Comparing remote and in-person interpretation experiences for clinicians and Spanish-speaking patients with limited English proficiency: a mixed methods study

Abstract

Objective

There is concern that remote medical interpretation is not as patient-centred as in-person interpretation, but limited evidence exists comparing interpreter service delivery methods. Using mixed methods, remote and in-person professional medical interpretation were examined from the perspectives of Spanish-speaking patients with limited English proficiency and community health centre (CHC) clinicians.

Design

Patient experience survey data from Spanish-speaking patients and interviews of primary care clinicians assessed their experiences of using remote versus in-person interpretation. Multivariable regression models estimated the association of the interpreter method with patient-reported experiences of (1) clinician communication and (2) interpreter support.

Setting

Three CHC organisations in California, USA.

Intervention

Remote versus in-person medical interpretation.

Primary outcomes

Patients’ and clinicians’ experiences of using in-person versus remote professional medical interpretation.

Results

We recruited 303 Spanish-speaking patients (mean age: 40.4, % female: 69.0%) to complete a survey assessing their experiences with professional medical interpretation and 19 clinicians who used professional medical interpretation for interviews. In regression analyses of patient experience survey data, no evidence of an association between the interpreter method used and patient-reported experiences of clinician communication or interpreter support was found. In interviews, however, clinicians strongly preferred in-person interpreters and highlighted operational and communication challenges associated with using remote interpreters. Interviews revealed six themes related to interpreter services delivery methods: (1) in-person interpretation supports effective communication and clinician-patient relationships, (2) in-person interpretation enhances operational efficiency, (3) cost-effectiveness of delivery methods depends on language demand and clinic needs, (4) in-person interpretation enhances quality control and reduces privacy risks, (5) considerations when integrating external personnel and (6) the availability of and limited use of audio-video medical interpretation.

Conclusions

To meet the operational needs of CHCs, policymakers and healthcare payers should consider expanding payment models that enable the provision of interpreter services using multiple methods.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Remote medical interpretation is often viewed as less patient-centred than in-person interpretation, but limited evidence exists comparing interpreter service delivery methods.

WHAT THIS STUDY ADDS

• Although patient-reported care experiences were not significantly associated with interpreter service delivery methods, clinicians strongly preferred in-person interpreters because of more effective communication, stronger clinician-patient relationships, greater operational efficiency and enhanced quality control and reduction of privacy risks.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Payment policies should allow for interpreter service delivery using multiple methods, prioritising in-person methods whenever operationally feasible.

Introduction

High-quality clinician communication is central to the provision of patient-centred care, especially for patients with limited English proficiency (LEP).^{1 2} Extensive research indicates that clinician communication is positively associated with improved patient outcomes, patient care experiences and quality of care.³ Compared with using ad hoc, untrained clinic staff as interpreters, using professional medical interpreters for patients with LEP is associated with improved clinician communication, better health outcomes, higher quality of care and fewer medical errors.^{4 5}

Remote interpreter services are commonly used by clinics, but there is concern that remote medical interpreter services may not be as patient-centred or effective as in-person interpreter services.⁶ While several research studies have examined interpreter service delivery methods, particularly in emergency departments,^{7 8} there remains limited evidence comparing these methods in primary care settings. A systematic review found that in-person interpretation was linked to better patient care experiences compared with remote interpretation.⁹ Other survey research studies, however, found no difference in patient-reported care experiences for remote and in-person interpreter services, as the care experiences of Spanish-speaking patients were overwhelmingly positive, irrespective of the interpretation method used.^{10 11} Existing evidence suggests that in-person interpretation can improve patient outcomes relative to remote interpretation, but the findings are mixed. Additional evidence is needed to understand differences in quality of care and patient outcomes between the use of remote versus in-person interpreter services in primary care settings.

Medical interpreters are a crucial link between clinicians and Spanish-speaking patients, facilitating patient access to services, including scheduling appointments, comprehending prescription instructions and completing follow-up care activities. We compare Spanish-speaking patients with LEP and community health centre (CHC) clinicians’ experiences of working with remote and in-person professional medical interpreters. Electronic health record (EHR) data were used to verify the documented need for interpretation services and the preferred language of patients.

Based on past evidence about the benefits of in-person medical interpretation,⁹ we hypothesised that patients receiving in-person interpreter services would report better clinician communication and interpreter support compared with patients receiving remote audio-only interpreter services. To supplement the patient experience surveys, semi-structured interviews were conducted with CHC clinicians to compare experiences of remote versus in-person interpretation for patients with LEP.

Methods

We used a mixed methods design that included quantitative and qualitative patient experience survey data from Spanish-speaking patients with LEP and qualitative interview data assessing CHC clinician experiences of using remote versus in-person medical interpretation. The study used a convergent mixed methods approach; quantitative and qualitative patient survey data and qualitative clinician interview data were collected separately, but analysed and interpreted together to more comprehensively understand barriers and facilitators of using in-person and remote interpreter services delivery methods.

Patient experience survey data

Patient survey data and encounter assessment data from the California Department of Healthcare Services (DHCS) Medical Interpreter Pilot Programme (MIPP) were analysed. Launched in March 2022, DHCS implemented MIPP to assess the provision of and identify strategies to improve the provision of professional medical interpreter services for Medi-Cal members with LEP.¹² Medi-Cal is California’s Medicaid programme, which pays for a variety of medical services for children and adults with limited income and resources. Through a competitive bidding process, DHCS selected three private language services companies to provide professional medical interpreter services and paired each company with a separate CHC organisation as a pilot site. MIPP introduced professional medical interpreters into CHC operations for Medi-Cal members with LEP. DHCS used public meetings with key stakeholders, including patients and public involvement, to prioritise in-person embedded on-site professional medical interpreters to be implemented and examined as part of MIPP. The MIPP staffing model aimed to address stakeholder priorities for patient-centred care by integrating professional medical interpreters into CHCs’ in-person operations. Interpreters would provide in-person and remote professional medical interpretation services for low-income patients with LEP insured by a Medi-Cal managed care plan (MCP), health insurance plans contracted by DHCS to manage the healthcare of low-income individuals and families.

