Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2020 Jun 5.
Published in final edited form as: J Subst Abuse Treat. 2020 Mar;112 Suppl:49–55. doi: 10.1016/j.jsat.2020.02.008

Service Utilization and Chronic Condition Outcomes among Primary Care Patients with Substance Use Disorders and Co-Occurring Chronic Conditions

Kari A Stephens 1,2,3, Imara I West 2, Kevin A Hallgren 2, Brenda Mollis 1, Kris (Pui Kwan) Ma 2, Dennis M Donovan 2,4, Brenda Stuvek 4, Laura-Mae Baldwin 1
PMCID: PMC7274163  NIHMSID: NIHMS1594316  PMID: 32220411

Abstract

Background:

Patients with a substance use disorder (SUD) often present with co-occurring chronic conditions in primary care. Despite the high co-occurrence of chronic medical conditions and SUD, little is known about whether chronic condition outcomes or related service utilization in primary care varies between patients with versus without documented SUDs. This study examined whether having a SUD influenced the use of primary care services and common chronic condition outcomes for patients with diabetes, hypertension, and obesity.

Methods:

A longitudinal cohort observational study examined electronic health record data from 21 primary care clinics in Washington and Idaho to examine differences in service utilization and clinical outcomes for diabetes, hypertension, and obesity in patients with and without a documented SUD diagnosis. Differences between patients with and without documented SUD diagnoses were compared over a three-year window for clinical outcome measures, including hemoglobin A1c, systolic and diastolic blood pressure, and body mass index, as well as service outcome measures, including number of encounters with primary care and co-located behavioral health providers, and orders for prescription opioids. Adult patients (N = 10,175) diagnosed with diabetes, hypertension, or obesity before the end of 2014, and who had ≥2 visits across a three-year window including at least one visit in 2014 (baseline) and at least one visit occurring 12 months or longer after the 2014 visit (follow-up) were examined.

Results:

Patients with SUD diagnoses and co-occurring chronic conditions were seen by providers more frequently than patients without SUD diagnoses (p’s<.05), and patients with SUD diagnoses were more likely to be prescribed opioid medications. Chronic condition outcomes were no different for patients with versus without SUD diagnoses.

Discussion:

Despite the higher visit rates to providers in primary care, a majority of patients with SUD diagnoses and chronic medical conditions in primary care did not get seen by co-located behavioral health providers, who can potentially provide and support evidence informed care for both SUD and chronic conditions. Patients with chronic medical conditions also were more likely to get prescribed opioids if they had a SUD diagnosis. Care pathway innovations for SUDs that include greater utilization of evidence informed co-treatment of SUDs and chronic conditions within primary care settings may be necessary for improving care overall for patients with comorbid SUDs and chronic conditions.

Keywords: substance use, primary care, chronic condition, diabetes, hypertension, obesity

1. Introduction

Managing chronic conditions (e.g., diabetes, hypertension, obesity) is a major component of primary care practice. Nearly half (46.9%) of all visits to primary care physicians by non-pregnant adults were made by those with hypertension (Ashman, Rui, Schappert, & Strashny, 2017). In one large sample, over 52 million Americans with high blood pressure (84%), and over 22 million Americans with diabetes (86%), visited a primary care physician in 2014 (Petterson, McNellis, Kilink, Meyers, & Bazemore, January 2018). The prevalence of hypertension, diabetes, and obesity is high as evidenced in a study of 148 primary care practices that included 667,379 active patients – 33.5% of these patients were diagnosed with hypertension, the most common condition, 11.9% with diabetes, and 11.9% with obesity (Ornstein, Nietert, Jenkins, & Litvin, 2013).

High need adults (i.e., those with more than three chronic diseases) have been found to have three times the healthcare expenditure than adults with only two chronic conditions (Hayes et al., August 2016). Primary care patients with substance use disorder (SUD) diagnoses often have multiple comorbidities associated with high clinical complexity and cost (Korthuis et al., 2017; Saitz & Daaleman, 2017; Shapiro, Coffa, & McCance-Katz, 2013), indicating that some of this cost may be due to comorbid SUDs. Prevalence rates of comorbid SUDs with chronic conditions in primary care vary. For example, in a study of patients with co-occurring diabetes and hypertension, 1.9% of patients also had opioid use disorder, 2.2% had cocaine use disorder, 1.1% had cannabis use disorder, and 8.8% had alcohol use disorder, although this may be an under-detections of SUDs, given these data were based on electronic health record (EHR) documented diagnoses that were provider identified, rather than comprehensively screened for or assessed (Winhusen, Theobald, Kaelber, & Lewis, 2019). In a sample of 2,000 adults, across five primary care settings in the U.S., rates of SUD were higher than the EHR study and exceeded 13% when using confidential interviews (Wu et al., 2017); and in a sample of adult patients with high risk diabetes, 48.3% had a comorbid SUD (Wu et al., 2018).

Further complicating chronic condition and SUD treatment are prescription opioids which are often prescribed in primary care for non-cancer chronic pain, a common condition seen in primary care that also co-occurs with diabetes, hypertension, and obesity, SUD, and other conditions (Guh et al., 2009; Mills, Torrance, & Smith, 2016; Morasco et al., 2011). However, guidelines for prescribing opioids recommend that patients with SUDs may be at higher risk for opioid related problems, such as opioid use disorder or relapse (Dowell, Haegerich, & Chou, 2016). Yet little is known about the rates of opioid prescribing within SUD populations in primary care who also have multiple chronic conditions.

