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.
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.
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.
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.
Non-English language preference (NELP), also referred to as limited English proficiency, is common in the United States: more than 25 million Americans speak English less than “very well.”1 Navigating the health care system with NELP poses a significant challenge to patients and their families. NELP is linked to a higher cost of care in the emergency department (ED),2 decreased understanding of care,3 and increased adverse events among hospitalized pediatric patients.4 NELP has also been associated with increased length of stay (LOS) in the ED,5 increased likelihood of admission,6 and more unscheduled return visits.7 Among children hospitalized with bacterial infections requiring prolonged intravenous antibiotics, NELP patients had an increased LOS and fewer home health referrals.8
Although there is a known relationship between NELP and worse outcomes in the ED and for specific diagnoses, more work needs to be done to understand the impact of NELP on pediatric hospital outcomes, particularly among the groups who prefer a language other than Spanish. One large study demonstrated that there are not increased readmissions in the NELP group; however, no distinction was made between different non-English language groups.9 Because of marked variation in various language-related resources within the health care system, it is worthwhile to analyze the Spanish speaking and non-Spanish speaking NELP cohorts separately.10,11 Better understanding the differences in outcomes of patients with NELP can be used to inform policy decisions and guide hospitals on potential interventions to improve care for these patients and their families.
The authors of this study aimed to explore the relationship between language preference and outcomes including hospital LOS, time of discharge (between 00:00 and 23:59), ED return visits, readmissions, and cost for general pediatric inpatients. An exploratory aim was to describe the interaction of technology dependence (as a proxy for medical complexity) with these outcomes in NELP patients, because children who are technologically dependent consistently have higher health care costs and increased risk of readmission.12,13
Methods
Study Design
We conducted a retrospective analysis at an urban, quaternary care, free-standing children’s hospital. The institutional review board approved the study. Patients ages 0 to 18 admitted to any of the general medicine services between January 1, 2017, and December 31, 2019, were included in the study. Exclusion criteria were patients >18 years of age, patients initially admitted to a surgical service, patients who did not spend at least 1 midnight in the hospital, and patients who died during hospitalization. We identified patients from the electronic health record (EHR) on the basis of admission service, discharge service, and date of admission. For patients with multiple visits during the study period, we selected the first visit within our inclusion dates as the “index” visit, and all data collected was associated with that index visit.
Our children’s hospital has in-person interpretation available for Spanish-speakers during daytime hours on both weekdays and weekends, and for Arabic-speakers during daytime hours on weekdays. In-person interpretation for other languages is available with advance notice. On-demand phone and video interpretation are available at all hours for > 200 languages. Document translation services are available for Spanish during business hours.14
Demographic Measures
Demographic information included age, race, ethnicity, insurance status, and language preference, which was obtained by self-report questionnaire at hospital registration and extracted via EHR query.
Language Preference
Patients were grouped into 3 language preference categories: English language preference (English speaking), Spanish-speaking NELP (Spanish speaking), or non–Spanish-speaking NELP. A patient’s preferred language was defined as the language that their parent or guardian requested as their preferred language during registration, or the language used by an interpreter with the parent or guardian during the patient’s hospitalization. Patients were defined as having NELP if they preferred any language other than English, and as Spanish-speaking if Spanish was their preferred language. Non–Spanish-speaking NELP patients were defined as patients with a preferred language other than English or Spanish. Patient families that did not indicate an alternative preferred language or use an interpreter during hospitalization were categorized as English-speaking. Manual chart review was used in cases in cases in which no preferred language was listed, or if there was discrepancy between a patient’s documented language preference between admissions.
Technology Dependence
We defined technology-dependent patients as those requiring a device to overcome a severe limitation in physiologic functioning (eg, tracheostomy, feeding tube, indwelling central line), and technology dependence was identified using the diagnosis list in the EHR and International Classification of Diseases, 10th Revision codes Z93.0, Z93. 1, Z95.828, and Z99. 89.13
Outcome Measures
LOS was defined as the time (in hours) between the hospital admission order and subsequent hospital discharge order. Time of discharge was calculated as the time of day (in hours) that the discharge order was entered on the day of discharge. Return visits to the ED or all-cause inpatient readmissions within 30 days of discharge were recorded and dichotomized as either present or absent. Hospital cost of admission was defined as the EHR-generated charges at the time of discharge, before any insurance adjustments, and was log transformed for analyses.
