Abstract
Background
Recognition of clinically deteriorating hospitalized patients with activation of rapid response (RR) systems can prevent patient harm. Patients with limited English proficiency (LEP), however, experience less benefit from RR systems than do their English-speaking counterparts.
Objective
To improve outcomes among hospitalized LEP patients experiencing clinical deteriorations.
Design
Quasi-experimental pre-post design using quality improvement (QI) statistics.
Participants
All adult hospitalized non-intensive care patients with LEP who were admitted to a large academic medical center from May 2021 through March 2023 and experienced RR system activation were included in the evaluation. All patients included after May 2022 were exposed to the intervention.
Interventions
Implementation of a modified RR system for LEP patients in May 2022 that included electronic dashboard monitoring of early warning scores (EWSs) based on electronic medical record data; RR nurse initiation of consults or full RR system activation; and systematic engagement of interpreters.
Main Measures
Process of care measures included monthly rates of RR system activation, critical response nurse consultations, and disease severity scores prior to activation. Main outcomes included average post-RR system activation length of stay, escalation of care, and in-hospital mortality. Analyses used QI statistics to identify special cause variation in pre-post control charts based on monthly data aggregates.
Key Results
In total, 222 patients experienced at least one RR system activation during the study period. We saw no special cause variation for process measures, or for length of hospitalization or escalation of care. There was, however, special cause variation in mortality rates with an overall pre-post decrease in average monthly mortality from 7.42% (n = 8/107) to 6.09% (n = 7/115).
Conclusions
In this pilot study, prioritized tracking, utilization of EWS-triggered evaluations, and interpreter integration into the RR system for LEP patients were feasible to implement and showed promise for reducing post-RR system activation mortality.
Supplementary Information
The online version contains supplementary material available at 10.1007/s11606-024-08678-x.
INTRODUCTION
Recognition of clinically deteriorating hospitalized patients with activation of rapid response (RR) systems can prevent patient harm. Language barriers, however, are a substantial obstacle for patients with limited English proficiency (LEP) navigating the US healthcare system, and may result in providers overlooking subtle indicators of impending clinical deterioration.1–3 Systemic inequities and implicit biases can also negatively affect the management of LEP patients. There is substantial evidence suggesting that patients with LEP have worse health outcomes overall, including both during and after events requiring immediate evaluation by healthcare providers, such as hypotension and hypoxia;4–10 this disparity persists even in the presence of a RR system.9 While negative outcomes from these events can be mitigated by consistent interpreter utilization, indicating the importance of reliable communication between patient and provider,9 few studies have evaluated effects of integrating interpreter support into the RR response system.
Despite widespread use of RR systems, there is significant variability in interpreter utilization for hospitalized patients.11,12 For example, in an Australian hospital, just 17.1% of patients reporting interpreter need actually received this service.13 Many health professionals have limited knowledge about laws and policies concerning interpreter services; therefore, interpreter utilization depends largely on often flawed, subjective evaluations of patient language needs.14 Faced with barriers to qualified medical interpretation, patients frequently rely on friends and family, introducing concerns about competency and confidentiality.3,15 Rather than placing the burden of language access on patients, interpreters should be seamlessly integrated into the care of patients with LEP.16 Providing culturally and linguistically competent interpreters is an organizational-level strategy to drive equitable access to healthcare.1,17
Early warning scores (EWSs) are predictive models built with objective patient data to help predict clinical deterioration; theoretically, this may help offset more subjective patient assessments that can be flawed or limited by poor communication between patient and provider. Many hospitals have incorporated EWSs into the electronic health record (EHR) to help providers accurately identify “at-risk” patients. In these models, objective data collected from routine assessments are used to calculate an EWS, which can subsequently be employed to display a patient’s clinical trajectory over the course of their hospitalization.
A systematic review associated the incorporation of an EWS-initiated protocol into the RR system with improved outcomes.18 Another group found that non-emergent EWS monitoring and associated interventions reduced mortality for patients at risk of deterioration by 22%.19 Before our project commenced, our hospital had integrated EWSs for all non-intensive care patients into the EHR; however, monitoring of the scores was provider dependent. Despite having a pre-existing RR system, having a high EWS was not a system activation criterion.
