes, both classes with psychiatric comorbidity had higher effect estimates of this association, with female sex having over 4 times higher adjusted odds of membership in Multimorbidity including psychiatric morbidity class compared to the Low Multimorbidity class. This relationship between sex and diagnosis of a psychiatric condition is consistent with national reporting of women having higher prevalence of any mental illness, depression/anxiety, and severe mental illness (Bose et al., 2018). This finding warrants further study to assess additional barriers to health care engagement, while accounting for multimorbid disease management. This study did not collect and could not assess experiences unique to women and gender non-conforming people, including gender-based violence; such data would inform future clinical assessments and primary care priorities (Muncan et al., 2020; Shirley-Beavan et al., 2020).

4.1. Limitations

This study has limitations. While the study measured diseases over a five-year period and most with a physiological or biological marker to confirm diagnosis, some disease states were classified by self-report alone. This is a particular limitation with severe psychiatric disorders, though the study based classification on self-report of diagnosis or prescription of a medication. An overlap also existed with measurement of diseases over the five-year period and the health care utilization and health event data, limiting the temporality of associations. However, this study’s purpose was to determine the holistic needs of an aging population of PWID and not infer causality of multimorbidity class on outcomes. Our study’s restriction to participants with two study visits between 2015 and 2019 limited the sample size, though sensitivity analysis with all participants during the time period showed a similar 4 class structure of multimorbidity. Additionally, participants excluded from this study due to not having a second study visit between 2015 and 2019 were significantly more likely to report injection drug use at baseline in the sensitivity analysis (p < 0.001), a potential factor related to being lost to follow up. Because these data came from an aging population of largely Black, male, urban dwellers, its findings may not be generalizable to the larger population of people who inject drugs. Future studies should attempt to classify based on severity of disease states and utilize longitudinal data to determine transitions of multimorbid disease management. Finally, we dropped HIV as an indicator given its conditional probability being similar in all four classes. This finding suggests potential effect measure modification and that future work should determine the burden of multimorbidity stratified by HIV status.

5. Conclusions

As current or former PWID seek health care, their needs extend beyond the traditional realms of substance use. This study demonstrates heterogeneity in multimorbidity for PWID with distinct latent classes. Three of these classes of multimorbidity indicate greater health care need and the need for interventions to improve engagement. Additionally, two classes had enhanced need for care because of co-occurring mental health disorders. Clinicians can capitalize on the implications for each class, tailoring follow-up and care services to best address the needs of multimorbidity classes. Holistic health care systems can address these needs best with integrated care systems that provide harm reduction, substance use and mental health treatment together, and wrap around social services, particularly as this population continues to age.

Supplementary Material

Supplemental Table

Acknowledgements

We thank the participants and staff of the ALIVE study for their contributions to this research.

Financial support

This work was supported by the National Institutes of Health (grants DA036297 and T32DA007292 Maher and Johnson MPI to O.H.).

Footnotes

Declaration of competing interest

The authors report no real or perceived vested interests related to this article that could be construed as a conflict of interest.

CRediT authorship contribution statement

Omeid Heidari: conceptualization, methodology, formal analysis, data curation, writing – original draft; Becky L. Genberg: conceptualization, writing – review & editing, funding acquisition; Nancy Perrin: conceptualization, methodology, writing – review & editing; Derek T. Dangerfield: methodology, writing – review & editing; Jason E Farley: methodology, writing – review & editing; Gregory Kirk: conceptualization, writing – review & editing, funding acquisition; Shruti H Mehta: conceptualization, writing – review & editing, funding acquisition.

Supplementary data to this article can be found online at https://doi.org/10.1016/j.jsat.2022.108806.

