Translating Discharge Instructions for Limited English-Proficient Families: Strategies and Barriers

Completeness and readability of GPT-4-generated multilingual discharge instructions in the pediatric emergency department

Objectives

The aim of this study was to assess the completeness and readability of generative pre-trained transformer-4 (GPT-4)-generated discharge instructions at prespecified reading levels for common pediatric emergency room complaints.

Materials and Methods

The outputs for 6 discharge scenarios stratified by reading level (fifth or eighth grade) and language (English, Spanish) were generated fivefold using GPT-4. Specifically, 120 discharge instructions were produced and analyzed (6 scenarios: 60 in English, 60 in Spanish; 60 at a fifth-grade reading level, 60 at an eighth-grade reading level) and compared for completeness and readability (between language, between reading level, and stratified by group and reading level). Completeness was defined as the proportion of literature-derived key points included in discharge instructions. Readability was quantified using Flesch-Kincaid (English) and Fernandez-Huerta (Spanish) readability scores.

Results

English-language GPT-generated discharge instructions contained a significantly higher proportion of must-include discharge instructions than those in Spanish (English: mean (standard error of the mean) = 62% (3%), Spanish: 53% (3%), P = .02). In the fifth-grade and eighth-grade level conditions, there was no significant difference between English and Spanish outputs in completeness. Readability did not differ across languages.

Discussion

GPT-4 produced readable discharge instructions in English and Spanish while modulating document reading level. Discharge instructions in English tended to have higher completeness than those in Spanish.

Conclusion

Future research in prompt engineering and GPT-4 performance, both generally and in multiple languages, is needed to reduce potential for health disparities by language and reading level.

Performance of ChatGPT and Google Translate for Pediatric Discharge Instruction Translation

BACKGROUND AND OBJECTIVES

Patients who speak languages other than English face barriers to equitable healthcare delivery. Machine translation systems, including emerging large language models, have the potential to expand access to translation services, but their merits and limitations in clinical practice remain poorly defined. We aimed to assess the performance of Google Translate and ChatGPT for multilingual translation of pediatric discharge instructions.

METHODS

Twenty standardized discharge instructions for pediatric conditions were translated into Spanish, Brazilian Portuguese, and Haitian Creole by professional translation services, Google Translate and ChatGPT-4.0, and evaluated for adequacy (preserved information), fluency (grammatical correctness), meaning (preserved connotation), and severity (clinical harm), along with assessment of overall preference. Domain-level ratings and preferred translation source were summarized with descriptive statistics and compared with professional translations.

RESULTS

Google Translate and ChatGPT demonstrated similar domain-level ratings to professional translations for Spanish and Portuguese. For Haitian Creole, compared with both Google Translate and ChatGPT, professional translations demonstrated significantly greater adequacy, fluency meaning, and severity scores. ChatGPT (33.3%, P < .001) and Google Translate (23.3%, P = .024) contained more potentially clinically significant errors (severity score ≤3) for Haitian Creole than professional translations (8.3%). Professional Haitian Creole (48.3%) and Portuguese (43.3%), but not Spanish (15%), translations were most frequently preferred among translation sources.

CONCLUSIONS

Machine translation platforms have comparable performance to professional translations for Spanish and Portuguese but shortcomings in quality, accuracy, and preference persist for Haitian Creole. Diverse multilingual training data are needed, along with regulations ensuring safe and equitable applications of machine translation in clinical practice.

A Cyclical Model of Barriers to Healthcare for the Hispanic/Latinx Population

As the Hispanic/Latinx population in the United States continues to grow, disparities in health outcomes for this population continue to widen. Worse health outcomes can be attributed to a variety of barriers to healthcare specific to this population, but scant literature exists that presents the connections among them. Many of these barriers to healthcare stem from institutional racism, disparities in socioeconomic status, and xenophobia. To synthesize barriers that hinder the Hispanic/Latinx population from receiving quality healthcare, we propose a cyclical model. The model begins with obstacles present before even visiting a healthcare provider, including structural inequalities, cultural beliefs, documentation status, scheduling, and transportation. Next, the model poses barriers during the visit with a healthcare provider, including problems with provider bias, cultural barriers, and interpretation services. Finally, the model presents barriers that occur after the medical encounter, including payment, medication adherence, indigent care, and follow-up appointments or referrals. Once the patient needs to schedule a follow-up appointment or referral, the cycle restarts with the barriers present before accessing care. All of these barriers to healthcare are areas for potential mitigation of the healthcare disparities that currently disadvantage the Hispanic/Latinx population. The proposed Hispanic Cyclical Healthcare Barrier (HCHB) model helps organize solutions to the barriers, illustrating the need for multiple interventions due to the interconnectedness of the barriers.

