Descriptive analysis of interpreter service mode costs & usage in Northwestern Wisconsin pre and peri COVID-19

Abstract

Background

Interpreter service mode (in person, audio, or video) can impact patient experiences and engagement in the healthcare system, but clinics must balance quality with costs and volume to deliver services. Videoconferencing and telephone services provide lower cost options, effective where on site interpreters are scarce, or patients with limited English proficiency (LEP) and/or interpreters are unable to visit healthcare centers. The COVID 19 pandemic generated these conditions in Northwest Wisconsin (NWWI). The objectives of our study were to describe visit characteristics for interpreter services pre and peri COVID-19 and how interpreter visits and costs differed by modality and language pre and peri COVID-19 at Mayo Clinic Health System (MCHS) Eau Claire and Menomonie, Wisconsin (WI).

Methods

We analyzed medical records and billing data from MCHS Eau Claire and Menomonie that included interpretation visit characteristics and aggregate costs. We performed descriptive analysis of visit characteristics and aggregate costs (dependent variables: cost, duration, language, modality, payer type, department, patient age, and patient biological sex) from June-December 2019, 2020 (independent variables). We used Pearson’s chi-square and t tests to test for significant differences in visit characteristics between time periods.

Results

Significant differences were found in distribution of all visit characteristics (language, p-value < .001; biological sex, p-value .011; mode, p-value < .001; payer type, p-value < .001; department, p-value < .001; age, p-value .016; visit duration in hours, p-value < .001) between 2019 and 2020 windows. Total costs of all interpretation modes increased between 2019 (in person, $188,109; audio, $15,734; video $15,881) and 2020 (in person, $388,500; audio, $44,376; video, $34,245) despite similar visit counts for both years (2019, 3750 visits; 2020, 3425 visits). Differences between years are consistent with COVID-19 protocols, department usage by language, and patient age.

Conclusions

Our results show similar counts in the number of overall visits and increase in interpreter service costs in all modes, positively reflecting upon the availability of multimodal language resources.

Background

American Sign Language (ASL), Hmong, and Spanish-speaking patients face many obstacles to healthcare access even before they enter the care center. Studies have shown that many LEP patients are unaware of what healthcare services are available and how to access them due to a lack of formal information provided in patients’ native languages [1, 2]. When LEP patients seek medical services, it is important to know which hospitals their insurance will cover, or which facility will best fulfill their needs. However, even though federally-funded care centers, including Mayo Clinic Health System Northwest Wisconsin (MCHS NWWI), are required to provide interpreter services for LEP individuals [3, 4], interpretation is not universally free to the patient in the United States, which can cause confusion and impede care [2]. In order to attend their appointment, patients are often required to check in at a front desk; however, this can also be a challenge since LEP patients may experience “difficulty following conversations with receptionists, providing proper documentation required for coordinating care, and booking and attending appointments” [1]. Once patients meet with a provider, errors in communication can impact the care that they receive, as they may not fully be able to express why they are at the hospital or their symptoms [1]. Further, unclear communication from the provider can leave patients with confusion regarding their health. According to one study that explored LEP patient appointment experiences, “about half” of survey participants “indicated that they did not know why they were seeing that specialty provider, did not understand ordered tests, test results, or the treatment plan” [5]. Additionally, it can be difficult for providers to communicate dose directions or how to acquire the necessary medications [1, 6], which can leave LEP patients unable to “understand what the medication was for or how to use it” [5]. Errors in communication, such as these, present potential dangers for LEP patients [7], as there is an increased risk of “inadequate communications, misdiagnosis, medication errors and complications, and even death” [1]. In hopes of remedying this challenge, some LEP patients purposely attempt “finding providers from their ethnic background,” however when these individuals “take it upon themselves to seek care from these providers,” it may “delay healthcare access” [1]. Meanwhile, “usage of professional interpreters has been shown to improve the quality of care for patients with limited English proficiency, resulting in higher patient satisfaction” [8].