Patients with chronic conditions and a co-occurring SUD are complicated to treat in primary care (M. Butler et al., 2008; Graham et al., 2017) and primary care settings are struggling to incorporate SUD related care (Center for Behavioral Health Statistics and Quality, 2016). To promote this incorporation, SUDs are being re-conceptualized as chronic medical conditions in need of evidence-based screening and treatment (Polydorou, Gunderson, & Levin, 2008; Shapiro et al., 2013); and for many patients with SUDs, screening and treatment can be delivered directly in primary care (Bray, Del Boca, McRee, Hayashi, & Babor, 2017; Gryczynski et al., 2017; Korthuis et al., 2017). Integrated behavioral health in primary care is an effective way to co-treat chronic conditions (e.g., diabetes) and mental health issues (e.g., depression) and an evidence base is growing to show it can also improve treatment for SUDs (Huang, Wei, Wu, Chen, & Guo, 2013; LaBelle, Han, Bergeron, & Samet, 2016; Watkins et al., 2017).

Despite the high potential for co-occurrence of chronic medical conditions and SUD, little is known about whether chronic condition outcomes or related service utilization in primary care varies for patients with or without documented SUDs. EHR systems in primary care capture service utilization and routine clinical data on clinical outcomes for chronic conditions of diabetes (hemoglobin A1c (HbA1c) values), hypertension (blood pressure measurements), and obesity (body mass index (BMI)). This study aimed to examine medical outcomes and service utilization for patients with diabetes, hypertension, and obesity who either have or have not been identified with a SUD diagnosis. Given the increased complexity in care that a SUD diagnosis infers, it is hypothesized, in a cohort of patients with chronic conditions, that patients with SUD diagnoses would have higher service utilization visit rates, lower prescription rates for opioids, and poorer chronic condition outcomes than patients without SUD diagnoses.

2. Materials and Methods

2.1. Setting

This study used data from the WWAMI (Washington, Wyoming, Alaska, Montana and Idaho) region Practice and Research Network (WPRN), a practice-based research network committed to research and quality improvement across this five state region. The WPRN is partnered with the Pacific Northwest Node of the NIDA National Drug Abuse Treatment Clinical Trials Network (CTN). Four independent small and medium sized community-based primary care organizations in the WPRN with 21 clinics comprise Data QUEST, a technical infrastructure that supports harmonization of EHR data across disparate primary care practices for the purpose of translational research (Cole, Stephens, Keppel, Estiri, & Baldwin, 2016; Stephens, Anderson, Lin, & Estiri, 2016; Stephens et al., 2012). The 21 primary care clinics within these organizations include community health clinics that provide primary care to patients with limited income, insurance, or other resources, as well as rural critical access hospital-associated clinics that mostly serve patients in small towns and rural areas. Each organization was given an information sheet about the study, and all four Data QUEST member organizations agreed to participate. The University of Washington’s Institutional Review Board considered this study exempt.

2.2. Study Population

Data QUEST was used to identify adult patients, 18 years or older, with at least two primary care encounters in the three-year study period (2014 through 2017), including at least one primary care office visit in 2014 (baseline) and at least one primary care office visit 12 months (or longer) after the 2014 office visit (follow-up). Patients were included in the analysis if they had at least one diagnosis of diabetes, hypertension, or obesity associated with an encounter or included on the problem list by the end of 2014 and had both baseline and outcome measurements for the outcomes associated with their respective diagnosis (described below). They were also grouped based on whether or not they had a SUD diagnosis (i.e., alcohol, stimulant, drug, cannabis, or opioid) by the end of 2014.

2.3. Demographics and Social Determinants

Gender, race, and ethnicity were extracted from the Data QUEST warehouse for each patient. Age was computed as of 2014 using the patient year of birth available in the warehouse. Rural or urban status was assigned using Rural-Urban Commuting Area Codes (RUCAs) (WWAMI Rural Health Research Center (RUCA), 2019) based on patient ZIP code designation. RUCAs are a Census tract-based classification scheme that utilizes standard Bureau of Census Urbanized Area and Urban Cluster definitions in combination with work commuting information to characterize all of the nation’s Census tracts regarding their rural or urban status, with a ZIP Code RUCA approximation. The Social Deprivation Index (SDI) is a composite measure that was assigned to each patient based on their ZIP code (Bazemore et al., 2016; D. C. Butler, Petterson, Phillips, & Bazemore, 2013; Robert Graham Center, 2019). SDI comprises eight measures collected in the American Community Survey (ACS) (i.e., percent Black, percent living in poverty, percent non-employed, percent with less than 12 years of schooling, percent single parent households, percent renter occupied housing, percent households with no car, and the percent living in overcrowded conditions) and one measure of high needs constructed from the ACS (percent under the age of 5 or a female between the ages of 15 and 44). The index ranges from 0 to 100 and a higher score indicates greater social deprivation.

