What role do GP factors have on decisions to investigate symptoms of possible cancer? A mixed-methods systematic review

Talk Code: 
Victoria Hardy
Victoria Hardy, Adelaide Yue, Stephanie Archer, Samuel Merriel, Matthew Thompson, Jon Emery, Fiona Walter
Author institutions: 
University of Cambridge, University of Exeter, University of Washington, University of Melbourne


Clinical guidelines of symptoms most predictive of cancer are used in many countries to improve the timely diagnosis of this heterogeneous disease by standardizing use of diagnostic tests and referral pathways. Variation in appropriate use of referral pathways has prompted concerns about suboptimal use of investigations contributing to diagnostic delay for many cancers. This has been interpreted as due to inconsistent GP adherence to guidelines. As clinical decision-making involves complex cognitive processes, factors related to the GP may play an important role in decisions to investigate possible cancer, but this has not been examined. We therefore evaluated the influence and perceptions of GP factors on decisions to investigate symptoms that might indicate an underlying cancer in primary care.


We searched MEDLINE, Embase, Scopus, CINAHL and PsycINFO between January 1990 and December 2019 for studies in developed countries that reported the influence of GP factors on testing and referral decisions for any cancer type. GP factors were defined as ‘attributes of the GP that they bring to the practice setting which have been the object of interest regarding performance and competence assessment’. Quality assessment and data extraction were undertaken independently by two authors. We used a convergent segregated approach to analyze quantitative and qualitative findings before combining them in a narrative summary.


Twenty-four studies (15 quantitative and 9 qualitative), predominantly in European countries, met review eligibility criteria. We identified a total of twelve GP factors. The most substantive body of evidence was found for suspicion of cancer, years of experience, gut feeling, and age. Other GP factors included gender, continuing medical education involvement, fear of malpractice, tolerance for uncertainty, assessment of cancer risk, first impressions, and attitudes to risk and gatekeeper role. Odds of non-urgent investigation was higher when GPs’ suspected cancer; urgent referral was less likely in the absence of ‘alarm’ symptoms, when GPs felt unsupported by referral criteria to act on a suspicion of cancer. GPs were more likely to investigate colorectal and ovarian cancer and became willing to act outside of clinical recommendations as years of experience increased. Gut feeling at referral predicted a subsequent cancer diagnosis and was a valued diagnostic tool for guiding management of non-specific symptoms. The influence of age on decisions to investigate different cancer sites was mixed.


GP factors influenced and were perceived to have an important and intersecting role in testing and referral decisions for symptoms suggestive of an underlying cancer, with suspicion of cancer, gut feeling, and years of experience most implicated. The utility of these factors for managing non-specific symptoms warrants focused attention. This has particular implications for developing strategies to optimize rapid diagnostic cancer testing pathways that promote more timely diagnosis of harder-to-detect cancers.

Submitted by: 
Victoria Hardy
Funding acknowledgement: 
This work arises from the CanTest Collaborative, funded by Cancer Research UK [C8640/A23385] under which Victoria Hardy is a PhD student, Dr. Samuel Merriel is a Clinical Research Fellow, Prof. Matthew Thompson and Prof. Jon Emery are Associate Directors, and Dr. Fiona Walter is Co-Director.