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.
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.
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).
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.
Children and families preferring languages other than English experience inequity in the emergency department. Undertriage, or the concept of underestimating a patient’s acuity, may have deleterious consequences. No previous studies have evaluated the association of language preference with undertriage.
Pediatric patients with caregivers preferring languages other than English are more likely to be undertriaged. Dedicated interventions to understand and mitigate inequity in the triage process are needed.
Each year, more than 30 million children are seen in emergency departments (EDs) in the United States.1 Triage, the initial assessment to determine sorting of patients by acuity and severity, is a critical first step in the ED encounter. The most widely used triage algorithm in the United States for both adult and pediatric patients is the Emergency Severity Index (ESI).2–5 The ESI functions by asking 2 fundamental, sequential questions: first, “Who should be seen first?” and second, “How can we optimize moving patients to their final dispositions based on anticipated resource use?”2,3 This model seeks to prioritize care for ill patients while streamlining the use of ED resources. Importantly, it has been validated for use in pediatric populations.6,7
Undertriage, or underestimating a patient’s acuity and/or resource requirements, can have important consequences. At best, it can result in inefficient deployment of ED resources; at worst, it can lead to morbidity and mortality because of delays in necessary care.8–12 The bulk of research regarding undertriage has been in relation to trauma care, primarily in adults, although there are some studies in pediatrics.10–12
Though we know disparities exist for pediatric patients and families preferring languages other than English (LOE) in the ED, little research has been conducted to evaluate if differences exist for this population in relation to triage. Previous work has shown that children preferring LOE in the ED experience longer lengths of stay,13 increased likelihood of revisits,14,15 underuse of interpretation services,16,17 and differing care,18–20 both in terms of diagnostics and treatment, compared with their counterparts preferring English. But what if some of these disparities in care can be traced back to initial disparities in triage?
Evaluating for disparities in undertriage is important because ESI level serves as an indicator to providers regarding a patient’s acuity and potential resource use, potentially leading to anchoring bias and impacting downstream care. We performed this study to determine if there are differences in undertriage for pediatric patients based on caregiver preferred language.
Methods
Study Design, Setting, and Population
This was a retrospective cross-sectional study of visits for patients aged younger than 21 years, presenting from January 1, 2019, to January 31, 2021, at our 2 affiliated pediatric EDs: an academic urban pediatric ED with an annual census of 90 000 patient visits and a satellite urban free-standing pediatric ED with an additional 36 000 annual patient visits. Patients with a documented caregiver preferred language in the electronic health record (EHR) were included in the study cohort. The institutional review board at our hospital approved this study.
Defining Undertriage
Our study team collaboratively developed a definition for undertriage, taking into consideration the ESI algorithm used in our EDs, as well as institutional practices.21 The ESI algorithm consists of 5 categories for patients, ranging from 1 to 5.3 Patients with ESI 1 require immediate life-saving intervention. Patients with ESI 2 are acutely or severely ill. If patients do not meet criteria for ESI 1 or 2, the algorithm asks how many different resources will be needed for the patient. If none, the patients can be triaged as ESI 5. If 1 resource will be needed, they can be triaged as ESI 4. If many resources will be needed and vital signs are not grossly abnormal, then these patients can be categorized as ESI 3. The ESI framework also provides details regarding what constitutes resources, which include laboratories, imaging, intravenous fluids, and nonoral medications.
Importantly, at our main hospital, patients triaged as a ESI 4 or 5 are routinely cared for in a separate area of the ED. This section is uniquely designed for patients who have minor illnesses and does not have physical and/or staffing resources to provide nebulization treatments, oxygen, or intravenous fluids/medications. Accordingly, we defined undertriage as encounters in which the patient was assigned an ESI score of 4 or 5 and required hospital admission or significant ED resources, including nebulized treatment, supplemental oxygen, or intravenous placement. Because undertriage may not be immediately apparent at the initial visit, an additional undertriage indicator was defined as patients with an ESI score of 4 or 5 who returned to the ED within 14 days and required admission.
Triage is a 2-step process at our hospitals. First, an initial quick sort is performed, lasting less than 2 minutes, and includes consideration of the reason for visit, rapid assessment of airway, breathing, circulation, and mental status, and any pertinent medical history such as oncologic disease or hemoglobinopathy. Initial ESI is assigned at this step. A second nurse then performs a full nursing assessment, including a more complete history and vital signs. The initial ESI can be modified at this step. This is considered the final ESI, and these data were used to define undertriage.