With a variety of language interpretation modes (in person, audio, and video), potential benefits and costs of each service can vary [8,9,10,11,12]. In one study that compared direct costs associated with language interpretive services (LIS) provided, it was found that in-person services generated 4.3% of direct costs followed by audio and video services at 3.6% [9]. Based on this data, we can observe that audio and video modes of interpretation were more cost-effective interpretation modes. However, “video and telephone interpreting limits the ability to recognize and respond to emotional and physical cues,” which may lead LIS users to the conclusion that “in-person interpreting, may be more acceptable” [8]. Furthermore, video interpretation may have advantages over audio-only, especially in complex educational or psychosocial encounters [13, 14]. This holds particularly true for ASL users. As American Sign Language is a visual language, in-person or video interpretation is necessary while interpreting via telephone is impossible. However, it has also been found that the extra time needed to acquire and work with interpreters can be exponential compared to an English-speaking patient’s visit [1, 2, 5, 15]. Global analyses beyond healthcare settings have identified similar concerns regarding the potential financial and qualitative benefits and challenges of remote interpretation, notably in legal procedures [16,17,18,19]. While each linguistic interpretive mode has qualities that may aid a patient, the underlying necessity is having access to high-quality healthcare language services, as language barriers are serious obstacles to obtaining meaningful healthcare resources. These obstacles are many and have been worsened by the COVID-19 pandemic.

In addition to the barriers to patient care that ASL, Hmong, and Spanish-speakers already face, the COVID-19 pandemic has made it harder for providers and their patients to communicate. For ASL speakers in particular, masks make lip-reading, often used in conjunction with or when someone does not understand sign language, impossible. The additional requirement of in-person interpreters or interpreter services for ASL patients may add time onto their visit. Additionally, new pandemic related information such as signage or pre-visit instructions may only be available in English or a limited number of additional languages. As a result, non-English speakers may not know about key hospital procedures until an interpreter or interpreter service is found after the patient is already on site. Isolation protocols create an additional burden on communication between patients and providers. In addition to impacting the healthcare experiences of the LEP patient community, the issues presented by COVID-19 pose potential impacts to the costs and usage of interpreter service modes.

The COVID-19 pandemic necessitated social-distancing measures, stay-at-home orders, and reduced travel. To prepare for future health emergencies and increased telehealth utilization, it is important for health systems to understand how interpreter services shifted in response to system demands during this crisis. Our research examines aggregate interpreter service costs and usage overall and how costs and usage of interpreter services shifted from one modality to another for LEP patients whose first languages are Spanish, Hmong, and ASL at the MCHS NWWI locations in Eau Claire and Menomonie before and during the outbreak of COVID-19. The objective of our study is to examine costs, usage, and modality of interpreter services before and during the COVID-19 pandemic. Included in our analysis are visit characteristics (language, biological sex, interpretation mode, payer type, department, age, and visit duration in hours) that provide context for usage and cost data. By identifying contexts in which interpreter service modes are utilized and costs are accrued, we aim to help clinics understand characteristics of interpreter service utilization and costs to meet the needs of departments, LEP patients, and immigrant and refugee groups in and beyond pandemic contexts. Given the existing and rapidly growing multilingual community in NWWI, the challenge that COVID-19 has posed to immigrant community health, and the centrality of the MCHS to NWWI health, this study fills a gap in our regional knowledge by providing a comprehensive description of outpatient interpretation costs and as the first analysis of interpreter service modes conducted at MCHS NWWI. It contributes more broadly to health services research by providing language- and modality-specific insights into the cost and provision of interpreter services. Furthermore, our comparative analysis serves as the first study aiming to enhance the understanding of how cost patterns within these ancillary services responded to conditions imposed by COVID-19.