References

  1. Andersen RM (1968). Families use of health services: A behavioral model of predisposing, enabling, and need components. Center for Health Administration Studies, University of Chicago. [Google Scholar]
  2. Artenie AA, Jutras-Aswad D, Roy E, Zang G, Bamvita J-M, Levesque A, & Bruneau J (2015). Visits to primary care physicians among persons who inject drugs at high risk of hepatitis C virus infection: Room for improvement. Journal of Viral Hepatitis, 22(10), 792–799. 10.1111/jvh.12393 [DOI] [PubMed] [Google Scholar]
  3. Barile JP, Mitchell SA, Thompson WW, Zack MM, Reeve BB, Cella D, & Smith AW (2015). Patterns of chronic conditions and their associations with behaviors and quality of life, 2010. Preventing Chronic Disease, 12, E222. 10.5888/pcd12.150179 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Barratt H, Rojas-Garcia A, Clarke K, Moore A, Whittington C, Stockton S, Thomas J, Pilling S, & Raine R (2016). Epidemiology of mental health attendances at emergency departments: Systematic review and meta-analysis. PloS One, 11(4), Article e0154449. 10.1371/journal.pone.0154449 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bose J, Hedden SL, Lipari RN, & Park-Lee E (2018). Key substance use and mental health indicators in the United States: Results from the 2015 National Survey on Drug Use and Health. Publication No. SMA 16-4984, NSDUH Series H-51. In Key substance use and mental health indicators in the United States: Results from the 2015 National Survey on Drug Use and Health. Publication No. SMA 16-4984, NSDUH Series H-51 (pp. 1–97). Substance Abuse and Mental Health Services Administration (SAMHSA). 10.1016/j.drugalcdep.2016.10.042. [DOI] [Google Scholar]
  6. Busija L, Lim K, Szoeke C, Sanders KM, & McCabe MP (2019). Do replicable profiles of multimorbidity exist? Systematic review and synthesis. European Journal of Epidemiology, 34(11), 1025–1053. 10.1007/s10654-019-00568-5 [DOI] [PubMed] [Google Scholar]
  7. Carleton R, Thibodeau M, Teale M, Welch P, Abrams M, Robinson T, & Asmundson G (2013). The center for epidemiologic studies depression scale: a review with a theoretical and empirical examination of item content and factor structure. PloS One, 8(3). 10.1371/journal.pone.0058067 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Celeux G, & Soromenho G (1996). An entropy criterion for assessing the number of clusters in a mixture model. Journal of Classification, 13, 195–212. [Google Scholar]
  9. Centers for Disease Control and Prevention. (2012). Integrated prevention services for HIV infection, viral hepatitis, sexually transmitted diseases, and tuberculosis for persons who use drugs illicitly: Summary guidance from CDC and the U.S. Department of Health and Human Services. MMWR, Morbidity & Mortality Weekly Report, 61(RR-5), 1–40. http://www.ncbi.nlm.nih.gov/pubmed/23135062. [PubMed] [Google Scholar]
  10. Chi FW, Parthasarathy S, Mertens JR, & Weisner CM (2011). Continuing care and long-term substance use outcomes in managed care: Early evidence for a primary care-based model. Psychiatric Services (Washington, D.C.), 62(10), 1194–1200. 10.1176/ps.62.10.pss6210_1194 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Cigolle CT, Lee PG, Langa KM, Lee Y-Y, Tian Z, & Blaum CS (2011). Geriatric conditions develop in middle-aged adults with diabetes. Journal of General Internal Medicine, 26(3), 272–279. 10.1007/s11606-010-1510-y [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Cullen W, O’Brien S, O’Carroll A, O’Kelly FD, & Bury G (2009). Chronic illness and multimorbidity among problem drug users: A comparative cross sectional pilot study in primary care. BMC Family Practice, 10, 1–10. 10.1186/1471-2296-10-25 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Dennis D, Lassiter M, Connelly WH, & Lupfer KS (2011). Helping adults who are homeless gain disability benefits: The SSI/SSDI outreach, access, and recovery (SOAR) program. Psychiatric Services, 62(11), 1373–1376. 10.1176/ps.62.11.pss6211_1373 [DOI] [PubMed] [Google Scholar]
  14. Gelberg L, Andersen RM, & Leake BD (2000). Healthcare access and utilization the behavioral model for vulnerable populations: Application to medical care use and outcomes for homeless people. Healthcare Access and Utilization, 34(6), 1273–1302. [PMC free article] [PubMed] [Google Scholar]
