Role Preferences in Medical Decision Making: Relevance and Implications for Health Preference Research.

Health preference research (HPR) is being increasingly conducted to better understand patient preferences for medical decisions. However, patients vary in their desire to play an active role in medical decisions. Until now, few studies have considered patients' preferred roles in decision making. In this opinion paper, we advocate for HPR researchers to assess and account for role preferences in their studies, to increase the relevance of their work for medical and shared decision making. We provide recommendations on how role preferences can be elicited and integrated with health preferences: (1) in formative research prior to a health preference study that aims to inform medical decisions or decision makers, (2a) in the development of health preference instruments, for instance by incorporating a role preference instrument and (2b) by clarifying the respondent's role in the decision prior to the preference elicitation task or by including role preferences as an attribute in the task itself, and (3) in statistical analysis by including random parameters or latent classes to raise awareness of heterogeneity in role preferences and how it relates to health preferences. Finally, we suggest redefining the decision process as a model that integrates the role and health preferences of the different parties that are involved. We believe that the field of HPR would benefit from learning more about the extent to which role preferences relate to health preferences, within the context of medical and shared decision making.

Caregiver-reported quality of communication in pancreatic and periampullary cancer.

BACKGROUND: Communication between caregivers and clinical team members is critical for transitional care, but its quality and potential impact on outcomes are not well understood. This study reports on caregiver-reported quality of communication with clinical team members in the postpancreatectomy period and examines associations of these reports with patient and caregiver outcomes. METHODS: Caregivers of patients with pancreatic and periampullary malignancies who had undergone pancreatectomy were surveyed. Instrument measures assessed care experiences using the Caregiver Perceptions About Communication with Clinical Team Members (CAPACITY) instrument. The instrument has two main subscales: communication, assessing the extent to which providers helped caregivers comprehend details of clinical visits, and capacity, defined as the extent to which providers assessed whether caregivers were able to care for patients. RESULTS: Of 265 caregivers who were approached, 240 (90.6%) enrolled in the study. The mean communication and capacity subscale scores were 2.7 ± 0.6 and 1.5 ± 0.6, respectively (range, 0-4 [higher = better]). Communication subscale scores were lower among caregivers of patients who experienced (vs. those who did not experience) a 30-day readmission (2.6 ± 0.5 vs. 2.8 ± 0.6, respectively; p = .047). Capacity subscale scores were inversely associated with restriction in patient daily activities (a 0.04 decrement in the capacity score for every 1 point in daily activity restriction; p = .008). CONCLUSIONS: After pancreatectomy, patients with pancreatic and periampullary cancer whose caregivers reported worse communication with care providers were more likely to experience readmission. Caregivers of patients with greater daily activity restrictions were less likely to report being asked about the caregiver's skill and capacity by clinicians. PLAIN LANGUAGE SUMMARY: This prospective study used a validated survey instrument and reports on the quality of communication between health care providers and caregivers as reported by caregivers of patients with pancreatic and periampullary cancer after pancreatectomy. In an analysis of 240 caregivers enrolled in the study, lower communication scores (the extent to which providers helped caregivers understand clinical details) were associated with higher odds of 30-day patient readmission to the hospital. In addition, lower capacity scores (the extent to which providers assessed caregivers' ability to care for patients) were associated with greater impairment in caregivers. The strikingly low communication quality and capacity assessment scores suggest substantial room for improvement, with the potential to improve both caregiver and patient outcomes.

Screening Second Victims for Emotional Distress: Assessment of the Clinimetric Properties of the WITHSTAND-PSY Questionnaire.

INTRODUCTION: Adverse events (AEs) are a leading cause of patient morbidity and mortality, greatly impacting healthcare providers' well-being (second victim (SV) phenomenon). Since it is not accurately captured by existing psychometric instruments, we developed a clinimetric instrument for assessing SVs' emotional distress before and after an AE. METHODS: Content validity and clinical utility of the WITHSTAND-PSY Questionnaire (WS-PSY-Q) were examined using cognitive interviews. Rasch analysis (n = 284) was applied for clinimetric assessment (i.e., construct, concurrent, and clinical validity, internal consistency), considering two crucial psychological facets of the SV phenomenon (1st: emotional impact of the AE, 2nd: current emotional state). RESULTS: The Rasch partial credit model was used. The 1st facet demonstrated overall acceptable clinimetric properties with the subscale anxiety meeting clinimetric threshold values (e.g., all items with ordered thresholds, Loevinger's coefficient h ≥ 0.40; Person Separation Reliability Index (PSI) = 0.7). The 2nd facet showed overall better clinimetric properties for both subscales (e.g., h ≥ 0.40, PSI = 0.82 and 0.79, respectively; receiver operating characteristic area of 0.80 and 0.86, respectively). For both datasets, item fit statistics, except those for item 19, were within the critical range (z-score < ±2.5), and meaningful differential functioning analysis was observed for only 4 (out of 24) items. Local dependency was not observed, except for two item couples in the depression subscales. CONCLUSIONS: The WS-PSY-Q is the first clinimetric tool assessing SVs' emotional distress. It should be regarded as part of the armamentarium used by clinicians to assess in-depth healthcare providers' psychological reactions in the aftermath of an AE to mitigate burnout and allostatic overload.