Prior to MIPP, only one of the three pilot clinic sites had an existing professional staff workforce to interpret medical encounters for patients with LEP. This clinic employed more than two dozen ‘cultural liaisons’ who speak a total of 16 different languages and perform patient navigation responsibilities in addition to medical interpretation. The two other pilot clinics primarily relied on their Spanish-speaking bilingual medical assistants to provide ad hoc interpretation in between fulfilling their core duties prior to MIPP. Bilingual medical assistants were not trained or certified to provide professional medical interpretation. All three pilot site clinics used remote MCP language services when bilingual clinic staff members were unavailable to interpret or when a language could not be accommodated by staff.

Despite having on-site bilingual clinical staff and the MCP’s language services, all three pilot sites had gaps in access to professional medical interpreters before MIPP. As a result, each clinic had existing contractual relationships with private language services companies when MCPs could not find an interpreter for a requested language or an interpreter could not be found in a timely manner. MIPP tested the implementation of integrating in-person and remote professional medical interpretation services that were tailored to each clinic’s patient population with LEP, bilingual staff capacity and operational needs. On-site Spanish language medical interpreters supported in-person visits and telemedicine encounters, while off-site remote interpreters were used for languages other than Spanish.

The documented need for interpreter services was determined based on EHR data, which identified patients as requiring interpretation through their documented preferred language and previous use of interpreter services during clinical encounters. Two out of the three pilot sites were staffed with an onsite Spanish-speaking medical interpreter. These two pilot sites used audio-only remote interpretation during telemedicine encounters, when the clinician was conducting the encounter remotely. Due to executive decisions made by leaders at one of the pilot clinic sites to prohibit external personnel, in-person professional medical interpretation was not possible. At this pilot clinic site, MIPP delivered services exclusively using audio-only remote interpretation.

Patient experience survey data

After clinician encounters, the medical interpreter asked the patient if they were interested in participating in a survey related to their clinic experience and receipt of interpreter services. Unless interpreters faced time constraints, which was only documented by interpreters for 6% of MIPP-supported encounters, all patients who received MIPP-supported interpreter services were informed of the survey and recruitment was attempted. Patients were provided with a printed or online study information sheet, informing them that the survey was part of an external evaluation, their responses were voluntary and confidential and their participation would not impact their care. Within 48 hours of completing the survey, an external research team member emailed or texted a US$10 electronic gift card to each participant. While interpreters introduced survey participation to patients, they did not administer it and were not present when patients responded to it. To reduce potential response biases, they completed the survey privately—either online by having the survey link sent to their email address, over the phone with a trained research team member from (redacted institution) within 1–2 weeks or in a designated private area within the clinic. The survey did not ask patients to evaluate individual interpreters, but rather to report their experiences of clinician communication and interpreter support.

During the study period, 1211 unique Spanish-speaking LEP patients were informed about the patient experience survey after an MIPP-supported encounter. Of these patients, 561 (46.3%) expressed interest in completing the survey either online or by phone, and 303 surveys were completed in Spanish for an overall survey response rate of 25.0%. Of the 303 Spanish surveys completed, 206 (68%) were completed over the phone with a researcher and 97 (32%) were self-administered online within 1–2 weeks of the encounter. English and Spanish versions of the patient experience survey are available in the online supplemental file, exhibits 1 and 2. Survey responses in Haitian Creole, Farsi, Karen, Pashto, Russian and Swahili were collected during the study timeframe (n=114); however, these patient groups exclusively received remote interpreter services and were excluded from the analyses because they had no variation in interpreter services delivery method, a key variable examined in the study.

Encounter data were collected via an encounter assessment data form that medical interpreters would fill out after each encounter. This form assessed patient demographics (including age and Medi-Cal eligibility) and interpreter service utilisation (including interpreter services delivery method and clinical services). Encounter data were then integrated with patient survey data to compare remote versus in-person interpreter services.

Patient experience survey measures

Main independent variable

The main independent variable is interpreter services delivery method (remote vs in-person) and is a binary variable of in-person interpreter services (1) or remote interpreter services (0) based on the interpreter encounter assessment form. Even though the language services companies had the capabilities to deliver audio-video remote interpreter services, all MIPP-supported remote interpreter services were delivered using only audio.

Outcome measures

Patient care experiences were assessed using two composite survey measures, each consisting of four questions: (1) clinician communication and (2) interpreter support. The questions were adapted from the visit version of the Clinician-Group version of the Consumer Assessment of Healthcare Providers and Systems (CG-CAHPS) survey.¹³ Adaptation entailed reducing language complexity so that questions were written below a sixth grade reading level based on Flesch-Kincaid readability tests.¹⁴

For clinician communication, patients were asked: (1) if the clinician listened carefully, (2) showed respect, (3) encouraged the patient to ask questions and (4) spent enough time with them. Patients had three response options: yes, definitely; yes, somewhat and no.^{15 16} Responses of ‘yes, definitely’ were scored as 100, ‘yes, somewhat’ was scored as 50 and ‘no’ was scored as zero. The four questions were equally weighted and then averaged to generate a 0–100 composite score for clinician communication.¹⁵ Survey composite measures were standardised to a 0–100 so that coefficients from regression models could be interpreted similarly across composite measures with different numbers of items within them. This scoring method is consistent with scoring guidance from the CG-CAHPS, a widely used and well-validated patient experience survey.¹⁷ The clinician communication composite had adequate internal consistency reliability (α=0.75).