  15. Genberg BL, Astemborski J, Treisman G, Anagnostopoulos A, Mehta SH, Kirk GD, & Abraham A (2019). Engagement in treatment for depression among people who inject drugs in Baltimore, Maryland. Journal of Substance Abuse Treatment, 106, 107–112. 10.1016/j.jsat.2019.09.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Goldman HH, Frey WD, & Riley JK (2018). Social security and disability due to mental impairment in adults. Annual Review of Clinical Psychology, 14(1), 453–469. 10.1146/annurev-clinpsy-050817-084754 [DOI] [PubMed] [Google Scholar]
  17. Gonsoulin ME, Durazo-Arvizu RA, Goldstein KM, Cao G, Zhang Q, Ramanathan D, & Hynes DM (2017). A health profile of senior-aged women veterans: A latent class analysis of condition clusters. InnovationinAging, 1(2). 10.1093/geroni/igx024 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Hall M, Dondo TB, Yan AT, Mamas MA, Timmis AD, Deanfield JE, Jernberg T, Hemingway H, Fox KAA, & Gale CP (2018). Multimorbidity and survival for patients with acute myocardial infarction in England and Wales: Latent class analysis of a nationwide population-based cohort. PLoS Medicine, 15(3), 1–18. 10.1371/journal.pmed.1002501 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Hartzler B, Dombrowski JC, Williams JR, Crane HM, Eron JJ, Geng EH, Mathews C, Mayer KH, Moore RD, Mugavero MJ, Napravnik S, Rodriguez B, & Donovan DM (2018). Influence of substance use disorders on 2-year HIV care retention in the United States. AIDS and Behavior, 22(3), 742–751. 10.1007/s10461-017-1826-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Hoots BE, Finlayson TJ, Broz D, & Paz-Bailey G (2017). Antiretroviral therapy use among HIV-infected people who inject drugs-20 cities, United States, 2009–2015. Journal of Acquired Immune Deficiency Syndromes, 75(Suppl. 3), S392–s396. 10.1097/qai.0000000000001416 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Islam MM, Topp L, Conigrave KM, White A, Haber PS, & Day CA (2013). Are primary health care centres that target injecting drug users attracting and serving the clients they are designed for? A case study from Sydney, Australia. The International Journal on Drug Policy, 24(4), 326–332. 10.1016/j.drugpo.2012.06.002 [DOI] [PubMed] [Google Scholar]
  22. Islam MM, Valderas JM, Yen L, Dawda P, Jowsey T, & McRae IS (2014). Multimorbidity and comorbidity of chronic diseases among the senior Australians: Prevalence and patterns. PloS One, 9(1), Article e83783. 10.1371/journal.pone.0083783 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Johnston MC, Crilly M, Black C, Prescott GJ, & Mercer SW (2019). Defining and measuring multimorbidity: A systematic review of systematic reviews. European Journal of Public Health, 29(1), 182–189. 10.1093/eurpub/cky098 [DOI] [PubMed] [Google Scholar]
  24. Juul-Larsen HG, Christensen LD, Bandholm T, Andersen O, Kallemose T, Jørgensen LM, & Petersen J (2020). Patterns of multimorbidity and differences in healthcare utilization and complexity among acutely hospitalized medical patients (≥65 years) - A latent class approach. Clinical Epidemiology, 12, 245–259. 10.2147/CLEP.S226586 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Khorrami Z, Rezapour M, Etemad K, Yarahmadi S, Khodakarim S, Mahdavi Hezaveh A, Kameli M, & Khanjani N (2020). The patterns of non-communicable disease multimorbidity in Iran: A multilevel analysis. Scientific Reports, 10(1), 3034. 10.1038/s41598-020-59668-y [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Krupski A, West II, Graves MC, Atkins DC, Maynard C, Bumgardner K, Donovan D, Ries R, & Roy-Byrne P (2015). Clinical needs of patients with problem drug use. The Journal of the American Board of Family Medicine, 28(5), 605–616. 10.3122/jabfm.2015.05.150004 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Kuwornu JP, Lix LM, & Shooshtari S (2014). Multimorbidity disease clusters in aboriginal and non-aboriginal Caucasian populations in Canada. Chronic Diseases and Injuries in Canada, 34(4), 218–225. [PubMed] [Google Scholar]
  28. Larney S, Peacock A, Leung J, Colledge S, Hickman M, Vickerman P, Grebely J, Dumchev KV, Griffiths P, Hines L, Cunningham EB, Mattick RP, Lynskey M, Marsden J, Strang J, & Degenhardt L (2017). Global, regional, and country-level coverage of interventions to prevent and manage HIV and hepatitis C among people who inject drugs: A systematic review. The Lancet Global Health, 5(12), e1208–e1220. 10.1016/S2214-109X(17)30373-X [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Larsen FB, Pedersen MH, Friis K, Glumer C, & Lasgaard M (2017). A latent class analysis of multimorbidity and the relationship to socio-demographic factors and health-related quality of life. A national population-based study of 162,283 Danish adults. PloS One, 12(1), Article e0169426. 