Examining trends in substance use disorder capacity and service delivery by Health Resources and Services Administration-funded health centers: A time series regression analysis.

BACKGROUND: The opioid epidemic and subsequent mortality is a national concern in the U.S. The burden of this problem is disproportionately high among low-income and uninsured populations who are more likely to experience unmet need for substance use services. We assessed the impact of two Health Resources and Services Administration (HRSA) substance use disorder (SUD) service capacity grants on SUD staffing and service use in HRSA -funded health centers (HCs). METHODS AND
FINDINGS: We conducted cross-sectional analyses of the Uniform Data System (UDS) from 2010 to 2017 to assess HC (n = 1,341) trends in capacity measured by supply of SUD and medication-assisted treatment (MAT) providers, utilization of SUD and MAT services, and panel size and visit ratio measured by the number of patients seen and visits delivered by SUD and MAT providers. We merged mortality and national survey data to incorporate SUD mortality and SUD treatment services availability, respectively. From 2010 to 2015, 20% of HC organizations had any SUD staff, had an average of one full-time equivalent SUD employee, and did not report an increase in SUD patients or SUD services. SUD capacity grew significantly in 2016 (43%) and 2017 (22%). MAT capacity growth was measured only in 2016 and 2017 and grew by 29% between those years. Receipt of both supplementary grants increased the probability of any SUD capacity by 35% (95% CI: 26%, 44%) and service use, but decreased the probability of SUD visit ratio by 680 visits (95% CI: -1,013, -347), compared to not receiving grants.
CONCLUSIONS: The significant growth in HC specialized SUD capacity is likely due to supplemental SUD-specific HRSA grants and may vary by structure of grants. Expanding SUD capacity in HCs is an important step in increasing SUD access for low income and uninsured populations broadly and for patients of these organizations.

Introduction

The opioid problem, identified as early as the 1990s, has escalated since 2010 with heroin use and since 2013 with synthetic opioids [1, 2]. The problem is now considered an epidemic with an estimated 20.8 million (7.8%) individuals ages 12 and older in the United States reported to have substance use disorder (SUD) in 2015 [3]. During the same year, the rates of opioid use disorders were 0.6% for heroin and 2.0% for prescription opioids [4]. From 2000 to 2014, overdose deaths involving prescription opioids and heroin increased 200% and overdose deaths from all opioid drugs increased 137% [5]. Opioid use disorders alone led to approximately 14.9 deaths, 296 emergency department visits, (both in 2017) and 225 hospitalizations per 100,000 individuals (in 2014) [6, 7]. In addition, there were 16.3 opioid-related deaths per 100,000 persons in 2015 [5]. The escalation in opioid use is a significant problem, but other SUDs including alcohol and methamphetamine are prevalent causes of morbidity and preventable mortality and require treatment capacity [3]. The proportion of those who report needing but not receiving SUD treatment overall is as high as 89% nationally [8]. Furthermore, primary care settings are well positioned and recommended to provide treatment for individuals with SUDs [9, 10].

There are disparities in opioid use disorders and access to SUD services. Low-income individuals have higher rates of opioids misuse and opioid use disorder than the general U.S. population [11]. Uninsured and Medicaid patients had significantly worse access to SUD and mental health services than Medicare and privately insured individuals [12]. The former group cited various barriers contributing to unmet need for SUD care, including not being able to afford SUD treatment, lack of availability of needed SUD services, and greater distance in geographic proximity to SUD services [13]. Furthermore, non-Hispanic African-Americans and Hispanic/Latinos were found to have greater barriers to accessing SUD treatment services than non-Hispanic Whites [14, 15]. Increasing access to SUD services in primary care settings for these populations is considered an important strategy to combat the nation’s opioid epidemic [16].In particular, medication-assisted treatment (MAT) delivered by health providers in primary care settings is essential to expanding access to opioid use treatment [17, 18].

Health Resources and Services Administration (HRSA)-funded health centers (HC) deliver primary care to racially/ethnically diverse and low-income and uninsured patients and can play a strategic and significant role in expanding access to SUD and MAT services to the underserved [19]. In 2017, 1,373 HC organizations with 11,056 delivery sites provided comprehensive and affordable primary care to over 27 million, or 1 in 12 Americans, 63 percent of whom are racial/ethnic minorities [20, 21].

In a 2014 national survey of patients served by HRSA-funded HCs, patients reported a slightly lower risk of SUD and opioid use disorder compared to national rates, with 6.4% of HC patients at moderate or high risk of SUD and 1.2% were estimated to abuse opioids and/or were dependent on them, though it is likely that much of the available SUD data are self-reported and underestimate SUD risk [22-24]. In 2018, nearly 70 percent of HC organizations provided SUD services in at least one delivery site, but on-site service provision was not universal as HCs can arrange for services to be provided through contracted providers or can refer patients to providers with informal agreements [18].

In March 2016, HRSA awarded $94 million through the Substance Abuse Service Expansion (SASE) supplemental grants to 271 or 20% of HCs, with supplementary funding ranging from $217,000 to $406,000. The grants were awarded to increase numbers of SUD providers and staff and increase access to SUD and MAT services [25]. HCs that received these grants were required to: 1) enhance or establish an integration primary care/behavioral health model; 2) increase patients screened through Screening, Brief, Intervention, and Referral to Treatment (SBIRT) model; and 3) increase access to patients by either adding at least 1.0 full-time equivalent SUD provider or by adding or enhancing existing SUD services; 4) coordinate services necessary for patients; and 5) provide training and educational resources to assist health professional in making informed prescribing decisions. In September 2017, HRSA awarded another $200 million through Access Increases in Mental Health and Substance Abuse Services (AIMS) supplemental grants to 1,178 or 86% of HCs, with grants ranging from $84,000 to $176,000 [26]. AIMS grants included supporting the expansion of SUD services with a focus on opioid abuse, but also sought to increase mental health services. Recipients of this grant were required to: 1) expand direct hire staff and/or contractors who will support mental health and substance abuse service expansion focusing on the treatment, prevention, and awareness of opioid abuse; 2) provide access to expanded mental health and substance abuse services; and 3) increase the number of mental health patients and/or substance abuse patients as a result of AIMS funding. SASE provided awards of up to $325,000 per year for two years (2016 through 2018), while AIMS awarded funding divided into one-time or ongoing supplements, of up to $75,000 per year.

This study seeks to understand the impact of these targeted funding mechanisms on capacity and ability to improve access to SUD services provided by specialized SUD staff. Several recent studies have examined behavioral health capacity at HCs and suggest gaps in current SUD capacity and service use at HCs [27-33]. In addition, a limited number of studies have directly examined the potential role of recent HRSA investments in SUD capacity and service use at HCs, with emerging evidence suggesting such investments were associated with addiction treatment capacity [34]. To address these gaps, we examined the trends in HCs in specialized SUD and MAT capacity, service use, and patients and visits per provider in general and the potential impact of the 2016 and 2017 HRSA grants on these indicators. We hypothesized that HRSA grants would increase the number of specialized SUD providers and staff who would deliver more visits to more patients. The results provide important information on SUD capacity in the primary care settings most commonly used by low-income and uninsured populations and further highlight how this capacity may be increased to address the opioid epidemic and the need for SUD services. Because SASE funding was distributed in March 2016 and AIMS funding was distributed in September 2017, we anticipated that the impact of SASE was discernible in 2016 and 2017, but the impact of AIMS would be discernible partly in 2017 with the full impact mostly observed in 2018 and subsequent years. We focused on specialized SUD staff because we could not measure SUD services provided by medical or mental health providers, other than MAT.

Methods

Data and sample

For this study, we used a time-series analysis using HC administrative data and publicly available data sources to assess the impact of HRSA supplemental funding on our outcomes. We used the 2010 to 2017 Uniform Data System (UDS) to examine HC staffing and delivery of MAT and SUD services. UDS is an annual cross-sectional administrative database maintained by HRSA. All HRSA-funded HCs are required to submit a UDS report annually including data on patient characteristics, staffing, utilization of services, and revenues from the previous calendar year. UDS captures aggregate information at the HC organizational level (i.e., parent organization or network level) rather than individual delivery sites that operate within the organization. Each HC organization operates multiple delivery sites.