For interpreter support, the patients were asked if (1) the interpreter helped explain how they were feeling to the clinician, (2) whether the interpreter helped them understand the clinician’s instructions and (3) whether the interpreter treated them with courtesy and respect. These three questions had three response options: ‘yes, definitely’, ‘yes, somewhat’ and ‘no’.¹⁸ To standardise the scoring with the clinician communication composite, the interpreter support response options were scored as: ‘yes, definitely’ was scored as 100, ‘yes, somewhat’ was scored as 50 and ‘no’ was scored as 0. The fourth interpreter support question asked patients to rate the interpreter on a numeric scale of 0–10, with 0 being the worst interpreter possible and 10 being the best interpreter possible. The question was scored by multiplying the rating by 10 to align with the 0–100 score range of the other three questions. The four item scores were then averaged to calculate the interpreter support composite score (internal consistency reliability, α=0.60).¹⁹

These quantitative composite measures were supplemented by open-ended response questions where patients could elaborate on their responses to close-ended questions. Specifically, they were asked about the ways their experience with MIPP interpreters differed from past experiences without a professional medical interpreter and if they had any recommendations for improving the provision of interpreter services.

Control variables

Regression analyses of the patient experience survey data controlled for patient age, sex, primary clinical service (primary care, health education, women’s health, other), survey mode (researcher administered vs online, self-administered), self-rated physical health (excellent/very good/good vs fair/poor) and self-reported English-language proficiency based on past research examining LEP patients’ experiences of care (very well/well vs not well, not at all).^{2 20}

Quantitative analyses of patient experience surveys

Descriptive statistics summarised the distribution of survey responses and interpreter services methods (remote vs in-person) were stratified to compare patient characteristics. To understand the association of covariates for regression modelling, correlation coefficients for all covariates were calculated. A correlation of 0.60 was used to identify high correlations and no covariates were removed based on this criterion.

Multivariable linear regression models were estimated to examine the association between interpreter services delivery method (remote vs in-person) and each of the patient care experience composite measures. The first regression model was for clinician communication and the second regression model was for interpreter support. Patient age, sex, primary clinical service, self-reported English proficiency, self-reported physical health and survey mode were adjusted for in both regression models. Both survey composite measures were log-transformed to correct for heteroskedasticity due to their skewed distribution.²¹ Both regression models accounted for the clustering of patients within clinics using robust SEs. After regression estimates, we assessed collinearity by calculating variance inflation factors (VIFs). None of the models had VIF values >2.0; therefore, no variables needed to be removed from the final models.

We also conducted sensitivity analyses to address selection effects using propensity score analyses. Propensity score analyses help balance measured patient characteristics for remote and in-person interpreter visits when comparing the delivery methods. R Statistical Package V.4.3.2 was used for all quantitative statistical analyses.²²

Qualitative analyses of open-ended patient survey responses

Responses to open-ended patient experience survey questions were recorded, transcribed verbatim, translated and then cleaned to remove any confidential and/or identifying information. Then, the transcripts were coded to identify recurring themes. The results were discussed and themes were identified and refined at bi-weekly team meetings. Thematic analysis related to the use of interpreter services delivery methods included quantifying the frequency of recurring themes, consolidating patient experiences and selecting representative quotes for each theme.

Clinician interview data

We conducted research interviews with CHC clinicians about their experiences of using remote and in-person interpreter services for Medi-Cal members with LEP. Clinicians from three CHCs were invited to participate in a single interview between 10 February 2023 and 26 April 2023. Research team members invited clinicians to participate via email and, when needed, coordinated time off with a clinic supervisor to facilitate their participation in the interview. The invitation requested clinicians to participate in a single, 30–45 min interview and included an information sheet outlining the study’s purpose, procedures and the voluntary nature of participation. All interviews were conducted via secure video conferencing through Zoom. Interviews were conducted by the Principal Investigator or trained research analyst. During the interview, verbal consent was obtained, and participants were reminded they could withdraw at any time without consequence. Interviews were audio-recorded with consent and transcribed verbatim. Participants were emailed a virtual US$25 gift card as a token of appreciation for participating. The recordings and transcripts were stored securely and anonymised to protect participants’ confidentiality.

Clinician interview measures

Guided by past research examining medical interpreter services,^{4 5} a semi-structured interview guide was developed. The interview guide included 19 questions organised into three sections: (1) experiences with interpreter services, (2) impact of interpreter services on patient care and (3) comparison of remote and in-person interpretation. Open-ended interview questions were used to facilitate an in-depth understanding of clinicians’ experiences and preferences of interpretation methods. The clinician interview guide is available in the online supplemental file, exhibit 3.

As part of the comparison of interpreter methods, clinicians were asked to rate six interpretation methods on a scale from 0 to 10: (1) in-person interpreter, (2) audio-video interpreter, (3) audio-only interpreter, (4) with a bilingual staff member (eg, medical assistant), (5) ad hoc interpretation using a family member or friend and (6) having no interpreter. For each method, they were asked to explain their rating based on the interpreter’s ability to accurately and clearly convey medical information, the interpreter’s respect for patient confidentiality, the effectiveness of communication between the interpreter and the patient and the time required to complete the interpretation.

