The Use of Interpreters in Health Centers: A Mixed-Methods Analysis

Abstract

Health centers serve millions of patients with limited English proficiency (LEP) through highly variable language services programs that reflect patient language preferences, the availability of bilingual staff, and very limited sources of third-party funding for interpreters. We conducted a mixed-methods study to understand interpreter services delivery in federally qualified health centers during 2009–2019. Using the Uniform Data System database, we conducted a quantitative analysis to determine characteristics of centers with and without interpreters, defined as staff whose time is devoted to translation and/or interpreter services. We also analyzed Medicaid-relevant policies' association with health centers' interpreter use. The qualitative component used a sample of 28 health centers to identify interpreter services models. We found that the use of interpreters, as measured by the ratio of interpreter full-time equivalents per patients with LEP, decreased between 2009 and 2019. We did not find statistically significant relationships between interpreter staffing and number of patients with LEP served, or in our examination of Medicaid-relevant policies. Our qualitative analysis uncovered homegrown models with varying program characteristics. Key themes included the critical role of bilingual staff, inconsistent interpreter training, and the reasonably smooth transition to virtual interpretation during COVID-19.

Health centers provide primary care to the nation's most vulnerable residents. Approximately 25% of the 30 million people who received care at health centers in 2019 have limited English proficiency (LEP) and need interpreters when clinicians and other staff are not fluent in their language.¹ Health centers are accustomed to caring for non-English speaking patients and are required by statute to provide language services (LS).² The commitment to LS is noteworthy from an equity perspective, since health centers serve nearly 30% of all people in the United States who are classified by the U.S. Census Bureau as LEP.³

Despite a half century of legal requirements to ensure access to vitally important LS,⁴ health care organizations do not always meet their obligations to provide linguistically appropriate care to their patients.^5–7 Since passage of Title VI of the Civil Rights Act in 1964, health care providers who receive federal funds must provide language access to people who are LEP.⁸ Subsequent guidance on the language requirements associated with Title VI mostly clarified and strengthened the requirements,⁹ and federal standards for Culturally and Linguistically Appropriate Services define the basic components of federal language access requirements related to health care organizations.¹⁰ Simply stated, patients must have timely access to LS provided by competent individuals; health care organizations bear the burden to inform the patient of the availability of these services and to offer them at no cost to the patient. Regulations issued under the Trump Administration related to the Section 1557 nondiscrimination provisions in the Affordable Care Act narrow the scope of proposed rules issued by the Obama Administration in 2016 and may weaken the obligation of health care organizations to provide each and every individual with LEP access to adequate LS.¹¹

The need for high-quality LS to facilitate communication has been studied extensively. Language services confer a broad range of health benefits (primarily because of improved communication between providers and patients), higher satisfaction for patients and providers, and a safer, higher-quality health care experience.^12–15 Yet despite the argument for including LS as a quality component in health care, coverage of interpreter services—whether delivered in person, by telephone, or through video applications—is not the standard of care in practice. Medicare does not pay for interpreter services, though the service certainly could be constructed as a Medicare-covered benefit.¹⁶ Most commercial insurers are reluctant to reimburse providers for LS. Fifteen state Medicaid programs provide some coverage for LS, though they tend to exclude health centers and hospitals from coverage, with the rationale that LS are part of these organizations' core service responsibility.¹⁷ Lack of a stable source of funding for interpretation constructs barriers to language access and creates an uneven mandate for health care organizations whose patient populations are disproportionately LEP.

While interpreters have been part of enabling services staff for many years, little information is available about rates of interpreter use in health centers, the models in place to deliver LS, and whether interpreter use has grown over the past decade, given substantial policy changes in the Affordable Care Act that greatly expanded health insurance coverage for health center patients. Very little is known, too, about the expectations for a trained interpreter workforce, and whether health centers require formal training or certification as a condition of employment.

This study uses a mixed methods approach to identify interpreter use in health centers over 2009–2019 and to describe models of service delivery commonly used to meet patients' language needs. First, we use quantitative data to answer the question: how has interpreter staffing in health centers changed over the period 2009–2019? Then, we use qualitative data for the question: how are health centers providing LS to their patients, and how did the delivery mode(s) change because of COVID-19? This study was deemed exempt by the George Washington University Institutional Review Board.

