Effect of Virtual Versus In Person Interpreting on Diabetes Outcomes in Non-English Language Preference Patients: A Pilot Study

Setting

This research was conducted at an urban primary care clinic in a Midwestern state that serves an ethnically, culturally, and linguistically diverse population.^26,33 A large proportion of patients were immigrants and refugees from Southeast Asia, East Africa, and Latin America. Based on 2021 statistics, 42% of clinic visits required an interpreter. The clinic is affiliated with a large state university and includes an interdisciplinary care team model. The clinic serves as a teaching site for undergraduate and graduate health professionals, including medical, pharmacy, and behavioral health trainees.

Study Sample

We employed inclusion and exclusion criteria to identify a study sample appropriate for our research question. Patients were eligible for inclusion if their record indicated: a diagnosis of T2DM; requirement of a medical interpreter; a hemoglobin A1c (HbA1c) laboratory value collected between January 1st, 2021 and January 1st, 2022; and at least 1 in person clinic visit between February 1st, 2021, and June 30th, 2021 with a diagnostic code for T2DM. Patients under age 18 years, those with type 1 diabetes, and pregnant patients were excluded.

Measures

Once the study sample was identified, we extracted multiple variables including patient demographics, language spoken, duration of diabetes diagnosis, HbA1c values, physician, presence of a clinical pharmacist visit, and the qualifying and the next subsequent diabetes visit dates and HbA1c value. In addition, certain variables required manual data extraction, including interpreter format and intervention complexity. One investigator manually reviewed visit notes in the electronic health record to determine interpreter format (in person vs virtual) present via nursing documentation.

To determine intervention complexity, we used an approach conceptually informed by the Medication Regimen Complexity Index (MRCI), a validated measure used to systematically classify the demands of medication regimens and previously used in adults with diabetes.^27,34 The complexity of diabetes intervention was classified within the clinic visit based on MRCI items as well as the author team’s clinical perception of patient burden. Clinic visits were manually coded from 0 to +6, as follows: no medication change (0), dose increase or decrease of 1 existing medication (+1), dose increase or decrease in ≥2 existing medications (+2), initiating a new oral medication (+3), starting ≥2 new oral medications (+4), starting a new glucose monitoring device (+5), or starting a new injectable medication (ie, insulin; +6; Figure 1). These visit scores were subsequently collapsed into 3 complexity categories: low (score of 0), moderate (scores of 1-3), and high (score of >3).

OPEN IN VIEWER

Figure 1.

Coding scale employed to classify complexity of interventions at clinic visits among a sample of limited English proficiency patients of an outpatient clinic with Type 2 Diabetes mellitus, 2021.

Data Analysis

We calculated univariate and bivariate statistics to characterize the sample and assess the extent to which measures of continuity (follow-up visit rate and time to follow-up visit), quality (change in HbA1c), and complexity (intervention complexity) were associated with interpreter type. We employed available-case analysis given sample size differences across variables. For example, in order to calculate the average time to follow-up visit, those without a follow-up visit were necessarily excluded. Pearson chi-square tests were used to compare categorical variables by interpreter type and independent samples t-tests were used for continuous variables. To account for non-normal distributions for certain variables, we also replicated analyses applying logarithmic transformation for time to follow-up visit and HbA1c. We conducted 2-tailed analyses in IBM SPSS Statistics for Windows, version 25.0 (IBM Corp., Armonk, NY), and set a significance level of 0.05. This study was exempt from human subjects review by the University Institutional Review Board.

Interpreter Type and Follow-Up Rate

Rates of follow-up and time to follow-up visit were similar between patients utilizing in person and virtual interpreters, though follow-up was more frequent when in person interpreters were used. Follow-up scheduling and visit attendance reflects multiple factors, including patient understanding of the care plan and follow-up recommendations, ability to easily schedule follow-up, and willingness to continue care with that clinic and clinician. In person interpreters, who may be able to assist with follow-up scheduling during the initial visit, likely improve follow-up scheduling and return rates.

Interpreter Type and Glucose Control

HbA1c values and change in HbA1c values were also similar between groups, with greater HbA1c improvement in the virtual group. This suggests that virtual interpretation is an effective means to evaluate patient diabetes control and implement changes to improve HbA1c. Patients with virtual interpreters were more likely to have visits with moderate complexity, versus low or high complexity, suggesting that gradual, step-wise changes in the care plan are an effective means to improve HbA1c, especially when using a virtual interpreter to communicate with patients.