Methods

Setting

Wisconsin is a hub of multilingual communication and immigration. As of the 2020 Census, the majority of foreign-born immigrants to Wisconsin come from Asia and Latin America, particularly Southeast Asia, Mexico, and Central America [20]. According to the Wisconsin Department of Health Services (WI DHS), the state has “the third largest Hmong population in the US” [21]. And immigrant Hispanic workers, primarily from Mexico and Central America, constitute the majority of the labor force in the state’s signature dairy industry [22]. Eau Claire sits an hour-and-a-half west of the Twin Cities and is the largest city in Northwest Wisconsin. Together, Eau Claire and Menomonie are the primary healthcare centers for the region’s rural, agricultural communities, which, alongside the Ashley Furniture headquarters in Arcadia (about an hour south), employ much of the region’s immigrant workforce. In keeping with industry needs, the state’s immigrant population has grown exponentially. The state’s Hispanic population grew by 46% overall between 2000–2015, with western and northern regions registering increases of 190.1% and 169.5% respectively [23]. The present and growing immigrant population in NWWI and the centrality of Eau Claire and Menomonie to regional health and wellbeing signify a need for LEP interpreter services, which may be provided in a variety of modes, to bridge linguistic barriers to health care.

Sample

The study sample consists of 7,175 outpatient visits that occurred at the Menomonie and Eau Claire, Wisconsin branches of the Mayo Clinic Health System (MCHS). De-identified data from 7,175 visits conducted in ASL, Hmong, and Spanish were provided by the Language Services Department at Mayo Clinic; however, we excluded 939 of the study sample patient visits from our analysis of age and 1552 from visit duration due to missing data or issues with the record such as visit never closed. Visits that occurred from June through December of 2019 (before the COVID-19 pandemic) and June through December of 2020 (during the COVID-19 pandemic) are included. Visits from January through May 2020 were omitted due to data entry challenges precipitated by rapidly-changing protocols during the first months of the pandemic. We strived to encompass a time frame that was adequate to study the effects before and during the pandemic [24]. Visits records were collected from three sources within the MCHS enterprise including the electronic medical record system (Epic), and their video interpretation contracts with language access networks My Accessible Real-Time Trusted Interpreter (MARTTI) and Language Line. This study was determined to be exempt from Institutional Review Board approval. No incentive was given to patients for participation.

Analytic methods

We conducted a descriptive analysis to describe the aggregate costs and characteristics of visits (dependent variables) occurring in each period (independent variable, 2019 or 2020). Characteristics analyzed included language, associated interpretation cost, modality (in-person, audio, or video), payer type, visit duration, department, patient age, and patient biological sex. MCHS interpretation contracts with language access networks MARTTI and Language Line provided us with additional information to analyze cost by language, service mode, such as audio or video, and sum of cost for interpreters. Additionally, we conducted Pearson’s chi-square and t tests where appropriate to test for significant differences between these time periods regarding visit characteristics.

Results

As shown in Table 1, significant differences were found in the distribution of visits by age, biological sex, language, mode of interpreter services, payer type, department, and visit duration of services between the 2019 and 2020 study windows. In-person interpreter visits account for over half of determined visits by mode in both years. Visits conducted with female patients accounted for nearly 60% of visits in both years, and visits conducted in Hmong accounted for over 60% of visits in both years. In 2020, in-person, female, Hmong, and video visits accounted for a smaller proportion of visits compared to 2019. Inversely, male, ASL, Spanish, and audio visits accounted for a larger proportion of visits compared to 2019.

Table 1 Sample characteristics of interpreter services patient visits, June – December 2019 and 2020, n(%)

Medicare was the most common payer type for Hmong, ASL, and total visits in both years. However, both Hmong and total visits listing Medicare as payer type decreased by over 10% between 2019 and 2020. Inversely, Hmong and total visits listing Medicaid as payer type grew by around 10% between 2019 and 2020. Medicaid, self-pay, and Worker’s Comp were most represented in Spanish interpretation visits.