Quantitative Study

Methods

Quantitative data

Data

We used data from the Uniform Data System (UDS), a standardized system used by the Health Resources and Services Administration (HRSA) to monitor and manage the health center program.¹⁸ All federally-funded health centers report information about their patients, staffing, performance, costs, and revenues to HRSA annually. Uniform Data System includes information on staff full-time equivalents (FTEs) who are classified as interpreters and defined as staff members whose dedicated time is "devoted to translation and/or interpretation services."¹⁸ Uniform Data System does not provide information about the mode of interpretation (for example, in-person, phone, or video) or the level of training associated with interpreter FTE. Contracted (non-staff) costs for interpretation and translation are not separately identifiable in UDS; they are reported along with other interpreter staffing costs or sometimes included in general enabling services calculations.

We chose interpreter use, as measured by interpreter FTE, as a proxy measure for the use of LS in health centers, though we recognize that use of interpreters is only one component of LS delivery in health centers and other health care organizations. Many health centers contract with telephone and video interpretation companies to supplement interpreter use or as their principal method of providing services. Data on use of these LS contracts, however, are generally not separately reported by health centers in their annual UDS reporting. Health centers also often use bilingual staff to provide LS, either through staff interpreting or through language-concordant care from bilingual clinicians; the UDS, however, does not collect information on the language proficiency of the health center workforce. For these reasons, we considered the use of interpreters as a reasonable measure, and that general increases or decreases in interpreter use over time could enhance understanding about the availability of LS in health centers.

For this study, we used data on health centers in 50 states and the District of Columbia for the years 2009–2019. After minor data cleaning related to missing data or interpreter staffing errors, a total of 13,528 observations (representing 1,405 health centers) were used in the analysis. We used Kaiser Family Foundation¹⁹ reports to identify state Medicaid expansions, and included a variable related to state Medicaid coverage of LS derived from two studies by Youdelman (2008 and 2017).^9,17

Analysis

Our key outcome variable of interest was the full-time equivalents (FTEs) for interpretation service as the measure of interpreter staffing, which is the only indicator of LS delivery included in annual UDS reporting. We analyzed health centers' characteristics including patient sex, age, race/ethnicity, income, and insurance types, as well as the size of staff in each type of service and revenues from each type of funding source.

We began with the trend analysis of the health center's average use of interpreters and the ratio of interpreter staff to the number of patients with LEP. Then, we conducted Welch's t-test for each type of characteristic using the 2019 cross-sectional data (n = 1,352) to compare characteristics in health centers with interpreters with those without interpreters. A chi-square test was used to examine whether the proportion of health centers serving rural or urban areas differed by having interpreters or not.

Next, we ran a multivariate linear regression, using the entire set of observations (n = 13,528) to determine the relationship between health centers' use of interpreters and each of their characteristics. We ran the same regression for a subgroup of health centers that reported a positive number of interpreter FTEs at least once throughout the study period (n = 4,895). To adjust our results for serial correlations between observations within the same health center across the study period and a potential heteroskedasticity issue, standard errors were clustered at the health center level for these regressions.

We also examined whether there was a shift in interpreter staffing before and after 1) adoption of Medicaid expansion, and 2) implementation of Medicaid reimbursement policy for interpreter services. For these regressions, we controlled for health centers' characteristics, and clustered standard errors at the state level to allow health centers' responses in the same state to be correlated. It is important to note that some states with Medicaid reimbursement policy carve health centers and hospitals out of coverage eligibility, with the expectation that LS delivery is a core component of their operational requirements.^17,19

We used center- and year-fixed effects for all our regression analyses. We adjusted all dollar values to the 2019 USD based on the Consumer Price Index.

Results

Trend analysis

About 25% of health centers provide at least some (non-zero) interpreter FTEs, a figure that has been consistent from 2009 to 2019. The proportion of centers with some interpreters increased from 24.6% in 2016 to 26.6% in 2019.

The ratio of interpreter FTE per 1,000 patients with LEP across all health centers decreased gradually from 0.42 FTE per thousand patients with LEP in 2009 to 0.36 in 2019 (Figure 1). This decrease was also found among centers that reported non-zero interpreter FTEs at least once between 2009 and 2019. In this group, while the overall trend fluctuated over time, the ratio reached a high of 0.996 FTEs per thousand patients with LEP in 2015 but decreased thereafter when the ratio was 0.79 FTEs per thousand patients with LEP in 2019.