To compare the size of HC SUD and MAT staffing with national and regional need for these services, we obtained publicly available opioid-related mortality rates from Centers for Disease Control and Prevention (CDC) Wide-ranging OnLine Data for Epidemiologic Research (WONDER) [4, 35]. We merged the CDC WONDER database with UDS data using the Federal Information Process Standards (FIPS) code associated with the address of the HC organization. If the FIPS code matched to more than one county, the county with the larger share of HC patients was selected. We also used the 2015 National Survey of Substance Abuse Treatment Facilities (N-SSATS) to assess the supply of drug and alcohol treatment facilities in the county where the HC was located. We merged the 2015 N-SSATS with UDS using FIPS code. We included all 1,375 HCs in the descriptive analyses, but for models we excluded 21 HCs that were not operational or did not report data in 2015. We also excluded 13 HCs that only received SASE funding from the models because the small sample size led to unreliable results. Our final 2015 sample for the models was 1,341 HCs.

Independent variables

The primary variable of interest was receipt of HRSA funding status. HRSA’s Bureau of Primary Health Care provided the list of HCs that received SASE and AIMS awards. We categorized HCs into those that received (1) no grants, (2) AIMS only, and (3) both SASE and AIMS (SASE/AIMS) grants. We controlled for several HC and market characteristics in 2015. HC characteristics included the size of the HC organization, indicated by whether the HC served more than 10,000 patients (vs. fewer), and use of electronic health records as an indicator of capacity to incorporate population health management [36]. Other HC organizational characteristics included urban/rural status of the HC organization and U.S. Census regions. The latter included New England, Middle Atlantic, East North Central, West North Central, South Atlantic, East South Central, West South Central, Mountain, and Pacific regions. We also included several measures of case mix and socioeconomic profile of HC patients. These included the proportion of patients who were racial/ethnic minorities, homeless, agricultural workers, below 100% of federal poverty guidelines, uninsured, or had Medicaid coverage. We identified whether the HC achieved patient-centered medical home (PCMH) recognition in the UDS, as an indicator of a comprehensive and integrated approach to care. PCMH recognition is reported at the HC-organization level, although recognition can be given to specific sites as well as to the whole organization. We also controlled availability of SUD services by the average number of facilities providing substance abuse services per 100,000 persons by county, extracted from the 2015 N-SSATS, and controlled for the opioid mortality rate per 100,000 persons by state, acquired from CDC WONDER.

Dependent variables

We measured specialized SUD capacity by (1) the proportion of HCs with at least one full-time equivalent (FTE) specialized SUD staff, (2) the average number of SUD staff per HC, and (3) the ratio of SUD staff per 1,000 patients. HCs report substance abuse workers, psychiatric or mental health nurses, psychiatric or clinical social workers, clinical psychologists, and other individuals providing alcohol or drug abuse counseling and/or treatment services as SUD staff. We did not include mental health or primary care providers that could deliver SUD services because UDS did not require HCs to report these provider’s provision of SUD services separately from medical or mental health services. We next measured SUD service use by (1) the proportion of FTE SUD staff to total HC patients and (2) the proportion of FTE SUD staff to total HC visits. We then measured panel size and visit ratio of SUD staff by calculating (1) the ratio of SUD patients per FTE SUD staff per year (panel size) and (2) the average number of SUD visits per FTE SUD staff. HCs report SUD visits provided by SUD staff and SUD patients as patients that had at least one SUD visit.

We also measured MAT capacity, service use, and provider panel size in 2016 and 2017. HCs reported providers with Drug Addiction Treatment Act of 2000 (DATA 2000) waivers, most of whom are likely to be primary care providers, for MAT starting in 2016. MAT allows providers to prescribe specific controlled medications for opioid dependency treatment, but the DATA waiver is not inclusive of all treatment possibilities (i.e., implantable formulations of naloxone/buprenorphine) [37]. Our measures of MAT capacity included the (1) proportion of HCs with MAT providers, (2) average number of MAT providers per HC, and (3) the ratio of MAT provider per 1,000 patients. We measured MAT service use by the proportion of total HC patients that were MAT patients. The UDS does not include the number of MAT visits. We measured MAT panel size by the ratio of MAT patients per MAT provider at each HC.

Analytic methods

We assessed the changes in our outcomes of interest from 2010 through 2017 for all HCs in the UDS in these years. We calculated the average annual percent change in SUD capacity, service use, and panel size and visit ratio from 2010 to 2015 to establish trends prior to 2016. We then calculated the percent change from 2015 to 2016 and 2016 to 2017 separately to assess increases or decreases since 2015 on our outcomes. We only measured changes in MAT capacity, service use, and panel size between 2016 and 2017 descriptively and conducted chi-square or t-tests as appropriate. We then constructed random-effects logistic, negative binomial, or Poisson regression models, as appropriate, controlling for HC patient and organizational characteristics and other likely confounders to assess the potential role of HRSA funding on outcomes. We included only complete data for all analyses presented in this paper. All analyses were conducted using Stata v.15 and Margins post-estimation command to report predicted probabilities for ease of interpretation. All statistically significant results with probability values of 0.05 or smaller were discussed.

Ethics statement

This research was granted exemption by the University of California Los Angeles Instructional Review Board (study number 16–001528) due to secondary analysis of de-identified and publicly available data.

Results

The national opioid-mortality rate increased at an average annual percent change of 11% from 2010 to 2015. This rate grew by 28% in 2016 and 12% in 2017 (Table 1). While the number of HCs grew from 1,124 in 2010 to 1,373 in 2017, the proportion of HCs with any SUD staff declined on average 1% per year from 2010 to 2015 but increased by 43% from 2015 to 2016 and 22% from 2016 to 2017. By 2017, more than one-third (35%) of HCs had any FTE SUD staff. The average number of SUD staff declined by 2% from 2010 to 2015 but increased by 21% in 2016 and by 17% in 2017. This trend was consistent with the ratio of SUD staff for 1,000 patients per HC. The proportion of SUD to total HC patients grew 1% on average from 2010 to 2015, declined 10% in 2016, and grew 11% in 2017. The proportion of SUD to total HC visits declined 4% on average from 2010 to 2015, but then grew 5% in 2016 and 1% in 2017. SUD panel size or the ratio of SUD patient to SUD staff declined by 4% from 2010 to 2015, increased by 19% in 2016, and declined by 10% in 2017. The same trends were observed for the ratio of SUD visits per SUD staff.

Table 1

Provision of Substance Use Disorder (SUD) and Medication-Assisted Treatment (MAT) services by federally-funded Health Centers (HCs) in the United States in 2010 and 2015 to 2017,.

	2010	2015	2016	2017	Average Annual Percent Change between 2010 to 2015c	Annual Percent Change 2015 to 2016	Annual Percent Change 2016 to 2017
Number of HCs	1,124	1,375	1,367	1,373	4%	-1%	0.4%
	Mean (SD) or % (n)	Mean (SD) or % (n)	Mean (SD) or % (n)	Mean (SD) or % (n)
SUD Capacity
Proportion of health centers with FTE SUD staff	20% (230)	20% (274)	28% (388)	35% (477)	-1%	43%	22%
Average number of FTE SUD staff	1.0 (4.8)	0.9 (4.6)	1.2 (4.7)	1.4 (4.9)	-2%	21%	17%
Average FTE SUD staff per 1,000 patients	0.11 (0.68)	0.10 (0.58)	0.11 (0.69)	0.12 (0.55)	-1%	13%	4%
SUD Service Use
Average proportion of total HC patients that were SUD patients	1.1% (3.8%)	1.2% (5.0%)	1.1% (3.2%)	1.2% (3.3%)	1%	-10%	11%
Average proportion of total HC visits that were SUD visits	1.7% (5.5%)	1.4% (4.7%)	1.4% (4.5%)	1.4% (3.8%)	-4%	5%	1%
SUD Panel Size and Visit Ratio
Average SUD patients per FTE SUD staff	318 (472)	249 (277)	295 (1,073)	264 (904)	-4%	19%	-10%
Average SUD visit per FTE SUD staff	1,580 (4,185)	1,128 (1,492)	1,263 (3,460)	1,129 (3,029)	-6%	12%	-11%
MAT Capacity
Proportion of health centers with MAT providers	—d	—	33% (453)	43% (588)§§§ e	—	—	29%
Average number of MAT providers	—	—	1.2 (3.8)	2.2 (5.6)§§§	—	—	73%
Average MAT providers per 1,000 patients	—	—	0.12 (0.7)	0.17 (0.5)	—	—	37%
MAT Service Use
Average proportion of total HC patients that were MAT patients	—	—	0.2% (1.0%)	0.4% (1.1%)§§ e	—	—	52%
MAT Panel Size
Average MAT patients per MAT provider	—	—	28.7 (44.8)	26.3 (40.1)	—	—	-8%
National Opioid-Mortality Rate
National Opioid-mortality rate per 100,000 persons	6.8 (0.05)	10.4 (0.06)	13.3 (0.07)	14.9 (0.07)	11%	28%	12%

a SUD = substance use disorder, MAT = medication assisted treatment; includes drugs and alcohol.

b Authors' analyses of data from the 2010 to 2017 Uniform Data System.

c The average percent change for 2010–2011, 2011–2012, 2012–2013, 2013–2014 and 2014–2015.

d MAT services were reported in the Uniform Data System starting in 2016.

e Statistically significant comparing 2017 to 2016 at §p<0.05,

§§p<0.01,

§§§p<0.001.

