Evaluating Familial History as a Phenotypic Screening Tool for Colorectal Cancer in the Australian General Practice Population

Abstract

Colorectal cancer (CRC) is the second most common cancer among males and third among females across the world. In Australia it is the second most prevalent and the second leading cause of cancer-related mortality with the number of CRC incidences doubling over the last decade. While there has been a reduction of the incidence-adjusted mortality of CRC, a significant number of CRC detections are made at either the intermediate or later stages of the disease progression despite the roll out of a population based screening program for individuals aged 50 and over. Data shows that ‘one size fits all’ nature of the program despite recommendations from the NHMRC to screen according to the familial risk of an individual and inappropriate colonoscopy referrals, may have led to over screening those at average risk while potentially under-screening and missing those at an increased risk. Furthermore this program may have missed individuals under the age of 50 that have a high familial lifetime risk of developing CRC and require earlier CRC screening with a colonoscopy. It was hypothesised that implementing an online familial risk tool that notified both patients (aged 25-74) and their GPs of their familial CRC lifetime risk would increase the uptake of risk-appropriate screening among the study population relative to controls that receive usual care, during the 12 month study period. In doing so, this thesis used a complex intervention aimed at improving the rate of risk-appropriate screening for colorectal cancer (CRC) among an Australian General Practice population. This intervention utilised an online evidenced -based familial history algorithm, that stratified participants into three Australian National Health and Medical Research Council (NHMRC) recommended relative risk groups for screening CRC. These categories are based on a strong body of evidence showing that familial phenotype as measured by family history is a significant and non-modifiable risk factor for an increased lifetime risk of CRC. The algorithm used in the online tool was adapted from the NHMRC guidelines but were also updated by utilising the most recent evidence-base in addition to consulting with a group of experts. This algorithm was then programmed into an online website called “Which test is best?”. This website notified participants of their familial risk in addition to faxing or emailing this information to their consulting General Practitioner (GP). The website was piloted among members of a cancer consumer group (n=26), before being amended to improve clarity and the website interface. It was then implemented in a clustered RCT to evaluate its effect on risk-appropriate CRC screening. The intervention was implemented at both the cluster (GP practice) (Intervention n=27, Control n=28) and participant (eligible patients aged 25-74 with no personal history of CRC and/or inflammatory bowel diseases) level (Intervention, n=836, Control n=726). Those in the intervention arm were given access to the online website with risk tool and their family history information. In addition to their familial risk category with NHMRC recommended screening guidelines were forwarded to their consulting GP, while the control group had usual care. Both groups were followed up 12 months later to obtain their self-reported CRC screening information using the online survey. Thereafter, the control group was immediately given access to the online website with risk tool so that their family history information could be recorded and the level of risk-appropriate screening could be calculated for both groups. The results from this study showed ,that there was no significant difference in risk-appropriate screening rates amongst participants allocated to the intervention group compared to the control group as there was no main effect of allocation when included as a predictor within a binomial logistic regression when modelled to the GEE. However, participants allocated to the intervention group that were designated as belonging to the potentially high-risk category did engage in significantly higher levels of risk-appropriate screening when compared to the control group at 12 month follow-up. This was observed by a significant interaction effect of both family history and allocation in predicting risk-appropriate screening the final GEE model. Specifically, potentially high-risk individuals that were allocated to the intervention group had higher odds (about five times) of engaging in risk-appropriate screening when compared to those at population level risk that were assigned to either control or intervention, when controlling for other variables. This suggests that the online familial risk tool was effective in changing the behaviour of participants from the intervention group that were categorised as having a family history consistent with a potentially high risk (defined as having lifetime relative risk three times or greater of the general population) of developing CRC in their lifetimes. GPs from participating clusters were followed up by a survey (n=66) to assess their attitudes, knowledge and perceived barriers on utilising family history to risk-appropriately screen their average risk patients. Three important findings emerged from this survey. Firstly it shows that the majority of GPs in this study regard family history as the most important factor in screening their asymptomatic patients for CRC. It also shows that these GPs in principle strongly support the NHMRC guidelines, continuing education and peer-reviewed evidence as the most important knowledge factors in evaluating their CRC screening recommendations for asymptomatic patients, while being somewhat less influenced by government policy and their patients’ personal perceptions about the efficacy of a particular CRC screening test. However GPs appear very sensitive to their patients’ fears and anxiety over CRC screening, assessing this factor as the most important barrier to screening for CRC followed by their subjective lack of experience with CRC screening and time constraints imposed during the consultation. Findings also showed a substantial level of dissonance between what GPs believe to be appropriate CRC screening for their asymptotic patients and what they may be likely to recommend with 77% GPs self-reporting that they still refer up to 10 average risk asymptomatic patients to a colonoscopy during a typical month. Taken together the findings from this thesis show that that an intervention which aims to include both the patient and GP improves the uptake CRC risk-appropriate screening for individuals with potentially high-risk. It shows that a tailored risk tool, that supports GP triage may be sufficient to improve uptake of CRC screening modalities across all risk groups but may not be sufficient to encourage risk-appropriate screening of those from average and moderate risk. This is mainly due to persistent over-screening in the average-risk group within our study sample. Future studies may need to examine and differentiate between under screeners and over-screeners in order to target and tailor interventions to those groups separately.