Descriptive statistics in 2019

We conducted cross-sectional bivariate analyses on the 2019 UDS data that compared health centers with interpreters with those without interpreters (Table 1). In 2019, 360 of the 1,352 CHCs reported interpreter FTEs more than 0. Health centers with interpreters reported an average of 3.46 FTE. These centers with interpreters showed higher expenditures in interpretation services, $210,000 in 2019, compared with an average of $20,000 for centers without interpreters. Expenses at centers without interpreters are likely to be associated with contracts for telephone or video LS.

Health centers with interpreters tend to have a larger number and proportion of

Figure 1.

Time trend of the average ratio of interpreter FTEs to 1,000 patients with limited English proficiency.

patients with LEP (7,210 versus 4,314; 23.7% versus 16.2%); they also had larger total patient volumes and employed more staff in total (26,956 versus 19,764; 227.62 FTEs versus 167.49 FTEs). Health centers with interpreters were more likely to serve rural areas (65.9% versus 56.3%). They also tended to have a greater share of non-Hispanic Asian patients but a smaller share of non-Hispanic Whites among their patients (4.3% versus 2.9%; 38.4% versus 43.2%). They also had proportionally more uninsured patients (28.5% versus 23.5%).

Regression analysis for 2009–2019

We found no statistically significant relationship between the interpreter staffing of health centers and the number of patients with LEP they served (Table 2). None of the demographic characteristics of patients they serve, including race/ethnicity, was associated with their use of interpreters, controlling for other health center characteristics and center- and year-fixed effects.

Medicaid expansion and Medicaid reimbursement

We conducted regressions for examining how implementation of Medicaid-relevant policies were associated with health centers' use of interpreters (Table 3). For states' implementation of both their Medicaid reimbursement policies for interpretation service or their expansion of the Medicaid eligibility, no statistically significant difference in the use of interpreters was found in the time period before and after their implementation of each of the policies.

Since the most recent information about Medicaid reimbursement policy for interpretation services in Kansas was as of 2009, we conducted the same regression analysis without observations in Kansas. However, we found no significant relationship between the states' reimbursement for interpretation services and health centers' interpreter staffing. As the reimbursement policies were coded based on data as of 2017,¹⁷ we also ran the same regression for the 2009–2017 data (Table 3), but found only a weak

Table 1.

SELECTED CHARACTERISTICS OF COMMUNITY HEALTH CENTERS BY USE OF INTERPRETERS, 2019^a

Table 2.

REGRESSION RESULTS FOR THE IMPACT OF COMMUNITY HEALTH CENTERS' CHARACTERISTICS ON INTERPRETER STAFFING, 2009–2019^a

negative association between the implementation of reimbursement policy and health centers' use of interpreters (β=–0.173; p<.1).

In summary, the quantitative analysis showed that a high proportion of health centers do not hire interpreters despite treating large numbers of patients with LEP. We found no statistically significant difference in the proportion of Hispanic patients between health centers with interpreters and those without interpreters. We found a null relationship between the use of interpreters and the number of Hispanic patients. Furthermore, our analysis indicates that health centers in states that expanded Medicaid and in states with some Medicaid reimbursement for interpreter services did not demonstrate higher rates of interpreter staffing than the other states.

Table 3.

CHANGE IN THE COMMUNITY HEALTH CENTERS' INTERPRETER STAFFING AFTER POLICY IMPLEMENTATION BY SELECTED POLICY, 2009–2019^a

Qualitative Study

Methods

Qualitative data

Our quantitative analysis leaves many questions about the delivery of LS in health centers unanswered—especially since health centers with high numbers of patients with LEP do not appear to have higher complements of an LS workforce. Because we recognize that interpreter FTEs are only one mode of LS delivery and that others such as bilingual staff and clinicians, and contracted telephone and video services, are not captured in the UDS, we include a qualitative component to identify models of LS commonly used in health centers.

