The association between normalising attributions and help-seeking for cancer alarm symptoms in the community: a cross-sectional survey

Abstract

BackgroundAttribution of early symptoms to non-pathological factors such as age or diet (normalising) is often identified retrospectively by patients with cancer as an explanation for delay in presentation. However, there have been no community-based studies. We aimed to assess associations between normalising attributions and help-seeking in a community sample of patients reporting cancer alarm symptoms. MethodsA questionnaire was mailed to 4858 adults (≥50 years with no cancer diagnosis) through primary care, asking about symptom experience in the previous 3 months. Respondents were asked, within a longer symptom list, whether they had experienced any of ten cancer alarm symptoms in the previous 3 months. Follow-up questions addressed perceived cause (free text), with responses categorised into normalisation (age, external factors), non-cancer disease, psychological factors, cancer, or “don't know”. Blank responses were treated as missing and excluded from analyses. Respondents were asked whether they had consulted a doctor (yes or no). FindingsResponse rate was 35% (n=1724). Normalising was common for four symptoms: persistent cough or hoarseness (81/296, 27%), persistent change in bowel habit (88/216, 41%), persistent change in bladder habit (67/166, 40%), and change in the appearance of a mole (27/71, 38%). Age (ie, an effect of getting older) was a common attribution for all symptoms except persistent cough, which was often attributed to smoking. Change in diet was a common attribution for change in bowel habit, and sun exposure for change in a mole. Missing attribution data (n=280) ranged from 15% (n=53) for cough to 42% (n=51) for change in mole. Logistic regression showed that normalising attributions were associated with being less likely to having contacted a general practitioner for persistent cough (odds ratio 0·46, 95% CI 0·27–0·79), after controlling for demographic characteristics, and the same effect was observed across symptoms. No sex or socioeconomic differences in help-seeking were seen. InterpretationAttributing symptoms to normal everyday causes was associated with delayed help-seeking for four cancer alarm symptoms. Attributions such as smoking and sun exposure, which are themselves risk factors for cancer, were among the attributions that undermined help-seeking. The substantial proportion of missing data for the attribution item suggests some reluctance to complete open text responses. More research is needed into how people make help-seeking decisions. FundingThis work was supported by a Cancer Research UK Postdoctoral Fellowship grant (C33872/A13216) awarded to KLW.