Varying Estimates of Social Workers in the United States: Which Data Source to Use?

Abstract

Behavioral health needs continue rise in the United States and constitute a key driver of health care utilization, costs, and outcomes. Social workers provide a wide range of services across health, behavioral health, and community settings, and while this heterogeneity in practice benefits care delivery, it complicates health workforce analyses. This analysis compares five commonly used national data sources and details similarities and differences in their estimates of the number, type, and practice characteristics of social workers. The analysis suggests that estimates vary significantly between data sets ranging from 282,425 to 1,022,859 social workers; as well as yield different findings of types of social workers in the United States, depending on the data set used. These differences have the potential to shape how researchers and policy makers assess the adequacy of the social work workforce and identify solutions to address the nation’s behavioral health and social care needs.

Emerging from the COVID-19 pandemic, the United States faces a marked behavioral health crisis with wide-ranging impacts. In 2021, overdose-related deaths were at their highest rate in the last 25 years (Spencer et al., 2022), 42% of youth reported feeling persistently sad or hopeless (Centers for Disease Control and Prevention [CDC], 2022), and moral injury, burnout and emotional exhaustion which were already problematic concerns for the health workforce, have intensified (Dzau et al., 2020). Despite the clear and urgent need for a range of behavioral health supports, researchers, policymakers, and educators, available data sources limit their ability to understand the nation’s behavioral health workforce, including the size, distribution, and practice characteristics (Beck et al., 2018; Hyde, 2013) of workers providing services. This lack of workforce data has stymied the development of policy interventions and investments in the behavioral health workforce (National Academies of Sciences, Engineering, and Medicine [NASEM], 2019). The result is that significant gaps remain in how best to increase the capacity of the behavioral health workforce to address the U.S. behavioral health crisis (Beck et al., 2018).

Even prior to the pandemic, there was growing recognition of the importance of addressing the psychosocial and behavioral health factors that impact physical health and the need to incorporate workforces that understand and are trained to address behavioral health and social care needs (NASEM, 2019). New payment models that shift from fee-for-service models to value-based care have highlighted the importance of building interprofessional teams that address physical, behavioral, and social needs (Fraher et al., 2022; Fraher & Ricketts, 2016). This shift requires researchers, policy makers, employers, and educators to understand the current and future adequacy of the workforce needed to deliver these services, including social workers who have not traditionally been counted as part of the health care workforce.

The combination of these forces—an increased focus on identifying and ameliorating behavioral health needs and social risks as well as changing payment models is shifting care from traditional health settings (e.g., hospitals) into community-based clinics, requires expanding settings and professions included in the health workforce ecosystem (Fraher & Ricketts, 2016). Social workers act as a boundary spanning profession that works across traditional health, behavioral health, and community-based settings (Fraser et al., 2018; NASEM, 2019; Steketee et al., 2017). Social workers serve as clinicians diagnosing and treating mental health and substance use conditions (Fraser et al., 2018), provide psychosocial support, (Lombardi et al., 2019; Martinez et al., 2019), act as care managers for vulnerable populations (Berrett-Abebe et al., 2020), and are often responsible for addressing social risk factors including food insecurity and creating linkages to community referrals (O’Brien, 2019). Because of the wide range of services social workers may provide, they are increasingly hired and embedded on teams delivering behavioral or social care in primary care and specialty care (NASEM, 2019; Zerden et al., 2020). According to the Bureau of Labor Statistics (BLS, 2023), the number of social workers employed in health care has increased by 15% since 2018 and is expected to rise faster than the average growth rate of all health care occupations. Specifically, social workers employed in substance use and mental health sectors are the fastest growing social work professions, with 11% expected job growth by 2032 (BLS, 2023).