10.1371/journal.pone.0169426 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Lesko CR, Keil AP, Fojo AT, Chander G, Lau B, & Moore RD (2019). Recent substance use and probability of unsuppressed HIV viral load among persons on antiretroviral therapy in continuity care. American Journal of Epidemiology, 188(10), 1830–1837. 10.1093/aje/kwz159 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Low AJ, Mburu G, Welton NJ, May MT, Davies CF, French C, Turner KM, Looker KJ, Christensen H, McLean S, Rhodes T, Platt L, Hickman M, Guise A, & Vickerman P (2016). Impact of opioid substitution therapy on antiretroviral therapy outcomes: A systematic review and meta-analysis. Clinical Infectious Diseases, 63(8), 1094–1104. 10.1093/cid/ciw416 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Maragh-Bass AC, Powell C, Park J, Flynn C, & German D (2017). Sociodemographic and access-related correlates of health-care utilization among African American injection drug users: The BESURE study. Journal of Ethnicity in Substance Abuse, 16(3), 344–362. 10.1080/15332640.2016.1196629 [DOI] [PubMed] [Google Scholar]
  33. McNeil R, Small W, Wood E, & Kerr T (2014). Hospitals as a “risk environment”: An ethno-epidemiological study of voluntary and involuntary discharge from hospital against medical advice among people who inject drugs. Social Science & Medicine, 1982(105), 59–66. 10.1016/j.socscimed.2014.01.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Mehta SH, Lucas GM, Solomon S, Srikrishnan AK, McFall AM, Dhingra N, Nandagopal P, Kumar MS, Celentano DD, & Solomon SS (2015). HIV care continuum among men who have sex with men and persons who inject drugs in India: Barriers to successful engagement. Clinical Infectious Diseases: An Official Publication of the Infectious Diseases Society of America, 61(11), 1732–1741. 10.1093/cid/civ669 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Miller WC, Hoffman IF, Hanscom BS, Ha TV, Dumchev K, Djoerban Z, Rose SM, Latkin CA, Metzger DS, Lancaster KE, Go VF, Dvoriak S, Mollan KR, Reifeis SA, Piwowar-Manning EM, Richardson P, Hudgens MG, Hamilton EL, Sugarman J Burns DN, … (2018). A scalable, integrated intervention to engage people who inject drugs in HIV care and medication-assisted treatment (HPTN 074): A randomised, controlled phase 3 feasibility and efficacy study. The Lancet, 392(10149), 747–759. 10.1016/S0140-6736(18)31487-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Mizuno Y, Purcell DW, Knowlton AR, Wilkinson JD, Gourevitch MN, & Knight KR (2015). Syndemic vulnerability, sexual and injection risk behaviors, and HIV continuum of care outcomes in HIV-positive injection drug users. AIDS and Behavior, 19(4), 684–693. 10.1007/s10461-014-0890-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Monroe AK, Chander G, & Moore RD (2011). Control of medical comorbidities in individuals with HIV. Journal of Acquired Immune Deficiency Syndromes, 58(5), 458–462. 10.1097/QAI.0b013e31823801c4 (1999). [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Morozova O, Dvoriak S, Pykalo I, & Altice FL (2017). Primary healthcare-based integrated care with opioid agonist treatment: First experience from Ukraine. Drug and Alcohol Dependence, 173, 132–138. 10.1016/j.drugalcdep.2016.12.025 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Muncan B, Walters SM, Ezell J, & Ompad DC (2020). “They look at us like junkies”: Influences of drug use stigma on the healthcare engagement of people who inject drugs in New York City. Harm Reduction Journal, 17(1), 53. 10.1186/s12954-020-00399-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Muthen LK, & Muthen BO (2011). Mplus user’s guide (6th ed.). Muthen & Muthen. [Google Scholar]
  41. Niedzwiecki MJ, Sharma PJ, Kanzaria HK, McConville S, & Hsia RY (2018). Factors associated with emergency department use by patients with and without mental health diagnoses. JAMA Network Open, 1(6), Article e183528. 10.1001/jamanetworkopen.2018.3528 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Nylund K, Asparouhov T, & Muthen B (2007). Deciding on the number of classes in latent class analysis and growth mixture modeling: A Monte Carlo simulation study. Structural Equation Modeling: A Multidisciplinary Journal, 4, 535–569. [Google Scholar]