Statistical Analysis
Descriptive statistics were calculated for the sample as a whole and included frequencies and percentages for categorical data and medians and ranges and for continuous variables. Differences in outcome measures by language preference were modeled using regression models on the basis of the distribution of the study data in which the outcomes were regressed, first on language and then adjusted for covariates including technology dependence. LOS analyses were modeled by using a γ distribution, and total encounters were modeled by using a Poisson distribution. Binary outcomes of ED return visits and hospital readmissions were modeled using logistic regression models. Time of discharge general linear regression models assumed a normal distribution. Because of the distribution of cost data, log transformation was used to normalize the data before linear regression. Differences in technology dependence by language preference were calculated and tested by using a χ2 test.
Model fit was assessed to ensure appropriateness of the sample distribution. Results are reported as model-based differences between groups, and are considered statistically significant at P < .05. All analyses were conducted by using SAS version 9.4.
Results
Demographics
A total of 4820 patients met eligibility criteria (Table 1). Of these patients, 648 (13.4%) had NELP, 550 (11.4%) were Spanish-speaking, and 98 (2.0%) had a non–Spanish-speaking NELP. Technology dependence was present in 401 (8.3%) patients, with a higher rate of technology dependence among the Spanish-speaking and non–Spanish-speaking NELP groups (7.7%, 12.0%, and 14.3% in English-speaking, Spanish-speaking, non–Spanish-speaking NELP, respectively, P = .0003; Table 2).
Characteristic . | N (%) . |
---|---|
Sex | |
Male | 2624 (55.4) |
Female | 2196 (44.6) |
Race | |
Asian | 234 (4.9) |
Black | 988 (20.5) |
Multiple | 147 (3.1) |
American Indian | 7 (0.2) |
Native Hawaiian or Pacific Islander | 3 (0.06) |
Other | 1582 (33.8) |
Unknown | 161 (3.3) |
White | 1698 (35.2) |
Ethnicity | |
Hispanic | 1619 (33.6) |
Not Hispanic | 3062 (63.5) |
Unknown | 139 (2.9) |
Language | |
English-proficient (EP) | 4172 (86.6) |
Spanish-speaking (SS) LEP | 550 (11.4) |
Non–Spanish-speaking (NSS) LEP | 98 (2.0) |
Insurance | |
Public | 2534 (52.6) |
Private | 1819 (37.7) |
Other | 467 (9.7) |
Characteristic . | N (%) . |
---|---|
Sex | |
Male | 2624 (55.4) |
Female | 2196 (44.6) |
Race | |
Asian | 234 (4.9) |
Black | 988 (20.5) |
Multiple | 147 (3.1) |
American Indian | 7 (0.2) |
Native Hawaiian or Pacific Islander | 3 (0.06) |
Other | 1582 (33.8) |
Unknown | 161 (3.3) |
White | 1698 (35.2) |
Ethnicity | |
Hispanic | 1619 (33.6) |
Not Hispanic | 3062 (63.5) |
Unknown | 139 (2.9) |
Language | |
English-proficient (EP) | 4172 (86.6) |
Spanish-speaking (SS) LEP | 550 (11.4) |
Non–Spanish-speaking (NSS) LEP | 98 (2.0) |
Insurance | |
Public | 2534 (52.6) |
Private | 1819 (37.7) |
Other | 467 (9.7) |
LEP, limited English proficiency.
Language . | Technology Dependence (N = 4419) (%) . | No Technology Dependence (N = 401) (%) . | χ-Squared . |
---|---|---|---|
English preference (N = 4172) | 321 (7.7) | 3851 (92.3) | P < .001 |
Spanish speaking (N = 550) | 66 (12.0) | 484 (88.0) | |
Non-Spanish speaking (N = 98) | 14 (14.3) | 84 (85.7) |
Language . | Technology Dependence (N = 4419) (%) . | No Technology Dependence (N = 401) (%) . | χ-Squared . |
---|---|---|---|
English preference (N = 4172) | 321 (7.7) | 3851 (92.3) | P < .001 |
Spanish speaking (N = 550) | 66 (12.0) | 484 (88.0) | |
Non-Spanish speaking (N = 98) | 14 (14.3) | 84 (85.7) |
Length of Stay
In unadjusted models, both patients who prefer Spanish and those who had a non-Spanish NELP had a longer LOS than English-speaking patients (P < .01 and P < .0001, respectively; Table 3). After adjusting for covariates, the average English-speaking patient LOS was 125.6 hours, with no difference in LOS between English-speaking patients and patients who prefer Spanish in the adjusted model (estimate = 0.99; 95% confidence interval (CI) [0.92–1.06]). However, in the fully adjusted model, the average length of stay for the non-Spanish speaking NELP group was 184.6 hours, which was 59 hours longer than the average LOS for English-speaking patients (odds ratio [OR]: 1.47, P < .0001, 95% CI [1.23–1.72]).