The expected growth of the social work workforce aligns with the significant growth in the social work educational programs. Today, there are 541 accredited Bachelor of Social Work (BSW) programs and 313 Master of Social Work (MSW) degree programs in the United States (Council on Social Work Education [CSWE], 2022). As the social work profession has grown, practice has shifted both in function and setting (Williams & Vieyra, 2018). Social workers are heterogeneous workforce with different degrees and licensure statuses that allow them to perform different functions. At the master’s level, social workers can seek licensure (after examination and supervised practice) to provide clinical services and practice independently. The Centers for Medicare and Medicaid Service (CMS) consider graduate-level licensed clinical social workers a reimbursable provider whose scope of practice includes assessment, diagnosis, psychotherapy, crisis intervention, and telehealth services (Page et al., 2017). Masters-level social workers who are licensed to practice independently by their respective state may have different scopes of practice based on state laws and different titles. For example, in 37 states, these social workers are called “LCSWs” (Licensed Clinical Social Workers) but may also be titled Licensed Independent Clinical Social Worker (LICSW) or Licensed Independent Social Worker (LISW). Graduate level social workers may also obtain a provisional license (while working toward independent licensure). Titles for these licenses also vary and can include Licensed Clinical Social Worker-Associates (LCSW-A), Licensed Social Worker (LSW), or Licensed Masters Social Worker (LMSW). Social workers may also be licensed at the bachelor’s level with varied state scopes of practice (but not independent practice or billing authority). Bachelors’ level titles also vary from state to state with most holding a Licensed Bachelor Social Worker (LBSW) title [More details about social work license and regulation can be found at the Association of Social Work Boards website].

Despite social workers’ growing role in the United States, health care system and their status as one of the largest groups of behavioral health providers in the country, there is a lack of adequate comprehensive and rigorous data on the size, geographic distribution, employment setting, educational preparation, and practice patterns of these professionals (Williams & Vieyra, 2018). The lack of data limits policymakers’ ability to understand if the current social work workforce is adequate in supply to meet demand, if enough new entrants are being trained in needed practice areas, and if the workforce is geographically distributed to meet the nation’s current and future health care needs. These data are needed to inform a range of decisions, including about whether to open or expand training programs in different locations, determine the numbers and types of social work graduates are needed to meet future demand, and gauge how well the social work workforce is prepared, now and in the future, to meet the U.S. health care system’s evolving needs (Fraher & Knapton, 2021; Ono et al., 2013).

Without intentional, decision-making informed by data, the United States may face future social work staffing shortages, resulting in long wait times for clients or clients’ inability to access needed behavioral health services. An ineffectively distributed social work workforce can exacerbate disparities in access to health care services, particularly in rural and underserved communities where patients often have higher health care needs and poorer health outcomes (Andrilla et al., 2021; Douthit et al., 2015; Holden et al., 2023; Sharma et al., 2024). Valid workforce data are critical to addressing important policy questions that could expand access to care, like what proportion of social work services are paid for by Medicaid or Medicare compared with private or out of pocket and what factors may increase participation in public insurance programs.

Unlike national data sets for physicians and dentists (American Medical Association [AMA], n.d.; Munson & Vujicic, 2021) and nationally representative data sources for nurses (e.g., National Sample Survey of Registered Nurses), few comprehensive, national data sets exist for social workers or for the behavioral health workforce more broadly (Beck et al., 2018). Despite calls to develop a standardized behavioral health workforce Minimum Data Set to address behavioral health sufficiency and workforce (mal)distribution issues, researchers have not reached consensus regarding how this Minimum Data Set should be structured, disseminated, and implemented across numerous types of behavioral health clinicians (Beck et al., 2018). Thus, in 2023, we have no definitive data source for the social work workforce, making it particularly challenging to develop, target and evaluate workforce development initiatives to meet the United States’ current and emerging behavioral health needs. Instead, researchers and policy makers need to rely on a mix of different state and national data sets that are limited in their ability to inform workforce planning.

This study analyzes five commonly used data sources and compares the availability and range of estimates on the total number of social workers in the workforce, demographic characteristics, practice location, employment setting, educational degree, and practice area. More specifically, the study assesses each data source’s ability to enumerate the multiple types of social workers that comprise the workforce, including bachelors prepared social workers, masters prepared social workers, and licensed clinical social workers.