Clinician interview analyses

Our qualitative analysis followed a rigorous thematic analysis approach, as outlined by the Standards for Reporting Qualitative Research (SRQR),²³ to ensure systematic identification, organisation and interpretation of key themes emerging from clinician interviews. Three researchers independently coded the first six transcripts, using inductive and deductive coding and analyses to capture emerging themes. Coding discrepancies were reconciled through weekly team discussions, and the final coding framework was refined through consensus. After achieving consensus on the coding approach, the remaining transcripts were divided among the three researchers, who coded them independently while maintaining regular check-ins to ensure consistency. New themes that emerged were reviewed collaboratively to determine whether the coding framework required adjustments. We adhered to the SRQR and incorporated key elements from the Consolidated criteria for Reporting Qualitative research, when applicable.²⁴ Data were analysed using NVivo V.12 software, which assisted in organising and managing qualitative data. NVivo’s analysis features were used to examine all transcript segments associated with each code. The identified themes were summarised with supporting evidence from the interviews and refined through discussion and iteration. Responses to interpreter services delivery methods ratings, on a scale from 0 to 10, were analysed, by calculated means and SD.

Results

Patient experience survey findings

Quantitative results

Among the 303 patient experience survey respondents (mean age: 40.4, % female: 69.0%), 236 (78%) received in-person interpreter services and 67 (22%) received remote interpreter services (table 1). Most patients reported speaking English ‘not at all’ or ‘not well’ and the levels were similar for patients receiving in-person interpreter services (83.5%) and those who received remote interpreter services (83.6%). Primary care was the most common primary clinical service provided (73.3%), followed by other services (paediatrics, dental, medication management and mental health) (18.2%), women’s health (7.6%) and health education (1.0%). Both in-person and remote interpreters were mainly used for primary care (75.9% and 64.2%, respectively).

Table 1. Descriptive statistics of Spanish-speaking patients with limited English proficiency, by interpreter services delivery method.

	All patients	In-person interpretation	Remote interpretation
Patient population (n, row %)	303 (100%)	236 (77.9%)	67 (22.1%)
Age (mean (SD))	40.4 (21.3)	39.7 (21.5)	43.0 (20.5)
Sex (n, column %)
Female	209 (69.0%)	152 (64.4)	57 (85.1)
Male	94 (31.0%)	84 (35.6)	10 (14.9)
English-language proficiency (self-reported) (n, column %)
Not at all	136 (44.9%)	102 (43.2%)	34 (50.7%)
Not well	117 (38.6%)	95 (40.3%)	22 (32.8%)
Well	38 (12.5%)	30 (12.7%)	8 (11.9%)
Very well	12 (4.0%)	9 (3.8%)	3 (4.5%)
Physical health (self-reported) (n, column %)
Excellent, very good or good	207 (68.3%)	156 (66.1%)	51 (76.1%)
Fair or poor	96 (31.7%)	80 (33.9%)	16 (23.9%)
Clinical service (n, column %)
Adult primary care	240 (79.2%)	184 (78.3%)	56 (83.6%)
Health education	3 (1.0%)	3 (1.3%)	0 (0%)
Women’s health	42 (13.9%)	36 (15.3%)	6 (9.0%)
Other service	18 (5.8%)	12 (5.1%)	5 (7.4%)
Survey mode (n, column %)
Researcher phone administered	195 (64.4%)	138 (58.5%)	57 (85.1%)
Online, self-administered	108 (35.6%)	98 (41.5%)	10 (14.9%)

Other services include paediatrics, dental, medication management and mental health.

High proportions of patients reported the most favourable response options across questions (online supplemental figures 1 and 2), resulting in ceiling effects.²⁵ The mean clinician communication composite score for in-person (mean=93.5, SD=15.9) and remote (mean=92.4, SD=15.7) interpreter methods was comparable and similarly high (online supplemental table 1). The mean interpreter support composite scores for patients receiving in-person (mean=99.3, SD=2.4) and remote (mean=98.2, SD=6.3) interpreter methods were very high. Most (87.7%) patients rated their interpreter 10 out of 10 points possible (online supplemental figure 3).

Adjusted regression results for the clinician communication and interpreter support models are summarised in table 2. Interpreter services delivery methods (remote vs in-person) were not significantly associated with clinician communication or interpreter support (p>0.05).

Table 2. Multivariable regression results: association of interpreter services delivery method and patient-reported care experiences.

	Clinician communication(n=303)	Interpreter support(n=303)
	Coefficient (95% CI)	Coefficient (95% CI)
Interpreter services delivery method (in-person vs remote)	0.0033 (−0.062, 0.069)	−0.012 (−0.032, 0.0093)
Age	0.0045*** (0.003, 0.006)	0.000 (−0.000, 0.000)
Female sex	0.056 (−0.016, 0.11)	−0.007 (−0.025, 0.012)
English-language proficiency (very well/well vs not well/not at all)	0.13*** (0.062, 0.20)	0.005 (−0.018, 0.027)
Physical health (excellent/very good/good vs fair/poor)	−0.019 (−0.078, 0.040)	0.004 (−0.015, 0.023)
Clinical service
Adult primary care	Reference	Reference
Health education	−0.087 (−0.35, 0.17)	0.020 (−0.062, 0.10)
Women’s health	0.036 (−0.063, 0.14)	−0.002 (−0.034, 0.029)
Other	0.051 (−0.017, 0.12)	0.001 (−0.021, 0.022)
Survey mode (researcher phone administered vs online self-administered)	−0.088** (−0.14, −0.031)	0.013 (−0.0051, 0.031)
Intercept	4.31***	4.59***

Regression model coefficient estimates of the log-transformed outcome measures are presented; 95% CI and p values.

^*P<0.05, **p<0.01, ***p<0.001.

For the clinician communication model, older patient age was associated with better clinician communication scores (β=0.004, 95% CI 0.003, 0.006, p<0.001). Having some level of English proficiency, compared with little to none, was associated with better clinician communication scores (β=0.13, 95% CI 0.06, 0.20, p<0.001). Phone surveys administered by researchers were associated with lower clinician communication scores (β=−0.06, 95% CI −0.14 to –0.03, p<0.05) compared with online, self-administered surveys.

