Effect of Behavioral Nudges on Serious Illness Conversation Documentation

Patients with cancer often experience physical and emotional distress, utilize unplanned acute care, and undergo medical interventions that are discordant with their wishes. Given the Covid-19 pandemic, these adverse outcomes are amplified, particularly for racial/ethnic minorities. Serious illness conversations (SICs) that elicit patients' values, goals, and care preferences, particularly early in the disease trajectory, are an evidence-based practice, improve patient mood and quality of life, and are recommended by national guidelines. Preliminary data suggests that SICs among patients with cancer are associated with improved quality of life, increased hospice utilization, and decreased acute care utilization. However, most patients with advanced cancer die without a documented SIC and there are well-documented health disparities in implementation for racial and ethnic minorities. Current strategies to promote SICs, including the multi-component strategies of the Serious Illness Conversation Program, focus primarily on clinician education and have marginally increased the timeliness and frequency of SICs and reduced patient anxiety and depression. While core elements of this program are transferable-such as its structured guide-clinical use remains low. For example, even after training, clinicians at Penn Medicine document SICs for fewer than 5% of patients with advanced cancer. There is critical need to develop, test, and disseminate strategies to improve the frequency of SICs.

Implementation strategies informed by behavioral economics are ideally suited to address this problem, which is fundamentally one of clinician and patient behavior change. Clinician barriers to initiating SICs include optimism bias, or the belief that one's own patient is unlikely to experience a negative event; uncertainty about prognosis and appropriate timing; and fear that bringing up end-of-life issues may be distressing to patients. Patient barriers to SIC initiation include fear of discussing the end of life and beliefs that SICs are not appropriate until late in the course of cancer. While previous studies have tested financial incentives for SIC documentation, little research has evaluated behavioral economics-informed strategies to align both clinicians and patients towards earlier SICs.

By intentionally modifying the way choices are framed, behavioral nudges can lead to desirable changes in clinician behavior while preserving clinician choice. The investigators' preliminary work demonstrates the effectiveness of an implementation strategy focusing on a clinician nudge, consisting of performance feedback and targeted text messages identifying patients at high risk of mortality based on a validated machine learning prognostic algorithm. This strategy led to a threefold increase in SIC documentation for high-risk patients, equitably across racial/ethnic minority subgroups, and is now in routine use across Penn Medicine practice sites. However, clinicians still did not document SICs for over half of patients, illustrating the limitations of a clinician-directed implementation strategy alone.

This study will expand on these preliminary findings to evaluate the synergy between clinician- and patient-directed nudges to increase SIC documentation. The main purpose of this research study is to evaluate the effectiveness of nudges to clinicians, to patients, or to both in increasing Serious Illness Conversation (SIC) documentation; and to identify moderators of implementation effects on SIC documentation. The investigators will employ rapid-cycle approaches to optimize the framing of nudges to clinicians and patients prior to initiating the trial and mixed methods to explore contextual factors and mechanisms. The investigators will conduct a four-arm pragmatic cluster randomize clinical trial to test the effectiveness of nudges to clinicians, nudges to patients, or nudges to both in increasing the frequency and timeliness of SIC documentation in cancer patients vs. usual care (UC).

Rationale for clinician nudge using mortality prediction and peer comparison: Due to optimism bias, clinicians routinely overestimate the life expectancy of patients with advanced cancer and delay SICs until too late in the disease course. In part because of this, clinicians reinforce a social norm that early SICs are not part of routine oncology care. Providing an objective assessment of predicted mortality risk may help counteract optimism bias among clinicians and help them identify patients most likely to benefit from SICs. Moreover, that individuals desire to conform to an approved behavior (an injunctive norm) and the behavior of others (a descriptive norm) may contribute to low observed SIC rates, and may also afford an opportunity for intervention. The investigators expect that periodically reminding clinicians of their own performance on SIC documentation, while providing both an injunctive norm (citing national and institutional guidelines) and a descriptive social norm (displaying the behavior of their best performing peers), will lead clinicians to conform more closely to these norms, as has been shown in studies conducted in other contexts.

Rationale for patient nudge using priming: Priming is a type of nudge that frames information to activate one's self-efficacy and willingness to engage in behavior change. This type of nudge has not previously been evaluated as a tool to promote SICs for patients with cancer. The investigators will test the added impact of a patient nudge designed to prime patients and, in turn, their clinicians to having a SIC.

Inclusion criteria

Clinician (M.D., P.A., or N.P.) participants must meet the following criteria:

• Provide care at least 1 clinic session per week for adult (age>18 years) patients with solid, hematologic, or gynecologic malignancies at a participating PennMedicine Implementation Lab site

Patient participants must meet the following criteria:

• Receive care for a solid, hematologic, or gynecologic malignancy from an eligible provider at a participating PennMedicine Implementation Lab site
• Have at least one scheduled Return Patient Visit (either in person or via telemedicine) with an eligible PennMedicine provider during the study period

Exclusion criteria

Clinicians will be ineligible for *any* of the following reasons:

• Provide exclusively benign hematology, survivorship, and/or genetics care

Patients will be ineligible for *any* of the following reasons:

• Previously documented SIC within 6 months
• Have a non-valid phone number