INFLUENCE OF DISEASE PREVALENCE AND AN ALTERNATIVE EXPLANATION ON EARLY CANCER DETECTION IN FAMILY MEDICINE

Sunday, October 20, 2013
Key Ballroom Foyer (Hilton Baltimore)
Poster Board # P1-8
Decision Psychology and Shared Decision Making (DEC)

Olga Kostopoulou, PhD, Miroslav Sirota, Shyamalee Samaranayaka and Thomas Round, MBChB, King's College London, London, United Kingdom
Purpose: Early detection of cancer is crucial for the successful management of the disease. As part of the National Awareness and Early Diagnosis Initiative (NAEDI) in the UK, we are investigating why family physicians might miss early presentations of cancer. In this study, we tested the influence of two factors: cancer prevalence and the availability of an alternative explanation for the patient’s symptoms.

Method: In an online study, family physicians saw two case vignettes, with early cancer as one of the possible diagnoses: colorectal in one case, lung in the other. In a 2x2 mixed factorial design, we manipulated two case attributes: cancer prevalence (high vs. low) and the availability of an alternative explanation for the symptoms (present vs. absent). Each patient visited twice, the second time with unchanged or slightly worsened symptoms. If respondents decided to refer to hospital at first visit, they did not see the second visit. Respondents gave their main working diagnosis and management decision at each visit separately.

Result: 300 family physicians participated. In a random-intercept, logistic regression model, both prevalence (OR 2.66 [95% CI 1.63-4.35], p<0.001) and alternative explanation (OR 1.80 [95% CI 1.09-2.96], p=0.021) were significant predictors of the likelihood that physicians would diagnose cancer. Physicians were more likely to do so at the second visit (OR 8.68 [95% CI 5.64-13.35], p<0.001) but this was less likely for lung than colorectal cancer (OR 2.59, p<0.001). The number of years in family medicine was positively associated with the likelihood of diagnosing cancer (OR 1.04, [95% CI 1.02-1.06], p<0.001).

Conclusion: Both an alternative explanation and cancer prevalence can influence whether family physicians will detect early cancers in patients. Cancer prevalence determines whether the cancer hypothesis will be entertained, in agreement with current memory models of hypothesis generation (Thomas et al. 2008). The presence of a plausible alternative explanation provides a ready-to-use shortcut in the reasoning process that saves time and cognitive effort, consistent with Kahneman’s view of people as lazy thinkers (2003). The strongest predictor of cancer detection was repeated visits. Presenting to the physician a second time is very likely to lead to referral. This has implications for providing advice to patients who may suffer prolonged symptoms but are reluctant to consult again following an earlier non-cancer diagnosis.