Caregiver Preferred Language Data
The first question that triage nurses ask of caregivers is their preferred language, which is then recorded in the EHR. Though pediatric patients, particularly teenagers, may feel comfortable sharing their medical information in English and/or providing interpretation for their caregivers, our standard institutional practice is to communicate directly with caregivers to obtain triage information, with the use of professional interpretation as needed.
Data Collection and Analysis
Patient demographic data were obtained from the EHR and included sex, age, race, ethnicity, preferred language, insurance type, and mode of arrival. Race and ethnicity were recategorized into a single variable: non-Hispanic white, non-Hispanic Black, Hispanic, and Other. Although standard of care is for the registration staff to document race and ethnicity by patient/caregiver self-report, we are unable to confirm this in practice by all registration staff. Preferred language was defined as the parent/caregiver’s stated primary language preference, which is collected during the initial triage step, as described previously. Language was categorized as either English, Spanish, or Other. Data pertaining to undertriage and triage nurses were also obtained from the EHR.
The 2 study sites share a linked EHR. Thus, when patients require transfer from the community hospital to the academic center, the same patient encounter is used, with a single, initial ESI. We also measured the proportion of patients who had return visits to either site, in addition to the proportion of unique triage nurses who were involved in these visits.
The outcome of interest was undertriage, and the primary independent variable of interest was caregiver preferred language. We used SAS software (SAS Institute Inc., Cary, North Carolina) to perform descriptive statistics and develop multivariable models, adjusting for sex, insurance type, and mode of arrival. Generalized estimating equations with an exchangeable correlation structure were used to account for clustering by triage nurse; nurses with 20 or fewer triage encounters were excluded. We performed secondary analyses to determine the association of preferred language with each subtype of undertriage (ie, hospital admission, significant ED resource use, or return visit with admission).
Results
Between January 1, 2019, and January 31, 2021, 124 775 patient encounters were triaged as an ESI of 4 or 5. Of these, 114 266 (91.6%) had a preferred language documented and were included in the analysis. Among these visits, 91 741 (80.2%) had caregivers who preferred English, 21 660 (19.0%) had caregivers who preferred Spanish and 865 (0.8%) had caregivers who preferred other languages (Fig 1, Table 1). There were 47 languages other than English or Spanish represented; the 5 most common were Amharic, French, Arabic, American Sign Language, and Mandarin. The median age of patients was 5 (interquartile range, 2, 10) years (Table 1). Teenagers, defined as those aged 13 to 21 years, comprised 17.2% of the sample (19 606 patients). Full adjusted multivariable models for overall undertriage, as well as each undertriage indicator, are provided in Supplemental Tables 4 and 5.
Characteristics of the Study Population
Demographic Characteristic . | Total N = 114 266, n (%) . | Undertriaged n = 4411, n (%) . | Not Undertriaged n = 109 855, n (%) . |
---|---|---|---|
Sex | |||
Female | 55 473 (48.5) | 2112 (47.9) | 53 361 (48.6) |
Male | 58 793 (51.5) | 2299 (52.1) | 56 494 (51.4) |
Age, y, median (IQR) | 5 (2–10) | 5 (2–12) | 5 (2–10) |
Race | |||
Non-Hispanic Black | 73 058 (63.9) | 2529 (57.3) | 70 529 (64.2) |
Non-Hispanic white | 4218 (3.7) | 280 (6.3) | 3938 (3.6) |
Hispanic | 27 540 (24.1) | 1220 (27.7) | 26 320 (23.9) |
Other | 7219 (6.3) | 327 (7.4) | 6892 (6.3) |