2.4. Documented Diagnoses

International Classification of Diseases (ICD) 9th or 10th edition codes were extracted from the EHR from fields associated with clinic visits and problem lists. The list of ICD codes used to indicate diabetes, hypertension, obesity, and SUDs (i.e., alcohol, stimulant, drug, cannabis, or opioid) were derived from Observational Medical Outcomes Partnership (OMOP) (Observational Health Data Sciences and Informatics (OHDSI), 2019) value sets developed from previous studies conducted with Data QUEST and in collaboration with the DARTNet Institute (DARTNet Institute, 2019). The list of ICD codes used to indicate SUDs was generated based on previous research involving EHR-based AUD diagnoses (Williams et al., 2017) and manual review of ICD-9 and ICD-10 codes for other SUD-related diagnoses. The list of ICD codes used to indicate common chronic non-cancer pain conditions was consistent with the United States National Pain Strategy (Mayhew et al., 2019).

2.5. Documented Outcomes and Service Utilization

2.5.1. Diabetes, Hypertension, and Obesity Outcomes

Outcome measures for each chronic disease group included HbA1c for diabetes, systolic and diastolic blood pressure for hypertension, and BMI for obesity. Outcomes were defined as the change in the measure from baseline to follow-up. When multiple measurements were available for the same patient, the first measure in 2014 was used for the baseline measurement and the first measurement at least one year (12 months) after baseline was used for the follow-up measurement.

2.5.2. Primary Care and Behavioral Health Visits

Visits were defined as the number of days with documented contact by a primary care or behavioral health provider in the two years (24 months) after the baseline measurement. Office visit CPT codes and provider type designations were used to identify primary care and behavioral health visits in the data.

2.5.3. Opioid Prescriptions

Medication orders were examined to determine if an opioid prescription was given to a patient within a 24-month period, after their baseline outcome was measured (Parchman et al., 2019).

2.6. Analysis

For each chronic disease group, patient characteristics were compared for those with and without a SUD diagnosis using Chi-squared tests and t-tests. Service utilization outcome measure comparisons were made for each chronic disease group and were assessed using either multiple linear regression or logistic regression adjusted for gender, age, race, ethnicity, rural status, and social deprivation. Change outcome measure comparisons were made for each chronic disease group and were assessed using multiple linear regression adjusted for baseline measurement, gender, age, race, ethnicity, rural status, and social deprivation. All analyses were performed using SAS 9.4 for Windows.

3. Results

3.1. Patient Descriptive Statistics

Patients (N = 10,175) were predominantly female (57%), 45 years or older (83.5%), White (93.5%), Non-Hispanic (88.8%), and urban (71.9%) (see Table 1). Patients with SUD diagnoses were significantly younger than patients without SUD diagnoses. Patients with both hypertension and SUD diagnoses were less likely to be male, to live in a rural area, and to have a lower social deprivation index score than those with hypertension and without SUD diagnoses. SUD diagnoses were also documented among patients from urban settings more frequently than those from rural settings. Patients with either obesity or hypertension and SUD diagnoses, in particular, had lower social deprivation compared to patients with obesity or hypertension and no documented SUD diagnosis. Over four out of five patients with a chronic condition of diabetes, hypertension, and / or obesity had at least one co-occurring diagnosis of a chronic pain condition.

Table 1.

Characteristics of patients with chronic diseases, at baseline, N = 10,175

Characteristics1 All Diabetes Hypertension Obesity
Overall n = 10,175 Any SUD n = 68 No SUD n = 1,071 p value2 Any SUD n = 589 No SUD n = 8,004 p value2 Any SUD n = 188 No SUD n = 2,622 p value2
n (%) n (%) n (%) n (%) n (%) n (%) n (%)
Gender ns <0.001 ns
Male 4,378 (43.0) 34 (50.0) 477 (44.5) 330 (56.0) 3,630 (45.4) 57 (30.3) 852 (32.5)
Female 5,797 (57.0) 34 (50.0) 594 (55.5) 259 (44.0) 4,374 (54.7) 131 (69.7) 1,770 (67.5)
Age 0.02 <0.001 0.001
18–44 1,686 (16.6) 16 (23.5) 153 (14.3) 82 (13.9) 766 (9.6) 90 (47.9) 950 (36.2)
45–64 4,087 (40.2) 37 (54.4) 520 (48.6) 363 (61.6) 3,150 (39.4) 72 (38.3) 1,078 (41.1)
65 + 4,402 (43.3) 15 (22.1) 398 (37.2) 144 (24.5) 4,088 (51.1) 26 (13.8) 594 (22.7)
Race ns ns ns
American Indian or Alaska Native 118 (1.2) 4 (5.9) 17 (1.6) 9 (1.5) 82 (1) 4 (2.1) 39 (1.5)
Asian 136 (1.3) 3 (4.4) 41 (3.8) 5 (0.9) 112 (1.4) 0 (0) 13 (0.5)
Black or African-American 122 (1.2) 2 (2.9) 12 (1.1) 10 (1.7) 90 (1.1) 4 (2.1) 31 (1.2)
White 9,512 (93.5) 58 (85.3) 946 (88.3) 558 (94.7) 7,495 (93.6) 174 (92.6) 2,466 (94.1)
More than One Race 6 (0.1) 0 (0.0) 0 (0.0) 2 (0.3) 3 (0.0) 1 (0.5) 1 (0)
Native Hawaiian or Other Pacific Islander 32 (0.3) 0 (0.0) 7 (0.7) 1 (0.2) 3 (0.0) 1 (0.5) 11 (0.4)
No Information 249 (2.5) 1 (1.5) 48 (4.5) 4 (0.7) 193 (2.4) 4 (2.1) 61 (2.3)
Ethnicity ns ns ns
Hispanic or Latino 885 (8.7) 8 (11.8) 144 (13.5) 34 (5.8) 599 (7.5) 15 (8) 292 (11.1)
Not Hispanic or Latino 9,036 (88.8) 57 (83.8) 885 (82.6) 538 (91.3) 7,223 (90.2) 165 (87.8) 2,261 (86.2)
No Information 254 (2.5) 3 (4.4) 42 (3.9) 17 (2.9) 182 (2.3) 8 (4.3) 69 (2.6)
Rural Status ns 0.005 ns
Rural 2,835 (27.9) 6 (8.8) 119 (11.1) 142 (24.1) 2,370 (29.6) 42 (22.3) 626 (23.9)
Urban 7,319 (71.9) 62 (91.2) 952 (88.9) 446 (75.7) 5,616 (70.2) 145 (77.1) 1,995 (76.1)
No Information 21 (0.2) 0 (0.0) 0 (0.0) 1 (0.2) 18 (0.2) 1 (0.5) 1 (0.04)
SDI Score ns <0.001 <0.001
SDI Mean (SD) 45.80 (25.5) 35.29 (21.0) 40.40 (24.3) 42.18 (24.6) 46.71 (25.6) 35.94 (20.8) 42.10 (24.1)
Presence of Common Chronic Non-Cancer Pain Conditions ns ns 0.002
Yes 8,403 (82.6) 923 (86.2) 54 (79.4) 504 (85.6) 6,605 (82.5) 173 (92.0) 2,190 (83.5)
No 1,772 (17.4) 148 (13.8) 14 (20.6) 85 (14.4) 1,399 (17.5) 15 (8.00) 432 (16.5)
Open in a new tab
1