Standard deviation or count in parentheses.

SUD, substance use disorder; MAT, medication assisted treatment; SD, standard deviation; HC, health center.

The proportion of HCs with MAT providers increased by 29% from 2016 to 2017. Similarly, the average number of MAT providers increased 73% from 1.2 MAT providers in 2016 to 2.2 in 2017. From 2016 to 2017, MAT service use as measured by the proportion of MAT patients to total HC patients grew 52%. MAT panel size decreased by 8% over the same time period.

The majority (68%) of HCs received an AIMS grant only, 19% received both AIMS and SASE, and 13% received neither (Table 2). Examining the SUD staff capacity of HCs in 2015 prior to grant distributions showed that those with both SASE/AIMS funding and those with AIMS only were more likely to have any SUD staff and a higher average number of SUD staff per HC than those without. However, the ratio of SUD staff per 1,000 patients, the ratio of SUD patients per SUD staff, and the ratio of SUD visits per SUD staff was not different by funding. Examining SUD service use showed that SASE/AIMS HCs were more likely to have a higher proportion of SUD visits, but the proportion of SUD patients did not differ by funding.

Table 2

Health center characteristics by SASE and AIMS grantee status in 2015.

	2015 (Baseline)
	Total	None	AIMS Only	SASE/AIMS	p-value
Number of HCs	1,341	13% (176)	68% (907)	19% (258)
	Mean (SD) or % (n)	Mean (SD) or % (n)	Mean (SD) or % (n)	Mean (SD) or % (n)
SUD Capacity
Proportion of health centers with SUD staff	20% (269)	10% (18)	17% (153)	38% (98)	0.000
Average number of SUD staff	1.0 (4.7)	0.6 (3.4)	0.8 (4.3)	1.9 (6.2)	0.010
Average SUD staff per 1,000 patients	0.1 (0.6)	0.1 (0.7)	0.1 (0.6)	0.1 (0.3)	0.870
SUD Panel Size and Visit Ratio
Average SUD patients per SUD staff	248.3 (278.0)	266.2 (313.3)	220.1 (250.4)	288.6 (308.1)	0.160
Average SUD visit per SUD staff	1123.6 (1497.4)	949.5 (923.2)	1039.9 (931.6)	1283.9 (2144.9)	0.401
SUD Service Use
Average proportion of total HC patients that were SUD patients	1.2% (5.1%)	0.9% (3.3%)	1.0% (5.7%)	1.8% (3.6%)	0.139
Average proportion of total HC visits that were SUD visits	1.4% (4.7%)	0.9% (3.4%)	1.2% (4.7%)	2.3% (5.5%)	0.006
Health Center Characteristics
More than 10,000 patients served last year	53.7% (720)	73.6% (80)	49.6% (450)	73.6% (190)	0.000
Urban (vs. rural)	44.8% (601)	50.4% (71)	44.1% (400)	50.4% (130)	0.071
Region
New England	7.7% (103)	15.4% (3)	6.6% (60)	15.4% (40)	0.000
Middle Atlantic	9.8% (132)	10.6% (16)	9.8% (89)	10.6% (27)
East North Central	12.6% (169)	18.5% (13)	11.9% (108)	18.5% (48)
West North Central	6.9% (93)	5.5% (7)	7.9% (71)	5.5% (14)
South Atlantic	16.7% (224)	13.4% (34)	17.2% (156)	13.4% (35)
East South Central	6.6% (88)	3.5% (16)	7.0% (63)	3.5% (9)
West South Central	10.5% (140)	3.5% (33)	10.9% (99)	3.5% (9)
Mountain	8.5% (114)	7.1% (10)	9.4% (86)	7.1% (18)
Pacific	20.8% (278)	22.4% (45)	19.4% (176)	22.4% (58)
Had Electronic Health Records	72.8% (976)	82.6% (105)	72.6% (658)	82.6% (213)	0.000
Percent with PCMH Recognition	68.4% (917)	85.7% (88)	67.0% (608)	85.7% (221)	0.000
Health Center Patient Characteristics
Percent of Patients that are Minority	55.6% (32.0%)	55.2% (33.9%)	55.0% (32.1%)	57.9% (30.6%)	0.424
Percent of Patients that are Homeless	7.3% (19.2%)	5.7% (18.3%)	6.5% (18.0%)	11.1% (23.1%)	0.002
Percent of Patients that are Migrant and Agricultural Workers	2.8% (10.5%)	3.4% (13.9%)	2.7% (10.5%)	2.5% (7.5%)	0.606
Percent of Patients that are less than 100% of the Federal Poverty Guideline	48.4% (24.4%)	46.3% (24.7%)	48.1% (24.2%)	50.7% (25.1%)	0.153
Percent of Patients that are Uninsured	27.0% (19.3%)	28.6% (21.5%)	27.9% (19.5%)	23.0% (16.1%)	0.001
Percent of Patients that are Medicaid	43.1% (19.9%)	39.8% (21.1%)	41.8% (19.9%)	50.0% (17.4%)	0.000
County Characteristics
Mean number of facilities per 100,000 persons providing substance abuse services	4.1 (2.1)	3.8 (2.3)	4.0 (2.1)	4.3 (2.0)	0.042
Opioid mortality rate per 100,000 persons	10.9 (6.4)	10.1 (6.2)	10.6 (6.2)	12.7 (6.8)	0.000

Standard deviation or count in parentheses. Analyses involved comparing independent and control variables by supplemental funding status using t-test or chi-square test, as appropriate.

SASE, Substance Abuse Service Expansion; AIMS, Access Increases in Mental Health and Substance Abuse Services; HC, health center; PCMH, patient centered medical home; SUD, substance use disorder.

Examining HC characteristics by funding status showed that those with supplemental funding were larger, located in the Pacific and New England census regions, and more likely to have electronic health records or PCMH recognition than those without funding. These HCs also had more patients that were homeless, uninsured, or Medicaid beneficiaries compared to HCs without funding. Similarly, these HCs were located in counties that had a higher average number of substance abuse facilities and a higher opioid mortality rate per 100,000 persons. HCs with SASE/AIMS funding were also more likely to have any SUD staff and be located in East North Central and New England census regions than those with AIMS only funding. Those with AIMS only funding were also more likely to be located in urban areas than those without funding.

Table 3 displays measures of SUD staffing, service use, panel size, and visit ratio from 2015 to 2017 and percent change in those measures during that same time period given HRSA funding status and after controlling for HC and county characteristics. Among HCs without funding, there was no increase or primarily a decline in staffing, service use, and panel size but an increase in visits per SUD staff from 2015 to 2017. However, HCs with SASE/AIMS funding were more likely to have added SUD staff (115%), more SUD staff (18%), and more SUD staff per 1,000 patients (63%) from 2015 to 2017. HCs with AIMS only showed a 49% growth in SUD staff and limited growth in other SUD capacity measures from 2015 to 2017. These HCs also showed a percentage change decline in service use. Regardless of funding status, all HCs saw a percentage change decrease in SUD panel size.

Table 3

Adjusted provision of Substance Use Disorder (SUD) services by federally-funded health centers in the United States between the year prior to grant distribution (2015) and the last year the grants were distributed (2017), by grantee status.