Language (English as reference) . | Unadjusted Effect (95% CI) . | Adjusted Effect (95% CI) . |
---|---|---|
Spanish speaking | 1.11a (1.03–1.20) | 0.99 (0.92–1.06) |
Non-Spanish speaking | 1.68a (1.42–1.99) | 1.47a (1.23–1.72) |
Language (English as reference) . | Unadjusted Effect (95% CI) . | Adjusted Effect (95% CI) . |
---|---|---|
Spanish speaking | 1.11a (1.03–1.20) | 0.99 (0.92–1.06) |
Non-Spanish speaking | 1.68a (1.42–1.99) | 1.47a (1.23–1.72) |
P < .05.
Time of Discharge
Patients who prefer Spanish and patients with a non-Spanish NELP were discharged later in the day compared to English-speaking patients in both unadjusted and adjusted models by using least square means (Table 4). After adjusting for technology dependence, English-speaking patients were discharged earliest, with a mean discharge time of 3:44 pm. Patients who prefer Spanish were discharged 0.5 hours later (P < .001) and patients with a non-Spanish NELP were discharged latest in the day, with an average discharge time 1.1 hours later in the day than English-speaking patients (P < .001).
Comparison . | Unadjusted Estimate (95% CI) in Hours . | Adjusted Estimate (95% CI) in h . |
---|---|---|
Spanish preference versus English preference | 0.51a (0.24–0.78) | 0.50a (0.23–0.77) |
Non-Spanish preference versus English preference | 1.11a (0.49–1.72) | 1.09a (0.47–1.71) |
Comparison . | Unadjusted Estimate (95% CI) in Hours . | Adjusted Estimate (95% CI) in h . |
---|---|---|
Spanish preference versus English preference | 0.51a (0.24–0.78) | 0.50a (0.23–0.77) |
Non-Spanish preference versus English preference | 1.11a (0.49–1.72) | 1.09a (0.47–1.71) |
P < .05.
Emergency Department Return Visits
Overall, 455 (9.4%) patients had at least 1 return visit to the ED within 30 days. There was no difference in the odds of ED return visits by language group (Spanish-speaking versus English-speaking OR = 1.32, 95% CI [0.99–1.74] and non-Spanish NELP versus English-speaking OR = 1.19, 95% CI [0.63–2.26]).
Hospital Readmissions
In total, 225 (4.7%) patients had an inpatient return visit within 30 days of the index visit. There was no significant difference in readmissions between the 3 patient groups. The odds of readmission for Spanish-speaking versus English-speaking was 0.92 (95% CI [0.60–1.41]) and for non-Spanish NELP versus English-speaking was 4.31 (95% CI [0.50–2.73]).
Cost
In the fully adjusted model, there was no statistically significant difference in cost between English-speaking patients ($52 170 95% CI [$49 608–$54 865]) and patients who prefer Spanish ($49 464 95% CI [$45 308–$54 003]) (P = .22). The estimated cost for admission for patients with a non-Spanish NELP was $62 686 (95% CI [$51 750–$75 933]), but it did not reach the threshold for statistical significance (P = .06).
Discussion
In this single-center study at a quaternary children’s hospital, non-English language preference was associated with prolonged LOS and discharge later in the day. These negative outcomes were attenuated for patients who prefer Spanish and more pronounced for patients with a non-Spanish NELP, perhaps because of more limited access to interpretation services for languages other than Spanish. The average patient with non-Spanish NELP stayed in the hospital > 2 days longer than the English-speaking patients, even after adjusting for technology dependence. This increased LOS may potentially put an already vulnerable patient population at increased risk for hospital-acquired conditions.8,15 Increased LOS also has significant repercussions for families, because parents may face additional financial and psychosocial stress.16,17 Patients with NELP were also discharged later in the day than English-speaking patients, with the non-Spanish NELP group again having the longest lengths of stay and latest discharge times. Time of discharge is an important metric to consider for families and hospital workflow. Families of children with medical complexity often prefer to be discharged earlier in the day to get adequately settled at home.18 Later discharges have been shown to adversely affect hospital workflow, because they can lead to ED and postanesthesia care unit crowding.19 Discharging patients early in the day is a goal for many hospitals,20 making the observed disparity in discharge time between English-speaking patients and patients with NELP antithetical to many institution’s priorities. Although prolonged LOS for the non–Spanish-speaking NELP group did not translate into a statistically significant increase in cost in this study, further investigation is warranted to assess the impact of disparities in LOS on costs of care for patients with NELP.