N’s and percentages are unadjusted for covariates.

2

p value based on Chi-Square or t-test. Significance at p < 0.05. ns = not significant.

3.2. Overlap in Chronic Conditions

3.2.1. Diabetes, Hypertension, and Obesity

Most patients (62.2%) in the overall sample had a diagnosis of hypertension without diabetes or obesity, followed by obesity only (12.6%) and hypertension and obesity (12.1%); and over three quarters of the patients (76.9%) in the overall sample had only one of these three conditions (see Figure 1).

Figure 1. Chronic disease overlap among all patients those with a SUD, with n’s by subgroup.

Figure 1.

Open in a new tab

3.3. Substance Use Disorder Diagnoses

Overall, 6.6% (n = 693) of the total sample had at least one comorbid SUD diagnosis –6.0% of patients with diabetes, 6.9% of patients with hypertension, and 6.7% of patients with obesity. Alcohol use disorders were the most commonly diagnosed category of SUD and co-occurred in over half of the patients (63.3%) with any documented SUD. The “Drug” SUD category, which could include illicit and prescription drug use disorders, polysubstance use disorder, or substance use disorders without a specific substance identified (e.g., other psychoactive substance related disorder, mixed or unspecified drug abuse, drug induced issue), was the second most common SUD diagnosis (25.7%). The other three SUD categories were documented at lower rates – stimulant (2.3%), cannabis (4.0%), and opioid (11.7%) use disorders. Documented rates of SUD categories were largely similar across chronic conditions of diabetes, hypertension, and obesity.

3.4. Chronic Condition Outcomes

Patients with SUD diagnoses were hypothesized to have poorer chronic condition outcomes than patients without SUD diagnoses. However, no differences in change scores of clinical outcomes were found between groups (see Table 2). These results were consistent regardless of whether patients were in “poor” vs. “good” control of their condition at baseline (i.e., poor was defined as HbA1c > 7%, systolic blood pressure ≥ 140 or diastolic blood pressure > 90, or BMI ≥ 30) (results not shown in tables).

Table 2.

Change in chronic condition outcome from baseline to 12+ months by chronic disease and SUD subgroups.

Outcomes N = 10,175 Diabetes Hypertension Obesity
Any SUD n = 68 No SUD n = 1,071 p value1 Any SUD n = 589 No SUD n = 8,004 p value1 Any SUD n = 188 No SUD n = 2,622 p value1
M (SD) M (SD) M (SD) M (SD) M (SD) M (SD)
Observation Outcomes2
Change in HbA1c 0.01 (1.80) −0.11 (1.70) ns - - - - - -
Change in SBP (mmHg) - - - −0.23 (22.06) −1.21 (20.11) ns - - -
Change in DBP (mmHg) - - - −0.86 (13.64) −0.83 (12.88) ns - - -
Change in BMI (kg/m2) - - - - - - 0.31 (4.05) −0.09 (3.32) ns
Open in a new tab
1

Linear regression p value. Adjusted for baseline measurement gender, age, race, ethnicity, rural status, and social deprivation index score.

