Background: 22% of US patients with cancer die in a hospital setting. As part of an effort to reduce unexpected inpatient (inpt) mortality, we reviewed records of all inpt cancer deaths at Stanford Hospital and reported findings to the treatment teams.

Methods: Deaths with a cancer related ICD 9/10 code between 5/2017 and 6/2019 were reviewed by a multidisciplinary team. Findings and potential opportunities for improvement were communicated to the pt’s primary outpt oncologist, inpt oncologists and other involved providers. Observed to expected (O:E) mortality for the year prior to the intervention (5/2016–4/2017), Year 1 (5/2017–4/2018) and Year 2 (5/2018–4/2019) of the intervention were compared with two sided t test, α=0.05 (Vizient Inc, Irving TX). Changes in supportive care utilization and end of life care between cases reviewed in Year 1 and Year 2 were compared with chi square analysis.

Results: There were 236 inpatient deaths reviewed. The median age was 64 years; 76% had solid tumors; 68% had metastatic disease; 33% had a previous inpt admission; 34% received chemotherapy in the last 2 weeks of life. Median length of stay was 7 days and 37% were admitted to the intensive care unit (ICU). The O:E mortality ratio significantly decreased between the year prior to intervention and Year 2 (0.95 vs. 0.69; p = .019), and Years 1 and 2 (0.90 vs. 0.69; p = .003). There was no noted difference in number of palliative care consults or resuscitation status at the time of death between Years 1 and 2. There was an increased frequency of advance care plan documentation on admission in Year 2 (p = .007).

Conclusions: Cancer pts who die in the hospital have high rates of recent hospitalizations, chemotherapy/radiation use in the last 2 weeks of life and ICU admissions. Decrease in O:E is likely multifactorial. Potential factors are improved documentation of comorbidities, increased access to palliative care services, and facilitation of hospice referrals which were partially driven by results of our reviews and resulting awareness around end of life care. Work is ongoing to standardize documentation of goals of care conversations in the electronic medical record and employ lay health workers for earlier end of life discussions.