	None			AIMS Onlyb			SASE/AIMSa,b			Percent Change 2015–2017
	2015	2016	2017	2015	2016	2017	2015	2016	2017	None	AIMS only	SASE/AIMS
Number of HCs	13% (176)			68% (907)			19% (258)
	Mean (SD) or % (n)			Mean (SD) or % (n)			Mean (SD) or % (n)
SUD Capacity
Proportion of health centers with SUD personnel	11% (19)	12% (21)	10% (17)	17% (157)	19% (174)	26% (235)	35% (91)	68% (175)	76% (195)	-10%	49%	115%
Average number of SUD staff	1.4 (0.4)	1.1 (0.4)	1.3 (0.4)	1.9 (0.3)	2.0 (0.3)	2.0 (0.3)	3.0 (0.3)	3.3 (0.3)	3.6 (0.3)	-6%	6%	18%
Average SUD staff per 1,000 patients	0.04 (0.01)	0.04 (0.01)	0.04 (0.01)	0.07 (0.01)	0.07 (0.01)	0.07 (0.01)	0.16 (0.03)	0.20 (0.03)	0.26 (0.04)	-9%	1%	63%
SUD Service Use
Average proportion of total HC patients that were SUD patients	1.5% (0.5%)	0.9% (0.4%)	1.1% (0.4%)	2.0% (0.4%)	2.0% (0.4%)	2.0% (0.4%)	5.0% (1.4%)	5.3% (1.5%)	5.8% (1.7%)	-22%	-1%	16%
Average proportion of total HC visits that were SUD visits	0.7% (0.3%)	0.6% (0.2%)	0.7% (0.3%)	2.1% (0.4%)	2.1% (0.4%)	1.8% (0.3%)	4.8% (1.4%)	5.2% (1.5%)	5.7% (1.7%)	-1%	-13%	18%
SUD Panel Size and Visit Ratio
Average SUD patients per SUD staff	1,050 (323)	818 (251)	696 (214)	342 (30)	296 (26)	302 (26)	472 (45)	392 (37)	352 (33)	-34%	-13%	-26%
Average SUD visit per SUD staff	1,152 (323)	1673 (469)	1755 (492)	1081 (81)	1101 (82)	963 (72)	1209 (104)	1201 (103)	1131 (97)	52%	-11%	-6%

a HRSA distributed $94 million in March 2016 through the Substance Abuse Service Expansion (SASE) program to 271 HCs ($217,000 to $406,000) with the goals of increasing SUD personnel, increasing number of patients screened and connected to SUD treatment, and increasing access to Medication Assisted Treatment (MAT) services.

b HRSA distributed $200 million in September 2017 through the Access Increases in Mental Health and Substance Abuse Services (AIMS) program to 1,178 HCs ($84,000 to $176,000) with the goals of increasing substance abuse services focusing on the treatment, prevention, and awareness of opioid abuse; increasing SUD personnel; and leveraging health information technology and training to increase and improve SUD services.

Standard deviation or count in parentheses.

SD, Standard Deviation; SUD, substance use disorder; AIMS, Access Increases in Mental Health and Substance Abuse Services; SASE, Substance Abuse Service Expansion; HC, health center; HRSA, Health Resources and Services Administration; MAT, Medication Assisted Treatment.

Table 4 presents the predicted probabilities for SUD capacity, service use, panel size, and visit ratio based on multivariate models controlling for covariates. The percentage change increase in SUD capacity among HCs with SASE/AIMS corresponded to significant predictive probability increases in service use indicators compared to HCs with AIMS only or no funding. For example, HCs with SASE/AIMS funding increased the probability of adding SUD staff by 26% (95% CI: 18%, 33%), the number of SUD staff by 0.4 FTE (95% CI: 0.2, 0.7), and more SUD staff per 1,000 patients by 0.10 FTE (95% CI: 0.05, 0.15), compared to HCs with AIMS only funding. Similar trends were seen when comparing predicted probabilities between HCs with SASE/AIMS funding to HCs without funding. When compared to HCs without funding, HCs with AIMS funding only increased the probability of hiring SUD staff by 10% (95% CI: 3%, 16%) and had no impact on other measures of SUD capacity. HCs with AIMS only funding decreased the probability of the proportion of SUD visits by 0.3% (95% CI: -0.4%, -0.2%) compared to HCs with no funding. However, HCs with both SASE/AIMS increased the probability of number of patients per SUD staff by 234 patients (95% CI: 16, 452) while HCs with AIMS only increased the probability by 314 patients (95% CI: 97, 532), compared to HCs with no funding. Visits per SUD staff increased the probability among HCs with SASE/AIMS funding by 41 visits (95% CI: 18, 63) compared to those with HCs with AIMS only funding and decreased the probability by 680 visits (95% CI: -1,013, -347) compared to HCs with no funding. The same pattern was observed for HCs with AIMS only compared to those without funding, a probability decrease of 720 visits (95% CI: -1,054, -387). The final regression models are displayed in the S1 Table.

Table 4

Predicted probabilities of Substance Use Disorder (SUD) services by federally-funded health centers in the United States between the year prior to grant distribution (2015) and the last year the grants were distributed (2017), by grantee status.

	Predicted Probability [95% CI]
	AIMS only vs. none	SASE/AIMS vs. none	SASE/AIMS vs. AIMS only
Number of HCs	13% (176)	68% (907)	19% (258)
SUD Capacity
Proportion of health centers with SUD personnel	10% [3%, 16%]***	35% [26%, 44%]***	26% [18%, 33%]***
Average number of SUD staff	0.2 [-0.3, 0.7]	0.6 [0.1, 1.2]*	0.4 [0.2, 0.7]***
Average SUD staff per 1,000 patients	0.00 [-0.02, 0.03]	0.10 [0.05, 0.16]***	0.10 [0.05, 0.15]***
SUD Service Use
Average proportion of total HC patients that were SUD patients	0.3% [0.1%, 0.5%]***	1.1% [0.6%, 1.7%]***	0.8% [0.4%, 1.3%]***
Average proportion of total HC visits that were SUD visits	-0.3% [-0.4%, -0.2%]***	0.9% [0.4%, 1.4%]**	1.1% [0.6%, 1.7%]***
SUD Panel Size and Visit Ratio
Average SUD patients per SUD staff	314 [97, 532]**	234 [16, 452]*	-80 [–104, –57]***
Average SUD visit per SUD staff	-720 [–1,054, –387]***	-680 [–1,13, –347]***	41 [18, 63]***

a HRSA distributed $94 million in March 2016 through the Substance Abuse Service Expansion (SASE) program to 271 HCs ($217,000 to $406,000) with the goals of increasing SUD personnel, increasing number of patients screened and connected to SUD treatment, and increasing access to Medication Assisted Treatment (MAT) services.

b HRSA distributed $200 million in September 2017 through the Access Increases in Mental Health and Substance Abuse Services (AIMS) program to 1,178 HCs ($84,000 to $176,000) with the goals of increasing substance abuse services focusing on the treatment, prevention, and awareness of opioid abuse; increasing SUD personnel; and leveraging health information technology and training to increase and improve SUD services.

Statistically significant at

*p<0.05;

**p<0.01;

***p<0.001.

SUD, substance use disorder; AIMS, Access Increases in Mental Health and Substance Abuse Services; SASE, Substance Abuse Service Expansion; HC, health center; CI, confidence interval.

Discussion

We did not find growth in specialized SUD capacity, service use, panel size or visit ratio prior to 2015. Annual percent change in all measures were significantly different after 2015 from prior, with growth in SUD capacity and service use but a decline in provider panel size and visits. Regression findings indicate changes from 2015 to 2017 seemed to be in response to SASE/AIMS funding efforts. The funding seemed to have promoted more HCs to hire SUD staff, and those with existing staff to hire more of such providers. Increased staffing corresponded to increased service delivery but produced fewer SUD visits compared to HCs that did not receive funding. Changes in MAT capacity, panel size, and service use were only measurable from 2016 to 2017 and all measures except the ratio of MAT patients per MAT provider and the ratio of MAT providers per 1,000 patients increased significantly. However, the MAT patient-to-provider ratio (26.3 patients per provider) remains below the initial 30-patient limit set for providers [38].