In a prior retrospective study at our institution,9 we found that Spanish-speaking patients had significantly higher disease severity (as measured by EWS) at the time of RR system activation when compared to English-speaking patients. Furthermore, being a Spanish-speaking patient was associated with a significantly longer post-RR event length of stay compared to English-speaking patients when controlling for disease severity. There was also a trend toward higher hospital mortality rates.9 We postulated that language-based delays in triggering the RR system may be impacting quality of care (see Fig. 1).
Figure 1.
Conceptual model of existing RR system for primary language Spanish patients.9
Thus, in this 1-year pilot study, we tested whether an intervention directed at reducing language- and communication-based issues could improve care for LEP patients, including length of hospitalization and mortality. The interventions included (a) using an EHR-based dashboard to display EWSs for LEP patients; (b) systematizing ongoing monitoring of EWSs by RR nurses; and (c) linking RR system activation to immediate engagement of a certified interpreter. As main outcomes, we aimed to test whether the interventions were associated with reduced patient acuity at the time of RR system activation, with reduced hospital length of stay after RR system activation, or with reduced mortality among patients experiencing RR system activation.
METHODS
Overview
This project was sponsored by our institution as part of the “Emerging Challenges in Biomedical Research – Health Equity” program. We developed the intervention through repeated Plan-Do-Study-Act cycles.20 First, we developed a dashboard within the EHR that trended the EWSs of all hospitalized, non-intensive care LEP patients to help readily identify those at risk of clinical deterioration. The RR nurses reviewed the dashboard every 12 h to monitor the scores of the target patient population. We then used the EWS to trigger standardized critical response nurse consultation or RR system activation. Lastly, we improved communication during RR system activations by reliably integrating interpreters into the RR team. We tested whether implementation of our interventions was associated with subsequent reductions in the group’s pre-RR system activation risk score, post-system activation length of stay, escalation in level of care, or mortality, using monthly electronic data reports. We used quality improvement statistics to determine whether changes in outcome over time were significantly associated with implementation of the interventions. The project goals aligned with the organizational priority to prevent patient harm, reduce hospital mortality, and improve equity of care. SQUIRE (Standards for Quality Improvement Reporting Excellence), V.2.0 was used to prepare this manuscript.21
Ethical Considerations
The proposed interventions were associated with minimal risk and the project was exempted by the UNC Institutional Review Board. Our project received full support and oversight by UNC Health entities, the Division of Hospital Medicine, and the North Carolina Translational and Clinical Sciences Institute.
Setting and Context
This project was conducted at the University of North Carolina Medical Center (UNCMC), a quaternary care center in Chapel Hill, NC with 950 beds and 37,000 patients annually. Compared with inpatients across all NC hospitals, UNCMC inpatients are 2.05 times as likely to be Hispanic. Four percent of all UNCMC inpatients self-identify as primary language Spanish (see Appendix 1).22 UNCMC provides interpreter services through 24-h access to onsite Spanish interpreters, as well as remote interpretation in more than 200 languages via an external partner organization.
In 2006, UNCMC introduced a RR system consisting of an afferent (recognition) limb and an efferent (response) limb. The afferent limb facilitates the prompt detection of patients at risk of clinical deterioration. The efferent limb represents the mobilization of the RR team to patient bedside, providing necessary resources to correct the imbalance of unmet needs; this constitutes the RR “event.” The most common reasons for RR system activation at UNCMC are hypotension, altered mental status, tachycardia, and hypoxia. At UNCMC, the RR team comprised a critical care-trained RR nurse, respiratory therapist, house supervisor, the patient’s primary bedside nurse, and a provider from the primary medical team. The RR nurse evaluates the patient and helps the primary medical team coordinate expedited medical care. In calendar year 2022, the RR team attended to 3281 RR system activations.
In addition to the RR system, UNCMC also has a process of critical response nurse consultation. The critical response nurse is the RR nurse serving in a different capacity. Any provider can activate a critical response nurse consult by placing an order in the EHR. Common reasons for use of this service include assistance with difficult intravenous access, blood draws, and clinical assessment, to name a few. The latter option is used by providers who may not feel confident in immediately activating the RR system and would prefer a second opinion.