Preventing Avoidable Rehospitalizations through Standardizing Management of Chronic Conditions in Skilled Nursing Facilities.

OBJECTIVES: This study evaluated the impact of standardized care protocols, as a part of a quality improvement initiative (J10ohns Hopkins Community Health Partnership, J-CHiP), on hospital readmission rates for patients with a diagnosis of congestive heart failure (CHF) and/or chronic obstructive pulmonary disease (COPD) after being discharged to skilled nursing facilities (SNFs). DESIGN: A retrospective study comparing 30-day hospital readmission rates the year before and 2 years following the implementation of the care protocol interventions. SETTINGS AND PARTICIPANTS: Patients discharged from Johns Hopkins Hospital or Johns Hopkins Bayview Medical Center to the participating SNFs diagnosed with CHF and/or COPD. METHODS: The standardized protocols included medical provider or nurse assessments on SNF admission, multidisciplinary care planning, and medication management to avoid unplanned readmissions to the hospital. Descriptive analyses were conducted to illustrate the 30-day readmission rates before and after protocol implementation. RESULTS: There were 1128 patients in the pre-J-CHiP cohort and 2297 patients in the J-CHiP cohort. About half of the patients with a recorded diagnosis of CHF without COPD had the standardized protocol initiated, whereas 47% of the patients with a recorded diagnosis of COPD without CHF had the standardized protocol initiated. Of patients with recorded diagnoses of COPD and CHF, 49% had both protocols initiated. A reduction in the readmission rate was observed for patients with COPD protocols, from 23.5% in 2011 to 12.1% in 2015. However, fluctuations in the readmission rates were observed for patients who initiated the CHF protocols. CONCLUSIONS AND IMPLICATIONS: There were improvements in the readmission rates in this study, especially for patients who had initiated standardized care protocols in the SNFs. Our findings demonstrate great value in standardizing care management and strengthening collaboration with chronic care settings to facilitate a smooth transition of medically complex patients discharged from large health care systems. Future interventions could consider assessing nonclinical factors that may impact preventable hospital readmissions.

The impact of emotional support on healthcare workers and students coping with COVID-19, and other SARS-CoV pandemics - a mixed-methods systematic review.

BACKGROUND: Pandemics such as COVID-19 pose threats to the physical safety of healthcare workers and students. They can have traumatic experiences affecting their personal and professional life. Increasing rates of burnout, substance abuse, depression, and suicide among healthcare workers have already been identified, thus making mental health and psychological wellbeing of the healthcare workers a major issue. The aim of this systematic review is to synthesize the characteristics of emotional support programs and interventions targeted to healthcare workers and students since the onset of COVID-19 and other SARS-CoV pandemics and to describe the effectiveness and experiences of these programs. METHOD: This was a mixed method systematic review. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were followed, and the review was registered on PROSPERO [CRD42021262837]. Searches were conducted using Medline, CINAHL, PsycINFO, Cochrane Library, and Scopus databases. The COVIDENCE systematic review management system was used for data selection and extraction by two independent reviewers. The JBI (Joanna Briggs Institute) critical appraisal tools were used to assess the quality of selected studies by two additional reviewers. Finally, data extraction and narrative analysis were conducted. RESULTS: The search retrieved 3161 results including 1061 duplicates. After screening, a total of 19 articles were included in this review. Participants in studies were nurses, physicians, other hospital staff, and undergraduate medical students mostly working on the front-line with COVID-19 patients. Publications included RCTs (n = 4), quasi-experimental studies (n = 2), cross-sectional studies (n = 6), qualitative interview studies (n = 3), and systematic reviews (n = 4). Most (63.4%) of the interventions used online or digital solutions. Interventions mostly showed good effectiveness (support-seeking, positive emotions, reduction of distress symptoms etc.) and acceptance and were experienced as helpful, but there were some conflicting results. CONCLUSION: Healthcare organizations have developed support strategies focusing on providing emotional support for these healthcare workers and students, but it is difficult to conclude whether one program offers distinct benefit compared to the others. More research is needed to evaluate the comparative effectiveness of emotional support interventions for health workers.