The stagnant growth in specialized SUD capacity prior to 2015 was in contrast to the rise in national opioid overdose deaths since 2010 [1]. However, the rapid growth in SUD capacity among HCs with SUD funding by 2017 indicates the likely effectiveness of the targeted effort by HRSA to promote capacity. This is consistent with previous research that indicates when such investments are made to HCs, an increase in staffing is observed [39]. The growth in SUD capacity was timely and resulted in 0.11 SUD FTE staff per 1,000 patients in 2016, compared to national estimates of 0.07 outpatient SUD treatment staff per 1,000 patients [40]. Despite this higher level of capacity, the estimated level of need in HCs indicates that this capacity is likely to be inadequate [41]. Further considerations should be given in continuous supplemental funding (as opposed to one-time funding) and for future targeted technical assistance to encourage providers to apply for patient limit waivers, further increasing MAT access for patients, and continue to remove cultural and societal stigmas associated with seeking SUD services [28, 42].

The combination of SASE/AIMS funding was associated with predicted probability increases in the proportion of patients and visits compared to receiving only AIMS funding or no funding. For example, the proportion of patients and visits that were associated with SUD services increased at HCs with SASE/AIMS compared to HCs receiving only AIMS funding or no funding. However, only the proportion of patients associated with SUD services increased at HCs with AIMS only compared to HCs without funding. Similarly, we found an increase in the number of SUD patients and visits for HCs with both SASE/AIMS compared to both HCs receiving AIMS only and an increase in SUD patients but a decrease in SUD visits for HCs with AIMS only versus HCs without funding (S2 Table). This difference may have been due to distribution of AIMS funds in September 2017 and our inability to assess the full impact of this grant mechanism. In other words, HCs with AIMS only funding experienced growth in SUD staffing and the number of SUD patients, but we did not capture growth in other measures of capacity and service use, potentially due to the limited maturity of the AIMS grant. It is possible HCs that received both SASE/AIMS were initially better equipped to expand SUD capacity and services, due to the requirements of SASE to establish an integrated behavioral health/primary care model, compared to HCs that received AIMS only. Those with AIMS only may not have been equipped or incentivized to sustain increased SUD capacity.

The growth in specialized SUD capacity corresponded to a decrease in productivity of SUD staff. As an expected result of supplemental funding, these improvements may have led to changes in practice patterns among HCs that received these grants, but the patient panel size for SUD providers is still substantially lower than other provider types [40]. While increased staffing may have reduced the burden of care on existing providers and permitted reductions in panel size and provision of visits, there appears to be a balance between managing patients with increasing complex conditions and addressing unmet SUD needs in communities [43].

The increase in MAT capacity and service use from 2016 to 2017 within HCs is noteworthy and highlights HC efforts to expand their ability to manage SUD patients by primary care providers. Together, increased co-location and expansion of SUD and MAT capacity in HCs highlight the progress towards on-site SUD service delivery in these primary care settings. This progress is likely to benefit HC patients and potentially combat the opioid epidemic and its consequences. Our findings indicate that room for developing additional capacity still exists, particularly at HCs without SUD and MAT capacity. Our findings also imply the importance of identifying solutions to sustaining this growth in the longer term.

Growth in specialized SUD capacity is dependent of availability of such workforce and ability of HCs to recruit and retain these providers. For example, a report to Congress indicated that rural areas have greater turnover and challenges in recruiting SUD providers, and these shortages have a direct impact on quality of care [41]. Given the diversity of their patients, HCs need SUD staff that are culturally and linguistically competent [41]. Recruiting and retaining such staff can be an additional challenge [41]. Similarly, challenges to increasing MAT capacity also exist. Physicians may require additional training and education to better serve SUD patients, particularly those with complex comorbid conditions who have a high rate of relapse [28]. Furthermore, having an on-site or nearby pharmacy to obtain the prescribed medication is critical to the success of delivery of MAT at HCs and help complete the continuum of comprehensive care [42]. In 2017, 43% of HCs had on-site pharmacy staff, but this number may need to grow [21, 44].

Limitations

The limitations of our study include potentially underestimating SUD capacity because HCs have the option to report SUD services provided by mental health staff; we did not include mental health service provision in our analysis. Furthermore, SUD services are potentially undercounted because those provided by primary care or mental health providers are not included in UDS reporting and this bias is likely to underestimate any association between HRSA funding and the outcomes. We could not assess provision of SBIRT services by providers because HCs were not required to report this data in the UDS. Additionally, HRSA funds could have been used to employ or train staff, expand health information technology, or for other purposes in alignment with funding opportunities, but we lacked data on exactly how funds were used by HCs. We were unable to assess the proportion of national supply of MAT or SUD providers attributable to HCs, as providers can choose not to be listed in registries of providers with DATA waivers. Between 2016 and 2017, the definition of MAT provider was expanded beyond physicians to include certified nurse practitioners and physician assistants that were eligible to prescribe MAT for the treatment of Opioid Use Disorders, potentially overestimating our association between MAT capacity, service use, and panel size with receipt of HRSA supplemental funding. Furthermore, the most recently available UDS data was 2017, restricting our ability to assess the impact of the SASE and AIMS grants on SUD and MAT staffing, service use, provider panel size and visit ratio in later years. Our findings demonstrate that funding may promote colocation and increase in SUD and MAT capacity and service use are generalizable to HRSA-funded HCs but are also relevant to other health centers or primary care settings which focus on low-income and uninsured populations with a high burden of SUDs.

Conclusions

Combined SASE and AIMS funding corresponded to colocation and increased SUD capacity and service use among HCs that received these grants. However, given the rapid escalation of the opioid epidemic and mortality rate, SUD and MAT capacity and service delivery by HCs has to continue to increase [8, 12, 13, 45]. Future research is needed to evaluate longer term outcomes of HRSA efforts to promote SUD and MAT capacity, how HCs invested these grants, and what best practices can be scaled up.

In 2018, HRSA awarded an additional $350 million to HCs through the funding opportunity Expanding Access to Quality Substance Use Disorder and Mental Health Services (SUD-MH) [46]. These funds are intended to implement and advance evidence-based strategies to expand access to integrated SUD and mental health services and reflect the continued commitment of HRSA in tackling the opioid epidemic and will likely lead to more growth in SUD and MAT capacity in HCs [19]. Colocation of SUD and MAT services is an important step in delivery of early intervention to avoid SUD morbidity and mortality and is crucial in battling the opioid epidemic. Initiating or expanding SUD capacity by other safety net providers might be challenging, but expanding MAT capacity among primary care physicians, in other settings, and increasing the patient cap to MAT providers are attainable and necessary national strategies to reduce opioid mortality [28, 47].

Full regression models of substance use disorder capacity, service use, and panel size and visit ratio by grantee status.

(DOCX)

Click here for additional data file.

Percent change and predicted probabilities of substance use disorder patients and visits by grantee status, from 2010 to 2017.

(DOCX)

Click here for additional data file.

(XLSX)

Click here for additional data file.

Transfer Alert

This paper was transferred from another journal. As a result, its full editorial history (including decision letters, peer reviews and author responses) may not be present.

17 Aug 2020

PONE-D-20-00610

Examining trends in substance use disorder capacity and service delivery by Health Resources and Services Administration-funded health centers: A time series analysis

PLOS ONE

Dear Dr. Pourat,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

==============================

Please revise the title to better reflect the nature of this study. Please explain what time series analysis methods were used. ​

==============================

Please submit your revised manuscript by Sep 27 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

George Liu, PhD

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Congratulations on this work! This is a very useful paper; time trends have not been examined recently, and CHCs are such a critical part of the addiction treatment infrastructure in the US. I note some issues that could be clarified, and the takeaway messages should be sharpened. The gap between need and capacity could be presented more prominently (how many more FTE’s are needed?). After those corrections are made, this paper is a useful contribution to the literature.

Opioids are a high issue, but alcohol kills more people every year. When talking about SUD in the Intro, please also mention other drugs, at least once. You mention unmet need for addiction treatment in general, but the focus is squarely on opioids and it should not be. Meth is on the rise in many parts of the country, so we really don’t want to build opioid treatment capacity only—we want to build SUD treatment capacity. There is MAT for alcohol, but it is woefully underutilized.