New Contributions

While the workforce characteristics of some health professionals, like physicians and nurses, are captured using a Minimum Data Set, data on the social work workforce is less systemically collected and reported. Without better data sources and systematic approaches to classify and enumerate the social work workforce, there continues to be wide variability in how this growing profession is estimated. As behavioral health and social care needs continue to rise and remain inextricably linked to drivers of health care utilization, costs, and outcomes, social work data sources are necessary to improve population health needs and future planning. This study provides workforce researchers, policy makers, and educators with an analysis of the strengths, weaknesses, and implications of using five existing data sources for analysis of the composition and distribution of the social worker workforce.

Methods

Existing Data Sources on the Social Work Workforce

Using descriptive analyses, this paper compares five data sources that include social workers: (a) the American Community Survey (ACS); (b) the BLS’ Occupational and Employment Wage Statistics; (c) the National Plan and Provider Enumeration System; (d) state licensure data; and (e) educational program data. We describe how these data sources identify, collect and report information on social workers, including estimates of the size of the SW workforce, subcategories of SWs, description of the SW workforce included, available demographic information, level of geography available of practice, and strengths and limitations of each source.

Results

ACS Data

The ACS is a nationally representative data source for the U.S. population administered yearly by the U.S. Census Bureau. The ACS asks respondents to self-report their occupation using the Standard Occupational Classification (SOC), a federal classification system that assigns workers into occupational categories. Beginning in 2018, the ACS included four categories of social work practice drawn from SOC: (a) Child, Family, and School; (b) Health care; (c) Mental Health and Substance Abuse (MH/SUD); and (d) All Other.

Prior work about the social work workforce has used two indicators within the ACS to identify the population of clinical social workers: an SOC designation of “social work” and an educational degree of master’s or higher (Salsberg et al., 2017). It is possible to obtain the major of a bachelor’s degree and identify bachelors level social workers in ACS who report working as social worker. The ACS estimates there are 1,022,859 total social workers with 409,532 at the master’s degree or higher (see Table 1).

Table 1.

Details on Available Social Work Data Sources.

Year of data estimate	American Community Survey	ASWB (Licensure Data)	NPPES	OEWS/BLS	IPEDS
	2019	2021	2021	2020	2020
Estimate of the number of social work workforce	1,022,859	541,635	282,425	715,600	54,569
Estimates available by social work types, categories, or education	Social Work with Masters degree or higher: 409,532	Clinical: 325, 442 Master’s: 143,051 Bachelor: 68,445 Associate: 4,697		Child & school: 335,300 Health: 184,900 MH/SUD: 124,000 All other: 71,400	Bachelor: 21,204 Master’s: 33,455
Description of social work sample included in data source	Individuals who report they work as a “social worker”; can limit sample by degree	All licensed social workers in the United States	Individuals who obtained a National Provider Identifier (NPI) and indicate taxonomy as “social worker”	Individuals employed as a “social worker” in the U.S. workforce	All graduates of U.S. postsecondary institutions with a bachelor’s or master’s degree in social work
Available variables to characterize the workforce	Age; race/ethnicity; gender; all ACS variables	_	Gender; free text credential; license number	Type of setting; mean wage	Race/ethnicity
Geographic data on the workforce	ZIP code	State	Practice address	State	State
Strengths of data source	• Includes SWs who may not have a clinical license but are practicing in field • Includes demographic data of workforce • Includes breakdown of SW by specialty (matches BLS/OEWS)	• Licensed SWs are likely the clinical SW workforce	• Likely the sample is practicing and billing clinically • Practice address/location available to analyze	• Includes SWs employed in the field (likely those that are clinically practicing) • Can differentiate MH/SUD from other types of practice	Can observe number of new entrants to the workforce
Limitations of data source	• Includes individuals without a SW degree • No information on specialty of MSW degree • Practice location may be different from location where individual lives (possibly lives/works across states)	• Some states do not promptly update ASWB • May double-count individuals licensed in two or more states	• Only includes those with NPI; may exclude SWs who do not bill for services (e.g., domestic violence shelter) • Unclear how often NPI practice address is updated	• Unclear if sample includes MSW or clinical license • Unclear how self-employed SWs are included • May not include those with a SW degree with a different title (e.g., therapist)	• Not all those who graduate with SW degree go into the field • Many BSWs go on to get MSW