UNCMC uses the Epic electronic health record with an embedded proprietary Deterioration Index (DI) as the principal EWS. The DI is a value from 0 to 100 generated every 15 min and represents the probability of an adverse event within the next 12 h or mortality within the next 36 h (see Appendix 2).
Quality Improvement Initiative Timeline and Participants
Between April 2022 and March 2023, a multidisciplinary project team comprised the manager of interpreter services, a project manager, two RR nurses, two medical students, and four faculty physicians, met to design, implement, and evaluate the study interventions. An information technologist assisted interaction with the EHR. The team developed a project charter to provide the framework for change including measures of success. During the project period, the team met bi-weekly to review the process, data, and feedback and make small changes to improve the intervention as part of Plan-Do-Study-Act cycles (see Fig. 2).
Figure 2.
Summary timeline of changes made throughout the project period.
Patient Participants
All adult patients with a preferred language other than English at admission who were admitted to a non-intensive care bed were continuously identified and eligible for study interventions. Among these, we identified the subset who experienced a RR system activation for more detailed medical record assessment for the evaluation. No patients identified as LEP were excluded.
Description of the Interventions
LEP Dashboard Development
We designed a dashboard that automatically identified new LEP admissions and auto-populated with the most recent DI score, change in DI score since last manual review, and relative contribution of clinical variables to the current DI score (vital signs, laboratory results, neurologic status, etc.). At our hospital, there is a standard process for determining a patient’s “Preferred Language” and if they “Need [an] Interpreter.” For inpatients, this process occurs at an initial registration hub (e.g., intake desk, emergency department), or at bedside (for patients who are directly admitted without first going through a registration hub). The process starts with asking the patient about their preferred language; the answer is then documented in the EHR. Since this process occurs at intake, preferred language and interpreter need may change between hospitalizations, but rarely intra-hospitalization. Currently, there is no way to track whether preferred language or interpreter need changes for a given patient.
Change in RR System “Tracking” for LEP Patients
RR nurses were directed to review the dashboard at least every 12 h to monitor the DI scores of patients with LEP.
Change in RR System “Triggering” for LEP Patients
As shown in Figure 3, if a patient’s DI score achieved a value ≥ 60 or exhibited a sustained change (≥ 4 h) of ≥ 10 since last review, a RR nurse examined the patient’s medical record and consulted with the patient’s primary bedside nurse. If any pre-intervention RR system activation criteria were met (see Fig. 3), staff immediately activated the RR system. If not, the RR nurse would instead engage interpreter services to assist with a critical response nurse consultation (as previously defined).
Figure 3.
DI for LEP inpatients intervention workflow and RR system activation criteria.
Change in Communication Practices
RR system activation for Spanish-speaking patients resulted in an additional overhead page for urgent in-person interpreter services. In cases of critical care nurse consults for Spanish-speaking patients, the encounter was facilitated by non-urgent in-person interpreter assistance. For language interpretation other than Spanish, the RR nurse called remote interpreters to assist with communication via phone or video.
Evaluation of the Intervention for LEP Patients Who Experienced a RR Event
Data Sources The start of every RR event is documented using a narrator tab in the patient’s EHR. During the event, the recorder documents numerous data in the narrator tab, including the names and roles of those in attendance, a clinical summary, and the diagnostics/interventions pursued. The end of the event is signified by the closing of the narrator tab. Every RR event requires the use of this narrator tab; this information can be retrospectively extracted for review. We generated cross-sectional monthly data reports from the medical record on all hospitalized adult, non-intensive care LEP patients (defined as any patient with a documented “Preferred Language” other than English) with a RR event pre- and post-intervention implementation, between May 1, 2021, and March 31, 2023. We also collected cross-sectional monthly data on the number of critical response nurse consults throughout this period; this was extracted from historical orders in the EHR.
To evaluate the qualitative impact of the interventions, a quick response (QR) code linked to a QualtricsXM survey response system (see Appendix 3) was developed. The QR codes were sent via email to RR nurses and Spanish interpreters to solicit responses from those who may have experienced a LEP RR system activation during the intervention period.
Measures
Process Measures
Measures to assess intervention activities and fidelity included the monthly number of RR critical response nurse consults; the monthly number of RR system activations; and monthly average DI score at the time of RR system activation.
