Objectives:

Using Kaiser Permanente Colorado’s (KPCO) established patient portal, My Health Manager (MHM), we conducted a secondary data analysis of the effects of portal utilization on important palliative care and end of life (PCEOL) outcomes. The objectives of this project include:

• Describe MHM utilization patterns and user characteristics among deceased patients and their caregivers (via a proxy log-in option) in the last 12 months of life.
• Conduct exploratory analysis to test the association between MHM utilization among deceased patients in the last 12 months of life and their proxy caregivers with important PCEOL outcomes including advance directive documentation, hospitalization near the end of life, and hospice referral/use. 

Methodology:

Design: Retrospective cohort study to investigate the utility of portals during PCEOL.

Sample: Deceased patient and caregiver MHM users over the age of 18 between 1/1/2016-6/30/2019. Patients and caregivers must have been registered for MHM during the 12 months prior to the patient’s death. Patients who did not die from a chronic or serious illness (i.e. accident) were excluded from analysis to accurately capture patient portal use at end of life.

Data Extraction: Data was extracted from KPCO’s integrated electronic medical record, KP HealthConnect—Epic Systems, Death Records—Research Data Warehouse, Health Trac and My Health Manager databases. Data extracted included patient characteristics (age, race, ethnicity, gender, diagnoses, rural/urban location, and socioeconomic status), Care Group flags (Advanced Illness, Serious Illness, Chronic Illness), date of flag, enrollment data, MHM utilization (number of log-ons, total time on site, and features used—Table 1), and PCEOL outcomes (advance directive documentation, hospitalization near end of life, and hospice referral/use)

Analysis: After cleaning the data and addressing data issues and concerns, univariate descriptive analysis was conducted. Frequencies, measures of central tendency, and percentages were used to create summary tables of utilization patterns for the number of log-ons and used MHM features per month, over the 12-month period as the patient nears death. Utilization differences by patient characteristics were tested by t-test and ANOVA as appropriate. We conducted exploratory analyses to examine associations between patient and caregiver proxy MHM utilization with PCEOL outcomes (advance directive documentation, hospitalization near end of life, and hospice referral/use).