Note. ACS = American Community Survey; ASWB = Association of Social Work Boards; BLS = Bureau of Labor Statistics; IPEDS = Integrated Postsecondary Education Data System; MH/SUD = Mental Health and Substance Use Disorder; NPPES = National Plan and Provider Enumeration System; NPI = National Provider Identifier; OEWS = Occupational Employment and Wage Statistics; SW = Social Work.

The ACS does not include information on the major of the master’s degree (i.e., a master’s degree in social work), and ACS-designated social workers may hold a master’s degree in any field. In addition, occupational information is self-reported, and individuals not trained in social work may identify as a social worker if they are employed in a similar field. Because the ACS interviewer is responsible for coding the occupations held by respondents, interviewers may code a respondent as a “social worker” when the respondent is in fact employed in an adjacent field (e.g., child welfare). This limitation may occur in states that do not have legislated title protection, which allows only individuals who possess a social work degree to use the title of “social worker.” Conversely, some social workers may professionally identify with titles such as “counselor” and “case manager” when in fact they are formally trained as social workers and hold a social work degree. The ACS does not include information on licensure type which limits the ability to use this data source to estimate the size of the social work workforce that can practice independently to provide clinical services.

ACS data include person-level reported demographic characteristics including race, ethnicity, age, and gender, and allow for analysis at the ZIP code level, will enable substate analysis of the characteristics of the social work workforce. In addition, ACS includes information on job change, age (to estimate retirement), and family characteristics that would assist in modeling the capacity of the workforce.

BLS Data

The BLS produces state and national estimates of the social work workforce gathered through the Occupation Employment and Wage Survey (OEWS). The OEWS is a semi-annual survey of employers from non-farm industries that, similar to the American Community Survey, uses the SOC to classify social workers across four domains: (a) Child, Family, and School Social Workers; (b) Health Care Social Workers; (c) MH/SUD; (d) All Other Social Workers. OEWs reports there are more than 715,600 social workers; with the largest proportion working as Child, Family, and School Social Workers (n = 335,300; Table 1).

The fact that worker’s occupations in the OEWS are classified by their employer can complicate efforts to accurately estimate the social work workforce. The data do not indicate whether the employee is providing services at the master’s or bachelor’s level, and do not indicate employees’ licensure status. It is also unknown how this data set classifies social workers who are employed under a different title in their position (e.g., behavioral health care manager). For instance, as shown in Table 1, BLS reports a large number of social workers in total, yet the number of MH/SUD social workers it reports is significantly lower than other data sources. This finding suggests that the OEWS’s MH/SUD category may undercount the number of social workers practicing in this focus area.

Furthermore, the OEWS does not collect information on self-employed individuals, making it difficult to determine the percentage of licensed clinical social workers who are self-employed. A recent survey drawn from members of the National Association of Social Workers (NASW) found approximately 65% of the sample of social workers worked in private practice; however, as this survey only included members of this professional organization, results may not represent the true percentage of social workers working in private practice nationally (Lombardi et al., 2022). As such, BLS data likely underestimates the number of social workers who provide clinical behavioral health services independently in private practice.

Unlike the ACS, BLS does not provide person-level data, and no sociodemographic information or practice address data are available. However, it does provide details on the industry (North American Industry Classification System [NAICS] code) in which social workers are employed suggesting future work could examine the types of settings social workers of different types work within.