  43. Okeke NL, Ostermann J, & Thielman NM (2015). In, December 2012. Enhancing linkage and retention in HIV care: A review of interventions for highly resourced and resource-poor settings (pp. 376–392). 10.1007/s11904-014-0233-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Olaya B, Moneta MV, Caballero FF, Tyrovolas S, Bayes I, Ayuso-Mateos JL, & Haro JM (2017). Latent class analysis of multimorbidity patterns and associated outcomes in Spanish older adults: A prospective cohort study. BMC Geriatrics, 17(1), 186. 10.1186/s12877-017-0586-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Park B, Lee HA, & Park H (2019). Use of latent class analysis to identify multimorbidity patterns and associated factors in Korean adults aged 50 years and older. PloS One, 14(11), Article e0216259. 10.1371/journal.pone.0216259 [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Park TW, Cheng DM, Samet JH, Winter MR, & Saitz R (2015). Chronic care management for substance dependence in primary care among patients with co-occurring disorders. Psychiatric Services (Washington, D.C.), 66(1), 72–79. 10.1176/appi.ps.201300414 [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Parthasarathy S, Chi FW, Mertens JR, & Weisner C (2012). The role of continuing care in 9-year cost trajectories of patients with intakes into an outpatient alcohol and drug treatment program. Medical Care, 50(6), 540–546. 10.1097/MLR.0b013e318245a66b [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Piggott DA, Muzaale AD, Mehta SH, Brown TT, Patel KV, Leng SX, & Kirk GD (2013). Frailty, HIV infection, and mortality in an aging cohort of injection drug users. PLoS ONE, 8(1). 10.1371/journal.pone.0054910 [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Piggott DA, Muzaale AD, Varadhan R, Mehta SH, Westergaard RP, Brown TT, Patel KV, Walston JD, Leng SX, & Kirk GD (2016). Frailty and cause-specific hospitalization among persons aging with HIV infection and injection drug use. Journals of Gerontology - Series A Biological Sciences and Medical Sciences, 72 (3), 389–394. 10.1093/gerona/glw142 [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Salter ML, Lau B, Go VF, Mehta SH, & Kirk GD (2011). HIV infection, immune suppression, and uncontrolled viremia are associated with increased multimorbidity among aging injection drug users. Clinical Infectious Diseases, 53(12), 1256–1264. 10.1093/cid/cir673 [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Shirley-Beavan S, Roig A, Burke-Shyne N, Daniels C, & Csak R (2020). Women and barriers to harm reduction services: A literature review and initial findings from a qualitative study in Barcelona, Spain. Harm Reduction Journal, 17(1), 78. 10.1186/s12954-020-00429-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Sibley KM, Voth J, Munce SE, Straus SE, & Jaglal SB (2014). Chronic disease and falls in community-dwelling Canadians over 65 years old: A population-based study exploring associations with number and pattern of chronic conditions. BMC Geriatrics, 14, 22. 10.1186/1471-2318-14-22 [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. StataCorp. (2013). Stata Statistical Software (Release 13). StataCorp LP. [Google Scholar]
  54. Thompson MA, Mugavero MJ, Amico KR, & Cargill VA (2018). Annals of internal medicine clinical guideline guidelines for improving entry into and retention in care and antiretroviral adherence for persons with HIV: Evidence-based recommendations from an International Association of Physicians in AIDSCare Panel. 156(11). [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Visconti AJ, Sell J, & Greenblatt AD (2019). Primary care for persons who inject drugs. American Family Physician, 99(2), 109–116. [PubMed] [Google Scholar]
  56. Vlahov D, Anthony JC, Munoz A, Margolick J, Nelson KE, Celentano DD, Solomon L, & Polk BF (1991). The ALIVE study, a longitudinal study of HIV-1 infection in intravenous drug users: Description of methods and characteristics of participants. NIDA Research Monograph, 109, 75–100. [PubMed] [Google Scholar]
  57. Westergaard RP, Hess T, Astemborski J, Mehta SH, & Kirk GD (2013). Longitudinal changes in engagement in care and viral suppression for HIV-infected injection drug users. AIDS, 27(16), 2559–2566. 10.1097/QAD.0b013e328363bff2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Westergaard RP, Kirk GD, Richesson DR, Galai N, & Mehta SH (2011). Incarceration predicts virologic failure for HIV-infected injection drug users receiving antiretroviral therapy. Clinical Infectious Diseases, 53(7), 725–731. 10.1093/cid/cir491 [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Whitson HE, Johnson KS, Sloane R, Cigolle CT, Pieper CF, Landerman L, & Hastings SN (2016). Identifying patterns of multimorbidity in older Americans: Application of latent class analysis. Journal of the American Geriatrics Society, 64(8), 1668–1673. 10.1111/jgs.14201 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplemental Table
NIHMS1884442-supplement-Supplemental_Table.docx (26.8KB, docx)

RESOURCES