The Volume and Cost of Quality Metric Reporting.

Importance: US hospitals report data on many health care quality metrics to government and independent health care rating organizations, but the annual cost to acute care hospitals of measuring and reporting quality metric data, independent of resources spent on quality interventions, is not well known. Objective: To evaluate externally reported inpatient quality metrics for adult patients and estimate the cost of data collection and reporting, independent of quality-improvement efforts. Design, Setting, and Participants: Retrospective time-driven activity-based costing study at the Johns Hopkins Hospital (Baltimore, Maryland) with hospital personnel involved in quality metric reporting processes interviewed between January 1, 2019, and June 30, 2019, about quality reporting activities in the 2018 calendar year. Main Outcomes and Measures: Outcomes included the number of metrics, annual person-hours per metric type, and annual personnel cost per metric type. Results: A total of 162 unique metrics were identified, of which 96 (59.3%) were claims-based, 107 (66.0%) were outcome metrics, and 101 (62.3%) were related to patient safety. Preparing and reporting data for these metrics required an estimated 108 478 person-hours, with an estimated personnel cost of $5 038 218.28 (2022 USD) plus an additional $602 730.66 in vendor fees. Claims-based (96 metrics; $37 553.58 per metric per year) and chart-abstracted (26 metrics; $33 871.30 per metric per year) metrics used the most resources per metric, while electronic metrics consumed far less (4 metrics; $1901.58 per metric per year). Conclusions and Relevance: Significant resources are expended exclusively for quality reporting, and some methods of quality assessment are far more expensive than others. Claims-based metrics were unexpectedly found to be the most resource intensive of all metric types. Policy makers should consider reducing the number of metrics and shifting to electronic metrics, when possible, to optimize resources spent in the overall pursuit of higher quality.

Profile of the Postoperative Care Provided for Patients With Pancreatic and Periampullary Cancers by Family and Unpaid Caregivers.

PURPOSE: Patients with pancreatic and periampullary cancers may experience significant reduction in their quality of life and often rely on family and unpaid caregivers for assistance after surgery. However, as caregivers are not systematically identified, little is known about the nature, difficulty, and personal demands of assistance they provide. We aim to assess the frequency and difficulty of specific assistance caregivers provide and identify potential interventions that could alleviate the caregiving demands. METHODS: This was a prospective, multi-institutional study of caregivers accompanying patients with periampullary and pancreatic cancer at their 1-month postpancreatectomy office visit. An instrument that drew heavily on the National Study of Caregiving was administered to caregivers. RESULTS: Of 240 caregivers, more than half (58.3%) of caregivers were the patients' spouse, a quarter (25.8%) were daughters or sons, 12.9% other relatives, and 2.9% nonrelatives. Caregivers least frequently provided assistance with transportation (14.6% every day) and most frequently provided assistance with housework (65.0% every day, P = .003) and diet (56.5% every day, P = .004). Caregivers reported the least difficulty helping patients with exercise (1.5% somewhat difficult). Caregivers reported significantly more difficulty with assisting with housework (14.5% somewhat difficult, P < .001) and diet (14.9% somewhat difficult, P < .001). Caregivers identified the immediate postpancreatectomy and early discharge periods as the most stressful phases. They also reported having received very little information about available services that could have supported their efforts. CONCLUSION: Caregivers of patients with periampullary or pancreatic cancer provide considerable assistance in the postoperative period and many reported difficulty in assisting with housework and diet. Work is needed to better prepare and support caregivers to better enable them to adequately care for patients with pancreas and periampullary cancer.

Patient interpretations vary for questions in the Patient-Reported Outcomes Measurement Information System Upper Extremity.

The Patient-Reported Outcomes Measurement Information System Upper Extremity (PROMIS-UE) is frequently used in research/clinical care, but how it is understood by patients is unknown. We conducted a qualitative study including 12 cognitive interviews with patients with hand/upper extremity conditions and those purposively sampled for mixed literacy. Using framework analysis, we identified six themes total: difficulty answering questions due to insufficient information; uncertainty about whether to respond with the ability to perform the task with the injured extremity alone, either the injured or healthy extremity, or both; lack of experience doing certain tasks; uncertainty about whether to answer questions based on ability to perform the task with or without adaptive techniques; answering questions based on limitations not related to upper extremity function; and uncertainty regarding whether to answer questions on the basis of ability or pain. This study demonstrates the challenges in completing questionnaires and that variability may limit the reliability, validity and responsiveness of the PROMIS-UE.