One important issue that readers likely aren’t aware of is the fact that each health center operates multiple sites (6 or so, last time I checked). The should be stated clearly at the beginning of the paper, since when you say that 70% of CHCs have SUD capacity (line 107), this doesn’t mean that 70% of sites have capacity; it means that at least one site within 70% of grantees have capacity. This is a huge difference, and as currently written, the paper greatly overstates SUD capacity on the ground in health centers. Line 143 mentions that UDS data are at the grantee level, but not that most grantees operate multiple delivery sites. On line 19, the number of health center organizations is mentioned, and the number of care delivery sites should be included there as well.

The takeaway messages could be clearer in the abstract. In addition, this sentence in the abstract can be clarified: “From 2010 to 2015, 20% of health centers had any SUD staff, one full-time equivalent SUD staff

employed on average, and did not report a growth in SUD capacity or service delivery.”

One thing that should be discussed as a shortcoming is the way SUD capacity is measured: “(1) the proportion of HCs with at least one full-time equivalent (FTE) SUD staff, (2) the average number of SUD staff per HC, and (3) the ratio of SUD staff per 1,000 patients.” (line 180). So, all types of capacity aren’t captured, such as care provided by a MH specialist or primary care provider. This is a huge issue and should be described and made clear for the reader. Maybe the authors should say “specialist SUD capacity” instead of “SUD capacity.” This issue is mentioned in the Limitations, but attempts should be made to clarify this for the readers before they get to that point in the paper.

Was increasing SUD capacity a requirement of receiving the grants (the requirements should be described briefly)? If so, it’s surprising to readers that don’t consider the fact that a lot of SUD care is provided by non-SUD specialists that “Receipt of both supplementary grants increased the probability of any SUD capacity by 35%.” All the more reason to be clear on what is meant by capacity.

The service use measures might also be troubling. If they are from Table 6 in the UDS, those figures are commonly under-reported. If the numbers for “SUD patients” and “SUD visits” are just those patients served by SUD specialists, then this is an undercount since SUD patients/visits might be served by MH staff or primary care physicians. This shortcoming should be noted.

Line 123: a recent paper examines the impact of one of the grants: https://ps.psychiatryonline.org/doi/abs/10.1176/appi.ps.201900409

Line 122: Several newer studies are available that should be mentioned if the older and non-peer-reviewed papers are cited here (and I’d recommend removing the two non-peer-review cites, since they aren’t needed). These are just the ones that I know about, so rechecking the lit review on this point might make sense.

• Jones E. Medication-assisted opioid treatment prescribers in federally qualified health centers: capacity lags in rural areas. Journal of Rural Health. 2018;34(1):14-22

• Jones E, Rieckmann T. On-site mental health and substance use disorder screening and treatment capacity in health centers. Journal of Drug Issues. 2018;48(2):152-164.

• Jones E, Zur J, Rosenbaum S. Homeless caseload is associated with behavioral health and case management staffing in health centers. Administration and Policy in Mental Health and Mental Health Services Research. 2017;44(4):492-500.

• Jones E, Ku L. Sharing a playbook: integrated care in community health centers. American Journal of Public Health. 2015;105(10):2028-34.

• Jones E, Zur J, Rosenbaum S, Ku L. Opting out of Medicaid expansion: impact on encounters with behavioral health specialty staff in community health centers. Psychiatric Services. 2015;66(12):1277-82.

That’s neat that the authors used WONDER data too! I haven’t seen this merged with the UDS before. This doesn’t come across in the Abstract; I’d consider adding it there. I see NSSATS was also merged with the UDS, which might be mentioned in the abstract.

On average, Section 330 funding from HRSA only comprise 17% of each health center’s revenue (last time I checked). So, the “HRSA-funded” and “federally-funded” in the title and elsewhere should be eliminated, since it gives an outsize importance to HRSA funding.

Minor comments:

Line 106: it says self-reported data is likely an undercount, but the other data sources for SUD prevalence are largely self-report as well.

Line 173: Is PMCH recognition from the UDS? Also, it should be clarified that PCMH recognition could be at the site level and thus not cover all of a grantee’s sites.

I might have missed it, but I didn’t see a description of how the multivariable models were fitted.

Reviewer #2: PLOS One review Pourat et al

Examining trends in SUD capacity and service delivery by HRSA funded health centers: a time series analysis

This paper examines a critical topic, SUD staffing and service delivery by health centers. I believe work like this is critical to expand access to treatment beyond specialty care settings. In particular, the effects of HRSA’s SASE and AIMS grants are examined, which can be useful to guide future federal efforts to combat the opioid overdose epidemic.

Overall I feel the authors have done a good job with this paper, but two main concerns arose as I read it. These are detailed below.

• The SASE+AIMS group clearly outperformed the AIMS-only group, and the authors suggest AIMS simply hasn’t had time to reach maturity and they weren’t able to assess its full impact (p. 22). Both of these are reasonable suggestions, but it appears there may be more to it than that. It looks like the high-performing SASE+AIMS group was a fairly selective group (only 19% of the sample) that may have been qualitatively different. While I am not an expert in SASE and AIMS, my understanding is SASE was only available to FQHCS (recipients of section 330 grants), not FQHC look-alikes. If so, this appears to be an important confound to be acknowledged. AIMS, on the other hand, was not limited to FQHCs. As the authors acknowledge, at baseline the SASE+AIMS group had significantly more SUD staff, SUD visits at baseline, were bigger, etc. (Table 2, p. 15). In other words, they were better equipped to expand SUD treatment. These may have included FQHCs that either already had SUD treatment as part of their scope of project, or they added it as part of their participation in SASE (Form 5A of the application functioned as a Change in Scope request if SASE funding was received). By contrast AIMS did not require or include a change of scope. It appears, then, that the SASE grant would have provided its recipients with a springboard to increase SUD services because they could use the grant to expand these services, then (critically) they could sustain these services and staff beyond the end of the grant by billing for these services within their scope of project. For the AIMS-only group, on the other hand, a change of scope is explicitly not included, so the sustainability of the services may end along with the grant, providing them with far less incentive to expand. I am not disagreeing with the authors’ conclusions that SASE+AIMS was effective, but I think an important policy-relevant piece might be getting left out. It looks to me like the SASE grant likely helped, but it would be incorrect for a reader (and policymakers) to conclude the same results from SASE will necessarily occur with AIMS or other future funding. The lack of the same effect so far in the AIMS group seems to support this so far. The structure of SASE, particularly with respect to the scope of service and sustainability, may have been decisive. To be fair, the authors would have been criticized if they tried to make this point too strongly with the data available, but I would urge them to consider adding it to the discussion, limitations, and/or abstract if they agree it is likely. Right now as the paper stands I fear a reader of the abstract or conclusions (which are the only things some people read) would come away with the possibly oversimplified conclusion that all funding is inherently equally good funding. Having said all this, although I have done some work with FQHCs I do not consider myself an expert in FQHC financing so if HRSA or the authors feel I am off base on some part of this argument, maybe I am.

• The paper vacillates between discussing general SUD variables and opioid-related ones specifically, which can lead to confusion. Opioids are certainly a part of the SUD problem, but opioid trends can and often do diverge from those of other substances.

o For example, on p. 10 WONDER opioid mortality rates are presented as a measure for “substance abuse need”. Although the term “substance abuse need” is imprecise, I interpret it as shorthand for need for substance use disorder treatment. If this is correct, opioid mortality is not the same thing for a number of reasons, including the exclusion of other substances (cannabis, stimulants, alcohol and others all having their own trajectories) and changing lethality of substances over time (e.g. due to fentanyl), which can happen independently of treatment need as defined by SAMHSA (e.g. in their NSDUH survey). One fix would be to look for a better measure, but I don’t know of any readily available in annual form at the county level. The easiest fix is probably to simply say opioid mortality was used as a control, and not try to present it as a broader measure of need.

o Similarly, N-SSATS is used “to assess the supply of drug and alcohol treatment facilities in the county where the HC was located.” If the focus on opioids, a more precise control may be to count only treatment programs aimed at opioids (question 12 on the 2015 N-SSATS). However, again, some of the dependent variables in this paper are in fact for general SUD while others are specifically for opioids, which complicates things. One approach may be to use a different control variable for each DV, depending on the focus (SUD or OUD) though admittedly that would complicate the analyses substantially. Perhaps performing a sensitivity analysis, and just mentioning it in a footnote if it makes no difference, would be sufficient.

