We conducted focus groups with ethnically and linguistically diverse patients to elicit perspectives on clinician EHR use during outpatient visits. Through posted flyers and in-person recruitment at 7 primary and specialty care clinic waiting rooms (family medicine, adult primary care, obstetrics or gynecology, diabetes, anticoagulation, cardiology, and gastroenterology or hepatology), we identified eligible patients: (1) English, Cantonese, or Spanish-speaking, (2) adults, and (3) receiving primary or specialty care from an urban safety-net hospital. This safety-net system cares for a population that is 16% African-American, 23% Asian, and 37% Latinx and insured predominantly by Medicaid (58%) or Medicare (22%) [12]. A prior study found nearly 50% of patients have inadequate health literacy [13]. There are 24/7 interpretation services via phone interpreters with video or in-person interpreters available during business hours. This system utilizes an Office of National Coordinator (ONC) certified EHR [14]. We collected sociodemographic characteristics by telephone, including use of validated questions to assess English proficiency and health literacy [15,16]. We provided participants US $35 for participation.

The focus group guide (Multimedia Appendix 1) was created by the study team using an iterative process based on analysis of prior studies. We used consensus to ensure questions would be accurately translated before inclusion in the final focus group guide, which was translated into Chinese and Spanish by fluent, bilingual team members. It included questions on perceptions about overall communication, clinician communication, EHR use during visits, multitasking and silent EHR use, and preferences for EHR use. We used videos of reenacted examples [10] to demonstrate silent versus multitasking EHR use to help participants distinguish between these types of use. In Cantonese, there is no single word to convey the idea of multi-tasking so we felt demonstration of these concepts through videos was necessary.

We conducted 6 (3 English, 2 Cantonese, 1 Spanish) in-person 90-min focus groups in patients’ preferred languages from November 2017 to January 2018. A total of 2 Spanish focus groups were planned, but 1 focus group became a one-on-one interview when all other participants did not attend the focus group. The Spanish and English focus groups were conducted by a bilingual research team member; the Cantonese focus groups were conducted by a different bilingual team member. We acquired verbal and written consent from participants before participation. The focus groups were audio-recorded then transcribed into English for analysis. The research team contained Spanish and Cantonese speakers who could consult the audio files if participants’ meaning was unclear in English transcripts.

Using a grounded theory approach [17,18], we (EK, GM, NR) independently coded the same 3 transcripts (1 from each language), then through consensus developed a codebook (Multimedia Appendix 2) that encompassed themes represented in these transcripts. One team member (GM) then applied this codebook to all transcripts using ATLAS.ti 7.0 (ATLAS.ti Scientific Software Development GmbH); no additional themes were identified during this process, suggesting primary thematic saturation [19]. We triangulated results with primary care clinicians through interactive presentations and determined findings most likely to impact clinical practice.

This study was approved by the University of California San Francisco Institutional Review Board.

There were 37 participants (Table 1). Of the 37, 11 were Cantonese-speaking and 5 Spanish-speaking. Mean age was 54, and 57% (21/37) were women. Nearly half (17/37, 46%) reported limited health literacy, 41% (15/37) reported infrequent personal computer use, and 41% (15/37) reported poor or fair health. Cantonese-speaking participants were more likely to report limited health literacy and feel their primary care providers (PCPs) do not know them well.

Table 2 contains themes and representative quotes. Themes were consistent across language groups and classified into 2 categories: perspectives on EHR use and strategies for patient-centered EHR use.

Participants generally accepted EHR use, recognizing that its use assisted with care. This acceptance was conditional on the assumption that EHR use was focused on their care. Participants felt similarly about multitasking and silent EHR use, reporting that each type of use was appropriate during different parts of the visit.

Despite the general acceptance of silent, multitasking, and frequent EHR use, participants had suggestions for how clinicians could exhibit more patient-centered EHR use.

Participants uniformly desired more eye contact during EHR use. In particular, some felt computer use is inappropriate when patients are attempting to show physical concerns:

All participants reported a desire for clinicians to communicate what they were doing in the computer. In particular, they asked clinicians to provide a warning (ie, signaling) before transitioning to silent EHR use.

As part of this desire for transparency about EHR use, participants—including LEP participants—desired screen sharing:

In this short report, we found consensus among linguistically diverse safety-net patients on several themes regarding clinician EHR use. Previously, when this setting employed an EHR that did not meet ONC-certification requirements [14], we found that non-English-speaking patients reported more computer use but less concern that PCPs listened less carefully because of computer use; moreover, Asian patients had higher odds of reporting that computers helped PCPs remember patient concerns [11]. The findings of this study, conducted 3 to 4 years after implementation of a more comprehensive, certified EHR, suggest that safety-net patients across multiple languages experience frequent clinician EHR use but recognize its value to their care even if clinicians multitask or use EHRs silently during visits.

Despite acceptance of EHR use, participants provided suggestions for clinicians to improve the patient experience during EHR use. Safety-net patients—including limited health literate and LEP patients whom clinicians may not expect to read the EHR—wanted clinicians to be transparent about EHR use and even engage them in the process of EHR use. In triangulating our findings with clinicians, we found clinicians felt reassured that patients accepted exam room EHR use and surprised that LEP patients wanted to be included in what clinicians were doing on the computer.

These findings in safety-net patients augment existing recommendations for patient-centered EHR use [20-23]:

This study is limited by a small sample size within each language from a single setting, lack of information about patient-clinician language concordance, inability to report a response rate, and use of a single individual to code transcripts after developing themes through consensus. Strengths of this study are a diverse, safety-net population and recruitment from both outpatient primary care and specialty clinics.

As educators develop communications curricula for patient-centered EHR use, these findings and other perspectives from diverse patients should inform the content [23,33,34]. Future efforts should investigate multilevel interventions to increase adoption of patient-centered EHR use strategies, including computer (EHR user interface or content), patient (activation or empowerment), environment (redesign or reposition equipment in rooms), and policy level (incentives) interventions.

Linguistically diverse safety-net patients accept the prevalence and utility of EHR use during outpatient visits, if focused on their care. However, there continues to be room for improvement for clinicians to adopt patient-centered strategies, including eye contact, signaling EHR use, and screen-sharing with safety-net patients.