Helping without harm: providing emotional support to health care workers in 2023.

Reflecting on the efforts to provide acute emotional support to health care workers (HCWs) before and after the COVID­ 19 pandemic, 3 guiding principles are proposed for health care organizations, with the aim to support their workers by an efficient combination of disciplines and resources: 1) normalize the use of support resources for HCWs; 2) assess actual needs rather than act on assumptions; 3) reduce barriers for HCWs to get the support they need. Each of these principles is described in terms of their usefulness and potential for further developments that might provide better emotional support for HCW in the future.

Responsiveness of the Patient-Reported Outcome Measure for Vascular Malformation Questionnaire in Patients with Low-Flow Vascular Malformations.

PURPOSE: To assess the responsiveness, defined as the ability to detect change in a patient's health or function, of the Patient-Reported Outcome Measure for Vascular Malformation (PROVAM) questionnaire in a cohort of patients with low-flow vascular malformations (VMs). MATERIALS AND METHODS: PROVAM was previously developed to assess symptoms, functional limitations, and social/emotional effects experienced by patients with VMs. This is a prospective cohort study of 56 patients with venous and lymphatic VMs who completed at least 2 PROVAM questionnaires, of whom 43 had undergone treatment with sclerotherapy in the interim between questionnaires. External responsiveness was assessed using a receiver operating characteristic (ROC) curve to ascertain whether a change in the total PROVAM score predicts whether patients reported symptom improvement and by correlating the change in the total PROVAM score and change in symptoms reported during clinic visit. Internal responsiveness was evaluated using Wilcoxon signed rank test, Cohen d effect size (ESp), and standard response mean difference (SRM). RESULTS: The total PROVAM score demonstrated excellent discrimination for symptom improvement with an area under the ROC curve of 0.856. There was a statistically significant, moderate positive correlation between the change in the total PROVAM score and the change in patient symptoms as determined from clinical visits (Spearman correlation coefficient [rs] = 0.67, P < .001). The total PROVAM score and all subdomain scores improved significantly after treatment (all P < .05). ESp and SRM were 0.80 and 0.83, respectively. CONCLUSIONS: PROVAM is responsive to improvement after treatment and may be useful to assess health-related quality of life in patients treated for VMs.

An Evidence and Consensus-Based Definition of Second Victim: A Strategic Topic in Healthcare Quality, Patient Safety, Person-Centeredness and Human Resource Management.

The concept of second victims (SV) was introduced 20 years ago to draw attention to healthcare professionals involved in patient safety incidents. The objective of this paper is to advance the theoretical conceptualization and to develop a common definition. A literature search was performed in Medline, EMBASE and CINAHL (October 2010 to November 2020). The description of SV was extracted regarding three concepts: (1) involved persons, (2) content of action and (3) impact. Based on these concepts, a definition was proposed and discussed within the ERNST-COST consortium in 2021 and 2022. An international group of experts finalized the definition. In total, 83 publications were reviewed. Based on expert consensus, a second victim was defined as: "Any health care worker, directly or indirectly involved in an unanticipated adverse patient event, unintentional healthcare error, or patient injury and who becomes victimized in the sense that they are also negatively impacted". The proposed definition can be used to help to reduce the impact of incidents on both healthcare professionals and organizations, thereby indirectly improve healthcare quality, patient safety, person-centeredness and human resource management.

Veterinary healthcare needs to talk more about error: For the wellbeing of our patients and medical teams.

Over the past 2 decades, patient safety has become an established priority in human healthcare. There is a large body of research in human medicine on harm caused by healthcare, its impact, and interventions to prevent it. There are also numerous guidelines, policies, and regulations to improve safety. An important realization has been that the same errors that harm patients can also harm members of the healthcare team. Empathetic handling of safety incidents can have positive effects on both the wellbeing of providers and their care of patients. An essential element in patient safety is the creation of a "culture of safety" within the health care team. A strong culture of safety describes a work environment where risk is acknowledged, individuals can report errors without fear of punishment, and the organization has a commitment to collaboratively implementing system changes to prevent future errors. A key element of safety culture is ensuring that healthcare team members are supported and asked to help create solutions for safer care. The principles of safety science and practices to improve safety have not yet been widely adopted in veterinary medicine. We describe a case of a serious medication error and how it was handled to illustrate key components of a culture of safety and a system-based approach to improvement. This case is timely as a recent review of patient safety events in 3 veterinary hospitals found medication-related errors to be the most frequently reported events. Open conversations about safety events and errors that can harm not only our patients but also our healthcare teams will help veterinary professionals learn from their mistakes, support members of the team, and prevent future harm.