The adjusted R² values of the regression models indicate that 13.6% of the variance (adjusted R²=0.14) was explained for in the clinician communication composite measure and 5.6% of the variance (adjusted R²=0.06) was explained for in the interpreter support composite measure.

Table 2 summarises the multivariable regression model estimates for the final models that examined the association of interpreter services delivery method (in-person or remote) with (1) clinician communication and (2) interpreter support. Interpreter services delivery methods were not significantly associated with either of the patient experience composite measures (table 2). Adjusted comparisons of clinician-patient communication and interpreter support for in-person and remote interpreter service delivery methods are illustrated in the online supplemental figure 4. Consistent with the main findings, results of the sensitivity analyses using propensity score methods (online supplemental table 2) found no differences between in-person and remote interpreter services.

Qualitative results

Overall, the patient experience survey respondents were very satisfied with the professional medical interpreter services and did not have many recommendations for improvement. Without prompting, however, 11 patients specifically expressed a preference for in-person interpretation services over audio-only remote services. There was also concern that the current technology being used for audio-only remote interpretation services was outdated, resulting in poor call quality and unstable connections. As reported by one Spanish-speaking patient who used in-person MIPP services:

I think [MIPP interpreters] are very good, since they are in-person. Now many services are robotic, in health care.

Clinician interview findings

Based on MIPP encounter records and clinic leader recommendations, clinicians with direct interaction with MIPP interpreters and Medi-Cal members with LEP were invited to participate in a research interview. Of 23 invited participants, 19 interviews were conducted, which is an 82.6% response rate. Participants included nurse practitioners (n=9), health educators (n=6), primary care physicians (n=3) and a dentist (n=1).

Clinician ratings of interpreter services delivery methods

An in-person medical interpreter was rated the highest by clinicians (mean score of 9.43 out of 10) from the six methods assessed (in-person, audio-video remote, audio-only remote, bilingual clinician or staff member, family or friend or no interpreter) (online supplemental table 3). Clinicians rated the bilingual staff member method as second best, with a mean score of 8.23 (SD=1.69). Audio-video and audio-only remote medical interpretation were rated similarly with means of 7.80 (SD=1.25) and 6.30 (SD=2.78), respectively.

Clinician interview themes focused on interpreter services delivery methods

Interviews revealed six themes related to interpreter services delivery methods: (1) in-person interpretation supports effective communication and clinician-patient relationships, (2) in-person interpretation enhances operational efficiency, (3) cost-effectiveness of interpreter methods depends on language demand and clinic needs, (4) in-person interpretation enhances quality control and reduces privacy risks, (5) considerations when integrating external personnel and (6) the availability of and limited use of audio-video medical interpretation. The six themes are detailed below and summarised in table 3.

Table 3. Clinician interview themes related to using professional medical interpreter services, by delivery method.