2

All means and standard deviations are raw values and reflect changes within an individual patient’s baseline and follow-up measures

3.5. Service Utilization

Patients with SUD diagnoses were hypothesized to have higher service utilization rates due to increased complexity and lower prescription rates of opioids based on guideline recommendations discouraging prescription of opioids to individuals with SUDs. As hypothesized, patients with SUD diagnoses were seen more frequently by providers than patients without a SUD diagnosis. However, contrary to our hypothesis, patients with SUD diagnoses received opioid prescriptions at a higher rate.

3.5.1. Behavioral health visits

Only three (14%) of the 21 clinics had a behavioral health provider (BHP) co-located at the practice; therefore, findings relating to BHP service utilization only included patients from those clinics. Patients’ total number of BHP visits and likelihood of getting seen by a BHP varied by condition and presence of SUD (see Table 3). Patients with documented SUD diagnoses generally trended towards being seen for more visits by a BHP compared to patients without a documented SUD diagnosis; however, this effect was only significant for patients with hypertension. Specifically, among patients with hypertension, BHP visits more than tripled for patients with a SUD diagnosis, and were nominally (but non-significantly) higher among patients with diabetes or obesity if they had a documented SUD diagnosis. The standard deviations across these BHP visit averages were high in all groups, reflecting the large range of number of visits per patient (i.e., 0–131 diabetes, 0–168 hypertension, 0–144 obesity). Less than half of all patients who could be seen by a BHP (i.e., patients in clinics that had co-located BHPs) were seen by a BHP (i.e., 20.1% for diabetes, 17.9% for hypertension, and 20.9% for obesity). Patients with a documented SUD diagnosis received BHP visits at higher proportions (i.e., 29.0% for diabetes, 32.4% for hypertension, and 31.0% for obesity) than patients without a SUD diagnosis (i.e., 19.5% for diabetes, 16.8% for hypertension, 20.2% for obesity). These higher proportions of BHP visits were significantly higher for patients with hypertension (p<.001) and obesity (p=.004).

Table 3.

Service utilization over a 24-month period for patients with diabetes, hypertension, and obesity, with or without a SUD diagnosis.

Utilization Outcomes1 Diabetes Hypertension Obesity
Any SUD No SUD Any SUD No SUD Any SUD No SUD
M (SD) n M (SD) n p value M (SD) n M (SD) n p value M (SD) n M (SD) n p value
Behavioral health visits in the 2 years after baseline2,3 2.6 (7.4) n = 62 2.3 (9.4) n = 903 ns 4.7 (15.6) n = 389 1.6 (7.2) n = 4,749 <0.001 4.5 (13.2) n = 126 2.5 (10.1) n = 1,694 ns
Primary care visit days in the 2 years after baseline2,3 17.8 (14.7) n = 68 14.8 (11.3) n = 1,071 0.01 12.7 (10.9) n = 589 11.1 (9.8) n = 8,004 <0.001 13.6 (12.4) n = 188 11.2 (9.4) n = 2,622 <0.001
Received opioid prescription (%) in the 2 years after baseline3,4 29.4% n = 68 21.7% n = 1,071 0.03 32.4% n =589 21.6% n = 8,004 <0.001 36.2% n = 188 22.7% n = 2,622 <0.001
Open in a new tab
1

All means and standard deviations reflect the unadjusted frequencies of visits or percentages.

2

Linear regression p value. Adjusted for gender, age, race, ethnicity, rural status, and social deprivation index score.

3

Baseline was the first date of their diagnosis.

4

Logistic regression p value. Adjusted for gender, age, race, ethnicity, rural status, social deprivation index score, and presence of common chronic non-cancer pain conditions.

3.5.2. Primary care visits

The average number of primary care provider visits during a two-year period after baseline ranged from 11.1 to 17.8 visits across chronic condition subgroups, evidencing that most patients with these chronic conditions were seen frequently and more if they had a co-occurring documented SUD diagnosis (see Table 3). Patients with documented SUD diagnoses had higher rates of primary care provider visits compared to patients without documented SUD diagnoses for each of the three co-occurring conditions.

3.5.3. Opioid Prescriptions

Higher proportions of patients with a documented SUD diagnosis (i.e., 29.4% diabetes, 32.4% hypertension, 36.2% obesity) received opioid prescriptions compared to patients without a SUD diagnosis (i.e., 21.7% diabetes, 21.6% hypertension, 22.7% obesity, see Table 3). These higher trends in opioid prescribing were significantly higher for the patients with in all three chronic conditions of diabetes, hypertension and obesity.

4. Discussion

Mixed support was found for the hypotheses. Patients with SUD diagnoses did not differ on chronic condition outcomes for diabetes, hypertension, and obesity. The lack of differences may have been due to under detection of SUD diagnoses, which may have resulted in patients with undocumented SUDs reducing clinical outcome changes for the non-SUD groups. Patients with diabetes, hypertension, and obesity in our sample had a rate of comorbid SUD (6.6%) that was lower than estimated national rates for the general adult population (2–12% depending on type of substance) (Merikangas & McClair, 2012). While integrated behavioral health approaches have been developed to address mental health issues like depression for patients with diabetes with high success (Kozlowska et al., 2017), they have largely not addressed substance use comorbidities, nor have they been used consistently to help promote effective screening for SUDs. A previous study of these clinics found that SUDs were likely under-detected and under-treated (Hallgren et al., 2020), potentially further complicating care for these complex patients.