Hand hygiene improvement of individual healthcare workers: results of the multicentre PROHIBIT study.

BACKGROUND: Traditionally, hand hygiene (HH) interventions do not identify the observed healthcare workers (HWCs) and therefore, reflect HH compliance only at population level. Intensive care units (ICUs) in seven European hospitals participating in the "Prevention of Hospital Infections by Intervention and Training" (PROHIBIT) study provided individual HH compliance levels. We analysed these to understand the determinants and dynamics of individual change in relation to the overall intervention effect. METHODS: We included HCWs who contributed at least two observation sessions before and after intervention. Improving, non-changing, and worsening HCWs were defined with a threshold of 20% compliance change. We used multivariable linear regression and spearman's rank correlation to estimate determinants for the individual response to the intervention and correlation to overall change. Swarm graphs visualized ICU-specific patterns. RESULTS: In total 280 HCWs contributed 17,748 HH opportunities during 2677 observation sessions. Overall, pooled HH compliance increased from 43.1 to 58.7%. The proportion of improving HCWs ranged from 33 to 95% among ICUs. The median HH increase per improving HCW ranged from 16 to 34 percentage points. ICU wide improvement correlated significantly with both the proportion of improving HCWs (ρ = 0.82 [95% CI 0.18-0.97], and their median HH increase (ρ = 0.79 [0.08-0.97]). Multilevel regression demonstrated that individual improvement was significantly associated with nurse profession, lower activity index, higher nurse-to-patient ratio, and lower baseline compliance. CONCLUSIONS: Both the proportion of improving HCWs and their median individual improvement differed substantially among ICUs but correlated with the ICUs' overall HH improvement. With comparable overall means the range in individual HH varied considerably between some hospitals, implying different transmission risks. Greater insight into improvement dynamics might help to design more effective HH interventions in the future.

Assessment of Caregivers' Burden When Caring for Patients With Pancreatic and Periampullary Cancer.

BACKGROUND: Family and other unpaid caregivers play an active role in the recovery of individuals with pancreatic and periampullary cancer after pancreatectomy. However, little is known about caregivers' experiences and how to better support them. METHODS: Caregivers accompanying patients to 1-month postpancreatectomy visits at 3 hospitals completed an electronic survey between November 2018 and February 2020. We examine measures of absenteeism and work productivity loss among the subset of caregivers who reported working for pay and comparatively assess caregiver experiences by employment status. All analyses were performed as 2-sided tests. RESULTS: Of 265 caregivers approached for study participation, 240 (90.6%) enrolled. Caregivers were primarily female (70.8% female, 29.2% male) and spouses (58.3%) or adult children (25.8%) of patients, with a median age of 60 years. Of the 240 caregivers included in the study, 107 (44.6%) worked for pay. Nearly half (44.4%) of working caregivers reported being absent from work because of caregiving amounting to a 14% loss in work hours. While at work, 58.9% of working caregivers reported increased work difficulty as a result of caregiving. Taken together, an estimated 59.7% loss in work productivity was experienced because of caregiving in the month following pancreatectomy. After adjustment for sociodemographic factors, working (vs nonworking) caregivers reported increased financial (odds ratio [OR] = 2.32; P = .04) and emotional (OR = 1.93; P = .04) difficulties and daily activity restrictions (OR = 1.85; P = .048). CONCLUSIONS: Working caregivers of patients with pancreatic and periampullary cancer experience negative impacts on work and productivity, and caregiving-related financial and emotional difficulties may be amplified. This study highlights the need for workplace policies to support unpaid cancer caregiving.

Effect of the STRIDE fall injury prevention intervention on falls, fall injuries, and health-related quality of life.