Data and analysis

We used the 2019 UDS to select a purposive sample of health centers that reported at least some (non-zero) interpreter FTE. We drew our sample for maximum variation by geographic location, patient volume, number of interpreters employed, percent of population with LEP, and proportions of patients by race, ethnicity, and coverage. We contacted chief executive officers (CEOs) via email to schedule a 30-minute video interview with the LS manager or other staff familiar with the delivery of LS. The group of participants was varied and represented health center leadership (CEOs, chief operation officers, vice presidents of service lines), clinical and service management (site managers and directors of quality improvement, behavioral health, and equity and inclusion), clinicians (physicians and nurse practitioners), and interpreters and interpreter managers. Participants received a $30 gift certificate in appreciation of their time. Interviews were video-recorded with permission, transcribed, and coded using the Dedoose qualitative software platform following standard protocols. The first three transcripts were coded by two study team members (MR and MK) and the rest were coded by one team member (MK). Program characteristics identified through coding were reviewed and discussed through a consensus process. A first round of illustrative quotations was identified through reading all transcripts and additional quotations were selected after coding was completed. Interview domains concerned LS models, funding, program characteristics, and challenges during COVID-19. We spoke with 33 individuals from a total of 28 health centers. Our sample includes health centers in 20 states covering the major regions of the U.S.

On average, health centers in our qualitative sample saw about 34,000 patients in 2019. Average number of interpreter FTE in the sample was 8.28 but ranged from less than 0.1 FTE to over 35 FTE. About 43% of participating health centers had fewer than five interpreter FTE, and another 28.8 % had between five and 10 (Table 4). The proportion of health center patients indicating a preference for care in a language other than English (the patient group with LEP) ranged from 4% to 69%, with a median of 41%. In one-third of health centers, more than 50% of patients indicated a preference for care in another language. The most common language preference was Spanish, with 85.7% indicating that this was the most commonly requested language among their patients who needed LS. Eighteen percent of our sample health centers was classified as rural.

Results

We identify nine key themes that emerged from our analysis of qualitative interview data. Box 1 includes a selection of illustrative quotations by theme.

1. 1. Our research did not uncover a standard model of LS in a health center setting. Language services are highly variable, with few established best practices guiding the development, management, and improvement of service delivery. Health centers had programs with substantial variation across LS delivery modes, with models ranging from 100% telephone interpretation to 80% bilingual staff interpretation. Several programs have multiple on-site interpreters for the one or two most spoken languages and rely exclusively on contracted telephone or video services for other languages spoken by their patients. Though all health centers we interviewed had a telephone and/or video system in place to provide LS, most often through external service contracts, several indicated that contracted services were used less frequently than in-person options. This may reflect our selection of health centers with at least some interpreters on staff.

2. 2.

   Bilingual staff were the backbone of LS provision in health centers. Two-thirds (18/28) of health centers indicated that substantial proportions of their patient-facing staff were bilingual, and all respondents said they used bilingual staff for at least some encounters for patients with LEP. Several said that all their medical assistants and front-office staff were bilingual. A few health centers offer enhanced pay for bilingual staff, generally $1 or more an hour. Bilingual employees were often referred to as interpreters, since they provide interpretation services in addition

   

   Table 4.

   CHARACTERISTICS OF HEALTH CENTER LANGUAGE SERVICES PROGRAMS (N=28)

   to their routine job responsibilities as medical assistants or schedulers, or in other supportive roles. These bilingual staff are not considered interpreters in UDS data. Several interviewees mentioned that language competency in Spanish was a key consideration in hiring, with various methods to assess language competence for interpreters and bilingual staff. A small number of health centers in our sample used external testing services to assess language competency among clinicians