National Plan and Provider Enumeration System

The Centers for Medicare and Medicaid Service (CMS) NPPES collects information on providers eligible to bill CMS for services through a unique identifier system called the National Provider Identifier (NPI). Having an NPI is a requirement for all providers who bill CMS for services. Each provider within NPPES is categorized into a provider specialty “type” known as a taxonomy code, which the provider selects when registering for an NPI number. There are three taxonomies for social workers within NPPES: (a) Social Worker—104100000X, (b) Clinical—1041C0700X, and (c) School—1041S0200X. NPPES includes the clinician’s practice address and respondents can also report gender, license number, and additional credentials. Social workers in the NPPES who have an NPI are likely individuals in clinical settings in which they or their agency or organization receives payments from CMS. The NPPES includes 285,000 social workers, 69% reported a primary taxonomy of “Clinical,” 30% reported a primary taxonomy of “Social Worker,” and 1% reported a primary taxonomy of “School.”

The NPPES relies on providers to voluntarily update their practice address if they move positions or clinical locations. In reality, such updates may never be made or may occur at irregular intervals from the original registration, leading to inaccurate data (DesRoches et al., 2015). NPPES includes a facility file with practice addresses of various facilities; however, because the provider file and facility taxonomies are not connected, the data do not allow analysts to link individual social workers to their employment settings.

As NPPES is an administrative dataset for clinicians eligible to bill CMS, it is likely that only includes social workers at the independent clinical level but may also include social workers at the master’s level who are billing “incident to” another provider on teams. It is unclear which proportion of licensed independent clinical social workers have an NPI. It is likely this data source underestimates the number of clinical social workers practicing in the United States.

Licensure Data

State licensure bodies collect administrative data to ensure the workforce meets educational and clinical practice requirements, avoids misconduct, and maintains professional competence. The Association of Social Work Boards (ASWB) also collects data on the licensed workforce in each state. ASWB collects data on four types of social work license: (a) Associate licenses (less than a college degree); (b) Bachelors level; (c) Masters level; (d) Independent clinical license. According to the Association of Social Work Boards, which sponsors the social work licensure exams, there are 541,635 social workers across licensure types: 325,442 clinical independent social workers; 143,051 Masters social workers, 68,445 bachelors level licensed social workers, and 4,697 associate level social workers.

The utility of social work licensure data is limited by the lack of a uniform licensure system: all 50 states, Washington, DC, and the U.S. territories have their own licensure boards, each with different scopes of practice and state-specific nuances of the Social Work Practice Act (Morrow, 2023). Furthermore, licenses issued for the same educational levels and scope of practice may have different titles across states. For instance, depending on the state, master’s-level independent social workers may be called Licensed Clinical Social Workers (LCSWs) (e.g., Arizona, California), Licensed Independent Social Workers (LISWs) (e.g., Iowa, New Mexico), Licensed Independent Clinical Social Workers (e.g., Alabama, New Hampshire), or Licensed Graduate Social Workers (LGSW) (e.g., Minnesota) (Kovacs, 2023). The Association of Social Work Boards (ASWB, 2023) keeps an updated registry of social work licensures and requirements by degree level on their website.

Although to practice independently (without any supervision) and bill CMS a social worker would need a clinical license, not all social workers are licensed. For example, a masters-level social worker working in a domestic violence shelter providing group therapy that does not bill insurance may not be licensed. As such, a limitation of using licensure data to enumerate the social work workforce is that it is unclear what percentage of all eligible social workers seek and maintain licensure. A recent study of master’s-level social work graduates found that 82% of the sample planned to pursue licensure (Salsberg et al., 2020). This finding provides a useful estimation of the proportion of future graduates who may become licensed. However, because social workers may be working in settings where they are not required to have a license, licensure data possibly underestimate the United States’ practicing social work workforce.

ASWB maintains the number of licensed social workers by state for all types of social work licenses. However, the data elements collected by ASWB between states are not uniform. For example, in some states the practice address may be recorded while in other states the address listed is the practitioner’s home address. Furthermore, some social workers are licensed in more than one state, meaning that licensure data may overestimate the number of licensed clinicians. Yet because no research has examined the proportion of social workers licensed in more than one state, the extent of this overestimation remains unclear.

