Using a cluster randomized design at the clinic level, this project will implement and test three real-world, scalable advance care planning interventions among primary care clinics across three University of California health systems. Seriously ill patients identified using data from the electronic health record will receive (1) an advance directive with targeted messaging, (2) intervention 1 plus prompting to engage with the Prepare For Your Care website, or (3) intervention 2 plus engagement from a clinic-based facilitator. A Research cohort of patients will provide complete surveys at baseline, 12 and 24 months. The main outcomes are advance directive completion among the population cohort and goal concordant care among the Research cohort at 12 months.

The goal of the project is to test, implement, and disseminate real-world, scalable advance care planning (ACP) interventions among busy, primary care clinics across three UC Health systems. The study aims are as follows:

Aim 1: With the input of primary care stakeholders (i.e., adults with serious illness, caregivers and clinical staff) at each site, determine the barriers, facilitators and best workflow plans to implement the three ACP interventions.

Aim 2: Implement the three ACP interventions across primary care clinics using a cluster randomized design across three UC Health systems: distribution of an advance directive (AD) (arm 1), distribution of the AD plus prompting the patient to engage with the Prepare For Your Care website (arm 2), and the AD plus engagement with the Prepare For Your Care website plus a Care coordinator intervention to engage clinicians (arm 3).

Aim 3: Compare the effectiveness of the three interventions on documentation of ACP (represented by an AD, POLST or out-of-hospital DNR form available in the medical record) on a population-based primary care cohort of seriously ill patients across three UC Health systems, and assess ACP engagement, quality of communication and goal-concordant care among a Research cohort subset of these patients.

The investigators hypothesize that there will be a dose response of additive improvement from arm 1 to arm 2 to arm 3 on AD completion (primary outcome) at the population level and receipt of goal-concordant treatment at the Research cohort level.

Seriously ill patients commonly receive treatments at the end of life that are inconsistent with or not guided by their values and goals. People with serious illness bear a considerable burden of suffering, and complexity of illness is strongly associated with increased hospitalization and healthcare utilization. Families of patients with serious illness who die often sustain substantial financial and emotional injury. Yet, this is not an irremediable result of advanced technology and an enlarging older population: ACP - the process of discussion of end of life care, clarification of values and goals, and embodiment of preferences through written documents and medical orders - is associated with less aggressive medical care near death and earlier hospice referral, which is then associated with better quality of life for patients and better bereavement adjustment for loved ones.

Experimental evidence supports the premise that ACP improves the treatment choices made by and for people with serious illness. A systematic review concluded that ACP interventions increase completion of ADs, increase the frequency of discussions about care preferences and increase concordance between patient's preferences and end-of-life care received. An earlier review found that multicomponent interventions were more likely to result in AD completion, yet much of this work focused on hospitalized patients and patients with cancer. A systematic review of the effects of 113 ACP interventions found decreased use of life-sustaining treatment and increased hospice and palliative care, but concluded that more experimental designs and community studies are needed. Many of these reviews noted a lack of head-to-head comparison of ACP interventions and a need for structures to implement ACP in standard care. The Institute of Medicine IOM critiqued studied ACP interventions for being targeted as one-time decisions rather than being conceptualized as incorporated into the routine of care.

A fundamental gap exists in whether and how ACP interventions can be implemented at a healthcare system level. Although ACP has shown the potential to help patients and clinicians achieve the "triple aim" of improving quality of care and patient satisfaction, improving population health and lowering costs by honoring patients' wishes, it has been slow to be prioritized in healthcare systems. Even among seriously ill patients, ACP remains uncommon. Although quality and efficiency are increasingly emphasized and several health systems are in early phases of adopting programs to improve ACP, few programs have successfully implemented ACP across their healthcare system. Implementing a system-wide program among a diverse patient population with primary care physicians (PCPs) who are already overburdened with the complex management of seriously ill patients remains a challenge.

This project will implement three system-wide advance care planning interventions and compare their effect. This quality improvement project will be implemented at the level of the health system with no direct patient contact and collection of only deidentified data for evaluation of the effect of the interventions. A second project nested in the population-based quality improvement project will enroll and evaluate a Research cohort that will provide patient and caregiver level information about the effect of the three ACP interventions. These components will be conducted in an identical fashion at UCLA, UCSF and UCI.

Using a computer algorithm developed and validated for this project, data from the electronic health record will be used to delineate a group of seriously ill patients cared for in primary care that require ACP and also identify whether these patients have ADs in the electronic health record in the past 3 years. All patients receiving primary care in the health system are eligible to be identified as seriously ill patients.

The project will implement 3 real-world ACP intervention comparators using a clustered randomized trial design. The interventions are: (arm 1) distribution of an AD with a tailored message via MyChart in the electronic health record and/or postal mail, (arm 2) arm 1 plus access to the Prepare For Your Care website that is a publicly available decision support tool for ACP, and (arm 3) arm 2 plus patient engagement by a clinic-based facilitator who will tee up AD completion, Prepare website use and coordinate patient ACP interactions with primary care physicians. The interventions will continue over a 24-month period, with an evaluation at 12 and 24 months.

Comparison of the 3 intervention arms will occur at the population level by collecting deidentified data from the electronic health record reflecting AD completion (existence of an AD or POLST or outpatient DNR order and the date of the document), vital status, and healthcare utilization (hospitalization, ICU use, ED use, and location of death).

A Research cohort will be enrolled from the seriously ill population-based cohort that do not have an AD or POLST within the past three years by asking patients via mail, telephone, or in person whether they are willing to participate in an observational survey study. Interested patients will be asked to provide written informed consent to participate in up to 3 surveys: baseline, 12 months and 24 months. The surveys may be completed on a paper-and-pencil hard copy, completed by interview over the telephone, completed in person or completed over a secure web link, depending on patient preference. The surveys will ask about experience with and readiness for ACP, health status, preferences for healthcare and decision making, and demographics. If a patient is unable to be surveyed at month 12 or month 24, a caregiver will be approached for survey. If the patient dies, a caregiver will be asked to complete a survey about goal-concordant treatment, perceptions about care and satisfaction with care and communication. Patients will also be asked to consent to medical record review to evaluate documentation about ACP and measure goal concordant care. The principle outcome is AD completion at the population level using administrative data from the electronic health record.

Population cohort

Inclusion Criteria

• At least 2 visits to a primary care clinic in the health system in the past 12 months
• Serious illness according to an algorithm using electronic health record data including patients with: advanced cancer, advanced heart failure, advanced chronic obstructive pulmonary disease, decompensated liver disease, end stage renal disease, amyotrophic lateral sclerosis, or was a vulnerable elder with a serious illness

Exclusion Criteria: None.

Research cohort: Derived from the population cohort plus:

Inclusion Criteria

• Speak English or Spanish
• Provide informed consent

Exclusion Criteria

• Too cognitively impaired to provide informed consent
• Primary care physician identifies that survey may be harmful

Statistician blinded to study arm