Theme	In-person professional medical interpretation	Remote (audio-only) professional medical interpretation	Illustrative quote
Theme 1: in-person interpretation supports effective communication and clinician-patient relationships	Clear communication because facial expression and body language can be assessed. In-person interpreters can build a stronger connection with the patient, translating to greater trust and credibility in the clinician and, therefore, the healthcare system in general. Patients feel they can express concerns fully. This can sometimes result in lengthy conversations, delaying clinician schedules.	Communication can sometimes be unclear because visual cues, such as body language and facial expressions cannot be assessed. The flow of the conversation may be lengthened because the clinician has to tell the interpreter when to start and stop talking rather than picking up on visual cues. Every encounter may be supported by a different interpreter. Without continuity, the interpreter is less able to tailor their approach based on the clinician’s communication style, for example, use of humour.	“…I want to ask them personal questions about how they’re feeling … And so when I have a medical interpreter, she says everything to them that I’m saying so that they get an idea of my personality and the provider that I am, and know that I want to get to know them. And then we go into talking about all these screenings they need in their chronic conditions and things like that so that they’re more … comfortable speaking with me … so if I’m able to build a patient relationship by being myself when I walk into the room and the medical interpreter helps me with that, I feel like there’s better outcomes because they are willing to listen to me and have warmed up to that we’re both there in the room to provide care to them. So they listen to why they need these screenings and this help with their medical care”.
Theme 2: in-person interpretation enhances operational efficiency	Greater ease of interpreter scheduling. On-demand interpretation can be accommodated in community health centres that take walk-in patients, experience unexpected high rates of missed appointments and have often run behind schedule. Alleviates the ad hoc interpreting responsibilities of medical assistants (MAs), who are then able to focus more on their MA duties. Bottlenecks can arise when multiple clinicians require an interpreter at the same time.	Access to multiple interpreters at once, alleviating the bottleneck of having to share one interpreter among too many clinicians. Given that scheduling remote interpreters usually must be done beforehand, using remote interpretation is easy for clinicians to use because they can ‘just pick up the phone’. Prescheduling of remote interpreters does not allow for the flexibility needed by community health centres.	“Definitely. I do agree with that. I believe, yeah, ever since I’ve started with the medical assistant role assisting, … the [interpreter] has made my work life easier … I’m able to … get stuff done probably faster than usual when [interpreter] is around. … We have hectic days … where I’m running around and and at the same time I’m needing to interpret and I mean do interpretation and I have other stuff too that need to be sent out or need to be run down in the lab. So definitely has made my life, my work life easier having [MIPP interpretation] in office”.
Theme 3: cost-effectiveness of interpreter services delivery methods depends on language demand and clinic needs	In-person interpreters can be cost-effective if there is sufficient, consistent interpretation need and non-interpretation time can be reallocated to patient care activities (eg, calling patients to help schedule appointments, provide appointment reminders, collect follow-up information, etc). Costly if the interpreter’s schedule is unbooked and cannot be repurposed to supporting other clinical care activities. Costly to hire an in-person interpreter if the interpretation demand for a language is low and sporadic.	Remote interpreters can be cost-effective if they are used on an ‘as-needed’ basis rather than hiring an in-person professional medical interpreter and managing their schedule and work. Access to a broader range of languages because it is often not feasible or cost-effective to have staff in-person interpreters for uncommon languages. Unless there is an established agreement with the language services company, it can be costly if the remote interpreter has to be prescheduled, patients miss their appointments or patients and/or clinicians are not available at the time of the scheduled appointment.	“These services, these interpreter services need to be on demand. Our patients … cannot do the prescheduled service. Even if they do, let’s say [someone] makes an appointment to go in for, let’s say family practice … And they’re there, he gets there on time. Maybe the provider’s running late because there was some urgent case that came in ahead of him and the interpreter’s on the line from [MIPP]. Well, that interpreter’s gonna sit there some, a lot of times paid for 45 minutes to wait … But then it isn’t just finished when they see that primary care doctor, because now the primary care doctor has referred them down to the lab, then the pharmacy, and perhaps imaging each one of those has to be pre-scheduled. So how does that work? It doesn’t”.
Theme 4: in-person interpretation enhances quality control and reduces privacy risks	More interpreter accountability, ease of performance monitoring (as opposed to the anonymity of providing services solely over the phone). Less risk for privacy concerns and interception of communication because conversations are happening privately when in-person.	Less managerial control and monitoring of interpreter performance, resulting in potential quality and challenges of managing privacy concerns.	“We had a horrible situation the other day that resulted from this, because there’s no way for us to verify that the translator is actually the translator that we’re supposed to have. And somebody else was using that person’s phone. … It was clear because the voice was different. The person was different than who was doing the translation because our staff noticed that … was a totally different person”.
Theme 5: considerations when integrating external personnel	Employed clinic personnel may not be receptive to having external personnel providing services on-site due to teamwork and patient confidentiality concerns. Clinics sometimes lack the physical space for on-site interpreters to work when not supporting clinic encounters.	No external personnel are on-site, limiting the need for direct supervision, clinic management and oversight of daily operations.	“We cannot take liability for an outside agency to come on our property and interact with our patients as part of our patient care. Who is not trained, is not an employee, doesn’t know about even where these other units are. They wouldn’t even know how to get a hold of any of our other units … We don’t know their training, … background, their qualifications. … so you would’ve taken liability and responsibility for somebody wandering around inside our organization”.
Theme 6: availability of and limited use of audio-video medical interpretation	Audio-only remote medical interpretation was often used for in-person medical encounters. Audio-video medical interpretation was not used for in-person encounters with patients.	Audio-video remote medical interpretation was considered useful for telemedicine encounters, but it was rarely used. Clinicians underscored the promise of remote audio-video interpretation to improve communication because of their ability to consider visual cues. Digital exclusion, limited access to devices and concerns about usability and privacy contribute to the low uptake of video-based services.	“The video, I like having the person’s face there. They’re able to see someone and it’s kind of the same thing. They’re able to see me, I’m able to see them, they can see my face. Body language has a lot to do with how I speak as well. So just them able to see my face and interpret the phone. Like I said, at times, my experience with the phone translating, and I don’t know if this is just related to our systems, but it gets choppy. I get a lot of words missed or skipped, can’t, it depends on the connection that can’t hear something, the patient can’t hear the phone as well”.

Theme 1: in-person interpretation supports effective communication and clinician-patient relationships

Clinicians unanimously favoured in-person interpreters over remote interpreters for effective patient communication and supporting their relationships with patients. When working with an in-person interpreter, clinicians reported that they could comprehensively communicate using non-verbal cues such as facial expressions and body language. With remote interpreters, clinicians reported that patients are not able to observe visual cues from the interpreter, and the flow of communication can be hindered. Interpreters are less able to assess natural breaks in the clinician’s speech, resulting in interruptions. Clinicians indicated that repeating the message, providing directions to the interpreter over the phone and cueing the interpreter to start and stop interpretation result in longer encounter times and may increase the likelihood of communication errors compared with in-person interpretation. As one clinician highlighted,

I think just having them be in the room and just being able to interpret word for word what you’re saying as a clinician to the patient because they are you know, they’re mirroring me and they’re, they’re my voice or my line of communication with that patient. And if I’m able to have a medical interpreter in a room, I’m able to talk to them the same way and for them to understand in the same way that any English-speaking person would for what I’m saying. And it makes a huge difference.

Through repeated interactions, in-person interpreters also become familiar with clinicians’ communication approaches, including their sense of humour, and are better able to assist the clinician in building rapport with patients. One clinician elaborated on this nuance, sharing:

Let’s say something is preventing [the patient] from their regular hike or bike riding. Someone has fallen, he’s not able to ride their bike. You know, I tried to see where his life has changed in terms of how his interests are doing. And then of course, you know help figure out a way to help them. And most of the time you also build kind of like jokes. …And going through an interpreter. If …I’ve known the interpreter long enough so the interpreter knows my style. That’s why I prefer a consistent interpreter because over time they would understand you know, when you’re joking and also convey that it also makes the patient understand that.

Moreover, in-person interpreters are better able to emulate clinicians’ physical gestures and interpersonal approaches to provide more authentic experiences for patients. A clinician described this unique advantage of working with in-person interpreters:

I think just having them be in the room and just being able to interpret word for word what you’re saying as a clinician to the patient because they are you know, they’re mirroring me and they’re, they’re my voice or my line of communication with that patient. And if I’m able to have a [professional] interpreter in a room, I’m able to talk to them the same way and for them to understand in the same way that any English-speaking person would for what I’m saying. And it makes a huge difference.