   Box 1. THEMES AND SELECTED QUOTATIONS FROM QUALITATIVE INTERVIEWS

   Theme	Quote(s)	Job Title
   Our research did not uncover a standard model of LS in a health center setting.	"We have a good mix of everything. We have in person, we have over the phone, and we have video … The goal is that every department has its own department interpreter, but then I guess to supplement that is we have a program called medical interpreter trainings. And that's where participants went through 40 hours of training. Employees who are already bilingual went through 40 hours of training to then get extra pay and be considered a medical interpreter. Those are mostly medical assistants that were already in the clinic."	Arizona, Executive Team Member
   Bilingual staff were the backbone of LS provision in health centers.	"We really have to focus on hiring individuals that can communicate with staff. We have a lot of learners coming through for many different programs, PA programs, NP programs, medical school residencies, pharmacy techs, you name it… But we say, if they don't speak Spanish, we can't take them."	Michigan, Chief Executive Officer
   	"We hire bilingual individuals that also act in interpretation as part of their role … We rarely hire any medical assistant that is not bilingual… We have several providers who are bilingual."	Arkansas, Chief Executive Officer
   	Our [interpreter] job doesn't simply entail interpreting but we also manage the phone lines for any incoming calls, assist with medical refills, lab result requests. Any task that the doctor or nurse will send us … a large variety of tasks, referrals, we get tons of messages regarding referrals."	West Virginia, Interpreter
   Most health centers relied upon multiple modalities to provide LS, but in-person interpreters are often preferred when patient-clinician language concordance is not achievable.	"I know for sure, all of us providers very strongly preferred to use our interpreters because they are so good at what they do. …'	Arizona, Interpreter
   	"We have a hierarchy, so what's the best way to deliver language services? Well, the best way, somebody speaks my language … The second best way is you have a person in the room with me interpreting. The next best way according to our patients is video, right, it's video when it works … Next level would be phone. Next level would be, forget you guys, I'll bring my kids, which is against every ethical principle."	Nebraska, Chief Operating Officer
   	"In those days, getting a patient, a provider, and interpreter to line up at the same time was like a lunar eclipse of something… And they could never stay on time, and everybody, including the patient, was frustrated. Now [after moving to a health centerwide telephone interpretation system], when I start my visit and I have an interpreter, I feel like she's there and I'm just speaking directly to my patient… The good news is we can do a lot more because we were a large campus. People were running up and down streets, not waiting for the elevator. So much time was wasted."	Massachusetts, Chief Behavioral Health Officer
   Third-party funding for LS at health centers was generally not available. Most funding for LS comes from general revenues and 330 grants.	"Interpretation services aren't reimbursable, right? We're not getting paid for them. Some of the different MCOs, of which there's six right now in Kentucky, it's all over the place. Some of them would allow us to order these face-to-face interpreters. Some of them have their own phone numbers that we could call and use. I would say, beginning late 2019 and early into 2020, a lot of those organizations really pulled back. They either cancelled their phone numbers, or they said we're not going to do these face-to-face interpretations for you all anymore. So I think we went from two or three of the MCOs that were providing some sort of direct service to one."	Kentucky, Director of Communications
   Third-party funding for LS at health centers was generally not available. Most funding for LS comes from general revenues and 330 grants.	"When [providing interpreter services] gets talked about, it's just almost as an expectation. That insurance companies, Medicaid, MCOs would expect that we can provide linguistically relevant and culturally relevant care. And so it's an expectation, but it's certainly not something that I feel like there's been much support to do."	Illinois, Chief Operating Officer
   	"So we have a combination (or LS options) but the long and short of it is from a technology standpoint. We have a combination of grants but our ultimate payment of our overhead comes from operations."	Arkansas, Chief Executive Officer
   	"I think financially we struggle to support [language services] in the contract… it's very expensive … I know some of the insurers have approached us … with the option to have iPads for their particular clients … But it's very difficult to segment the population based on insurance type … We'll only call this translator if you're with [specific Medicaid plan] … we'll call that translator if you're with [other health plan] … We appreciate the acknowledgment that the cost is not covered."	Texas, Clinician and Service Director
   	"I have to say that [the cost of language services] is not something our administration ever put on us or burdens us with. It's always make sure that you're getting the patient care they need. I've literally never heard them say, you're costing us money with your translation services."	Rhode Island, Clinician
   The use of metrics to track LS use, or gaps in services was limited in health centers.	"Right now we based our staffing off of not hardcore metrics, which is more what were trying to get than how often is a waiting room or the provider delayed because we don't have an interpreter … What we go through now is just through our metrics that we have in our EMR, so we will be improving that."	Iowa, Director of Operations
   The use of metrics to track LS use, or gaps in services was limited in health centers.	"Besides just the different languages I don't think [we collect any metrics]."	Illinois, Vice President of Community Relations
   	"We collect statistics on every single call that we do, and we can report on a weekly basis … But we could do monthly or any type of reports based on language, based on type of service. Meaning that if it was done through us or if it was through the vendor, type of visit. We do specific metrics for the ER … We can report by doctor too … We collect the time of the visit, that's the start of the visit, the end of the visit."	Massachusetts, Director of Interpreter Services Department
   	"So right now [demonstrating computer tracking of interpreter requests], you could see anything in red means you need an interpreter. Anything in yellow means in process. So that's just the main dispatch system and anybody in the institution can just see without even logging in. So we can see that there are five Cambodian staff in process … It shows you in terms of the provider where they are in terms of the queue."	Massachusetts, Director of Performance Improvement
   Relatively few health centers required certification/prior training for interpreters, in part because of the cost of training and lack of recognition of need for national standards/quality.	"You must have a Bridging the Gap or some other 40-hour certified interpretation course, and then we also use a language and interpretation proficiency test… We'll have you take that test to check for fluency. You can't work bilingually unless you have passed that test with a certain score."	Kentucky, Director of Communications
   	"We will hire people that are not certified. But I think we want to have a culture where we're encouraging people to become certified."	Illinois, Chief Operating Officer
   The need for affordable, effective, and efficient LS is growing in health centers, with little engagement from national leaders to facilitate the need.	"We're looking for expand [LS] right now because we've actually hired quite a few more providers, who do need them … I think we have two providers who rely completely on interpreter services and only two FTEs."	Arizona, Interpreter
   	"Every provider in our region thinks that we should provide interpretation services for our patients [even at locations not affiliated with the health center]. We've actually been told by several providers, if you don't send an interpreter with this patient, we don't want to see your patient. And then so I have to pull out the CLAS standards … It's a constant fight. We have hospitals that don't have interpreters. And they're like, the phone's locked up at night. And I say, unlock it."	New York, Chief Executive Officer
   	"The biggest challenge currently right now is the need to grow … The second challenge I would say when we get requests to have interpreters accompany patients to their specialist appointments. I had to start putting up some barriers because it's already tough to staff in house and then now I have to worry about trying to send an interpreter there and I'm losing three, four hours."	New York, Language Services Manager
   The lack of a career path for LS staff in health centers created staffing gaps.	"At least since I've been here, we've had two medical interpreters that have moved into medical assistants. We have one that moved into referral coordinator."	Indiana, Front Office Manager
   	"That's probably one of the things I would say is unfortunate about coming as an interpreter and being really good. We often end up losing people to other departments because there isn't a lot of mobility unless you would move into my position. We don't have different scaled interpreters too."	Indiana, Interpreter Coordinator
   COVID-19 created less disruption for interpreter services than was originally anticipated.	"When this COVID happened we were so scared because they gave us like a day of training how to use [it]. That was very challenging for all of us, our attention was so high, our stress level was so high … But thankfully with the interpreters we have in here in house, we did very well in terms of learning how to use the zoom … A lot of the patients they liked it because … they didn't want to come to the office because of the COVID." "We got the job done. Our priority was, OK, we have a Swahili, we have a Russian patient that needs this and that. We get on it and we work on it. This is one of the greatest things that I have to say about interpreters … We're still here. We're still interpreting."	New Hampshire, Interpreter
   	"Before COVID, we were always in-person interpreting, and once COVID really started, I had spoken to my supervisor because they had brought a bunch of boxes with medical records into our office, and we were unable to move our desks to property socially distance … So that was when they were able to obtain iPads in order for us to interpret via iPad."	West Virginia, Lead Interpreter
   	"It's a little more complex with the video visits. But there is a way to do that, we've had to find a couple of workarounds. …"	California, Chief Nursing Officer