Dialysis (DLS) was the most utilized department for interpretation services by number of visits, although visit numbers and proportion of total visits decreased between 2019 and 2020. The second-most utilized department for interpretation services per year and overall was radiology, followed by pathology, family medicine, and ophthalmology, all of which experienced an increase in proportion of interpreter services visits between 2019 and 2020. The mean duration of visits was significantly higher in 2020, but duration varied greatly in both years and was right-skewed with the 2019 median at 1.73 h (IQR: 0.80, 5.26) and the 2020 median at 1.44 h (IQR 0.69, 5.43). Duration increased across all languages between 2019 and 2020.

As shown in Table 2 and consistent with descriptive findings in number of visits, in-person interpretation also had the highest costs. Similarly, when looking at cost breakdowns by language and mode, Hmong interpretation accounted for both the most visits and highest costs in both years; however, Hmong in-person interpretation costs fell and while virtual rose in 2020. While ASL recorded the fewest number of visits in both years, it accounted for over a third of total interpretation costs. Spanish accounted for the highest virtual interpretation costs in both years, despite Hmong having the most virtual visits.

Table 2 Aggregate interpretation costs, June – December 2019 and 2020

Consistent with the change in number of visits and an increase in in-person interpretation for the language, ASL total costs rose between 2019 and 2020. Spanish, the second-most utilized language by number of visits per year and total, accrued the highest virtual interpretation and least in-person costs in both years. Total costs of all interpretation modes increased between 2019 and 2020 overall despite similar visit counts for both years.

Discussion

Our analysis yielded a variety of both anticipated and surprising results. The increase in proportion of audio interpreter visits between 2019 and 2020 was anticipated, due to COVID protocols. However, the decreased proportion of video visits was unexpected and runs counter to findings from other health care centers, such as the Royal Melbourne Hospital in Australia, where video interpreting appointments increased from 10–15 appointments per month (before COVID-19) to 100–200 a month in 2020 [25]. This may be due to logistical difficulties with video interpretation due to masking guidance or data entry error. Similarly, the decrease in dialysis visits observed in 2020 could potentially reflect pandemic-motivated changes to at-home dialysis protocols or personal decisions to perform at-home dialysis. The decrease in Hmong interpretation visits between years were likewise notable. The significant difference between the number of visits by biological sex, with women visiting more frequently, aligns with national trends in healthcare usage [26].

The increase in total in-person interpreter costs in 2020 runs counter to our expectations. While virtual interpreter service aggregate costs rose as expected in 2020, the audio interpretation aggregate costs increased notably and unexpectedly over video costs between 2019 and 2020. Aggregate cost increase between 2019 and 2020 is consistent with visit duration nearly doubling between the two timeframes. The aggregate cost increase of audio over video could relate to the potential for logistical difficulties with video interpretation due to pandemic personal protective equipment (PPE) protocols reducing usage and thereby cost of video.

Interpreter costs and usage by language also fluctuated in unanticipated ways between the two time periods. The contrasting rise in ASL in-person costs and decline in Hmong in-person costs in 2020 was unanticipated, as was our observation that the aggregate costs by language do not directly map to the number of visits for each language. For example, Hmong interpretation accounts for about two-thirds (67%) of visits but less than half of the aggregate costs (39%), and Spanish constitutes about 16% of virtual visits but 50% of virtual visit costs. The ASL aggregate cost increase may be linked to the mean visit duration increase across all interpretation language groups in 2020, potentially due to COVID protocols lengthening appointment times and PPE hindering communication.