Theme 2: in-person interpretation enhances operational efficiency

Clinicians reported both benefits and drawbacks of remote and in-person interpreter service methods in terms of flexibility and integration into operational workflows. When using in-person interpreters, bottlenecks can arise when an interpreter is shared among multiple clinicians. Clinicians noted, however, that in-person interpreters have greater scheduling flexibility because their physical presence facilitates their ability to respond to on-demand interpretation requests. Additionally, when scheduled patients miss their appointments, clinicians reported that in-person interpreters were often able to use that time productively by supporting patient care—such as making reminder phone calls or translating patient-facing written materials. In contrast, when using remote interpreters, prescheduled time slots were frequently missed due to patient delays and the need to accommodate walk-ins, which limited their responsiveness. Prescheduled remote interpretation can work well for appointments that run on time and when multiple clinicians need access to interpreters across a broad range of languages. Overall, clinicians emphasised that on-demand access to interpretation was the most critical success factor for responsive interpreter services, regardless of whether the method was in-person or remote.

On-site professional medical interpreters enabled the delivery of on-demand interpreter services, which reduced the interpretation burden on medical assistants. Prior to having on-site MIPP interpreters, medical assistants were often interrupted from their main job responsibilities to help accommodate unanticipated interpretation demands for Spanish-speaking patients. Incorporating in-person professional medical interpreters into operational workflows allowed for medical assistants to focus on their core duties and improve the clinic’s operational efficiency. As one clinician explained:

And so when [MIPP interpreter] was able to do that in her role, the medical assistant was free to really optimize their medical assistant role. And when that’s happening, they can just do their job. And so when that happened, the productivity numbers were higher, and there was better satisfaction with the services that were provided and medical assistance weren’t so frustrated about having to do these dual roles. And as far as turnover, I feel like there has been some turnover, but probably not as much as there was in previous years. And that might be multifactorial, but definitely the presence of medical interpreters from the MIPP project reduced that.

Theme 3: cost-effectiveness of delivery methods depends on language demand and clinic needs

In-person and remote interpretation were both perceived as having potential for being cost-effective for CHCs under the right conditions. If there is a high patient demand for a common language, such as Spanish, clinicians indicated that it made financial sense to integrate an in-person professional medical interpreter on a part-time or full-time basis. Any interpreter downtime can be used to help complete other patient care activities, such as conducting outreach calls to schedule follow-up appointments and to assist with population health management activities. However, if interpretation demands are sporadic and scattered across multiple languages, using remote contracted interpreters is considered more cost-effective and helps avoid operational inefficiencies associated with the suboptimal use of interpreter time.

Theme 4: in-person interpretation enhances quality control and reduces privacy risks

Regarding quality control and privacy regulations, the performance of in-person interpreters can be monitored in real time, which was reported by clinicians to foster a strong sense of accountability and reduced risks of potential privacy breaches. In contrast, monitoring remote interpreters’ performance and conducting quality control activities was reported to be more challenging because of inherent privacy drawbacks of transmitting confidential information over the phone compared with in-person. This was described by a key stakeholder:

We had a horrible situation the other day because there’s no way for us to verify that the interpreter is actually the interpreter that we’re supposed to have. And somebody else was using that person’s phone. And it was clear because the voice was different. Our staff noticed that the person with the name that was supposed to be interpreting had interpreted before. The person on the phone was a totally different person.

Managing an in-person interpreter workforce was also reported to present some challenges related to integrating them with the workplace culture, managing relationships with third-party language services companies and overcoming physical space constraints. Remote interpretation was perceived as potentially simplifying these issues because CHC leadership can focus on managing external company relationships, while factors such as securing space for interpreters and integrating them into workplace culture can be avoided.

Theme 5: considerations when integrating external personnel

Clinicians emphasised important considerations when integrating external personnel into clinic operations. Having external interpreters work on-site introduced administrative complexity. Unlike employed staff, externally employed interpreters do not always have required screening and background checks completed, creating potential issues with team cohesion, patient confidentiality and interpreter accountability. Employed staff, however, require ongoing performance monitoring and management, necessitating additional administrative resources for supervision.

Remote interpreters can alleviate some of the challenges associated with in-person interpretation and have the benefit of not requiring physical space at the clinic. Clinicians reported that some examination rooms do not have adequate space for another person to be present during encounters and that can only accommodate remote interpretation service methods.

Clinician feedback emphasised that the integration of external personnel requires sufficient physical space, a team-oriented organisational culture and managerial capacity.

Theme 6: availability of and limited use of audio-video medical interpretation

Although two of the three MIPP language services companies offered audio-video remote medical interpretation, none of the clinicians interviewed reported firsthand experiences of using video-based interpretation during patient encounters. As a result, clinician ratings of video interpretation were based on their perceptions rather than direct experience. Despite this, clinicians reported video interpretation as a promising alternative to audio-only remote interpretation, citing its potential to enhance communication by incorporating visual cues such as facial expressions and body language.

Clinicians reported that the absence of visual elements when using audio-only interpretation limits their ability to fully convey meaning, assess patient engagement and establish rapport. Several clinicians suggested that expanding access to video-based interpretation could bridge some of the communication gaps associated with phone-based services, particularly for more complex medical discussions. Concerns were raised, however, regarding the feasibility of implementing video-based interpretation in CHC settings, including technology limitations, clinic infrastructure constraints and patient digital literacy barriers. Future efforts to improve interpreter services should examine how video-based remote interpretation can be integrated into operational workflows to optimise patient-clinician communication while maintaining accessibility and ease of use.

