The geography of genetics: an analysis of referral patterns to a cancer genetics service

Of 10,878 referrals considered, 25% were withdrawn or did not return the family history questionnaire, so did not continue with the referral. Only 43% had a risk level assessed and categorised as moderate or high risk, the remainder were at no greater than population risk, or their risk status was undetermined. One possible reason for a quarter of referrals not being pursued is that the patient did not request the referral, but was referred by a hospital consultant or their GP without their knowledge. Another is that the family history questionnaire is long and complicated, so some patients are put off by the amount of time it takes to complete, or perhaps lack the necessary language or comprehension skills. This might also, in part, explain the differences found in service access and use between most and least deprived areas.

Protheroe et al. (2006) found that CGSW guidelines placed some women referred for a family history of breast cancer into a higher risk category than the National Institute for Health and Clinical Excellence (NICE) guidelines. From their sample of 75 referred women, 23% would have been placed into a lower risk category if NICE guidelines had been adopted, which would have reduced the percentage of referred patients at moderate or high risk in this study even further.

Referred patients classed as high risk were more likely to come from secondary care than primary care sources, while moderate risk referrals are more likely to come from primary care. It is possible that further assessments undertaken in secondary care lead to a patient being considered at higher risk than was previously the case e.g. if a woman is in for a mammogram after having discovered a breast lump. In secondary care there are more resources immediately available than in primary care, with fewer patients to be seen per health care professional, and therefore more time to spend on assessing each patient. This greater level of resources, especially time, means that in secondary care it is possible that an appropriate high risk conclusion is reached before referral, and that someone at moderate risk, who can be managed in the community or hospital, is not referred on. It is also possible that specialists in secondary care have a greater knowledge of cancer genetics risk factors than generalists in primary care simply by virtue of seeing more such cases. Watson et al. (2001) reported that GPs were likely to raise the issue of family history with patients, even if those patients did not meet the referral guidelines from the regional genetics service, which might explain the higher level of moderate referrals from primary care.

Over 91% of referred patients were female. There are a number of factors that could account for this preponderance of female referrals. It is known that for breast cancer 99% of cases are in women (Cancer Research UK 2007), which is borne out by these data, with only 0.7% of breast cancer referrals being male. An additional explanation could be the reluctance of men to take care of their health (Conrad and White 2007). Those men that are referred are less likely to be at high or moderate risk than women. The fact that men are more likely to have an unknown or uncertain risk assessment could be due to the types of cancer that affect men. Breast and ovarian cancers combined account for 68.5% of all the CGSW referrals, and the screening tools for breast cancer are well validated and accurate, at least for older women.

Ranks of referral rates per 10,000 head of population by unitary authority showed that distance to a CGSW centre is not a predicting factor. While Powys, which is the furthest unitary authority from any CGSW centre in terms of distance has the lowest referral rate, Newport, which is adjacent to Cardiff, home of the largest CGSW centre, has the third lowest referral rate. The Vale of Glamorgan, which is also adjacent to Cardiff, has the highest referral rate. This is supported by previous work on referral patterns which show that distance is not necessarily the most influential factor (Martin and Williams 1992; Field and Briggs 2001). Powys shares a long border with England, and Jones (2002) noted referrals being made to England. For this study, some of the genetics services on the English side of the border were contacted. One service in the North West received between five and ten cancer genetic referrals per year from Wales, all of which were returned with advice to make a referral to the CGSW. Another service in the West Midlands accepted about 25 referrals per year. As noted above, there were 98 referrals from England on the ISCO database that were removed during data cleaning. Therefore the number of patients crossing the border for cancer genetics services is small in comparison with the total number of referrals. McDonald et al. (2004) reported a correlation between distance of unitary authority from a CGSW centre and referral rates on a database of just over 4,000 referrals from 1998 to 2002. However, that pattern no longer holds true for the 10,878 referrals received between 1998 and 2006. This might be due to the rapid rise of primary care referrals as opposed to the more gradual increase of secondary care referrals.