Outcome Measures
Our main outcome measures were monthly mortality rate among patients experiencing a RR system activation, monthly average escalation of care (defined as transferring to a higher acuity level of inpatient care) post-RR system activation, and monthly average length of hospitalization post-RR system activation.
Balancing Measures
As a balancing measure, we assessed the average monthly duration of RR events, defined as the difference between the start and end times of the event in the narrator documentation tab.
Analysis
Pre- and post-intervention X bar and R control charts were developed with Microsoft Excel QI Macros Statistical Process Control Software. Special cause variation was identified using the Nelson Rules (see Appendix 4).23 Charts represent measures derived from monthly cross-sectional samples of all hospitalized general medical patients with LEP and occurrence of a RR system activation. On X bar charts, special cause variation indicates a significant shift in the mean of a target measure due to reasons that are not explained by common or inherent sources of variation.24 The X bar chart derives control limits from the data shown in the R chart as the average range of values. If a corresponding R chart’s values are out of control, the X bar chart control limits are inaccurate. Special cause variation on an X bar chart but not on the corresponding R control chart indicates “special cause variation,” meaning variation due to a specific, identifiable event or factor that occurred in time with the change in the mean of the target measure. We also report pre- versus post-average values across all months for process and outcome variables to provide context for analytic findings.
RESULTS
Throughout the combined pre- and post-intervention periods (from May 1, 2021, to March 31, 2023), a total of 222 individual LEP patients experienced 302 RR system activations (with some patients experiencing more than one RR system activation). In the pre-intervention period (May 1, 2021, to April 30, 2022), there were 135 RR system activations and 68 critical response nurse consultations for LEP patients. Subsequently, in the post-intervention period (May 1, 2022, to March 31, 2023), LEP patients experienced 167 RR system activations and 60 critical response nurse consultations.
Process Measures
Between the pre- and post-intervention periods, neither the average monthly number of critical response nurse consults for LEP patients (5.67/month pre- to 5.45/month post-intervention) nor the average monthly number of RR system activations (11.3/month pre- to 15.7/month post-intervention) showed special cause variation. Average disease severity (DI scores) at the time of RR system activation showed special cause variation by Nelson rule 5 (two out of three consecutive points more than 2 standard deviations from the center line in the same direction) on the X bar chart, with a slight rise in the average DI scores per month for the post-intervention period (41.65 pre- to 42.97 post-intervention, Fig. 4), and corresponding stable, consistent variability in the R-chart.
Figure 4.
Average disease severity (DI score) at the time of RR system activation pre- and post-intervention.
Outcome Measures
LEP patients experiencing RR system activation post-intervention compared to those experiencing RR system activation pre-intervention showed special cause variation indicating a decreased mortality rate (X bar chart, Fig. 5) by Nelson rules 2 and 3. Rule 2 specifies nine consecutive points on one side of the centerline while rule 3 specifies six or more consecutive points steadily decreasing. The R chart confirms special cause variation, showing no systematic variation in the per month mortality rate. These findings strongly suggest that the observed decreased mortality (7.42% [n = 8/107] pre- to 6.09% [n = 7/115] post-intervention) after RR system activation is due to the intervention.
Figure 5.
Rate of post-RR system activation mortality pre- and post-intervention.
While LEP patients experienced an average increased rate of escalation of care after RR system activation (43.75% or 59/135 escalated pre- versus 45.73% or 102/222 post-intervention), and decreased length of hospitalization post-RR system activation (an average of 20.45 days per patient pre- versus 19.52 days per patient post-intervention), we found no special cause variation that linked these changes to intervention implementation.
Balancing Measures
There was no special cause variation in average RR event duration for LEP patients (0:48:24 min pre- to 0:56:56 min post-intervention).
The QualtricsXM stakeholder survey yielded 17 total responses (14 from interpreters and three from RR nurses; see “Limitations” section). Qualitative information from frontline stakeholder comments is provided in Table 1 (there were no negative comments and neutral comments were not included).
Table 1.