Not documented | 2231 (2.0) | 55 (1.3) | 2176 (2.0) |
Language | |||
English | 91 741 (80.2) | 3369 (76.4) | 88 372 (80.4) |
Spanish | 21 660 (19.0) | 991 (22.4) | 20 669 (18.8) |
Other | 865 (0.8) | 51 (1.2) | 814 (0.8) |
Insurance | |||
Private | 14 650 (12.8) | 807 (18.3) | 13 843 (12.6) |
Public | 88 476 (77.4) | 3291 (74.6) | 85 185 (77.5) |
Self | 10 933 (9.6) | 303 (6.9) | 10 630 (9.7) |
Not documented | 207 (0.2) | 10 (0.2) | 197 (0.2) |
Mode of arrival | |||
Ambulance, transport, police | 5967 (5.2) | 473 (10.7) | 5494 (5.0) |
Personal | 108 299 (94.8) | 3938 (89.3) | 104 361 (95.0) |
Demographic Characteristic . | Total N = 114 266, n (%) . | Undertriaged n = 4411, n (%) . | Not Undertriaged n = 109 855, n (%) . |
---|---|---|---|
Sex | |||
Female | 55 473 (48.5) | 2112 (47.9) | 53 361 (48.6) |
Male | 58 793 (51.5) | 2299 (52.1) | 56 494 (51.4) |
Age, y, median (IQR) | 5 (2–10) | 5 (2–12) | 5 (2–10) |
Race | |||
Non-Hispanic Black | 73 058 (63.9) | 2529 (57.3) | 70 529 (64.2) |
Non-Hispanic white | 4218 (3.7) | 280 (6.3) | 3938 (3.6) |
Hispanic | 27 540 (24.1) | 1220 (27.7) | 26 320 (23.9) |
Other | 7219 (6.3) | 327 (7.4) | 6892 (6.3) |
Not documented | 2231 (2.0) | 55 (1.3) | 2176 (2.0) |
Language | |||
English | 91 741 (80.2) | 3369 (76.4) | 88 372 (80.4) |
Spanish | 21 660 (19.0) | 991 (22.4) | 20 669 (18.8) |
Other | 865 (0.8) | 51 (1.2) | 814 (0.8) |
Insurance | |||
Private | 14 650 (12.8) | 807 (18.3) | 13 843 (12.6) |
Public | 88 476 (77.4) | 3291 (74.6) | 85 185 (77.5) |
Self | 10 933 (9.6) | 303 (6.9) | 10 630 (9.7) |
Not documented | 207 (0.2) | 10 (0.2) | 197 (0.2) |
Mode of arrival | |||
Ambulance, transport, police | 5967 (5.2) | 473 (10.7) | 5494 (5.0) |
Personal | 108 299 (94.8) | 3938 (89.3) | 104 361 (95.0) |
There were 72 755 unique patients and 23 112 (31.8%) had additional visits during the study time frame. There were 275 unique nurses who assigned triage, of whom 245 (89%) had 20 or more triage encounters. Among these nurses, rates of undertriage ranged from 0% to 22.7%. Of the patients who were undertriaged and had a repeat visit, 1386 (6.4%) were triaged by the same nurse.
We also examined the most common diagnosis categories for undertriaged visits, including return visits within 14 days. The most common diagnoses were skin and soft tissue pathology, other gastrointestinal pathology, unspecified viral infections, upper respiratory pathology, and asthma (Table 2).
Most Common Diagnosis Categories of Undertriaged Patients, n (%)
Skin and soft tissue pathology | 613 (11.2) |
Other gastrointestinal pathology | 563 (10.2) |
Unspecified viral infection | 402 (7.3) |
Upper respiratory pathology | 383 (7.0) |
Asthma | 367 (6.7) |
Neurologic pathology | 360 (6.5) |
Other infectious pathology | 306 (5.6) |
Other diagnoses | 292 (5.3) |
Pneumonia, bronchiolitis | 272 (5.0) |
Dehydration, malnutrition | 237 (4.3) |
Skin and soft tissue pathology | 613 (11.2) |
Other gastrointestinal pathology | 563 (10.2) |
Unspecified viral infection | 402 (7.3) |
Upper respiratory pathology | 383 (7.0) |
Asthma | 367 (6.7) |
Neurologic pathology | 360 (6.5) |
Other infectious pathology | 306 (5.6) |
Other diagnoses | 292 (5.3) |
Pneumonia, bronchiolitis | 272 (5.0) |
Dehydration, malnutrition | 237 (4.3) |
A total of 4411 of 114 266 patient visits (3.9%) were undertriaged. In comparison with those with caregivers preferring English, patients with caregivers preferring Spanish or other languages were more likely to be undertriaged (3369 of 91 741, 3.7% [English] versus 991 of 21 660, 4.6% [Spanish] versus 51 of 865, 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) (Table 3). These differences remained after multivariable adjustment (Spanish versus English: adjusted odds ratio [aOR], 1.3; 95% CI, 1.3–1.5, and other languages versus English: aOR, 1.6; 95% CI, 1.2–2.2) (Table 3).