   

   and other staff, especially for languages other than Spanish or not represented by in-person interpreters. Language fluency for non-native Spanish-speaking clinicians and staff was mostly commonly assessed through conversations with Spanish-speaking interpreters or interpreter services managers.

3. 3. Most health centers relied upon multiple modalities (in-person interpreters, phone and video services) to provide LS, but health center representatives indicated that in-person interpreters are often preferred by providers and patients when patient-clinician language concordance is not achievable. In-person interpreters—defined as health center staff whose principal responsibility was to provide interpretation for patients with LEP—were used by all but three (89%) of the health centers in the sample. This high percentage of in-person interpreters was not surprising, since we purposefully selected a sample with at least some interpreter FTE to learn as much as possible about the use of interpreter services in health centers. All the programs with in-person services had Spanish-speaking interpreters, and many had on-site interpreters who spoke other languages. Some sites used the term "interpreter" to refer to the person doing the interpreting, rather than the classification of the person doing the interpreting; this definition could include non-interpreter dual-role staff whose primary job is a providing front-desk support or serving as an MA, but who also fill in as interpreters when needed. All health centers reported some use of telephone and video services, often used to support less common languages or when interpreters were unavailable. Common complaints about the use of telephone and video interpretation (using iPads) related to lack of functioning equipment and long wait times for infrequently needed languages. Our findings about the preference for in-person interpreters, however, may not reflect the experiences of health centers that have chosen to provide LS exclusively through contracted services, given that our sample was selected from health centers whose UDS reports indicated at least some interpreter staffing.