The dip in Hmong visits between 2019 and 2020 and unanticipated use-cost ratios of Hmong and Spanish visits may relate to differences in department and patient age by language. Although dialysis and end-stage renal disease (ESRD) are more common with Hispanic than non-Hispanic communities in the United States of America [27], at MCHS NWWI Eau Claire and Menomonie Hmong interpretation visits account for 94.7% of total DLS interpretation visits between 2019 and 2020, while Spanish-speaking patients were most represented in Obstetrics and Gynecology (OB/GYN) (58%) and Pediatrics (72.4%) departments. This deviation from national trends in DLS usage and could stem from differences in patient age by language; the average age of Hmong patients in our sample was 58 years, whereas Spanish and ASL interpretation patients averaged 33 and 48 years-of-age, respectively. The overwhelming representation of Hmong interpretation in DLS visits could contribute to lower costs than anticipated for the language due to the high frequency of revisits within this appointment type resulting in increased patient comprehension and decreased need for interpretation in each subsequent visit. Inversely, the much higher representation of Spanish in OB/GYN and Pediatrics departments could correlate to increased need for interpretation and consequently longer visit duration and costs due to the nature of the visit. The drop in percentage of Hmong interpretation visits between 2019 and 2020 is also consistent with DLS department data. In 2019, DLS accounted for 29.6% of Hmong visits; in 2020, 23.1%. The 6.5% decrease in Hmong DLS visits and 4.8% decrease in total Hmong visits between 2019 and 2020 may stem from Hmong patients electing to perform at-home DLS during the pandemic and/or from Hmong patient mortality for which average age (58) and predominant department (DLS) would indicate increased risk over other language groups.

The higher costs-to-visit ratio for Spanish versus Hmong interpretation could also relate to community demographics in Eau Claire, Dunn (in which Menomonie is the county seat), and the surrounding counties. As above-demonstrated, Hispanic and Latinx populations are increasing at the fastest rate in the northwestern region of Wisconsin, and while 2020 census data records higher percentages of Asian alone communities (4.2% Eau Claire, 3.2% Dunn) than Hispanic/Latino communities (2.9% Eau Claire, 2.3% Dunn) in Eau Claire and Dunn counties, nearby Trempealeau County – which depends on health services in Eau Claire and Dunn – recorded the inverse, with 12.9% Hispanic/Latino as opposed to only 0.4% Asian alone [28]. Further, while Mayo Clinic Health System Eau Claire & Menomonie receive a higher number of Hmong interpretation visits than Spanish, there is a greater need for Spanish interpretation at regional free clinics. Hispanic community members comprised 37% of the total number of patients at the Chippewa Valley Free Clinic in Eau Claire, WI in 2021; Asian, 4% [29]. The influx of Hispanic and Latinx community members in the region could correlate to a higher need for interpretation services than in the Hmong community. Inversely, indicators of community establishment in the state may relate to a reduction in interpreter service usage by Hmong patients. As of 2010, most Hmong community members in Wisconsin are either born in the USA or have lived in the country for at least a decade, and 80% of Hmong community members that speak Asian American or Pacific Islander (AAPI) languages also speak English well [30].

Payer type and age differences also reflect to differing levels of community establishment between LEP language groups. Frequent usage of Medicare for Hmong and ASL visits correlates to the older average age of those patient groups. The younger age of patients at Spanish interpreter visits (under 35 in both years), indicative of more patients of working age and less financial stability among Hispanic patients [31], is consistent with their higher use of Medicaid, Worker’s Comp, and Self Pay.

Finally, the overall stability of interpretation services usage between 2019 and 2020 was unexpected and contradicts pandemic trends in interpretation usage. One study conducted by CSA research reported 55% of language service providers reported a decrease in business since the pandemic began [32]. This was not reflected in our research. Our results show minimal decrease in number of overall visits and increase in interpreter service costs in all modes, positively reflecting upon the availability of multimodal language resources.

Conclusions

The objective of our study was to analyze costs, usage, and modality of interpreter services at the Mayo Clinic Health System in Eau Claire and Menomonie, Wisconsin before and during the COVID-19 pandemic. Through our descriptive analysis, we identified common characteristics of patients that most utilized interpreter services: female, speakers of Hmong, 51 average years old, using in-person interpretation at dialysis visits with Medicare payer type for healthcare appointment. Our study also correlated age differences between language groups to trends in departments and payer types. We found the number of visits by language to be disproportionate to costs with Hmong having a percentage of usage by number of visits than costs, whereas Spanish had a much higher percentage of virtual costs than usage. We determined changes in interpretation mode and patient demographics during the first year of the COVID-19 pandemic, including a decrease in Hmong visits, increase in virtual interpretation costs, and increased visit duration in 2020. Finally, we discovered only a slight dip in the number of total interpretation visits and an increase in interpretation costs between 2019 and 2020.