Exploring the use of mobile translation applications for culturally and linguistically diverse patients during medical imaging examinations in Australia - a systematic review

Australian healthcare provides services to a vast culturally and linguistically diverse (CALD) population. Professional interpreters are the gold standard for medical interpretation during healthcare interactions with CALD patients with limited English proficiency (LEP). However, accessing interpretation services can be difficult and may not be appropriate when timely translation is needed. Mobile translation applications (MTAs) have been suggested as a way for healthcare workers (HCWs) to provide timely translation when engaging with CALD patients. This systematic review aimed to investigate the potential for MTAs to be used in Australian medical imaging (MI) departments to enhance communication and safety for CALD patients and HCWs. Enablers and limitations of MTAs were appraised for use in MI and important design considerations suggested. Results found that MTAs may enhance communication between CALD patients and MI professionals and uphold safety by more accurately performing procedure matching and healthcare assessments. MTAs also offer readily available translation during out of hours care, emergency scenarios and everyday care. However, reliability of free-input translation and patient confidentiality were flagged as important limitations of MTAs that need to be addressed should a safe MTA be designed for MI purposes. MTAs also need to be designed with consideration for CALD patients who have low literacy levels and mental impairment. Devices should be installed close to the point of care to enable MI professionals to easily retrieve and use the MTA. MTAs used in this way can potentially improve care of CALD patients in MI when professional interpreters are absent.

Spanish Translation and Cultural Adaptations of Physical Therapy Parent Educational Materials for Use in Neonatal Intensive Care

Background

A paucity of Spanish language, culturally relevant parent education materials in the healthcare setting results in suboptimal care for Latinx families and further perpetuates health disparities. The purpose of this article is to describe the process for Spanish translation and cultural adaptations to parent education materials of a parent-centered physical therapy program designed to support maternal mental health and infant development during Neonatal Intensive Care (NICU).

Methods

Two bilingual physical therapy (PT) students translated educational materials from English to Spanish and were proofread by a professional translator. Next, we conducted a materials review with 5 members of the Latine Community Review Board (CRB), a "standing" advisory group of natively Spanish-speaking, Latine North Carolinians who contract with research teams under the coordination of the Inclusive Science Program (ISP) of the North Carolina Translational and Clinical Sciences Institute (NC TraCS). Review session recruitment, facilitation, and data analysis were conducted by bilingual NC TraCS project managers and the primary investigator for the main feasibility study. Readability analyses were performed at the final stage of translation and adaptation.

Results

Themes from CRB review sessions for improvement included to 1) use parent-friendly language, 2) use the plural masculine form of gendered language for caregivers to include all gender identities in this neonatal context, 3) address challenges with direct translation, and 4) use written education materials to supplement in-person, hands-on training with parents and their infants. All translated materials received a grade level of 5 on the Crawford grade-level index.

Conclusion

Based on CRB feedback and readability analysis, the translation and cultural-adaptation process resulted in comprehensible written parent education materials for Spanish-speaking families. Review meetings with the CRB reinforced the need for Spanish materials in the healthcare setting. Further assessment of these materials with Spanish-speaking families in the NICU setting is needed.

Management of Discharge Instructions for Children With Medical Complexity: A Systematic Review

CONTEXT

Children with medical complexity (CMC) are at risk for adverse outcomes after discharge. Difficulties with comprehension of and adherence to discharge instructions contribute to these errors. Comprehensive reviews of patient-, caregiver-, provider-, and system-level characteristics and interventions associated with discharge instruction comprehension and adherence for CMC are lacking.

OBJECTIVE

To systematically review the literature related to factors associated with comprehension of and adherence to discharge instructions for CMC.

DATA SOURCES

PubMed/Medline, Embase, Cochrane Central Register of Controlled Trials, PsycInfo, Cumulative Index to Nursing and Allied Health Literature, Web of Science (database initiation until March 2023), and OAIster (gray literature) were searched.

STUDY SELECTION

Original studies examining caregiver comprehension of and adherence to discharge instructions for CMC (Patient Medical Complexity Algorithm) were evaluated.

DATA EXTRACTION

Two authors independently screened titles/abstracts and reviewed full-text articles. Two authors extracted data related to study characteristics, methodology, subjects, and results.

RESULTS

Fifty-one studies were included. More than half were qualitative or mixed methods studies. Few interventional studies examined objective outcomes. More than half of studies examined instructions for equipment (eg, tracheostomies). Common issues related to access, care coordination, and stress/anxiety. Facilitators included accounting for family context and using health literacy-informed strategies.

LIMITATIONS

No randomized trials met inclusion criteria. Several groups (eg, oncologic diagnoses, NICU patients) were not examined in this review.

CONCLUSIONS

Multiple factors affect comprehension of and adherence to discharge instructions for CMC. Several areas (eg, appointments, feeding tubes) were understudied. Future work should focus on design of interventions to optimize transitions.