4. 4. Third-party funding for LS at health centers was generally not available. Most funding for LS comes from general revenues and 330 grants. Twenty health centers in our sample did not receive any funding for LS from third-party payers. The sole migrant health center in our sample received additional funding from HRSA that could be used for LS. Three health centers currently use LS provided through one or more Medicaid managed care contracts: one health center in California negotiated with a Medicaid managed care organization to provide LS call services for patients in the plan; another in Indiana negotiated in-person LS for prenatal care patients; and a third in Kentucky received scheduling and clerical LS support. Three additional health centers had used some Medicaid managed care funding to provide LS, but those funds either were discontinued, or the services were deemed to be of a lesser quality than what their providers had become accustomed to.

5. 5. The use of metrics to track LS use, quality, or gaps in services was limited in health centers. Like other elements of LS in health centers, tracking performance varies substantially. At least one program developed a sophisticated dashboard to track delivery of LS by interpreter, clinician, patient language, and many other factors, with routine review to upgrade the program. A few other programs monitored wait times on a less systematic basis or tracked the number of interpretations by language requested or other characteristics. Most health centers, however, did not incorporate quality improvement and metrics review into their daily routines. Most interviewees felt that they were doing a very good job meeting the language needs of their patients despite lacking data to back up this claim.

6. 6. Relatively few health centers required certification/prior training for interpreters, in part because of the cost of training and lack of recognition of need for national standards/quality. Many viewed training as an aspirational goal they could not prioritize over more pressing programmatic needs. Only three health centers used formal training programs: a health center in Iowa worked with Bridging the Gap, a well-known 40-hour interpreter training program originally developed in 1995;²⁰ an Idaho health center used a 30-hour training course run by a local college for new interpreter staff; and a health center in Indiana offered Bridging the Gap training if the employee earned a high score on a non-English language proficiency test. Several health centers provided informal training with a few covering the cost of more formal training on an ad hoc basis.

7. 7. The need for affordable, effective, and efficient LS is growing in health centers, with little engagement from national leaders to facilitate the need. Throughout our interviews we heard that the need for additional LS resources is growing, both in overall volume and diversity of languages spoken, yet health centers were working on their own to figure out ways to address growing language-related patient needs. Most interviewees did not work collaboratively with neighboring health systems to share LS resources; on the contrary, several said they operated in a competitive interpreter services market, where hospitals lured interpreters with higher salaries and benefits. Further, health centers did not turn to their primary care association for assistance.

8. 8. The lack of a career path for LS staff in health centers created staffing gaps. Interviewees reported a substantial amount of movement across jobs with bilingual staff and interpreters, with no specific career ladder for interpreters to remain in their jobs and advance in the organization. At some health centers, bilingual medical assistants could be promoted to interpreters, and at others, vice versa. The pattern of "up and out" in terms of the LS position was a common theme, whether interpreters were promoted out of LS into other health center positions, or they left the health center for new LS opportunities. Sixteen health centers provided information about interpreter salaries; starting salaries for interpreters were in the $12.50–$22.00 per hour range, with the majority under $20.00. Interpreter wages were generally pegged to medical assistant wages.

9. 9. COVID-19 created less disruption for interpreter services than was originally anticipated. Health centers described a rapid, complex, and exhausting move to telehealth. Yet according to most interviewees, after an initial shutdown or curtailment in LS activity, transitioning from in-person to on-line interpreting was reasonably smooth amidst the chaos of the early weeks of the pandemic. Only two health centers reported furloughing any LS staff. Health center LS programs participated in mass testing activities, often conducted in health center parking lots or other sites. As telehealth processes were developed, LS staff participated in patient video visits and communicated frequently with patients about scheduling and prescriptions. According to several interviewees, LS staff became accustomed to working remotely. Some health centers had interpreters work from the clinic site that was closed to patient visits; when patients started to return to the office, several health centers decided to maintain remote communications with the interpreter to limit COVID exposure.

   COVID-19 led to staff shortages and interpreting equipment supply shortages, especially for health centers unaccustomed to telephone interpreting before the pandemic. When working from their homes, interpreters sometimes had to scramble to find places with privacy for the clinical visit. Though challenging, health centers praised interpreters for their flexibility and commitment and talked about a gradual resumption to making sure patients got the services they needed, including interpretation and language supports. The move to telehealth was problematic for patients who did not have the technology, internet access, or computer skills to participate in virtual appointments. For these patients, interpretation and health care visits required the use of phone services only, which may have affected quality of care.