Minor items:

p. 21 “The growth in SUD capacity was timely and corresponded to increased service delivery, resulting in 0.11 SUD FTE staff per 1,000 patients in 2016, compared to national estimates of 0.07 outpatient SUD treatment staff per 1,000 patients.” “Delivery” is probably not the best term here, since staffing is being discussed rather than service delivery.

p. 23 “services provided by primary care providers are not included in UDS reporting.” What about SBIRT?

p. 24 “SASE and AIMS funding corresponded to colocation and increased SUD capacity and service use among HCs that received these grants.” Technically correct, but this could lead some readers to assume AIMS on its own has shown strong results. Consider rewording it to something along the lines of “combining SASE and AIMS funding . . . ”

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: Yes: Darren Urada

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

25 Sep 2020

Reviewer #1

Reviewer Comment

Congratulations on this work! This is a very useful paper; time trends have not been examined recently, and CHCs are such a critical part of the addiction treatment infrastructure in the US. I note some issues that could be clarified, and the takeaway messages should be sharpened. The gap between need and capacity could be presented more prominently (how many more FTE’s are needed?). After those corrections are made, this paper is a useful contribution to the literature.

Response to Reviewers

Thank you for your comments. We have revised the paper in response to reviewer’s comments.

Location in Manuscript

No change

Reviewer Comment

Opioids are a high issue, but alcohol kills more people every year. When talking about SUD in the Intro, please also mention other drugs, at least once. You mention unmet need for addiction treatment in general, but the focus is squarely on opioids and it should not be. Meth is on the rise in many parts of the country, so we really don’t want to build opioid treatment capacity only—we want to build SUD treatment capacity. There is MAT for alcohol, but it is woefully underutilized.

Response to Reviewers

We broadened this discussion as suggested.

Location in Manuscript

Introduction, Page 6, line 140-145

Reviewer Comment

One important issue that readers likely aren’t aware of is the fact that each health center operates multiple sites (6 or so, last time I checked). The should be stated clearly at the beginning of the paper, since when you say that 70% of CHCs have SUD capacity (line 107), this doesn’t mean that 70% of sites have capacity; it means that at least one site within 70% of grantees have capacity. This is a huge difference, and as currently written, the paper greatly overstates SUD capacity on the ground in health centers. Line 143 mentions that UDS data are at the grantee level, but not that most grantees operate multiple delivery sites. On line 19, the number of health center organizations is mentioned, and the number of care delivery sites should be included there as well.

Response to Reviewers

The authors have clarified and included the number of delivery sites that are operated by health center organizations throughout the manuscript.

Location in Manuscript

Introduction, Page 7, line 169; lines 176-177

Methods, Page 10, line 245

Throughout manuscript

Reviewer Comment

The takeaway messages could be clearer in the abstract. In addition, this sentence in the abstract can be clarified: “From 2010 to 2015, 20% of health centers had any SUD staff, one full-time equivalent SUD staff employed on average, and did not report a growth in SUD capacity or service delivery.”

Response to Reviewers

The authors have clarified this sentence. In addition, the authors have edited the Conclusions section of the Abstract.

Location in Manuscript

Abstract, Page 4, line 61-63, line 69-122

Reviewer Comment

One thing that should be discussed as a shortcoming is the way SUD capacity is measured: “(1) the proportion of HCs with at least one full-time equivalent (FTE) SUD staff, (2) the average number of SUD staff per HC, and (3) the ratio of SUD staff per 1,000 patients.” (line 180). So, all types of capacity aren’t captured, such as care provided by a MH specialist or primary care provider. This is a huge issue and should be described and made clear for the reader. Maybe the authors should say “specialist SUD capacity” instead of “SUD capacity.” This issue is mentioned in the Limitations, but attempts should be made to clarify this for the readers before they get to that point in the paper.

Response to Reviewers

We included further clarification for focusing on SUD staff and inserted the word “specialized” several times to remind readers of the scope of this analysis.

Location in Manuscript

Introduction, Page 9, line 228-230

Throughout manuscript

Reviewer Comment

Was increasing SUD capacity a requirement of receiving the grants (the requirements should be described briefly)? If so, it’s surprising to readers that don’t consider the fact that a lot of SUD care is provided by non-SUD specialists that “Receipt of both supplementary grants increased the probability of any SUD capacity by 35%.” All the more reason to be clear on what is meant by capacity.

Response to Reviewers

We have added additional detail on grant requirements to the text. These details indicate that screening, brief intervention, referrals, and care coordination, and collaboration with SUD providers were intended outcomes. We could not measure if SUD treatment by primary care providers increased, though we agree that primary care providers may have increased such activities.

We have added the term specialized SUD capacity when possible in response to this comment.

Location in Manuscript

Introduction, Page 8, line 189-195; line 199-204

Throughout manuscript

Reviewer Comment

The service use measures might also be troubling. If they are from Table 6 in the UDS, those figures are commonly under-reported. If the numbers for “SUD patients” and “SUD visits” are just those patients served by SUD specialists, then this is an undercount since SUD patients/visits might be served by MH staff or primary care physicians. This shortcoming should be noted.

Response to Reviewers

Our service use measures are from Table 5a. We had previously noted this in the Limitations section and expanded on this matter.

Location in Manuscript

Limitations, Page 24, line 477-478

Reviewer Comment

Line 123: a recent paper examines the impact of one of the grants:

https://ps.psychiatryonline.org/doi/abs/10.1176/appi.ps.201900409

Response to Reviewers

Thank you for this citation. We have included it in the Introduction.

Location in Manuscript

Introduction, Page 9, line 217-218

Reviewer Comment

Line 122: Several newer studies are available that should be mentioned if the older and non-peer-reviewed papers are cited here (and I’d recommend removing the two non-peer-review cites, since they aren’t needed). These are just the ones that I know about, so rechecking the lit review on this point might make sense.

• Jones E. Medication-assisted opioid treatment prescribers in federally qualified health centers: capacity lags in rural areas. Journal of Rural Health. 2018;34(1):14-22

• Jones E, Rieckmann T. On-site mental health and substance use disorder screening and treatment capacity in health centers. Journal of Drug Issues. 2018;48(2):152-164.

• Jones E, Zur J, Rosenbaum S. Homeless caseload is associated with behavioral health and case management staffing in health centers. Administration and Policy in Mental Health and Mental Health Services Research. 2017;44(4):492-500.

• Jones E, Ku L. Sharing a playbook: integrated care in community health centers. American Journal of Public Health. 2015;105(10):2028-34.

• Jones E, Zur J, Rosenbaum S, Ku L. Opting out of Medicaid expansion: impact on encounters with behavioral health specialty staff in community health centers. Psychiatric Services. 2015;66(12):1277-82.

Response to Reviewers

We have updated our cited literature to more recent studies as suggested.

Location in Manuscript

Introduction, Page 9, line 215

Reviewer Comment

That’s neat that the authors used WONDER data too! I haven’t seen this merged with the UDS before. This doesn’t come across in the Abstract; I’d consider adding it there. I see NSSATS was also merged with the UDS, which might be mentioned in the abstract.

Response to Reviewers

The authors have included in the Abstract mention of these two data sources. However, the name of these data sources were not included there due to word limits.

Location in Manuscript

Abstract, Page 4, line 60-61

Reviewer Comment

On average, Section 330 funding from HRSA only comprise 17% of each health center’s revenue (last time I checked). So, the “HRSA-funded” and “federally-funded” in the title and elsewhere should be eliminated, since it gives an outsize importance to HRSA funding.

Response to Reviewers

We have used the term “HRSA-funded” to highlight the fact that some health centers do not receive 330 funding. We agree with the reviewer that this amount is a small proportion of HC funding on average, but the proportion varies and it is higher for some HCs. It is also an important factor in the amount of care they can provide to uninsured patients. We have kept this language only in three locations.

Location in Manuscript

Throughout manuscript

Reviewer Comment

Minor comments:

Line 106: it says self-reported data is likely an undercount, but the other data sources for SUD prevalence are largely self-report as well.

Response to Reviewers

The authors have acknowledged this point in the text.

