Author ORCID Identifier

https://orcid.org/0009-0004-0389-3518

Semester

Fall

Date of Graduation

2025

Document Type

Dissertation

Degree Type

PhD

College

School of Public Health

Department

Epidemiology

Committee Chair

Toni M. Rudisill

Committee Co-Chair

N/A

Committee Member

Christa L. Lilly

Committee Member

Danielle M. Davidov

Committee Member

Courtney Pilkerton

Abstract

ABSTRACT

Pancreatic Cancer in Appalachia: An Examination of Detection, Associated Factors, and Provider Experiences with a Focus on West Virginia

Melissa M. Elder

Introduction: The Appalachian population accounts for a substantial proportion of the United States and are uniquely susceptible to health care disparities. The Appalachian Regional Commission has reported on the inequities found within the Appalachian region as part of its formal establishment in 1965. When compared to the Non-Appalachian U.S., Appalachia experiences a higher incidence of chronic disease such as diabetes, cancer, lung, and liver disease; mortality from diseases of despair constituting prescription drug, alcohol, illegal drug use, and suicide. Concerning rurality of this region, economic distress including higher poverty and lower income levels are also markedly higher, especially when compared to rural Non-Appalachia. Considering all-cause mortality among 15–64-year-olds, Appalachia continues to surpass Non-Appalachia U.S. at significantly higher rates. Surpassing the national cancer mortality rate by 17%, West Virginia (WV) experiences higher proportions of risk factors such as smoking, obesity and lower rates of breast cancer screenings likely contributing to this contrast. Consequently, more than 26 million residents in the Appalachian region are likely to experience an acute illness such as pancreatic cancer differently from the rest of the nation. However, limited published literature has investigated detection of pancreatic cancer (PC), factors associated with diagnosis, and primary care provider experiences with this disease.

Methods: This research entailed a systematic review with meta-analysis to comprehensively review and analyze cohort studies detecting PC among high-risk groups in Appalachian states, including detection methods used, and calculating the number-needed-to-screen (NNS) to detect one pancreatic adenocarcinoma. Secondly, a cross-sectional study was conducted using patients treated for pancreatic cancer from 1993 - March 2023 who were included in the West Virginia Cancer Registry to determine what factors were associated with early vs. late detection of disease. Multivariable logistic regression models were used and included variables such as rurality, age, race, gender, and tumor location, as well as first courses of treatment at the time of diagnosis. The third study employed semi-structured interviews conducted with Primary Care Providers (PCPs) from April – June 2025. Participants must have been a practicing healthcare provider within the state of WV at time of study. Participants were recruited through email, online advertisements, and social media using non-probability sampling techniques. Interviews were conducted using an IRB approved script and were recorded and transcribed. Qualitative description was performed using deductive and inductive coding to establish themes.

Results: Eight cohort studies representing 3,948 patients met the criteria for inclusion in the meta-analysis. One study was conducted at a facility located within the Appalachian region, as opposed to within an Appalachian state. Two of the six studies reported the exact diagnosis stage. Among cases of pancreatic cancer (n=555), the estimated effect of being high risk suggested harm (Prevalence = 0.11, 95% CI, 0,0.52), corresponding to a number-needed-to-screen (NNS) of 9 to identify 1 pancreatic lesion. Six studies (476 cases) used Endoscopic Ultrasound (EUS) alone or in combination with other methods. Statistically significant heterogeneity (Q = 1,508.29, p = 0.00) and large inconsistency (I2 = 99.5%, 95% CI, 99.4 to 99.6%) were observed. The 95% prediction interval (PI) was 0.00 to 0.77. Risk of bias, based on ROBINS-E ranged from some concerns to high, with moderate certainty of evidence based on the GRADE assessment.

Among the 4,687 individuals who met inclusion criteria in the cross-sectional analysis of cancer registry data, 30% were diagnosed with early-stage PC. The majority of those diagnosed with PC lived in non-rural counties at the time of diagnosis and most patients did not receive surgery (82%), radiation (89%), chemotherapy (52%), nor immunotherapy (99%) as part of the first course of treatment. Radiation was performed more frequently with an early-stage diagnosis than late-stage (15.6% versus 9.1%, p < 0.0001). Several factors were associated with increased odds of early-stage diagnosis including age (aOR = 1.20; 95% CI 1.05, 1.36), females higher than males (aOR = 0.85; 95% CI 0.74, 0.98), and head of the pancreas as the primary tumor site (aOR = 2.25; 95% CI 1.95, 2.60). There were no statistically significant associations between the diagnosis of early-stage disease and rurality (aOR=0.94; 95% CI 0.76,1.16).

Thirteen semi-structured interviews were conducted with PCPs practicing in WV. The interviews were conducted via video conferencing technology, in person, or via telephone. Interviews ranged from 14 – 36 minutes. Practice experience ranged from 5 – 33 years and all were currently providing services in ten different counties throughout West Virginia (rural and non-rural). Most participants were women (n = 9, 69%) and of White race (n = 12, 92%). Provider type was fairly balanced among Nurse Practitioners, Physicians Assistants (n = 7, 54%) and Physicians (n = 6, 46%). Findings of the thematic analysis highlighted four themes: Practicing in Appalachia; Characteristics of Screening and Detection; Features of Managing Pancreatic Cancer; and Perceived Training, Education and Knowledge of PC among PCPs. In the first theme, three sub-categories emerged: ‘no experience diagnosing PC’, ‘participation in the WV Cancer Registry’, and ‘challenges’. Distinct sub-categories of screening and detection surfaced for ‘signs and symptoms’; ‘methods of detection’; ‘practice guidelines’; ‘motivations for seeking care’; and ‘willingness to be screened’. The theme ‘Features of Managing Pancreatic Cancer’ was supported by sub-categories: ‘Identification of patterns or risk factors for PC’; ‘aspects of referring patients to specialists’; and ‘supporting patient and family pre/post diagnosis’. Perceived Training, Education and Knowledge of PC among PCPs presented as a stand-alone theme.

Conclusion:

Our findings suggest that high-risk individuals in Appalachian states have a higher prevalence of pancreatic cancer compared to the US population overall. However, these findings need to be interpreted with respect to the high risk of bias and inconsistency observed in the included studies. Research intentionally aimed at screening and detection in high-risk groups in geographically diverse regions such as Appalachia, are needed to support clinical guideline development and would benefit from the availability of local, individual level data in future surveillance and detection research.

Findings from the WV Cancer Registry suggest a high proportion of this population did not receive any surgery, radiation, chemotherapy, nor immunotherapy as first course of treatment. Older adults in this population were also more likely to be diagnosed with early-stage disease, disputing current research and questioning underlying motivators for seeking care in this population. These findings may inform the diagnoses of early-stage PC in West Virginia.

With respect to interviews, results highlight the vagueness of detection and diagnosis, along with difficulties practicing in WV regardless of rurality given the state’s poor social determinants of health. Interventions need to focus on education, training and retention of PCPs in WV given the lack of specialists. Public health initiatives aimed at educating not only the public but PCPs and residents on aspects of pancreatic cancer, including risk factors, signs and symptoms and early primary care visits are needed to reduce PC morbidity and mortality.

Download

Share

COinS