Discussion

This mixed methods research study aimed to understand differences in patient and clinician experiences of using in-person and remote medical interpretation in CHCs. Among Spanish-speaking patients with LEP, we found that remote professional medical interpretation was not associated with worse patient-reported experiences of clinician communication and interpreter support compared with in-person interpretation. After controlling for potential confounding variables including patient age, self-reported English proficiency, health status and the method of survey completion, we found no statistically significant differences between remote and in-person interpreter services delivery methods. These findings align with past research indicating that patients with LEP are equally satisfied with in-person and remote interpretation.^{10 11}

While no statistically significant differences in patient-reported communication quality or interpreter support between in-person and remote interpretation were identified, analyses of the open-ended survey responses indicated that patients had strong preferences for in-person interpretation over remote services. Patients reported concerns about the phone technology used for remote interpretation, describing issues such as poor call quality, unstable connections and outdated equipment. While these concerns were only proactively reported by a few patients, their feedback revealed the limitations of remote interpreters using audio-only technology.

Our findings also align with past evidence indicating that patients with LEP and clinicians prefer in-person interpreters over remote interpreters due to improved rapport-building, trust and perceived communication quality.1,3 Our study builds on past evidence by demonstrating that, despite patient-reported experiences of clinician communication and interpreter support being no different for remote and in-person interpretation, clinicians and patients reported strong preferences for in-person interpretation. The results highlight the importance of ongoing financial support for CHCs to assure the provision of responsive interpreter services using audio-only, in-person and video-audio service delivery methods. There are important trade-offs of using different delivery methods, so developing and using patient decision aids²⁶ could help clinicians and patients make more informed decisions.

We found that clinicians rated ad hoc interpretation from bilingual medical assistants or other clinic staff as their second preferred interpreter services delivery method. This preference was described by clinicians as driven by the immediate availability of bilingual staff members, who can often provide on-demand interpretation without external scheduling requirements or coordination. Reliance on bilingual clinic staff to interpret on an ad hoc basis, however, raises important concerns, including their lack of formal training in medical interpretation and equity concerns associated with being pulled away from their primary job responsibilities, which can also disrupt the flow of clinic operations and decrease staff job satisfaction.

Using remote, externally contracted medical interpreters on an as-needed basis has some benefits, including scheduling flexibility, reduced human resources expenses compared with employment and broad language coverage. Despite these benefits, clinicians reported that audio-only interpreters were less able to foster rapport and patient trust compared with in-person interpreters.⁹ Relying on external language services companies to provide interpreter services may be misaligned with the missions of CHC organisations serving high proportions of immigrant and refugee populations. Hiring dedicated staff interpreters when there is not sufficient demand for a specific language or when interpreters are not trained to support patient care activities, however, could risk using limited CHC resources inefficiently. It may be more efficient for CHCs to improve the medical interpretation skills of existing staff, such as community health workers, patient navigators and cultural liaisons, who can then provide medical interpretation as part of their job responsibilities.

Innovation is needed to support the increased use of video-based interpreter services because it offers the benefits of both in-person and audio-only interpretation, while addressing their drawbacks. Although video-based interpretation enables visual cues that are essential to building rapport and improving communication clarity, clinicians reported limited or no use of this method and expressed interest in expanding access. This gap between the availability and implementation of video-based interpretation highlights that patients with LEP often face systemic barriers to engaging with video technology because of digital exclusion, limited access to devices and concerns about usability and privacy.²⁷ The low uptake of video-based services is not due to their lack of interest. Efforts to make video-based interpretation more feasible for CHCs to routine use should prioritise addressing digital literacy gaps, broadband access, culturally and linguistically appropriate communication tools and community-informed implementation.²⁸ Artificial intelligence-based medical interpretation solutions also hold promise to support clinician communication for patients with LEP, although they have not been rigorously tested among patients served by CHCs.²⁹

Limitations

The study results should be interpreted considering some limitations. First, although the patient experience survey response rate is comparable with other surveys of low-income populations,³⁰ non-response bias may have limited variation in patient experiences reported compared with a more representative sample. This could potentially bias the results towards the null hypothesis of finding no significant differences between in-person and remote interpreter services delivery methods. Second, limited variation for the study outcome measures could have impacted the regression estimates.³¹ A large and more diverse sample of patients could have increased variation in the patient experiences assessed and statistical power to detect differences by interpreter services delivery method. Third, internal consistency reliability for the interpreter support composite measure was relatively low (α=0.60), so low measurement precision may have biased results towards the null hypothesis of finding no differences between interpreter methods. Fourth, the interpreter’s role in introducing the survey may have influenced patients’ willingness to participate and may have resulted in socially desirable responses, biasing our results towards the null hypothesis of finding no differences between interpreter services delivery methods. The early introduction of the survey enabled phone surveys of patients to be administered within 2 weeks of the clinical encounter, reducing recall bias.³² Finally, we interviewed a modest sample of CHC clinicians, possibly limiting the identification of implementation barriers and facilitators related to interpreter service delivery methods. Future research should compare interpreter methods in a broader set of CHCs and in multiple languages. Ethnographic research can also lend insight about how to improve the responsiveness of medical interpreter services for patients with LEP.³³

Conclusion

Since 2000, US federal law requires that healthcare organisations receiving federal funding provide professional medical interpretation when language concordant care is not possible.³⁴ Despite this, few CHCs employ in-person professional medical interpreters. In-person interpreters were strongly preferred by CHC clinicians, but patients reported comparable experiences of clinician communication and interpreter support irrespective of the interpreter services delivery method used. Remote interpreters can offer practical advantages in terms of efficiency and accessibility under the right conditions, but clinicians and patients strongly prefer in-person interpretation. Policymakers and healthcare payers should consider expanding payment models that enable the provision of interpreter services using multiple methods to meet the operational needs of CHCs. Reforms that incentivise improved access to on-demand professional medical interpretation can also improve the provision of patient-centred primary care to patients with LEP.³⁵

The content is solely the responsibility of the authors and does not represent the official views or policies of the California Department of Health Care Services.

Data availability statement

No data are available.