Responses from Frontline Stakeholders on the QualtricsXM Survey
| Role | Comment |
|---|---|
| Interpreter | “Great teamwork and communication amongst the MDs and nursing.” |
| Interpreter | “From an interpreter’s standpoint, it’s helping us to increase our awareness and readiness to timely respond.” |
| Interpreter | “Helpful” |
| Interpreter | “Rapid responses can be scary for patients and families. Better communication can only improve things, even if it’s putting minds at ease that each party has the best information possible.” |
| Interpreter | “The process greatly improves the outcome of the situation dealt with; easier for patients to have a clear understanding and expectations, makes the patients feel more comfortable and they’re able to communicate their needs in a clear and concise way.” |
| Interpreter | “The process increases the level of care for patients and also the trust towards providers when the patient and their family members are listened to.” |
| Interpreter | “Without an interpreter present, it is extremely difficult to assess a patient's deterioration.” |
| RR nurse | “Keep this momentum. This makes rapid response events more productive.” |
| RR nurse | “Patient’s father actively took part in all aspects of rapid response.” |
| RR nurse | “It’s helpful to help me determine how ill the patient is per this index.” |
DISCUSSION
Barriers in communication can magnify preexisting systemic disparities in patient evaluation and treatment, leading to delays in recognition of clinical deterioration and, consequently, increased patient morbidity and mortality. In response to data from our hospital showing worsened outcomes among patients with LEP who experienced RR system activation, we undertook modifications to the workflow of the RR system to prioritize the monitoring and evaluation of this at-risk population aimed at prompt clinical recognition, assessment, and improved access to interpreters. We showed that the interventions were feasible to implement and found promising impacts of the interventions on mortality rates following a RR system activation.
We found that the proportion of LEP patients experiencing RR system activation who died in the hospital afterward decreased during the months after implementation of our intervention, showing special cause variation based on QI statistics.24 Overall, mortality after RR system activation decreased among LEP patients from 7.42% (8/107) during the pre-intervention period to 6.09% (7/115) during the post-intervention period. We did not find post-intervention special causes variation in the number of critical response nurse consults, number of RR system activations, escalation of care, length of hospitalization, or duration of RR events over time. Interestingly, we found that the RR system was activated for LEP patients at a slightly higher disease severity (as measured by the average monthly DI score) during the post-intervention period compared to the pre-intervention period, with special cause variation. We had expected that the intervention would result in RR system activation earlier, at lower DI scores, due to increased clinical monitoring. The increase in the DI score at the start of a RR system activation accompanied by a decrease in mortality afterward suggests the need for further investigation of the clinical effects of the intervention. For example, it is possible that in some cases, during the post-intervention period, activation of the RR system for patients with lower DI scores was averted by use of more routine clinical interventions, based on better clinician-patient communication. Additionally, the lack of an increase in length of hospitalization during the post-intervention period, despite increased acuity at the time of the RR system activation and accompanied by lower mortality, may suggest positive intervention effects on recovery of LEP RR activation survivors. These and other possibilities should be explored in future research. We were not able to carry out a full evaluation of the perceptions of our intervention among the clinicians and patients who were exposed to it.
Our quality improvement project had several strengths. First, the interventions responded directly to a serious problem affecting vulnerable patients at our institution. The development of the interventions was based both on prior evidence and on identification of the root causes for observed disparities in patient outcomes, and was refined, along with evaluation measures, through repeated Plan-Do-Study-Act cycles. Our interventions were designed to maximize feasibility and maintenance by making use of existing organizational capabilities, while redesigning them to focus on ensuring more equitable clinical care and outcomes. While our organization’s resources for recognizing and managing acute clinical deterioration are significant, the workflow we created is likely feasible to implement within smaller, less resource-intensive hospital environments that have existing RR systems, an EHR, and access to interpreter services.
This project aligns with prior research on successful use of DI scores to proactively trigger RR system activation for immediate evaluation of and interventions for deteriorating patients.19 Furthermore, our results are consistent with research highlighting the positive impacts of consistent interpreter use on RR system outcomes among patients with LEP, including on patient satisfaction and patient-provider communication.25 We know of no prior study, however, that uses an EHR-based warning system to trigger prompt use of interpreters for patients with LEP in coordination with the RR team.