BACKGROUND: Falls are common in older adults and can lead to severe injuries. The Strategies to Reduce Injuries and Develop Confidence in Elders (STRIDE) trial cluster-randomized 86 primary care practices across 10 health systems to a multifactorial intervention to prevent fall injuries, delivered by registered nurses trained as falls care managers, or enhanced usual care. STRIDE enrolled 5451 community-dwelling older adults age ≥70 at increased fall injury risk. METHODS: We assessed fall-related outcomes via telephone interviews of participants (or proxies) every 4 months. At baseline, 12 and 24 months, we assessed health-related quality of life (HRQOL) using the EQ-5D-5L and EQ-VAS. We used Poisson models to assess intervention effects on falls, fall-related fractures, fall injuries leading to hospital admission, and fall injuries leading to medical attention. We used hierarchical longitudinal linear models to assess HRQOL. RESULTS: For recurrent event models, intervention versus control incidence rate ratios were 0.97 (95% confidence interval [CI], 0.93-1.00; p = 0.048) for falls, 0.93 (95% CI, 0.80-1.08; p = 0.337) for self-reported fractures, 0.89 (95% CI, 0.73-1.07; p = 0.205) for adjudicated fractures, 0.91 (95% CI, 0.77-1.07; p = 0.263) for falls leading to hospital admission, and 0.97 (95% CI, 0.89-1.06; p = 0.477) for falls leading to medical attention. Similar effect sizes (non-significant) were obtained for dichotomous outcomes (e.g., participants with ≥1 events). The difference in least square mean change over time in EQ-5D-5L (intervention minus control) was 0.009 (95% CI, -0.002 to 0.019; p = 0.106) at 12 months and 0.005 (95% CI, -0.006 to 0.015; p = 0.384) at 24 months. CONCLUSIONS: Across a standard set of outcomes typically reported in fall prevention studies, we observed modest improvements, one of which was statistically significant. Future work should focus on patient-, practice-, and organization-level operational strategies to increase the real-world effectiveness of interventions, and improving the ability to detect small but potentially meaningful clinical effects. CLINICALTRIALS: gov identifier: NCT02475850.

Listening to the Patient Voice Adds Value to Cancer Clinical Trials.

Randomized clinical trials are critical for evaluating the safety and efficacy of interventions in oncology and informing regulatory decisions, practice guidelines, and health policy. Patient-reported outcomes (PROs) are increasingly used in randomized trials to reflect the impact of receiving cancer therapies from the patient perspective and can inform evaluations of interventions by providing evidence that cannot be obtained or deduced from clinicians' reports or from other biomedical measures. This commentary focuses on how PROs add value to clinical trials by representing the patient voice. We employed 2 previously published descriptive frameworks (addressing how PROs are used in clinical trials and how PROs have an impact, respectively) and selected 9 clinical trial publications that illustrate the value of PROs according to the framework categories. These include 3 trials where PROs were a primary trial endpoint, 3 trials where PROs as secondary endpoints supported the primary endpoint, and 3 trials where PROs as secondary endpoints contrast the primary endpoint findings in clinically important ways. The 9 examples illustrate that PROs add valuable data to the care and treatment context by informing future patients about how they may feel and function on different treatments and by providing clinicians with evidence to support changes to clinical practice and shared decision making. Beyond the patient and clinician, PROs can enable administrators to consider the cost-effectiveness of implementing new interventions and contribute vital information to policy makers, health technology assessors, and regulators. These examples provide a strong case for the wider implementation of PROs in cancer trials.

Prospective, multi-site study of patient outcomes after implementation of the TREWS machine learning-based early warning system for sepsis.

Early recognition and treatment of sepsis are linked to improved patient outcomes. Machine learning-based early warning systems may reduce the time to recognition, but few systems have undergone clinical evaluation. In this prospective, multi-site cohort study, we examined the association between patient outcomes and provider interaction with a deployed sepsis alert system called the Targeted Real-time Early Warning System (TREWS). During the study, 590,736 patients were monitored by TREWS across five hospitals. We focused our analysis on 6,877 patients with sepsis who were identified by the alert before initiation of antibiotic therapy. Adjusting for patient presentation and severity, patients in this group whose alert was confirmed by a provider within 3 h of the alert had a reduced in-hospital mortality rate (3.3%, confidence interval (CI) 1.7, 5.1%, adjusted absolute reduction, and 18.7%, CI 9.4, 27.0%, adjusted relative reduction), organ failure and length of stay compared with patients whose alert was not confirmed by a provider within 3 h. Improvements in mortality rate (4.5%, CI 0.8, 8.3%, adjusted absolute reduction) and organ failure were larger among those patients who were additionally flagged as high risk. Our findings indicate that early warning systems have the potential to identify sepsis patients early and improve patient outcomes and that sepsis patients who would benefit the most from early treatment can be identified and prioritized at the time of the alert.