Graduation Data

Social work program graduation data can be used to enumerate the new entrants into the workforce and evaluate which programs and states are producing the most graduates. Currently, two sources of data on social work graduates include (a) the Integrated Post-Secondary Education System (IPEDS) and (b) CSWE. IPEDS is a publicly available data source that collects information directly reported by colleges and universities. Institutions that receive Title IV funding are required to submit graduation information; those that do not receive federal funding may voluntarily submit it. Because IPEDS requires race and ethnicity data reporting, this database may contain useful information about the future diversity of and demographic trends within the social work workforce. IPEDS reports United States conferred 21,204 BSW degrees and 33,455 MSW degrees in 2020 (see Table 1).

CSWE collects counts of graduates of accredited social work programs through its Annual Survey of Social Work Programs. In 2020, CSWE reported receiving data on 90.7% of all master’s-level accredited social work programs. Beyond the count of social work degrees conferred, the CSWE survey gathers information on the number of social work students graduating in specialty training programs or certificates (i.e., child welfare; substance use, school social work), data on graduating student demographics (e.g., race/ethnicity; gender; and age), and information on the amount of student loan debt upon graduation.

However, graduation data may have limited utility in calculating the size of—let alone emerging trends within—the social work profession, as not all graduates pursue employment in direct clinical practice or in a position that requires a social work degree. For instance, a study of new MSW graduates found that 23% worked in a position that did not require a social work degree or did not work in the field of social work (Salsberg et al., 2020). Graduation data does not account for movement and transition between where one received their degree and in what state they practice, further complicating the workforce analyses.

This analysis compared five data sources to better understand the data available for estimating the number, type, and practice characteristics of social workers in the United States. These five data sources fundamentally differ in their method of identifying social workers, largely due to differences in who is counted as a social worker practice based on their degree, license, and ability to bill for services, as well as the source of information used to identify the workforce (i.e., self-report, educational administrative data, and employers). Due to these differences, estimates of the U.S. social work workforce vary from 280,000 to more than 1 million. Not all data sources have valid data to describe substate variations (e.g., county) in the number and distribution of social workers. These discrepancies across data sources greatly impede their utility in accurately identifying shortage areas within states.

Discussion

The health workforce crisis brought by the COVID-19 pandemic renewed focus on the value of workforce data to estimate the availability and distribution of health care workers to support workforce policy and planning. State and federal policy makers, educators, health systems, licensure bodies, and researchers need data to enact changes in education, regulation, programming, and payment and to make informed recommendations about how to best position the health workforce to address patient and community health needs.

This analysis finds that estimates of the total number of social workers, the number providing direct behavioral health care, their geographic location, demographic characteristics, and practice patterns vary widely depending on the data source used. Studies employing any of these data sources need to be aware of how each data set may influence findings and recommendations for education, practice, and policy. Health workforce projection models require accurate, timely, and rigorous data on the current supply, distribution, and practice of the workforce to project how well future supply will meet demand (Health Resources and Services Administration [HRSA], 2020). Findings from this analysis could produce very different estimates of future supply of the total number of social workers, and their distribution by setting, geography, and specialty.

A consistent dataset that identifies demographic, education, licensure, and practice characteristics—the minimum data needed to effectively plan for state and national workforce needs is greatly needed (Armstrong & Moore, 2015; Beck et al., 2018). Social work licensure boards could support the use of a minimum data set and collect information to produce both state and national estimates of the workforce. Additional information could be gathered when practitioners renew their licensure could include the settings where they work, the populations they serve, education and training specialty, current work status, and even work satisfaction. Some states produce robust information on the social work (and other health workforces) through licensure data. For example, the Virginia Department of Health Professions (VADHP) created a Behavioral Health Workforce Dashboard that incorporates demographic, geographic, practice setting, full-time status, payment types accepted, and workforce satisfaction as key datapoints to track and evaluate the behavioral health needs within state (VADHP, 2023). Yet at present, there is no single agreed-upon minimum data set collected across states, making it difficult to collate data across the United States.

