Health informatics approaches to reduce missed opportunities in diagnosis of pancreatic cancer

Background. Pancreatic cancer is a highly lethal cancer with increasing incidence among Veterans. Early diagnosis offers the best chance of survival, yet only 30% of Veterans with pancreatic cancer are diagnosed at early stages. Our proposal focuses on optimizing early diagnosis of patients with clinical findings suggestive

of pancreatic cancer who are at risk for delayed diagnosis through identifying and reducing missed opportunities in diagnosis (MODs). MODs are defined as instances in which post-hoc judgement indicates that alternative decisions or actions could have led to a more timely diagnosis. Significance/Impact. This proposal is centered around the VA HSR&D priority areas of quality and

safety and health care informatics. There are few systematic efforts to improve quality of cancer care for Veterans with pancreatic cancer. This proposal is one of the first large-scale efforts to examine the gaps in care for these patients. We propose to develop electronic trigger tools that, when implemented, will have the potential to identify

possible missed opportunities to either initiate or complete the diagnostic process for earlier pancreatic cancer diagnosis. Our results will also serve as foundational for developing informatics-based diagnosis tools for several other high-risk conditions, including other cancer types. Innovation. The development of novel decision support tools (e-triggers) to reduce diagnostic delays in

pancreatic cancer represents the first e-trigger tool that is centered on pancreatic cancer care and built around

a constellation or combination of “red flags” as opposed to a single “red flag” as in all prior works on the topic. Our study is also conceptually innovative because it represent a first step towards moving e-trigger tools to reduce diagnostic delays in pancreatic cancer from the research environment into the clinical setting.

Specific Aims. Our goal is to improve the diagnostic process for pancreatic cancer by developing effective decision support tools. We have 3 aims. Aim 1: To identify missed opportunities for pancreatic cancer diagnosis. Aim 2: To develop electronic algorithms (i.e., e-trigger tools) to identify high-risk Veterans with

potential missed opportunities for pancreatic cancer diagnosis. Aim 3: To conduct a formative assessment of e- trigger tools to guide future testing and implementation in clinical practice. Methodology. In Aim 1, we will randomly select 500 pancreatic cancer cases from an already established national VA cohort for structured review of their medical records to identify the frequency and types

of MODs as well as key patient-, provider-, and system-specific factors linked to these missed opportunities. In Aim 2, based on the findings of Aim 1 and what is known from the literature, we will develop up to 5 e-trigger tools to identify patients at high-risk for potential MODs through automated identification of specific patterns in

clinical electronic health record data. In Aim 3 we will perform formative assessment through stakeholder interviews with Veterans, health care providers, and operational personnel to identify barriers and facilitators to implementing developed e-triggers in clinical practice, as well as a host of socio-technical factors.

Implementation/Next Steps. The proposed research will develop health care informatics tools to identify opportunities for earlier pancreatic cancer diagnosis among high-risk patient subgroups who warrant diagnostic evaluation. In the next steps, we will examine the effectiveness of our informatics tools (e-trigger tools) to improve

quality of pancreatic cancer care in the VA through a hybrid type I effectiveness-implementation study. This multidisciplinary research, along with career development and mentoring plans, will increase knowledge of earlier cancer diagnosis opportunities in the VA and enhance my ability to transition to an independent health services

researcher with expertise in health care informatics. The VA is the largest provider of cancer care in the U.S., thus providing the ideal setting for developing and implementing new ways to optimize diagnostic processes in cancer care.