Frequency and Association of Undertriage Indicators, by Caregiver Preferred Language
Language . | n (%) . | OR (95% CI) . | aORa (95% CI) . |
---|---|---|---|
Overall undertriaged | |||
English | 3369/91 741 (3.7) | Reference | Reference |
Spanish | 991/21 660 (4.6) | 1.3 (1.2–1.4) | 1.3 (1.3–1.5) |
Other | 51/865 (5.9) | 1.6 (1.2–2.2) | 1.6 (1.2–2.2) |
Undertriage indicator: ESI 4 or 5+ hospital admission | |||
English | 781/91 741 (0.9) | Reference | Reference |
Spanish | 327/21 660 (1.5) | 1.8 (1.6–2.0) | 2.0 (1.7–2.3) |
Other | 16/865 (1.9) | 2.2 (1.3–3.6) | 2.2 (1.3–3.7) |
Undertriage indicator: ESI 4 or 5+ ED resourcesb | |||
English | 2975/91 741 (3.2) | Reference | Reference |
Spanish | 866/21 660 (4.0) | 1.2 (1.2–1.3) | 1.4 (1.2–1.5) |
Other | 46/865 (5.3) | 1.7 (1.2–2.3) | 1.7 (1.2–2.3) |
Undertriage indicator: ESI 4 or 5+ return within 14 d and admission | |||
English | 428/90 960 (0.5) | Reference | Reference |
Spanish | 136/21 333 (0.6) | 1.4 (1.1–1.7) | 1.4 (1.2–1.7) |
Other | 5/849 (0.6) | 1.3 (0.5–3.0) | 1.3 (0.5–3.0) |
Language . | n (%) . | OR (95% CI) . | aORa (95% CI) . |
---|---|---|---|
Overall undertriaged | |||
English | 3369/91 741 (3.7) | Reference | Reference |
Spanish | 991/21 660 (4.6) | 1.3 (1.2–1.4) | 1.3 (1.3–1.5) |
Other | 51/865 (5.9) | 1.6 (1.2–2.2) | 1.6 (1.2–2.2) |
Undertriage indicator: ESI 4 or 5+ hospital admission | |||
English | 781/91 741 (0.9) | Reference | Reference |
Spanish | 327/21 660 (1.5) | 1.8 (1.6–2.0) | 2.0 (1.7–2.3) |
Other | 16/865 (1.9) | 2.2 (1.3–3.6) | 2.2 (1.3–3.7) |
Undertriage indicator: ESI 4 or 5+ ED resourcesb | |||
English | 2975/91 741 (3.2) | Reference | Reference |
Spanish | 866/21 660 (4.0) | 1.2 (1.2–1.3) | 1.4 (1.2–1.5) |
Other | 46/865 (5.3) | 1.7 (1.2–2.3) | 1.7 (1.2–2.3) |
Undertriage indicator: ESI 4 or 5+ return within 14 d and admission | |||
English | 428/90 960 (0.5) | Reference | Reference |
Spanish | 136/21 333 (0.6) | 1.4 (1.1–1.7) | 1.4 (1.2–1.7) |
Other | 5/849 (0.6) | 1.3 (0.5–3.0) | 1.3 (0.5–3.0) |
Adjusted for sex, insurance, and mode of arrival, with clustering by individual triage nurse.
ED resources include supplemental oxygen, intravenous fluids/medications, or nebulization treatments.
Secondary analyses focused on specific undertriage indicators and demonstrated similar disparities in undertriage based on preferred language. A total of 1124 (1.0%) patient encounters met criteria for undertriage via being triaged as an ESI 4 or 5 and requiring hospital admission. Patients with caregivers preferring Spanish or other languages had higher likelihood of experiencing this type of undertriage, compared with those with caregivers preferring English (aOR, 2.0; 95% CI, 1.7–2.3, and aOR, 2.2; 95% CI, 1.3–3.7, respectively) (Table 3).
A total of 3877 patient visits (3.4%) with an ESI of 4 or 5 required significant ED resources, including nebulized treatment, supplemental oxygen, or intravenous placement. Again, patients with caregivers preferring either Spanish or other languages had a higher likelihood of experiencing this type of undertriage compared with those with caregivers preferring English (aOR, 1.4; 95% CI, 1.2–1.5, and aOR, 1.7; 95% CI, 1.2–2.3, respectively) (Table 3).
A final category of undertriage evaluated for patients with an ESI 4 or 5 who had a return visit to the ED within 14 days and required hospital admission; 569 (0.5%) patients met these criteria. Compared with those with caregivers preferring English, patients with Spanish-preferring caregivers had increased likelihood of this type of undertriage (aOR, 1.4; 95% CI, 1.2–1.7). There was no significant difference for this undertriage indicator for those preferring other languages (Table 3).
Discussion
In this large study at an academic children’s hospital ED, we found that patients accompanied by caregivers preferring LOE are more likely to experience undertriage compared with patients accompanied by caregivers who prefer English. Specifically, children with caregivers who prefer languages other than English are more likely to be assigned a triage acuity that does not align with the care they will ultimately need, which includes hospital admission, significant ED resource use (nebulized medication, intravenous placement, or supplemental oxygen), or return to the ED with admission within 14 days. Although one could define undertriage by ESI criteria in terms of more than 1 resource use, we chose to use a more conservative definition: the requirement of significant additional medical care. Our study highlights this disparity for patients with Spanish-speaking caregivers as well as for patients with caregivers who prefer languages other than English or Spanish.