At least two previous studies (Quinn et al. 2004; Murugan et al. 2005) have shown that referrals from secondary care outnumber those from primary care. This was the case with the CGSW initially, but that situation has now reversed. More referrals from primary care mean that patients are entering the CGSW system at an earlier stage, so if they are moderate or high risk, will come into contact with the surveillance services earlier that would otherwise have been the case if they had come via secondary care. One reason for patients referred from secondary care being at higher risk might be because they have already developed symptoms which could indicate an inherited cancer. For those patients who would benefit from screening, being referred from primary care means that they will have access to that screening sooner, which is beneficial.

The juxtaposition of high and low deprivation areas is clearly identified in cities, but rural areas appear to be more homogenous on maps. While it is possible that rural areas are indeed homogenous, it is also likely that the analysis failed to display any pockets of deprivation in deep rural areas. It is possible that, as suggested by Jordan et al. (2004) any small pockets of deprivation are in fact being “diluted”, or masked by larger, less deprived parts of the same geographical areal unit.

Jones (2002) investigated potential reasons for variations in referral decisions using a semi-structured questionnaire approach to investigate referral behaviour of GPs in eight practices in Powys, a predominantly rural area of Wales. As well as gauging the level of awareness of cancer genetics services in general, Jones’ study also investigated the extent to which GPs and practice nurses were aware of referral guidelines issued by the CGSW. During the course of the study, it became evident that several of the GPs had referred patients to centres across the border in England. Findings from the interviews with doctors and nurses suggested that patients in this rural area were just as likely to want to be referred to such tertiary services and that the distance patients would need to travel to access a cancer genetics centre was not an influence on their decision to refer but it might have any effect on where to make the referral. Thus, due to the topography and road and rail networks, patients in this locality were more likely to be referred to Birmingham (75 miles by road) than Swansea (105 miles). However, findings from the study suggest that “rurality is not an issue when referral is important for nurses or doctors.” (Jones 2002, p. 27).

Reasons for variations in the centres to which referrals are made are speculative but could include referrals to Rhyl reaching a plateau, or a saturation point where the number of referrals likely to be received each year will remain more or less constant in the future. Conjecturally, reasons for this might be that all families with inherited cancers have been identified in the north Wales catchment area, or all relevant health care professionals in that area are now fully aware of the CGSW. There might alternatively be evidence of an increasing affluence within the Swansea catchment area of mid and south west Wales, as more affluent people are more likely to be referred. A further reason could be changes in staffing levels across the centres due to natural wastage and new recruitment.

The correlation between the CGSW centre and the unitary authority in which a referred patient is resident is to be expected given CGSW’s informal area-based referral policy. Where there are variations, these might be due to personal circumstances of the referred patient, or the preferences of the person making the referral. The catchment areas were set up for management purposes, so do not necessarily reflect the situation on the ground. For example, patients resident in north Powys live closest to the CGSW centre in Rhyl, those resident in south east Powys live closest to the centre in Cardiff, yet all Powys residents should be referred to Swansea.

Referral rates increase as deprivation decreases. This might be because less deprived people include those who better understand how to utilise the services available to them and are more vocal when they believe they are not receiving the care they deserve. This is a likely hypothesis given that Pell et al. (2000) demonstrated that affluent patients are more likely to have their cases classed as urgent. Affluent patients are less likely statistically to become seriously ill, therefore it can be concluded that, rather than simply having more urgent illnesses, they are better able to communicate a sense of urgency when being assessed by their doctors. More affluent people might be more aware and informed about the risks of familial cancer, therefore are more likely to seek a medical opinion. Also, since only 43% of referrals to CGSW in this study were in fact found to have a higher than population risk, it can be argued that the referral system is susceptible to influence by those who are more able to argue their case, that perhaps some patients overstate the urgency or have a higher perceived sense of risk.

Research conducted in north Derbyshire by Maheswaran et al. (2006) studied the effects of deprivation and travel distance on breast cancer screening uptake. They found that attendance for screening was inversely proportional to socioeconomic deprivation and that increasing distance from a screening unit was associated with decreasing attendance levels, concluding that socioeconomic inequality in breast screening uptake continued to exist and that there was a small distance decay effect. In contrast, other studies have found that the relationship with distance is not uniformly linear. Field and Briggs (2001) for example, in a study of patients accessing services in Northamptonshire, found that utilisation did not decline linearly with distance but that instead the relationship tended to be ‘U shaped’, suggesting a more complex relationship. In this paper, there is no discernable linear distance decay effect, which also suggests a more complex relationship.