Undertriage for Children With Caregivers Preferring Languages Other Than English

OBJECTIVES

Undertriage, the underestimation of acuity, can result in delayed care and potential morbidity in the emergency department (ED). Although inequities in ED care based on language preference have been noted, little is known about its association with undertriage. We evaluated for differences in undertriage based on caregiver language preference.

METHODS

This was a retrospective cross-sectional study of patients aged younger than 21 years, triaged as Emergency Severity Index (ESI) level 4 or 5 (nonurgent), to the pediatric ED from January 1, 2019, through January 31, 2021. Indicators of undertriage were defined as hospital admission, significant ED resource use, or return visits with admission. We used logistic regression with generalized estimating equations to measure the association of preferred language with undertriage.

RESULTS

Of 114 266 ED visits included in the study, 22 525 (19.8%) represented patients with caregivers preferring languages other than English. These children were more likely to experience undertriage compared with those with caregivers preferring English (3.7% [English] versus 4.6% [Spanish] versus 5.9% [other languages]; Spanish versus English: odds ratio [OR], 1.3; 95% confidence interval [CI], 1.2-1.4] and other languages versus English: OR, 1.6; 95% CI, 1.2-2.2). Differences remained after adjusting for sex, insurance, mode of arrival, and clustering by triage nurse (Spanish versus English: adjusted OR, 1.3; 95% CI, 1.3-1.5) and other languages versus English: adjusted OR, 1.6; 95% CI, 1.2-2.2).

CONCLUSIONS

Children accompanied by caregivers preferring languages other than English are more likely to be undertriaged in the pediatric ED. Efforts to improve the triage process are needed to promote equitable care for this population.

The Impact of Non-English Language Preference on Pediatric Hospital Outcomes

OBJECTIVES

To characterize the relationship between non-English language preference (NELP) and hospital outcomes including length of stay (LOS), time of discharge, emergency department return visits, readmissions, and cost for pediatric general medicine inpatients.

METHODS

We conducted a retrospective analysis at an urban, quaternary care, free-standing children's hospital. Patients ages 0 to 18 admitted to any general medicine service between January 1, 2017, and December 31, 2019 were included. Patients were divided into 3 language preference categories: English, Spanish, and non-Spanish NELP. Single and multifactor regression analysis was used to model differences in outcome measures by language preference adjusted for technology dependence.

RESULTS

A total of 4820 patients met criteria. In adjusted models, the average LOS for English-speaking patients was 126 hours; LOS for patients who preferred Spanish was not significantly different, whereas LOS for patients with non-Spanish NELP was 50% longer (P < .001). English-speaking patients were discharged earliest in the day (mean 3:08 pm), with patients who preferred Spanish discharged 0.5 hours later and patients with non-Spanish NELP discharged 1.1 hours later than English-speaking patients (P < .001). Patients with NELP were found to be technology-dependent more frequently (P < .001) than the English preference group. Emergency department return visits, readmissions, and cost were not significantly different between groups.

CONCLUSIONS

NELP was associated with longer length of stay and discharges later in the day. The most pronounced differences occurred in patients with non-Spanish NELP who also had more frequent technology dependence and more limited access to interpreters.

Improving Patient Experience Scores Using Simultaneous Interpretation on Family-Centered Rounds

OBJECTIVES

Patients speaking a primary language other than English face barriers to equitable care, particularly patient-provider communications. There is no gold standard for providing inpatient medical interpretation on family-centered rounds (FCR). We aimed to implement simultaneous, in-person interpretation of FCR for Spanish-speaking families and hypothesized improved satisfaction in care.

METHODS

In-person, Spanish Equipment-Assisted Simultaneous Medical Interpretation (EASMI) was implemented in March 2018 on FCR. Child Hospital Consumer Assessment of Healthcare Providers and Systems (Child HCAHPS) experience scores on communication domains were analyzed for Spanish and English-speaking families pre- (n = 118) and postimplementation (n = 552). Postimplementation, we conducted medical team surveys (n = 104) and semistructured interviews with Spanish-speaking families (n = 25) to determine satisfaction with interpretation modalities (phone, video, and EASMI).

RESULTS

Spanish-speaking families exhibited statistically significant improvements in Child HCAHPS top box scores compared to English-speaking families in multiple communication and informed care-related domains. For example, "How often did your child's doctors explain things to you in a way that was easy to understand?" top box scores improved from 58% to 95% for Spanish-speaking families, compared to 85% to 83% for English speakers, with the differential effect of the intervention showing statistical significance (P = .001). Medical team surveys demonstrated high satisfaction with EASMI. Qualitative themes from interviews and open-ended survey responses emphasized multiple care benefits with EASMI, including a perceived reduction of communication errors and increased family participation.