Our study encountered several limitations due to missing data, particularly regarding visit duration and patient characteristics, which affected the reliability of cost analyses for different interpreter modes. These include interpretation visits eliminated from “n” such as those listing age as “deceased” or visits including no check-in or check-out time. In addition to duration visits eliminated from “n,” billing-level data does not capture whether the intensity of interpretation varied (by language, department, etcetera) and may underestimate actual interpreter time spent with patient. Mode-level limitations include potential data mis-entry: for instance, leaving mode blank, listing “appointment provider,” or “language line” selection defaulting to audio instead of prompting providers to select audio or video. The resulting number of undetermined visits by mode generated by data entry error may have skewed the findings on audio versus video interpretation usage, complicating the analysis of these service modes, for instance, by preventing our study from reliably determining cost of visit by mode or language or to confirm whether differences in mode related to differences in costs by year. LEP patients that refused interpretation or used family/friends as interpreters would also not appear in our data.

To further understand underlying reasons for the differences in costs and usage of interpreter service modes by language outlined in this descriptive analysis, our next step would be to propose a qualitative survey of LEP patient, interpreter, and healthcare provider interpretation experiences within a department that records a high usage of interpreter services, such as dialysis, and/or consistent usage across all language groups, such as radiology and pathology. This study would have the potential to reveal barriers to healthcare access by language and level of establishment in the community that, among other potential impacts, may be limiting the involvement of the region’s growing Hispanic/Latinx community from engaging MCHS NWWI.

The findings of the present study have the potential to affect clinical practice regarding interpreter service utilization. Our data demonstrates a sustained need for in person interpretation and, overall, a continued need for interpreter services across all modes and languages studied. Data regarding which departments and languages need more interpretation may be used toward the strategic hiring of multilingual providers—such as in-person interpreters of Hmong in DLS—in these areas to reduce costs. Taking into consideration the potential for video to provide higher quality interpretation than audio, our findings on the comparatively low engagement in video interpretation as opposed to audio during the pandemic shift to telehealth suggests a need to improve access to video interpretation during remote telehealth servicing. Previous qualitative research identified barriers to clinical use of interpreter services such as cumbersome access related to both processes and equipment, organizational investment in remote interpreter technologies, and staff training, which suggests areas for further investigation or quality improvement [33, 34]. Finally, our results suggest the need for rural health systems to monitor shifting payer characteristics, such as the growing utilization of interpretation services reimbursed by Medicaid and the decreasing use of interpretation services reimbursed by Medicare. Due to the variability in reimbursement by payer and the importance of adequate reimbursement to promote health equity, systems must make informed budgeting and hiring decisions regarding interpreter services [35]. As other health centers servicing rural communities—including Hospital Sisters Health System (HSHS) Sacred Heart and Prevea Health in NWWI—close, citing insufficient Medicaid funding [36,37,38], financial sustainability of interpreter services is a priority.

In conclusion, our study has contributed data to existing public health literature essential for identifying the departments most in need of LEP language services and the most utilized interpreter service modes and languages in NWWI, and for anticipating the future interpretation needs of LEP patients. The stability of overall usage and costs pre and peri COVID-19 found by our study reflects favorably upon the impact of multimodal interpretation services at MCHS NWWI on accessibility to healthcare for LEP patients in and beyond pandemic contexts. Although the COVID-19 pandemic is now over, telehealth usage remains high nationwide which makes our study’s findings relevant to current healthcare providers [39]. Lastly, given the growing multilingual community in NWWI and across the nation, our study may be used to inform future pandemic responses both regionally and nationally.