As hypothesized, however, patients with identified SUD diagnoses were seen more often by both primary care medical and behavioral health providers. Patients with chronic conditions and SUD, on average, received one to five more visits than patients with chronic conditions without SUDs, indicating increased treatment opportunities that may have contributed to the lack of chronic condition outcome differences. Patients with SUD diagnoses did engage in care, confirming that treatment of co-occurring SUDs in primary care may be feasible. Given that the type of care was not evaluated (i.e., problems addressed at each visit) during these visits, it is unknown whether this increase in visits addressed these chronic conditions, SUDs, or other issues. Care pathways that address improving SUD treatment in primary care will need to incorporate existing chronic condition care pathways.

Opioid prescriptions were evaluated as a key form of service utilization because of the risk that prescription opioids pose to patients with SUDs. Patients with SUD diagnoses proportionally received more opioid prescriptions (approximately one in four patients versus one in five), which warrants further investigation to determine if guideline recommendations for prescribing opioids can be improved in this population. Primary care providers may not have accounted for the increased risk of opioid related problems for patients with SUDs and they may benefit from improved strategies for addressing chronic pain using non-opioid treatments, particularly given the high co-occurrence of common chronic pain conditions in these patients.

SUD treatment in primary care will also need to address higher rates of SUDs in males and younger populations while addressing social deprivation related issues. Differences in demographics of patients with SUD diagnoses were consistent with the 2016 National Survey on Drug Use and Health, which found people who are male reported more drug use than females overall and most people with a SUD start drug use before age 18 and develop a disorder by age 20 (Center for Behavioral Health Statistics and Quality, 2017; National Institute on Drug Abuse (NIH), 2014; Poudel & Gautam, 2017). Social deprivation was found to be better overall in our SUD sample population rather than worse. Future studies are needed that leverage methods to reduce selection bias (e.g., matched controls or propensity scoring) which may help control for these differences and other observed patterns of heterogeneity (e.g., gender, age, rural status) among patients with comorbid chronic conditions with and without SUDs. Future studies may also utilize such methods to facilitate deeper investigations of the impact of provider visits on patient outcomes, as such analyses would require correction for selection bias that would be present in naturalistic settings (e.g., patients with worse health problems are typically exposed to more treatment).

4.1. Limitations

Our findings are observational and do not involve matched controls. One study found that SUD was associated with greater risk for complications of diabetes and all-cause mortality (Winhusen et al., 2019) using a matched control design. Using matched controls in future studies may show worse chronic medical condition related outcome changes for patients with SUD diagnoses, particularly given the differences in patient characteristics between those with and without SUDs. Our sample size was fairly small, particularly for diabetes, and many of the “no SUD” patients may have had an undiagnosed SUD. These patients likely include individuals with health disparities that may have also impacted health outcomes and service utilization and future studies need to address the impact of disparities. Despite these limitations, the use of EHR data allowed examination of a cohort of patients in routine care clinic settings and elucidated important differences in service utilization that may inform future treatment interventions.

5. Conclusions

Routine data collected in primary care via the EHR were examined to explore differences in chronic condition (i.e., diabetes, hypertension, and obesity) outcomes and service utilization in primary care among patients with and without SUD diagnoses. While no clinical outcome differences were detected, service utilization patterns indicated that patients who have co-occurring SUDs do get seen more, but may be at risk for more complications from opioid medications due to the higher frequency of prescription rates. Despite the higher visit rates, a majority of patients with SUDs and chronic medical conditions in primary care did not get seen by co-located behavioral health providers embedded in primary care clinics and therefore, may not be getting optimal care for SUDs. Care pathway innovations for SUDs that include behavioral health providers within primary care settings may improve care overall for patients with comorbid SUDs and chronic conditions.

Acknowledgements:

This work was supported by the National Institute of Drug Abuse’s National Drug Abuse Treatment Clinical Trials Network (CTN) Pacific Northwest Node (3 UG1DA013714)

