Analysis of Emergency Pancreatic Cancer Diagnoses to Inform Interventions to Reduce Diagnostic Delays

Start Date Nov 2021

Code B3-Aff

Status Ongoing

Project Lead
Natalia Khalaf
Senior Lead


Pancreatic cancer is increasing in incidence worldwide and is currently the number five cancer killer in the U.K. and number three cancer killer in the U.S. Although cancer-related deaths from most cancer types (lung, breast, colorectal, prostate) have fallen over the past several decades due to better prevention, diagnosis and treatment, pancreatic cancer has remained a major cause of cancer-related death, and additional study of this disease is needed. Nearly 85% of patients with pancreatic cancer have incurable (locally advanced or metastatic) disease at diagnosis, suggesting a need for systematic efforts to improve the timing of pancreatic cancer diagnosis. Our study team has found that 58% of U.S. Veterans with pancreatic cancer are diagnosed with stage IV disease, 77% are diagnosed while admitted to the hospital, and 28% report their first cancer-associated sign or symptom in the emergency department setting, highlighting a knowledge gap for which further study is needed.


Veterans Affairs Health Services Research & Development Career – $1,090,647

Aims & objectives

Aim 1: To identify determinants of emergency cancer diagnosis among Veterans with pancreatic cancer.

Aim 2: To identify disparities in emergency cancer diagnosis among Veterans with pancreatic cancer (e.g., race, age, education level, distance from VA, comorbid states).

Aim 3: To compare cancer-associated outcomes (stage, receipt of treatment, survival) between pancreatic cancer patients with and without emergency presentations.


Using a retrospective cohort study design, we will extract various patient-, cancer-, and health-systems-related variables from the electronic medical record among an already established cohort of 265 Veterans diagnosed with pancreatic adenocarcinoma from 2007-2019 at the Houston Veterans Affairs Medical Center. Inclusion criteria: all patients diagnosed with incident pancreatic adenocarcinoma (confirmed by histology review) from 2007 to 2019. Exclusion criteria: no primary care encounter within the 2 years prior to cancer diagnosis. Our primary outcome of interest is ECD, which is a cancer diagnosis preceded within 30 days by an emergency event (defined as an unplanned hospitalization or emergency department visit).

We will use chi-squared or Fisher’s exact tests, as appropriate, to evaluate the differences in predictor variables between pancreatic cancer patients with and without ECDs (Aim 1). We will also compare cancer-associated outcomes of stage, receipt of treatment and survival between patients with and without ECDs (Aim 3). Multivariable logistic regression will be used to investigate the strength of association between diagnosis characteristics and sociodemographic and medical factors and ECD incidence (Aim 2).

Outputs & impact

In previous work, we found 40% of Veterans with pancreatic cancer had missed opportunities in diagnosis, and ECDs represent an important subset of cases. Enhanced understanding of the process breakdowns that lead to ECDs are needed to inform meaningful interventions. As the largest provider of cancer care in the U.S., the Veterans Affairs Health Administration provides an ideal setting for developing and implementing pathways to optimize diagnostic processes in cancer care.

Next steps

This work can lead to development of tailored interventions and potential strategies to shorten diagnostic intervals, missed opportunities in diagnosis, and emergency presentations among pancreatic cancer patients, thus improving overall quality of care and timing of cancer diagnosis.

Back to Research