Project to Improve Communication About Serious Illness--Hospital Study: Pragmatic Trial (Trial 1)

OVERVIEW: This is a large pragmatic trial of the Jumpstart intervention designed to promote goals-of-care discussions for older, seriously ill, hospitalized patients. The trial recruits consecutively eligible patients from three UW Medicine hospitals. The Jumpstart Guide is a communication-priming intervention that addresses hospitalized patients' goals of care. It includes information, drawn from the electronic health record (EHR), identifying the dates and locations of prior advance care planning documents (e.g. living wills, healthcare directives, durable power of attorney for healthcare, and Physician Orders for Life Sustaining Treatments (POLST)) and patients' code status. This information is provided by email to patients' clinicians. The Jumpstart Guide includes tips to improve this communication. The trial will assess the effectiveness of the EHR-based clinician Jumpstart as compared with usual care.

This current study is "Trial 1" of the R01 Award funding this trial. Trial 2 was initiated following the completion of recruitment for Trial 1.

SPECIFIC AIM 1 (for Trial 1): Evaluate the efficacy of the EHR-based clinician Jumpstart compared to usual care for improving quality of care.

TRIAL 1 has three components.

Component 1- Subject Identification/Recruitment/Randomization: The investigators will use automated methods with EHR data, including natural language processing/machine learning (NLP/ML) approaches, to identify hospitalized patients with serious illness during the first 2-3 days of this specific admission. Screening reports will be produced daily and include all likely eligible patients. Study staff will use these daily screening reports to review individual records of eligible patients for inclusion and exclusion criteria (i.e. verify eligibility). Patients will not be approached for consent. Eligible patients will be assigned to intervention or comparator in a 1:1 ratio. Patients are randomized using variable size blocks and stratified for hospital and ADRD vs. no ADRD.

Component 2- EHR-based Clinician Jumpstart Guide: The Jumpstart guide is developed by applying NLP/ML algorithms to both inpatient and outpatient EHR notes (e.g., progress notes, specialty consult notes, alerts and care plans) preceding the current hospitalization. It summarizes the presence/absence of POLST, advance directives and DPOA (durable power of attorney) documentation and the patients' code status. It also provides general recommendations to initiating goals of care discussions. Jumpstart guides are prepared for the intervention group only.

Component 3- Delivery of the intervention: For the intervention group, the investigators deliver the Jumpstart guide to the primary hospital team (attending and resident physicians and advanced practice providers) via secure email. The Jumpstart guides are delivered within 1 business day of patient randomization. Study staff monitor the care team of the patient, and if there are any changes, ensure that any new providers receive the Jumpstart guide as well.

Comparator: The hospital teams for patients in the control group do not receive Jumpstart guides.

Outcome Assessment. Outcomes are obtained from the EHR and use, in part, NLP/ML methods. The primary outcome is EHR documentation of goals-of-care discussions in the 30 days following randomization. Secondary outcomes are measures of intensity of care, including utilization metrics (i.e., ICU admissions, ICU and hospital lengths of stay, 30-day hospital readmissions), costs of care during hospital admission, and estimated costs of implementing the intervention.

SPECIFIC AIM 2 (Trial 1): Evaluate implementation of the Jumpstart Guide and identify barriers and facilitators to future implementation.

Qualitative evaluation:

Clinician recruitment: Study staff will recruit clinicians who were involved with the study to participate in a short interview after the clinician's study involvement with the enrolled patient has ended. Clinician participants will be selected using purposive sampling to ensure a diverse group (e.g., age, race/ethnicity, gender, specialty, year of training).

Interview: Using an interview guide developed specifically for this project, interviewers will assess respondents' experience with the intervention and gather suggestions for ways to improve the intervention's content, delivery and implementation, including implementation outcomes (e.g., acceptability, fidelity, penetration, maintenance) that will guide future dissemination of the intervention.

Assessment: Interviews are audio recorded, transcribed, and analyzed using thematic analytic methods.

Quantitative evaluation:

Quantitative evaluation for Specific Aim 2 will be guided by the RE-AIM framework. We will evaluate the intervention's: 1) Reach (% of all identified eligible clinicians, patients and families that participated in the study); 2) Adoption (% of eligible services that participated); and 3) Implementation (the proportion of HTML Jumpstart Guides that were opened by a clinician, proportion of patients for whom an HTML JS guide was sent and for whom a JS was opened by a clinician).