References

  1. Ashman JJ, Rui P, Schappert SM, & Strashny A (2017). Characteristics of visits to primary care physicians by adults diagnosed with hypertension. National Health Statistics Reports, Number 106 Retrieved from https://www.cdc.gov/nchs/data/nhsr/nhsr106.pdf [PubMed] [Google Scholar]
  2. Bazemore AW, Cottrell EK, Gold R, Hughes LS, Phillips RL, Angier H, … DeVoe JE (2016). “Community vital signs”: Incorporating geocoded social determinants into electronic records to promote patient and population health. Journal of the American Medical Informatics Association, 23(2), 407–412. doi: 10.1093/jamia/ocv088 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Bray JW, Del Boca FK, McRee BG, Hayashi SW, & Babor TF (2017). Screening, brief intervention and referral to treatment (sbirt): Rationale, program overview and cross-site evaluation. Addiction, 112 Suppl 2, 3–11. doi: 10.1111/add.13676 [DOI] [PubMed] [Google Scholar]
  4. Butler DC, Petterson S, Phillips RL, & Bazemore AW (2013). Measures of social deprivation that predict health care access and need within a rational area of primary care service delivery. Health Services Research, 48(2pt1), 539–559. doi: 10.1111/j.1475-6773.2012.01449.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Butler M, Kane RL, McAlpine D, Kathol RG, Fu SS, Hagedorn H, & Wilt TJ (2008). Integration of mental health/substance abuse and primary care. Evid Rep Technol Assess (Full Rep)(173), 1–362. [PMC free article] [PubMed] [Google Scholar]
  6. Center for Behavioral Health Statistics and Quality. (2016). Key substance use and mental health indicators in the united states: Results from the 2015 national survey on drug use and health. (HHS Publication No. SMA 16–4984, NSDUH Series H-51). Retrieved from https://www.samhsa.gov/data/sites/default/files/NSDUH-FFR1-2015/NSDUH-FFR1-2015/NSDUH-FFR1-2015.pdf [Google Scholar]
  7. Center for Behavioral Health Statistics and Quality. (2017). 2016 national survey on drug use and health: Detailed table. Retrieved from https://www.samhsa.gov/data/sites/default/files/NSDUH-DetTabs-2016/NSDUH-DetTabs-2016.pdf [Google Scholar]
  8. Cole AM, Stephens KA, Keppel GA, Estiri H, & Baldwin LM (2016). Extracting electronic health record data in a practice-based research network: Processes to support translational research across diverse practice organizations. EGEMS (Wash DC), 4(2), 1206. doi: 10.13063/2327-9214.1206 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. DARTNet Institute. (2019). About dartnet institute. Retrieved from http://www.dartnet.info/AboutDI.htm [Google Scholar]
  10. Dowell D, Haegerich TM, & Chou R (2016). Prescribing opioids for chronic pain--reply. JAMA, 316(7), 774–775. doi: 10.1001/jama.2016.9242 [DOI] [PubMed] [Google Scholar]
  11. Graham K, Cheng J, Bernards S, Wells S, Rehm J, & Kurdyak P (2017). How much do mental health and substance use/addiction affect use of general medical services? Extent of use, reason for use, and associated costs. Canadian Journal of Psychiatry. Revue Canadienne de Psychiatrie, 62(1), 48–56. doi: 10.1177/0706743716664884 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Gryczynski J, McNeely J, Wu LT, Subramaniam GA, Svikis DS, Cathers LA, … Schwartz RP (2017). Validation of the taps-1: A four-item screening tool to identify unhealthy substance use in primary care. Journal of General Internal Medicine, 32(9), 990–996. doi: 10.1007/s11606-017-4079-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Guh DP, Zhang W, Bansback N, Amarsi Z, Birmingham CL, & Anis AH (2009). The incidence of co-morbidities related to obesity and overweight: A systematic review and meta-analysis. BMC Public Health, 9, 88–88. doi: 10.1186/1471-2458-9-88 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Hallgren KA, Witwer E, West I, & Baldwin LM, Donovan D, Stuvek B, Keppel GA, Mollis B, & Stephens KA (2020). Prevelance of documented alcohol and opioid use disorder diagnoses and treatments in a regional primary care practice-based research network. Journal of Substance Abuse Treatment, 110, 18–27. doi: 10.1016/j.jsat.2019.11.008 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Hayes SL, Salzberg CA, McCarthy D, Radley DC, Abrams MK, Shah T, & Anderson GF (August 2016). High-need, high-cost patients: Who are they and how do they use health care—a population-based comparison of demographics, health care use, and expenditures,… Retrieved from https://www.commonwealthfund.org/publications/issue-briefs/2016/aug/high-need-high-cost-patients-who-are-they-and-how-do-they-use [PubMed] [Google Scholar]
  16. Huang Y, Wei X, Wu T, Chen R, & Guo A (2013). Collaborative care for patients with depression and diabetes mellitus: A systematic review and meta-analysis. BMC Psychiatry, 13, 260–260. doi: 10.1186/1471-244X-13-260 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Korthuis PT, McCarty D, Weimer M, Bougatsos C, Blazina I, Zakher B, … Chou R (2017). Primary care-based models for the treatment of opioid use disorder: A scoping review. Annals of Internal Medicine, 166(4), 268–278. doi: 10.7326/m16-2149 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Kozlowska O, Solomons L, Cuzner D, Ahmed S, McManners J, Tan GD, … Rea R (2017). Diabetes care: Closing the gap between mental and physical health in primary care. The British journal of general practice : the journal of the Royal College of General Practitioners, 67(663), 471–472. doi: 10.3399/bjgp17X692993 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. LaBelle CT, Han SC, Bergeron A, & Samet JH (2016). Office-based opioid treatment with buprenorphine (obot-b): Statewide implementation of the massachusetts collaborative care model in community health centers. Journal of Substance Abuse Treatment, 60, 6–13. doi: 10.1016/j.jsat.2015.06.