Data limitations presented in this analysis have implications for a range of decision-makers. Prospective students rely on projections to determine whether to enter the profession, and educational programs consult projections to inform decisions about whether to expand or reduce educational enrollments in social work programs. Health workforce data are also important for funders to evaluate the effectiveness of investments to improve the supply and distribution of the workforce. For example, the Health Resources and Services Administration (HRSA) has invested millions of dollars in training and expanding the social work behavioral health workforce. Beginning in 2014, HRSA’s Behavioral Health Workforce Education and Training grants began funding an initial cohort of 62 social work programs and expanded to include other types of behavioral health providers (Kepley & Streeter, 2018). From 2014 to 2022, more than 17,600 new behavioral health professionals—including 10,738 social workers—have received BHWET funding and training (HRSA, 2023). Workforce data are needed to help federal policymakers further invest in workforce development and training programs. Likewise, evaluation efforts to assess the degree to which these investments have increased the capacity of the social work workforce to meet health needs, particularly in underserved communities, is warranted.

Future research on the social work workforce should focus on the type of social workers investigators are interested in studying and the geographical unit of interest. For example, licensure data may be the best source for analyzing the set of social workers who provide direct clinical behavioral health services that are eligible to bill to CMS. Licensure data can also be used to calculate state-level estimates of licensed clinical social work workforce. In contrast, because the NPPES include practice address data, it may be the best data source for understanding substate practice variation among social workers working in clinical settings across health, behavioral health, and schools.

Notably, there are other potential data sources that we did not examine in-depth that could potentially be utilized to enumerate and conduct research that could be used to shape policy interventions to increase access to behavioral health care. For example, claims data could increase our understanding of the types of services provided by social workers, as well as investigating the proportion of behavioral health services performed by social workers. Medicaid and Medicare claims data could assist in understanding the effectiveness of interventions delivered by social workers in improving health outcomes for beneficiaries. However, a limitation of claims data is that social workers who are unlicensed and cannot bill directly will be inadequately captured particularly under “incident to billing.” This issue is similar to advanced practices nurses who are billing under the license of a physician (Patel et al., 2022). Professional organizations like NASW also provide data on the number and types of practicing social workers among its approximate paid membership of social workers, but it does not gather information on non-members. Professional organization data could allow for research to understand global factors impacting social workers like burnout and intended attrition from the workforce. Electronic health record (EHR) data could be used to identify practice behaviors of social workers, types of patients and settings in which they work, and observe if these characteristics vary with licensure level (as licenses are typically included in the signature of notes). Some previous work has explored EHRs to describe social work practice and identified that social workers notes are not always easily identifiable when they are not the primary billing provider (Zerden et al., 2021). Furthermore, EHRs could allow the field to understand the clinical capacity of social workers (i.e., ratio of social workers to patients), but would be limited in enumerating the workforce. Other federal data sources—the Veterans Association, HRSA-funded Federally Qualified Health Centers, or those who participate in the National Health Service Corp Loan Repayment program—could be used to describe the types of services provided by social workers in these settings. Finally, with valid and reliable social work data researchers could link multiple data sources to test team-based models of care inclusive of social workers and the role of social worker interventions to improve health care quality and access.

Understanding the availability, utility, and limitations of data sources can help researchers and policymakers recognize how different data sources can impact the findings of analyses and workforce projections drawing on those data sources and influence subsequent decisions about future investments in training. All available data sources have strengths and limitations that should be detailed in studies that use them to allow stakeholders to understand the direction and extent to which the data may under- or overestimate the supply of the social work workforce. Without more accurate and clearly understood data to estimate workforce needs, data-driven solutions to ensure social work behavioral health workforce adequacy will remain elusive.