Search

Background Social media and online surveys are commonly used to recruit and collect data from patients and physicians about GI diseases – they are efficient, convenient, and less resource intensive compared to traditional recruitment approaches and paper surveys. However, online data fraud is increasing and difficult to identify. Online data fraud can include intentional duplicate responses/straight-lining/inattention, bots/malicious software, and professional survey takers who provide fraudulent responses to meet study eligibility. Purpose 1) Illustrate challenges of identifying fraudulent respondents through an algorithm and verification process we developed for our survey in IBD. 2) Demonstrate potential impact of fraudulent respondents on data and results. Method Online survey of Canadian adults (>18 years) with IBD about healthcare processes for managing IBD hosted using Qualtrics. Recruitment was done in clinic and online (mailing lists, social media). A $25 giftcard was offered for participation due to low response after 3 months in field, after which a large influx of ‘respondents’ occurred. Most were fraudulent although not obvious at first. To mitigate further fraudulent responses, we added the following to our survey: reCAPTCHA score, repeated question (year of IBD diagnosis), duplicate ID score, fraud score and honeypot question. Our algorithm to identify fraudulent responses included 13 binary ‘red flag’ variables: age 3 were categorized as likely fraudulent. Respondents with scores ≤3 were reviewed individually. Respondents flagged as likely real or unsure were emailed and asked to verify their age; those who correctly verified age were considered likely real and included in the final sample. Result(s) Of the 4334 ‘respondents’ who started the survey, based on fraudulent response score we identified 75% (n=3258) as likely fraudulent, 17% (n=727) as unsure and 8% (n=349) as likely real. After age verification, 76% (n=3297) were considered likely fraudulent, 14% (n=592) remained unsure, 10% (n=442) were considered likely real, and Conclusion(s) Despite convenience, social media and online surveys can be prone to fraudulent responses, especially when incentives are offered. We developed an algorithm and verification process to identify fraudulent responses using an IBD survey example. Given that only 10% of the full sample was considered likely real, researchers using social media and online surveys should carefully examine data for fraudulent responses and apply strategies to mitigate risks. Please acknowledge all funding agencies by checking the applicable boxes below CCC Disclosure of Interest None Declared

BACKGROUND: Patients from lower vs. higher socioeconomic classes may have less opportunity to participate in health research due to: less flexible work hours, working multiple jobs, less likely to be self-employed, less education and lower understanding of the importance of research. AIMS: To increase our understanding of the concerns and preferences for care of patients with inflammatory bowel disease (IBD) with less than a university degree. METHODS: A cross-sectional study was undertaken at the McGill University Health Centre IBD Clinic (July-August 2018). Inclusion criteria were: out-patients, aged 18 years or older, English- or French-speaking, less than a university degree and diagnosed with ulcerative colitis or Crohn’s Disease for at least one year. A research assistant conducted semi-structured interviews using a questionnaire that addressed: 1) characteristics of health care providers and the system, 2) what health care providers do in delivering care and 3) patient outcomes and goals. The questionnaire was created based on a focus group guide that had been developed for our larger study aimed at eliciting patient preferences for care in IBD. Descriptive statistics were used to characterize the study population and to quantify the number of participants who endorsed various elements of care. Qualitative data were analyzed by identifying themes and selecting exemplars to support them. RESULTS: In total, 23 individuals (median age=41; 12 females) participated, of whom 21 (91.3%) did not attend university and 2 (8.7%) attended university without completion. Fifteen (65.2%) participants rated access to an IBD specialist as the most important aspect of care; 8 (34.8%) and 9 (39.1%) participants said holistic and complementary and alternative medicine, respectively, were important for their care. The 3 most commonly reported experiences in the health care system were: good care (52.2%), lengthy delay until diagnosis (26.1%) and non-receptive staff (13.0%). The 3 most commonly reported experiences with health care providers were: kind staff (34.8%), receptive IBD doctor (26.1%) and non-receptive IBD doctor (21.7%). The 3 most helpful aspects in managing IBD were: IBD specialist, family and friends and treatment that reduces symptoms. The 3 most difficult aspects of living with IBD were: urges/pain/anxiety, social aspect and medications and side effects.The top-ranked goal was reducing symptoms. CONCLUSIONS: Access to a receptive IBD specialist was the most important aspect of receiving good quality care. Many patients used a multidisciplinary approach to achieve disease remission that included use of non-traditional therapies. These findings may be used to enhance the quality of patient care. FUNDING AGENCIES: Ferring Canada

BACKGROUND: The Tianjin Colorectal Cancer (CRC) Screening Program uses a questionnaire (HRFQ) to determine risk status and the initial screening test: stool testing (FIT) for average-risk and colonoscopy for high-risk individuals. AIMS: To determine agreement between colonoscopy-detected CRC and pathology-confirmed CRC. METHODS: A retrospective cohort study was conducted using the data from the 2012 Tianjin CRC Screening Program in Tianjin, China. Participants were aged 60 to 74 and residents of Tianjin who completed both the HRFQ and colonoscopy. Demographics and clinical data were obtained as well as FIT, colonoscopy and pathology results. Cohen’s Kappa was used to determine agreement between colonoscopy-detected and pathology-confirmed CRC. RESULTS: In 2012, 19,096 individuals completed the HRFQ and colonoscopy, of which 10,907 (57.1%) were HRFQ positive and 7,986 (41.8%) were colonoscopy positive. Colonoscopy positive findings included 7160 polyps, 728 adenomas and 102 CRC. Of all colonoscopy positive findings, 1,256 (15.7%) were sent to pathology where 326 were confirmed CRC. Only 59 (18.1%) of the pathology-confirmed CRC were detected at colonoscopy. Agreement between CRC detected at colonoscopy and CRC confirmed by pathology was Kappa=0.22 (95%CI=0.16–0.27). CONCLUSIONS: Poor agreement was found between colonoscopy-detected and pathology-confirmed CRC, suggesting that many CRCs are undetected and untreated. Effectiveness of the screening program would be improved by sending all removed tissues to pathology. FUNDING AGENCIES: Fonds de recherche du Québec- Santé