To our knowledge, this is the first study to explore the impact of language preference on undertriage, and our results have significant implications for the triage process in EDs. A key initial step is to engage with caregivers preferring LOE to better understand the triage experience from their perspective. Previous qualitative work in this space has focused primarily on the inpatient population and has yielded important findings, patterns, and areas for improvement.22–24
There are likely multiple factors underlying our findings. Importantly, we noted large variability in individual nursing undertriage rates. This suggests that there may be differences in how individual nurses triage patients based on caregiver preferred language. One related factor could be the underuse or misuse of interpretation by those in triage. Importantly, federal law mandates the use of interpretation when needed in various settings, including health care.25–27 Despite this law, interpreter services are underused in many areas of clinical care, including the ED.16,17,28–30 It is possible that interpreters are underused or inappropriately used in triage, especially when considering the environment, which is fast-paced and relies on brief interactions to quickly determine acuity. Examples could include not using interpreters, using nonqualified clinicians as interpreters, or using family members as interpreters. We suggest that a lack of appropriate interpreter use could serve as an important driving force behind undertriage.
An additional driving force may be the presence of bias among health care providers. Several previous studies have demonstrated clinician bias against Hispanic patients, a finding that may contribute to the undertriage of patients with Spanish-preferring caregivers in our study.31–34 Individual clinician biases, whether explicit or implicit, could lead to the inaccurate triage of patients and families who prefer languages other than English.
Our study also contributes to a growing body of literature focused on understanding disparities in hospital patient safety based on patient and family preferred language. For example, patients preferring LOE are more likely to experience adverse events while hospitalized and often receive discharge instructions not in their preferred language.35–38 Undertriage in and of itself is a safety issue because it indicates that patients are being assigned ESI levels that are not reflective of their true acuity. This has the potential to lead to significant additional downstream safety concerns because the underestimation of a patient’s severity of illness could lead to inappropriate or delayed care.
Our study has several limitations. First, it was conducted at 2 EDs within a single city, and thus results may not be generalizable. Second, we were unable to measure health care use of patients outside of our study sites. Additionally, we acknowledge that defining language preference in pediatrics is difficult because caregivers and patients, in particular, older patients, may have different preferences and levels of fluency. It is our institutional practice that triage information is obtained from caregivers, in the caregiver’s preferred language, with professional interpretation as needed, though it is possible that older pediatric patients with English fluency may have discussed their medical information directly with triage providers. Future research endeavors should aim to elucidate and evaluate potential impacts of language preference differences between older pediatric patients and their caregivers. Finally, we could not measure use of interpretation during triage because use of interpreter services is not linked to patient medical record numbers at our institution.
Future work should prioritize a multicenter evaluation of undertriage to determine its scope, as well as the use of quality improvement methodologies to mitigate inequity, with a focus on triage practices. Along these lines, EDs should allocate resources and funding to ensure that interpretation is always easily available in triage, either via in-person interpreters or remote options. Key to this effort is the development of mechanisms to track and quantify interpreter use. To address potential clinician implicit bias, we also advocate for transparency regarding undertriage events via direct and solution-oriented feedback. Finally, although there has been enormous progress in the realm of technologic interpretation modalities in the past couple of decades, we contend that there is a need for additional creative and accessible solutions. One example could include institutional incorporation of a mobile interpretation application that families can use on their personal devices. This solution would also have the added benefit of creating a system in which families themselves can initiate interpretation, an important shift in a communication dynamic that is often in the hands of the provider.
Conclusions
Pediatric patients with caregivers who prefer LOE are more likely to be undertriaged in the pediatric ED. These novel results provide important insights into the systemic inequities that exist for this population in health care. This study also has important implications for future disparities studies using triage acuity to control for severity because differences in triage acuity itself may be associated with language preference. Subsequent work will focus on evaluating for differences in downstream effects of undertriage, including adverse events, as well as engaging in quality improvement efforts to mitigate disparities in undertriage based on preferred language.
Drs Berkowitz, Chamberlain, Cohen, and Goyal developed the initial study concept, supervised data collection, and contributed to manuscript revision; Dr Rojas contributed to the study concept, assisted with data collection and data analysis, and drafted the initial manuscript; Ms Badolato and Dr Cook completed data analyses and contributed to manuscript revision; and all authors approved the final manuscript as submitted.
FUNDING: No external funding.
CONFLICT OF INTEREST DISCLOSURES: The authors have no conflicts of interest to disclose.
Comments