Discussion

Despite the important role they play in the care of patients with LEP, health centers approach the delivery of LS in a myriad of ways, with little evidence that the amount, type, or implementation of LS staffing meets the needs of their populations. About one in four health centers across the country employ some interpreters—a proportion that has changed very little over the past decade, despite growth in the numbers of patients with LEP who use health centers after expansions associated with the Affordable Care Act. Our data indicate that interpreters have not kept pace with patient need over the past decade, with the ratio of interpreter FTE per 1,000 LEP patients across all health centers gradually decreasing by 25% over the 2009 to 2019 period. The prevalence of bilingual staff found in the qualitative analysis offers insights into the quantitative finding that the proportion of patients with LEP receiving care at the health center was not a significant predictor of interpreter FTE. This could result from bilingual staff not counting as interpreters in UDS reporting, or from inadequate interpreter staffing at health centers, or both.

Despite clear legal requirements to provide LS in health care settings, the way these requirements are met is left largely up to individual health care organizations. The result is a panoply of options rather than a more limited array of successful models that are suitable in the context of health center care and could be customized to local circumstance.

As always, discussions about the delivery of LS must turn to the lack of sustainable funding for interpreters and the omission of reimbursement from most private and public health plans and insurance programs. Our qualitative interviews did not identify any health centers that received third-party payment for Medicare enrollees with language needs. Future studies should identify ways to improve access to LS for the nation's elderly population, especially through innovative payment models that seek to incentivize patient-centered, high-quality, and lower-cost delivery options. Our study demonstrates that Medicaid's limited coverage of interpreter services in selected states appears to be inconsequential, perhaps because health centers are typically carved out of coverage altogether.

Future research can address ways to reduce the variability across health centers, with a movement toward identifying models of excellence that can be replicated—with technical support from HRSA—within different health center contexts. Whether health centers are providing the right resources in the right amounts to advance patient care should be investigated using quality improvement, periodic measurement, and opportunities for collaboration at the state or national level. Previous work in this area can serve as a blueprint for health centers to develop a customized approach to LS delivery with the goal to meet every patient's communications needs.²¹

The pandemic compelled health centers to reconsider in-person interpreting as the only way to assure high quality LS delivery. Though preferred by many, in-person interpreting is not always available and can and should be supplemented by other modalities that address the LS needs of patients. Health centers need support to purchase and upgrade equipment, train staff, and develop technological solutions to ensure that patients who prefer care in another language get the services they need.

Limitations

Our quantitative analysis had several limitations. First, we looked at the interpreter FTEs reported on the UDS, but some centers counted multi-role staff, who also provided interpretation service, towards a service other than interpretation, leading to potential understatement of interpretation services provided by health centers. Second, the UDS does not capture the bilingual staff FTEs, so we were not able to determine if a health center provides enough interpretation services. Third, the UDS does not capture specific amounts spent by health centers on contracted telephone and video services, which may represent an important component of LS delivery. Thus, measuring the use of interpreters in health centers captures only one important delivery mode to address language needs and may understate the extent to which health centers are meeting the language needs of their patients. Fourth our study did not imply any causality due to potential omitted variable bias by the nature of using observational data without an experimental design, although we used panel data with fixed effects to reduce such a bias. Our qualitative analysis also had limitations, most notably the lack of generalizability associated with any limited sample of cases. In the study, we attempted to maximize variation in terms of key characteristics of health centers and LS delivery modes, but we are likely to have missed some variance that may have illuminated the understanding of the health center approach. We may also overstate the importance of on-site interpreters, since the sample was chosen in part because of at least some interpreter FTE staffing. The nature of LS operations and oversight means that many different types of professionals participated in the interviews, which may reflect variations in perspectives about the value and role of interpreter services activities separate and apart from the delivery of LS itself. Nevertheless, the diversity of roles appears to accurately reflect the locally-determined, or "home-grown," nature of LS in health centers.

Disclaimer

Supported by the Bureau of Health Workforce (BHW), National Center for Health Workforce Analysis (NCHWA), Health Resources and Services Administration (HRSA) of the US Department of Health and Human Services (HHS) as part of an award totaling $450,000, with 0% financed with nongovernmental sources. The contents are those of the author(s) and do not necessarily represent the official views of, nor an endorsement by HRSA, HHS or the US Government.