CONCLUSIONS

EASMI was associated with significant improvements in Child HCAHPS scores in communication domains and increased medical team and family members' satisfaction with interpretation. EASMI presents a novel method for equitable FCR for Spanish-speaking families.

Hospital Discharge Instructions: Characteristics, Accessibility, and National Guideline Adherence

OBJECTIVES

The content of pediatric hospital discharge instructions is highly variable. This study aimed to describe the characteristics, accessibility per patient literacy level and language, and national guideline adherence of pediatric hospital discharge instructions.

METHODS

This retrospective study assessed discharge instructions at a tertiary children's hospital. Patient and instruction characteristics, including patient health literacy level, grade level of instructions, patient language preference, and language of instructions were collected via chart review and electronic medical record query. Standard admission processes assessed health literacy via Brief Health Literacy Screening. The association between demographic and clinical characteristics and adherence to Solutions for Patient Safety guidelines was analyzed by using unadjusted and adjusted analysis methods.

RESULTS

Of 240 discharge instructions, 25% were missing at least 1 recommended content area: signs of worsening, where to seek help, or medication reconciliation. A patient health literacy deficit was identified in 15%; the mean grade level of instructions was 10.1. Limited English proficiency was reported among 17% one quarter of whom received language concordant instructions. Use of discharge instruction templates and discharge services were associated with improved guideline adherence (P <.001). Almost one-half of the study population had a complex medical history, which was associated with decreased guideline adherence (P = .04).

CONCLUSIONS

One-quarter of discharge instructions for this predominantly medically complex population failed to meet national standards. Accessibility was often limited by the reading grade level or discordant language of instructions. Templates may be a valuable tool for improving discharge instruction content, accessibility, and adherence to national guidelines.

Barriers to Follow Up Care in the South Asian Immigrant Population at High Risk of Acute Coronary Syndrome at Elmhurst Hospital Center

We studied South Asian immigrant patients who did not return to Elmhurst Hospital Center (EHC) after emergent cardiac catheterization in order to propose interventions to improve follow up care. We identified 74 eligible patients, interviewed 30 about follow up practices, and analyzed findings. Most patients are Bangladeshi and 77% preferred a foreign language. Some were visiting the US during the admission without intent to follow up. Half were dissatisfied with EHC providers, complications, and inadequate care at follow up appointments. Some patients were unaware of scheduled appointments or the necessity of follow up. Most follow with private providers due to language accessibility, availability, and proximity. We found that language barriers contribute to loss to follow up and the true loss to follow up rate is lower than reported at EHC. This can inform practices at hospitals with immigrant populations, minimize resource waste, and improve quality of care.

Breaking through barriers: the need for effective research to promote language-concordant communication as a facilitator of equitable emergency care

Individuals with limited English proficiency (LEP) are at high risk for adverse outcomes in the US health care system. This is particularly true for patients with LEP seeking care in the emergency department (ED). Although professional language interpretation improves the quality of care for these patients, it remains underused. The dynamic, discontinuous nature of an ED visit poses distinct challenges and opportunities for providing equitable, high-quality care for patients with LEP. Evidence-based best practices for identifying patients with LEP and using professional interpretation are well described but inadequately implemented. There are few examples in the literature of rigorous interventions to improve quality of care and outcomes for patients with LEP. There is an urgent need for high-quality research to improve communication with patients with LEP along the continuum of emergency care in order to achieve equity in outcomes.

Improving Discharge Instructions for Hospitalized Children With Limited English Proficiency

BACKGROUND AND OBJECTIVES

Patients with limited English proficiency (LEP) have increased risk of adverse events after hospitalization. At our institution, LEP families did not routinely receive translated discharge instructions in their preferred language. Our objective for this study was to increase the percentage of patients with LEP on the hospital medicine (HM) service receiving translated discharge instructions from 12% to 80%.

METHODS

Following the Model for Improvement, we convened an interdisciplinary team that included HM providers, pediatric residents, language access services staff, and nurses to design and test interventions aimed at key drivers through multiple plan-do-study-act cycles. Interventions addressed the translation request process, care team education, standardizing discharge instructions for common conditions, and identification and mitigation of failures. We used established rules for analyzing statistical process control charts to evaluate the percentage of patients with translated discharge instructions for all languages and for Spanish.

RESULTS

During the study period, 540 patients with LEP were discharged from the HM service. Spanish was the preferred language for 66% of patients with LEP. The percentage of patients with LEP who received translated discharge instructions increased from 12% to 50% in 3 months and to 77% in 18 months. For patients whose preferred language was Spanish, the percentage increased from 16% to 69% in 4 months and to 96% in 18 months.

CONCLUSIONS

Interventions targeting knowledge of the translation process and standardized Spanish discharge instructions were associated with an increased percentage of families receiving translated discharge instructions. Future work will be used to assess the impact of these interventions on postdischarge disparities, including emergency department revisits and readmissions.