Location in Manuscript

Introduction, Page 7, line 175-176

Reviewer Comment

Line 173: Is PMCH recognition from the UDS? Also, it should be clarified that PCMH recognition could be at the site level and thus not cover all of a grantee’s sites.

Response to Reviewers

Yes, PCMH recognition is obtained from UDS data and the authors have mentioned this as the data source. The authors have also clarified that PCMH recognition can be obtained at the site level.

Location in Manuscript

Methods, Page 11, line 273-276

Reviewer Comment

I might have missed it, but I didn’t see a description of how the multivariable models were fitted.

Response to Reviewers

We have described the needed information on how the models were fitted in the methods. However, we did not include fit statistics, because we are using random-effect models and cross-sectional longitudinal data. There are no readily available fit statistics for such models. However, in response to reviewer’s question, we examined the residuals, concordance, and correlation between predicted and observed values of our outcomes of interest manually. We found high correlation and high concordance between our predicted and observed values for all our outcome variables, indicating our models are appropriately fitted.

Location in Manuscript

Methods, Page 13, line 316

Reviewer #2

Reviewer Comment

This paper examines a critical topic, SUD staffing and service delivery by health centers. I believe work like this is critical to expand access to treatment beyond specialty care settings. In particular, the effects of HRSA’s SASE and AIMS grants are examined, which can be useful to guide future federal efforts to combat the opioid overdose epidemic.

Overall I feel the authors have done a good job with this paper, but two main concerns arose as I read it. These are detailed below.

• The SASE+AIMS group clearly outperformed the AIMS-only group, and the authors suggest AIMS simply hasn’t had time to reach maturity and they weren’t able to assess its full impact (p. 22). Both of these are reasonable suggestions, but it appears there may be more to it than that. It looks like the high-performing SASE+AIMS group was a fairly selective group (only 19% of the sample) that may have been qualitatively different. While I am not an expert in SASE and AIMS, my understanding is SASE was only available to FQHCS (recipients of section 330 grants), not FQHC look-alikes. If so, this appears to be an important confound to be acknowledged. AIMS, on the other hand, was not limited to FQHCs. As the authors acknowledge, at baseline the SASE+AIMS group had significantly more SUD staff, SUD visits at baseline, were bigger, etc. (Table 2, p. 15). In other words, they were better equipped to expand SUD treatment. These may have included FQHCs that either already had SUD treatment as part of their scope of project, or they added it as part of their participation in SASE (Form 5A of the application functioned as a Change in Scope request if SASE funding was received). By contrast AIMS did not require or include a change of scope. It appears, then, that the SASE grant would have provided its recipients with a springboard to increase SUD services because they could use the grant to expand these services, then (critically) they could sustain these services and staff beyond the end of the grant by billing for these services within their scope of project. For the AIMS-only group, on the other hand, a change of scope is explicitly not included, so the sustainability of the services may end along with the grant, providing them with far less incentive to expand. I am not disagreeing with the authors’ conclusions that SASE+AIMS was effective, but I think an important policy-relevant piece might be getting left out. It looks to me like the SASE grant likely helped, but it would be incorrect for a reader (and policymakers) to conclude the same results from SASE will necessarily occur with AIMS or other future funding. The lack of the same effect so far in the AIMS group seems to support this so far. The structure of SASE, particularly with respect to the scope of service and sustainability, may have been decisive. To be fair, the authors would have been criticized if they tried to make this point too strongly with the data available, but I would urge them to consider adding it to the discussion, limitations, and/or abstract if they agree it is likely. Right now as the paper stands I fear a reader of the abstract or conclusions (which are the only things some people read) would come away with the possibly oversimplified conclusion that all funding is inherently equally good funding. Having said all this, although I have done some work with FQHCs I do not consider myself an expert in FQHC financing so if HRSA or the authors feel I am off base on some part of this argument, maybe I am.

Response to Reviewers

We agree with the reviewer and have acknowledged these points in the Discussion section.

Please note that our data only included 330 grantees so our findings are not confounded by differences between HCs with and without 330 funding.

Location in Manuscript

Discussion, Page 23, lines 439-443

Reviewer Comment

• The paper vacillates between discussing general SUD variables and opioid-related ones specifically, which can lead to confusion. Opioids are certainly a part of the SUD problem, but opioid trends can and often do diverge from those of other substances.

o For example, on p. 10 WONDER opioid mortality rates are presented as a measure for “substance abuse need”. Although the term “substance abuse need” is imprecise, I interpret it as shorthand for need for substance use disorder treatment. If this is correct, opioid mortality is not the same thing for a number of reasons, including the exclusion of other substances (cannabis, stimulants, alcohol and others all having their own trajectories) and changing lethality of substances over time (e.g. due to fentanyl), which can happen independently of treatment need as defined by SAMHSA (e.g. in their NSDUH survey). One fix would be to look for a better measure, but I don’t know of any readily available in annual form at the county level. The easiest fix is probably to simply say opioid mortality was used as a control, and not try to present it as a broader measure of need.

Response to Reviewers

The authors have examined the available data and could not find publicly available data to measure substance use disorder mortality at the county level. We have reframed as the reviewer suggested to describe opioid mortality as a control measure.

Location in Manuscript

Methods, Page 11, line 278

Reviewer Comment

o Similarly, N-SSATS is used “to assess the supply of drug and alcohol treatment facilities in the county where the HC was located.” If the focus on opioids, a more precise control may be to count only treatment programs aimed at opioids (question 12 on the 2015 N-SSATS). However, again, some of the dependent variables in this paper are in fact for general SUD while others are specifically for opioids, which complicates things. One approach may be to use a different control variable for each DV, depending on the focus (SUD or OUD) though admittedly that would complicate the analyses substantially. Perhaps performing a sensitivity analysis, and just mentioning it in a footnote if it makes no difference, would be sufficient.

Response to Reviewers

We agree that this control variable is not specific to opioid treatment supply. We tested using the variable suggested by the reviewer and found our main findings to be unchanged. We did not replace our original variable because the opioid treatment programs are aggregated at the state-level which is less precise than our measure of supply. Therefore, we kept the original variable in the models.

Location in Manuscript

No change

Reviewer Comment

Minor items:

p. 21 “The growth in SUD capacity was timely and corresponded to increased service delivery, resulting in 0.11 SUD FTE staff per 1,000 patients in 2016, compared to national estimates of 0.07 outpatient SUD treatment staff per 1,000 patients.” “Delivery” is probably not the best term here, since staffing is being discussed rather than service delivery.

Response to Reviewers

We have revised this sentence for clarity.

Location in Manuscript

Discussion, Page 22, line 417

Reviewer Comment

p. 23 “services provided by primary care providers are not included in UDS reporting.” What about SBIRT?

Response to Reviewers

HCs report SBIRT overall in Table 6 but do not indicate who conducted the screening or intervention. So we did not examine SBIRT. We have added this limitation.

Location in Manuscript

Limitations, Page 25, line 480-481

Reviewer Comment

p. 24 “SASE and AIMS funding corresponded to colocation and increased SUD capacity and service use among HCs that received these grants.” Technically correct, but this could lead some readers to assume AIMS on its own has shown strong results. Consider rewording it to something along the lines of “combining SASE and AIMS funding . . . ”

Response to Reviewers

We have revised this sentence as suggested.

Location in Manuscript

Conclusions, Page 25, line 497

Academic Editor

Reviewer Comment

Please revise the title to better reflect the nature of this study. Please explain what time series analysis methods were used.

Response to Reviewers

We have edited our title to include our methodology and clarified our methods in the Methods section.

Location in Manuscript

Title

Methods, Page 13, line 316

Submitted filename: Response to Reviewers - SUD capacity-final.docx

Click here for additional data file.

3 Nov 2020

Examining trends in substance use disorder capacity and service delivery by Health Resources and Services Administration-funded health centers: A time series regression analysis

PONE-D-20-00610R1

Dear Dr. Pourat,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

George Liu, PhD

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Great job addressing the comments! I have no further comments.

Reviewer #2: The authors have done a good job addressing previous comments. I am satisfied with their responses.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: Yes: Darren Urada

17 Nov 2020

PONE-D-20-00610R1

Examining trends in substance use disorder capacity and service delivery by Health Resources and Services Administration-funded health centers: A time series regression analysis

Dear Dr. Pourat:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. George Liu

Academic Editor

PLOS ONE