Giving Information Systematically and Transparently in Lung and GI Cancer Phase 2

Despite exciting recent advances in cancer treatments, there still, ultimately, comes a time when advanced disease progresses, and patients can reliably be expected to have months, not years, left to live. For patients with metastatic cancers studied in the investigator's Coping with Cancer NCI R01s, this comes after progression on 1st- or 2nd-line therapy -- be it chemo-, immune-, or targeted therapy. Prior studies conducted by the investigator have found that oncologists can reliably predict when patients have only months to live (e.g., remarkable agreement between oncologist estimates of months to live shared with patients and patients' actual survival of months). By contrast, patients appear largely unaware of their prognosis. For example, 5% of patients a median of 5 months from death, accurately understood they had incurable, late/end-stage, terminal cancer, and likely only months to live. Dying cancer patients appear to lack the prognostic understanding needed to make informed choices.

Patients who grasp that they are dying (e.g., the 8.6% who "get the gist" that they likely have months to live), relative to those who do not, have been shown to have: a) higher rates of advance care planning (ACP), b) receive less burdensome, unbeneficial care (e.g., fewer intensive care unit, ICU, stays, less cardiopulmonary resuscitation, CPR), and c) more value-consistent care. The investigator has found that patient prognostic understanding is improved by oncologist discussions of life-expectancy, but despite 71% of patients wanting to discuss prognosis with their oncologists (83% adult cancer patients thought prognostic information was extremely/very important), only 17.6% of cancer patients within months of death reported that they had discussed prognosis with their oncologist. Not only do oncologists appear to discuss prognosis less than patients want them to, but even when prognostic discussions do occur, the investigator has found that some approaches (e.g., matter-of-fact) are more effective than others (e.g., vague) for promoting patients' prognostic understanding. Thus, prior work identifies a need to improve communication to promote patient prognostic understanding in a way that oncologists will likely learn, accept, use, and possibly implement more broadly in clinical practice.

To address this need, the investigator developed the "Giving Information Simply \&Transparently" (GIST), Oncolo-GIST intervention -- a manualized oncologist communication intervention that simplifies how to impart prognostic information by focusing on 4 basic steps: 1) Giving scan information, 2) Informing prognosis, 3) Strategizing sensitively, and 4) Transparently asking what the patient heard. Unlike traditional emphasis on numerical or medical details, the Oncolo-GIST approach is based on Reyna's Fuzzy-Trace Theory of decision-making, which emphasizes the need for an understanding of the bottom-line gist of a situation. The Oncolo-GIST approach distills prognostic discussions to clear communication of end-of-life (EoL) decision-making essentials (e.g., life-expectancy). 3 specific aims of the Oncolo-GIST approach will be tested in 2 phases:

Phase 1 will consist of two parts: 1) An interview of key stakeholders/key informants regarding Oncolo-GIST Version 1.0 in order to inform refinements to produce Oncolo-GIST Version 2.0. 2) An open trial of Oncolo-GIST Version 1.0 to inform refinements to produce Oncolo-GIST Version 2.0.

Phase 2 will involve a cluster randomized controlled trial (RCT) of Oncolo-GIST Version 2.0 on 50 patients with metastatic cancers worse on at least 1 line of therapy (chemo-, immune-, targeted), whose oncologists do not expect them to survive 12 months. Patients will be assessed in the week prior to their scheduled scan, within 1 week of the clinic visit in which progressive scan results are discussed, and then 2 and 4 months later to explore intervention effects on primary and secondary outcomes, respectively. Oncologists will be assessed in the week following that same clinic visit to obtain their impressions of the discussion of prognosis and the patient's prognostic understanding.

In Phase 2, for the pilot cluster RCT, the investigator will recruit (n=50) adults with metastatic GI or lung cancers with scan results that reveal progression (worsened disease) on an initial systemic treatment; that is, patients whose life-expectancy can reliably be estimated to be months, as opposed to years. Medical oncologists (n=4) who care for these patients will also be consented for study participation and half (n=2) will be randomized to receive the Oncolo-GIST training. The investigator expects 12-13 patients will be clustered within each of the 4 oncologists. Hierarchical Linear Modeling (HLM) techniques will be employed to address the non-independence of patient assessments within each cluster. Patients (n=25) will be seen by either an Oncolo-GIST trained oncologist or an oncologist not trained in the intervention; that is, usual care (n=25). Patients in both arms will have met the same eligibility criteria (i.e., have similar prognoses).

Patients will be assessed by trained research staff in the week prior to a scheduled meeting with their oncologist to discuss the scan results. This will provide patients' baseline levels of prognostic understanding and enable the investigator to determine how the intervention relates to pre-post scan visit changes in prognostic understanding. Patients will be assessed post-scan within a week of that progressive scan visit. Although not all patients are expected to die within the study observation period, given a median life expectancy of ~4-5 months from baseline, the investigator expects nearly half of the enrolled patients will die 4 months from baseline, and that the vast majority will die during the study observation period of 12 months. Thus, for all patients enrolled in this study, the medical care that they receive can reasonably be considered end-of-life care, whether they die during the study observation period or not.

The primary outcome is the patient's degree of prognostic understanding, measured using the investigator's validated 4-item assessment. The investigator will determine if the patient understood the scan results to be "worse" and their understanding of expected outcomes of treatments proffered (re: curability, survival, quality of life). Outcomes will also include whether a DNR order was completed for the patient, the McGill Quality of Life measure, performance status (e.g., Eastern Cooperative Oncology Group, ECOG), and care received (e.g., anticancer, intensive, palliative care). Treatment preferences will be assessed using the SUPPORT question regarding quality vs. quantity of life, which will be used to compare with actual EoL care received to operationalize "value-consistent" care. The investigator's validated Human Connection scale will assess therapeutic alliances from both the patient and oncologist perspective, and the investigator will assess oncologists' sense of how the scan discussion went (e.g., degree to which they think they communicated effectively, and that the patient understood them and had an accurate prognostic understanding). Demographic/background information (e.g., age, race/ethnicity, sex, education) and DNR documentation will be obtained from subject self-report at baseline and the patient's medical records. Previously validated measures will assess potential confounding influences such as time from diagnosis, prior discussions of prognosis, and health literacy using the REALM. Preferences regarding medical decision-making (e.g., an active vs. passive role in deciding the best course of treatment), patients' Religious Beliefs in EoL Care (RBEC), and the question "If your doctor knew how long you had left to live, would you want him/her to tell you?" will be assessed.

Hierarchical Linear Modeling (HLM) will be used to evaluate intervention effects. HLM is statistically appropriate because it accounts for the clustering of patients within oncologists, creating non- independence of clustered assessments. HLM will model oncologists as a random effect as has been done in prior RCTs. Baseline covariates known to affect study outcomes (e.g., patient health literacy) will be included in models to increase the precision of effect size estimates. This will provide a preliminary effect size estimate of Oncolo-GIST Version 2.0's ability to improve patients' prognostic understanding for a future, larger study. Linear and logistic regression models will estimate effects of the Oncolo-GIST intervention on secondary and exploratory outcomes.

The details for Phase 1 of the study are enumerated in a separate record marked "Giving Information Systematically and Transparently in Lung and GI Cancer Phase 1" (Oncolo-GIST P1) with NCT # NCT04158908.