Limitations
These initiatives were implemented at a single medical center, and thus may not be generalizable to other institutions. Future research should evaluate the effects of utilizing objective tracking and triggering methods for early RR system activation within other inpatient settings, such as in smaller community/rural hospitals or hospitals with fewer resources, and for other patient populations. Our findings are also limited by the study’s modest population size and constrained project period. Additionally, despite our best efforts, we were not able to comprehensively assess either implementation fidelity or user perceptions of our interventions. Finally, the project’s design offers only a limited capacity to assess causal relationships between the intervention and the observed outcomes. However, our use of the combination of X bar charts with R chart controls over 12 months of observation pre- and 11 months post-intervention implementation to evaluate study outcomes strengthens our assessment of the relationships between our findings and the implementation of our intervention.
CONCLUSIONS
Our current work builds on prior work by our team and others that highlights the vulnerability of LEP patients to negative health outcomes based on both systemic biases and unreliable or inaccurate patient-provider communication. The results from this pilot project show promise for future testing of interventions designed to improve the equity of RR system activation and outcomes for hospitalized LEP patients. Our results should stimulate further investigation of methods for using EHR dashboard warning alerts to initiate RR system activation for LEP patients, and to link the alerts to rapid engagement of interpreters along with the RR team.
Supplementary Information
Below is the link to the electronic supplementary material.
Acknowledgements
The authors acknowledge Brittany Davis, RN, Ellenita Kornegay, RN, BSN, Jennifer Mack, MHA, MBA, BSN, RN, CCRN, Mary Jo Barfield, MBA, Kelly Reilly, MEd, the UNC Medical Center interpreters and rapid response teams, and the North Carolina Translational and Clinical Sciences Institute. In October 2023, this work was recognized as a Top Finalist for The Joint Commission and Kaiser Permanente Bernard J. Tyson National Award for Excellence in Pursuit of Healthcare Equity.
Data Availability
The data supporting the findings of this study are available upon request from the authors.
Declarations:
Conflict of Interest:
The authors declare that they do not have a conflict of interest.
Footnotes
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Rashoka FN, Kelley MS, Choi JK, Garcia MA, Chai W, Rashawka HN. "Many people have no idea": a qualitative analysis of healthcare barriers among Yazidi refugees in the Midwestern United States. Int J Equity Health. 2022;21(1):48. doi: 10.1186/s12939-022-01654-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Lee JS, Pérez-Stable EJ, Gregorich SE, et al. Increased access to professional interpreters in the hospital improves informed consent for patients with limited english proficiency. J Gen Intern Med. 2017;32(8):863–870. doi: 10.1007/s11606-017-3983-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Morris MD, Popper ST, Rodwell TC, Brodine SK, Brouwer KC. Healthcare barriers of refugees post-resettlement. J Community Health. 2009;34(6):529–538. doi: 10.1007/s10900-009-9175-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Divi C, Koss RG, Schmaltz SP, Loeb JM. Language proficiency and adverse events in US hospitals: a pilot study. Int J Qual Health Care. 2007;19(2):60–67. doi: 10.1093/intqhc/mzl069. [DOI] [PubMed] [Google Scholar]
- 5.Hines AL, Andrews RM, Moy E, Barrett ML, Coffey RM. Disparities in rates of inpatient mortality and adverse events: race/ethnicity and language as independent contributors. Int J Environ Res Public Health. 2014;11(12):13017–13034. doi: 10.3390/ijerph111213017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.John-Baptiste A, Naglie G, Tomlinson G, et al. The effect of English language proficiency on length of stay and in-hospital mortality. J Gen Intern Med. 2004;19(3):221–228. doi: 10.1111/j.1525-1497.2004.21205.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Khan A, Yin HS, Brach C, et al. Association between parent comfort with english and adverse events among hospitalized children. JAMA Pediatr. 2020;174(12):e203215. doi: 10.1001/jamapediatrics.2020.3215. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.McDade JE, Olszewski AE, Qu P, et al. Association between language use and ICU transfer and serious adverse events in hospitalized pediatric patients who experience rapid response activation. Front Pediatr. 2022;10:872060. doi: 10.3389/fped.2022.872060. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Raff L, Moore C, Raff E. The role of language barriers on efficacy of rapid response teams. Hosp Pract (1995). 2023;51(1):29–34. doi: 10.1080/21548331.2022.2150416. [DOI] [PubMed] [Google Scholar]
