Feasibility of a Patient Diagnosis Vetting Method to Identify and Discard Fraudulent Online Patient Survey Responses: A Case Study in Cholangiocarcinoma.
Author(s)
Keilson JM1, Berk A2, Abdel-Wahab R3, Cornew S2, O'Brien C2, Lindsey S4, Bachini M4, Maithel SK5
1Emory University School of Medicine, Atlanta, GA, USA, 2Invitae, Palo Alto, CA, USA, 3Cholangiocarcinoma Foundation, houston, TX, USA, 4Cholangiocarcinoma Foundation, Herriman, UT, USA, 5Winship Cancer Institute, Emory University, Atlanta, GA, USA
OBJECTIVES
: This case study describes the analysis of a large, rapidly recruited cholangiocarcinoma patient survey dataset, with the objective of determining the veracity of patient self-reported diagnoses and discarding fraudulent responses. The feasibility of utilizing two external sources as a vetting mechanism for validating patient responses is evaluated.METHODS
: Investigators recruited cholangiocarcinoma patients into a patient-reported outcomes study via email and social media. A $25 gift card was offered for participation. The survey included reCAPTCHA fraud and abuse protection service. 2392 survey responses were received within 3 weeks; 1079 were analyzed after eliminating duplicate or incomplete responses. Respondents were then vetted against two known sources of cholangiocarcinoma patients: registration with the Cholangiocarcinoma Foundation (CCF) or Ciitizen, a patient-mediated health records and RWE platform.RESULTS
: Self-reported mean age was 45 years, whereas the known mean age of cholangiocarcinoma diagnosis is 65 years. Patient self-reported treatment features did not align with known treatment patterns for this disease, with 108 (10%) patients self-reporting liver transplantation and over 50% self-reporting neoadjuvant chemotherapy(n=572) and neoadjuvant radiotherapy(n=560). When the study population was restricted to those registered with CCF or Ciitizen, the sample size decreased to 208. This vetted sample more accurately reflected known age distributions for the disease, with a mean age of 58 years. The vetted sample also matched known treatment patterns more accurately, with n=1 (0.5%) reporting liver transplantation, n=42 (20%) reporting receipt of neoadjuvant chemotherapy, and n=15 (17%) reporting neoadjuvant radiotherapy.CONCLUSIONS
: Fraudulent responses are common among online patient surveys and there are limited methods available to verify patient diagnoses. This study utilized registration with a CCF or Ciitizen to validate survey responses. With this vetting structure applied, responses were more consistent with known disease and treatment patterns, suggesting that this method is feasible for validating survey respondents as actual members of the study's targeted patient population.Conference/Value in Health Info
2022-05, ISPOR 2022, Washington, DC, USA
Value in Health, Volume 25, Issue 6, S1 (June 2022)
Code
SA24
Topic
Methodological & Statistical Research, Patient-Centered Research, Real World Data & Information Systems
Topic Subcategory
Data Protection, Integrity, & Quality Assurance, Patient-reported Outcomes & Quality of Life Outcomes, PRO & Related Methods, Survey Methods
Disease
Oncology