The etiology of this disparity in length of stay and time of discharge on the basis of language preference is likely multifactorial and merits further exploration. Although patients with NELP have a legal right to access health care in their preferred language,21 professional interpretation is known to be underutilized in the hospital setting, and insufficient use of professional interpreting services likely contributes to the disparities in outcomes in the population with NELP. A common driver of inpatient communication challenges is the difficulty accessing interpreter services.22 At our institution, the number 1 cited barrier to consistent interpreter use is the time associated with interpreter use.23 Consistent use of interpretation services has been linked to improved patient outcomes such as fewer communication errors and increased patient comprehension,24–27 which may impact length of stay. Professional interpretation services are also positively received by patients and families,25–27 and improved patient satisfaction and communication may potentially improve patient family and the medical team collaboration for discharge planning. Additionally, families with a non-English language preference may also experience systemic bias within the health care system related to race, socioeconomic status, country of origin, or other identities.28 In focus groups, Spanish-speaking patients have reported feeling discriminated against because of their language preference.29 Ways to mitigate the effect of bias on patient care includes acknowledgment, respectful communication, building rapport, appropriate use of language services, individualized care, and leveraging technology to limit unconscious bias in health care.28–30
To our knowledge, this is the first study to illustrate the impact of NELP on pediatric hospitalizations, specifically in patients with NELP with preferred language other than Spanish. Most previous studies have characterized NELP patients as a single entity; however, fewer resources are available for families with a non-Spanish NELP. At our institution, patients and their families with non-Spanish NELP receive different access to interpretation resources, which is the case at other institutions as well.10,11 For example, although written discharge instructions improve patient outcomes, many children’s hospitals report that there is difficulty providing written instructions for languages other than Spanish,31,32 which may impact the time of day at which discharge occurs. As a result of the differences in interpreter and translator availability, the non-Spanish NELP patient population has a distinct health care experience from the Spanish-speaking patient population and may require dedicated quality improvement efforts to improve clinical outcomes.
The patients with NELP included in our study were also disproportionately likely to have technology dependence. This is a novel finding, but the reason behind this association is not clear. It may be that families with NELP with technology-dependent children living outside of our institution’s typical catchment area are willing to travel to seek subspecialty care at our institution. This is an important issue because technology-dependence and the associated complex chronic medical conditions are known risk factors for higher cost of care and more pediatric readmissions11,33 ; when patients are both medically complex and have NELP, there may be a synergistic negative effect that warrants attention.
Our study has important limitations. First, it was a single-site study, and we did not capture readmissions or return ED visits at other institutions; however, our institution is the largest children’s hospital in the region. Because there was a high proportion of children with technology dependence in our study, our results are most likely to be generalizable to other large, urban, academic institutions. Although we had a large sample size, including a large sample of patients who prefer Spanish, only 2.3% of the participants had a non-Spanish NELP. Although we did identify statistically significant and important differences in outcomes in that group, we may have been underpowered to detect other significant differences, specifically, the difference in admission cost between patients who prefer English and patients with a non-Spanish NELP. It is also difficult to capture a family’s language preference status. Although we categorized language preference as a categorical variable, in reality, it is a spectrum with variations even among individuals in the same family. Patients with a NELP may come from socioeconomic backgrounds or have different rates of health literacy that influence outcomes, which may have confounded our results. Finally, general medicine patients have a large spectrum of illness severity. Although there may be diagnosis-specific differences in the measured outcomes that we did not capture, we were able to adjust for technology dependence as a proxy for medical complexity. Because we saw a significantly higher proportion of children with technology-dependence in the NELP groups, more work is required to elucidate the relationship between technology dependence, medical complexity, and NELP.
Conclusions
Compared to English-speaking patients, patients with NELP experienced longer LOS and prolonged time to discharge. The notable disparity in outcomes for patients with non-Spanish NELP requires more direct efforts to improve services for patients with NELP who prefer languages other than Spanish and work toward a more equitable health care system for all.
FUNDING: No external funding.
CONFLICT OF INTEREST DISCLOSURES: The authors have no conflict of interest relevant to this article to report.
Dr Pilarz conceptualized and designed the study, drafted the initial manuscript, and reviewed and revised the manuscript; Dr Rodriguez contributed to study conceptualization and design, assisted with the initial draft of the manuscript, and critically reviewed the manuscript for important intellectual content; Ms Jackson provided statistical analysis, interpreted data, and critically revised the manuscript; Dr Rodriguez conceptualized and designed the study and critically reviewed the manuscript for important intellectual content; and all authors approved the final manuscript as submitted and agreed to be accountable for all aspects of the work.
Comments