Patients build experience and understanding of diabetes over time, including knowledge of diet, medications, and blood sugar monitoring techniques. Medical decision making in patients with diabetes often includes small adjustments based on objective lab and self-monitoring values versus subjective patient-reported symptoms. T2DM treatment involves titration of medications along evidence-based guidelines and is effective in English-speaking patients when provided virtually.^35,36 Previous work also indicates the value of in person interpreters in both developing relationships^16,37,38 and providing patient education. ¹⁶ Comparing the time of diabetes diagnosis and duration of clinician-patient relationship could further evaluate this impact. In person interpreters may have greater impact in patients with newly diagnosed diabetes, or new to a clinic or clinician, where significant education about the disease, monitoring, diet, and treatment options is needed.

Limitations and Future Directions

Our results provide preliminary evidence from a single urban primary care clinic with a family medicine training program. They reflect the impact of interpreter format at a single visit in the context of long-term chronic disease management. Our study focused on the presence and timing of a patient’s next follow-up visit but did not look at the total number of patient contacts during the time period, nor interpreter format utilized at follow-up visits.

Although in person and virtual groups in this study were similar in age, sex, and baseline HbA1c, other factors may have contributed to differences between groups. Virtual versus in person groups were not random; health system policy during the study period required providers to submit written requests, indicating a patient met criteria for use of an in person interpreter. Common indications include communication challenges, such as limited hearing, cognitive deficits such as dementia, severe mental health diagnoses, or limited health literacy. Our study did not compare overall medical complexity, social complexity, or cognitive function variability between patients in virtual and in person groups, which affects the challenge and complexity of interpreting work ³⁷ These patients may have additional challenges with taking medications, especially incorporating new medications or medication changes, and/or have higher target HbA1c goals.

Our study did not measure social determinants of health or health literacy levels between groups nor assess its impact on interpreter format or diabetes outcomes. Effective diabetes care plans are multifactorial, including blood sugar monitoring, diet, and exercise and heavily impacted by patient social determinants of health and health literacy. Access to income supports, healthy food, exercise facilities, transportation, and neighborhood safety all impact a patient’s ability to follow up and their A1c levels. Relatedly, we did not evaluate the influence of care coordination or interdisciplinary interventions on study outcomes, which could be valuable in future research. For example, at this clinic most patient education around glucometers and injectable diabetes medications was provided by clinical pharmacists, while diet and lifestyle education were provided by nurses, dieticians, and behavioral health specialists.

75% of in person interpreter visits included Karen or Burmese speaking patients, an ethnic group newer to the United States, and with increased barriers to accessing healthcare services and knowledge of Western medicine.^39,40 The majority of these visits utilized 3 well-known, frequently-used in person interpreters of the clinic. These interpreters often serve patients with limited health literacy, mental health diagnoses, and cognitive or memory concerns, often with looser HbA1c goals, and increased barriers to tight glycemic control. It is likely that the in person group reflected those with greater complexity and lower health literacy compared to the virtual group, although baseline HbA1c was similar across groups. When set up effectively, even dementia assessments can be conducted via telemedicine, but in person is preferred. ¹⁹ A randomized study would help ensure similar study groups and better assess the impact of interpreter modality, as well as clarify which patients benefit most from specific formats. Further research should focus on elucidating which patient factors, medical or social needs might predict a favorable response to each format.

Clinical Implications

Study results should be interpreted thoughtfully, as they may not be generalizable across communities, clinics, language, specialty, or clinical situation. Diabetes visits, like many chronic diseases, include minimal diagnostic ambiguity, a limited scope of physical exam maneuvers, and rarely require procedural intervention, situations where the use of an in person interpretation may have additional benefits. Primarily discussion-based visits with a defined diagnosis simplifies communication between patient, clinician, and interpreter as compared to other visit types and diagnoses.

Chronic disease management, such as in T2DM, may be an effective model for the use of virtual interpreters. This care is often provided in the context of a longstanding patient-clinician relationship, where rapport already exists. Primary care providers (PCP) often have a pre-existing knowledge of a patient’s cultural background, health literacy level, social context, and views on medication and interventions. This established rapport likely improves communication effectiveness in the virtual environment.

Patients presenting with new complaints, or requiring more complex history taking and physical exam maneuvers, may benefit from the use of an in person interpreter. More complex diagnostic work, include ED visits may further benefit from in person interpretation.^18,20 Patients with new or high complexity diagnoses, when initiating high risk medications and treatments, or visits with a large educational component may benefit from in person interpretation. Visits with interdisciplinary team members, such as nutritionists, clinical pharmacists, and behavioral health specialists may also benefit from in person interpreters.