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Mayhew M, DeBar LL, Deyo RA, Kerns RD, Goulet JL, Brandt CA, & Von Korff M (2019). Development and assessment of a crosswalk between icd-9-cm and icd-10-cm to identify patients with common pain conditions. The journal of pain : official journal of the American Pain Society, 20(12), 1429–1445. doi: 10.1016/j.jpain.2019.05.006 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Merikangas KR, & McClair VL (2012). Epidemiology of substance use disorders. Human Genetics, 131(6), 779–789. doi: 10.1007/s00439-012-1168-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Mills S, Torrance N, & Smith BH (2016). Identification and management of chronic pain in primary care: A review. Current psychiatry reports, 18(2), 22–22. doi: 10.1007/s11920-015-0659-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Morasco BJ, Gritzner S, Lewis L, Oldham R, Turk DC, & Dobscha SK (2011). Systematic review of prevalence, correlates, and treatment outcomes for chronic non-cancer pain in patients with comorbid substance use disorder. Pain, 152(3), 488–497. doi: 10.1016/j.pain.2010.10.009 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. National Institute on Drug Abuse (NIH). (2014). Principles of adolescent substance use disorder treatment: A research-based guide. Retrieved from https://www.drugabuse.gov/publications/principles-adolescent-substance-use-disorder-treatment-research-based-guide/introduction [Google Scholar]
  25. Observational Health Data Sciences and Informatics (OHDSI). (2019). Omop common data model. Data Standardization. Retrieved from https://www.ohdsi.org/data-standardization/the-common-data-model/ [Google Scholar]
  26. Ornstein SM, Nietert PJ, Jenkins RG, & Litvin CB (2013). The prevalence of chronic diseases and multimorbidity in primary care practice: A pprnet report. Journal of the American Board of Family Medicine, 26(5), 518–524. doi: 10.3122/jabfm.2013.05.130012 [DOI] [PubMed] [Google Scholar]
  27. Parchman ML, Penfold RB, Ike B, Tauben D, Von Korff M, Stephens M, … Baldwin LM (2019). Team-based clinic redesign of opioid medication management in primary care: Effect on opioid prescribing. Annals of Family Medicine, 17(4), 319–325. doi: 10.1370/afm.2390 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Petterson S, McNellis R, Kilink K, Meyers DS, & Bazemore A (January 2018). The state of primary care in the united states: A chartbook of facts and statistics. Retrieved from https://www.graham-center.org/content/dam/rgc/documents/publications-reports/reports/PrimaryCareChartbook.pdf [Google Scholar]
  29. Polydorou S, Gunderson EW, & Levin FR (2008). Training physicians to treat substance use disorders. Current psychiatry reports, 10(5), 399–404. doi: 10.1007/s11920-008-0064-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Poudel A, & Gautam S (2017). Age of onset of substance use and psychosocial problems among individuals with substance use disorders. BMC Psychiatry, 17(1), 10. doi: 10.1186/s12888-016-1191-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Robert Graham Center. (2019). Social deprivation index (sdi). Retrieved from https://www.graham-center.org/rgc/maps-data-tools/sdi/social-deprivation-index.html [Google Scholar]
  32. Saitz R, & Daaleman TP (2017). Now is the time to address substance use disorders in primary care. Annals of Family Medicine, 15(4), 306–308. doi: 10.1370/afm.2111 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Shapiro B, Coffa D, & McCance-Katz EF (2013). A primary care approach to substance misuse. American Family Physician, 88(2), 113–121. [PubMed] [Google Scholar]
  34. Stephens KA, Anderson N, Lin CP, & Estiri H (2016). Implementing partnership-driven clinical federated electronic health record data sharing networks. International Journal of Medical Informatics, 93, 26–33. doi: 10.1016/j.ijmedinf.2016.05.008 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Stephens KA, Lin CP, Baldwin LM, Echo-Hawk A, Keppel GA, Buchwald D, … Tarczy-Hornoch P (2012). Lc data quest: A technical architecture for community federated clinical data sharing. AMIA Jt Summits Transl Sci Proc, 2012, 57–62. [PMC free article] [PubMed] [Google Scholar]
  36. Watkins KE, Ober AJ, Lamp K, Lind M, Setodji C, Osilla KC, … Pincus HA (2017). Collaborative care for opioid and alcohol use disorders in primary care: The summit randomized clinical trial. JAMA Intern Med, 177(10), 1480–1488. doi: 10.1001/jamainternmed.2017.3947 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Williams EC, Gupta S, Rubinsky AD, Glass JE, Jones-Webb R, Bensley KM, & Harris AHS (2017). Variation in receipt of pharmacotherapy for alcohol use disorders across racial/ethnic groups: A national study in the u.S. Veterans health administration. Drug and Alcohol Dependence, 178, 527–533. doi: 10.1016/j.drugalcdep.2017.06.011 [DOI] [PubMed] [Google Scholar]
  38. Winhusen T, Theobald J, Kaelber DC, & Lewis D (2019). Medical complications associated with substance use disorders in patients with type 2 diabetes and hypertension: Electronic health record findings. Addiction, 114(8), 1462–1470. doi: 10.1111/add.14607 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Wu L-T, McNeely J, Subramaniam GA, Brady KT, Sharma G, VanVeldhuisen P, … Schwartz RP (2017). Dsm-5 substance use disorders among adult primary care patients: Results from a multisite study. Drug and Alcohol Dependence, 179, 42–46. doi: 10.1016/j.drugalcdep.2017.05.048 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Wu LT, Ghitza UE, Zhu H, Spratt S, Swartz M, & Mannelli P (2018). Substance use disorders and medical comorbidities among high-need, high-risk patients with diabetes. Drug and Alcohol Dependence, 186, 86–93. doi: 10.1016/j.drugalcdep.2018.01.008 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. WWAMI Rural Health Research Center (RUCA). (2019). Ruca data. Zip code ruca approximation. Retrieved from https://depts.washington.edu/uwruca/ruca-contact.php [Google Scholar]

RESOURCES