This study will compare a web-based advance directive to a standard advance directive.

Patients with GI and thoracic malignancies will be randomized to either a standard paper advance directive or the web advance directive (Ourcarewishes.org). Patients will be approached in the Cancer Center and provided the AD information to complete on their own. They will complete a survey at enrollment and a follow up survey administered by email. The primary endpoint will be proportion with new documentation in the EHR at 8 weeks. Second endpoints will include the change in satisfaction with end-of-life plans, AD acceptability, self-reported rates of completion, self-reported rate of sharing with their preferred surrogate decision-maker, proportion with new documentation of preferred decision-maker in the chart, and qualitative feedback. Patients in both groups will receive follow up reminder emails encouraging them to complete the advance directive.

Inclusion Criteria (any of the following)

• Pancreatic cancer, any stage except stage 0 (C25),
• NSCLC, stage IIIA or higher (C34),
• SCLC, any stage (C34),
• Colon cancer, stage IIIB or greater (C18),
• Rectal cancer, stage IIB or greater (C21),
• Esophageal cancer (adeno or squamous), any stage (C15),
• Stomach cancer, stage IIB or greater (C16),
• Gallbladder cancer, stage I or greater (C23),
• Cholangiocarcinoma, any stage (C22.1),
• HCC, any stage (no plan for transplant) (C22.0),
• Malignant carcinoid of GI tract, stage IV (C7A.019).

Exclusion criteria:

• Does not have an active patient portal (MyPennMedicine) account,
• Does not use email at least once per month,
• Has already completed a living will,
• Does not speak fluent English,
• Unwilling to be contacted by email,
• Determined to be too ill or inappropriate for participation by patient's oncologist.

Patients will be blinded to the comparison being made, but not to the intervention in their arm.