- 10.van Rosse F, de Bruijne M, Suurmond J, Essink-Bot ML, Wagner C. Language barriers and patient safety risks in hospital care. A mixed methods study. Int J Nurs Stud. 2016;54:45–53. doi: 10.1016/j.ijnurstu.2015.03.012. [DOI] [PubMed] [Google Scholar]
- 11.Benda NC, Fairbanks RJ, Higginbotham DJ, Lin L, Bisantz AM. Observational study to understand interpreter service use in emergency medicine: why the key may lie outside of the initial provider assessment. Emerg Med J. 2019;36(10):582–588. doi: 10.1136/emermed-2019-208420. [DOI] [PubMed] [Google Scholar]
- 12.López L, Rodriguez F, Huerta D, Soukup J, Hicks L. Use of interpreters by physicians for hospitalized limited English proficient patients and its impact on patient outcomes. J Gen Intern Med. 2015;30(6):783–789. doi: 10.1007/s11606-015-3213-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Blay N, Ioannou S, Seremetkoska M, et al. Healthcare interpreter utilisation: analysis of health administrative data. BMC Health Serv Res. 2018;18(1):348. doi: 10.1186/s12913-018-3135-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Lundin C, Hadziabdic E, Hjelm K. Language interpretation conditions and boundaries in multilingual and multicultural emergency healthcare. BMC Int Health Hum Rights. 2018;18(1):23. doi: 10.1186/s12914-018-0157-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Medicaid CfMa. 87 FR 47824: Nondiscrimination in Health Programs and Activities. Office for Civil Rights, Office of the Secretary, Health and Human Services. Accessed 28 June 2023, https://www.federalregister.gov/documents/2022/08/04/2022-16217/nondiscrimination-in-health-programs-and-activities
- 16.Bayer-Oglesby L, Zumbrunn A, Bachmann N, Team S Social inequalities, length of hospital stay for chronic conditions and the mediating role of comorbidity and discharge destination: a multilevel analysis of hospital administrative data linked to the population census in Switzerland. PLoS One. 2022;17(8):e0272265. doi: 10.1371/journal.pone.0272265. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Cohen AL, Rivara F, Marcuse EK, McPhillips H, Davis R. Are language barriers associated with serious medical events in hospitalized pediatric patients? Pediatrics. 2005;116(3):575–579. doi: 10.1542/peds.2005-0521. [DOI] [PubMed] [Google Scholar]
- 18.Credland N, Dyson J, Johnson MJ. Do early warning track and trigger tools improve patient outcomes? A systematic synthesis without meta-analysis. J Adv Nurs. 2021;77(2):622–634. doi: 10.1111/jan.14619. [DOI] [PubMed] [Google Scholar]
- 19.Cacciaglia A. Saving Lives with AI: Using the Deterioration Index Predictive Model to Help Patients Sooner. Accessed 30 June 2023, 2023. https://epicshare.org/share-and-learn/saving-lives-with-ai
- 20.Langley GL, Moen R, Nolan K, Nolan T, Norman C, Provost L. The Improvement Guide: A Practical Approach to Enhancing Organizational Performance (2nd edition). Jossey-Bass Publishers; 2009.
- 21.Goodman D, Ogrinc G, Davies L, et al. Explanation and elaboration of the SQUIRE (Standards for Quality Improvement Reporting Excellence) Guidelines, V.2.0: examples of SQUIRE elements in the healthcare improvement literature. BMJ Qual Saf. 2016;25(12):e7. doi: 10.1136/bmjqs-2015-004480. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.The Cecil G. Sheps Center for Health Services Research, University of North Carolina at Chapel Hill. NC Inpatient Patient Characteristics by Hospital 2021. Accessed 27 June 2023. https://www.shepscenter.unc.edu/wp-content/uploads/2023/07/ptchar_all_and_by_hosp_2021_and.pdf.
- 23.Nelson LS. The shewhart control chart—tests for special causes. J Qual Technol. 1984;237–239.
- 24.Provost LP, Murray SK. The health care data guide: learning from data for improvement. John Wiley & Sons; 2022.
- 25.Chua WL, See MTA, Legio-Quigley H, Jones D, Tee A, Liaw SY. Factors influencing the activation of the rapid response system for clinically deteriorating patients by frontline ward clinicians: a systematic review. Int J Qual Health Care. 2017;29(8):981–998. doi: 10.1093/intqhc/mzx149. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data supporting the findings of this study are available upon request from the authors.