People in the UK, which has relatively poor cancer survival, had lower awareness that the risk of cancer increases with age and reported more barriers to symptomatic presentation (especially worry about wasting the doctor's time) than other countries. However, people in Denmark, who also have relatively poor cancer survival, had high awareness of age-related risk of cancer and few barriers to symptomatic presentation. Awareness of cancer symptoms was high and beliefs about cancer outcomes were positive for all participating countries and differences between the countries were small. We found no evidence that the overall pattern of cancer awareness and beliefs followed the overall pattern of 1-year cancer survival across the countries.
To our knowledge, this is the first study examining differences in cancer awareness and beliefs between high-income countries. It is unlikely that methodological limitations explain the results. We developed harmonised versions of the ABC to measure the constructs in each country and harmonised the survey methods, using a single provider to carry out the sampling and interviews. We achieved samples that were representative, in terms of age, sex and education, of the underlying populations. Controlling for any differences in age, sex and education distribution between populations and doing sensitivity analyses to allow for under-representativeness of certain demographic groups made very little difference to the results. The large sample sizes meant that we were confident about the precision of our estimates, although it also meant that small differences were statistically significant even where actual differences were not likely to be of great public health significance. Moreover, where proportions reporting an outcome are high (e.g., in beliefs about cancer outcomes), odds ratios become unreliable as measures of relative risk (Davies et al, 1998
). In view of this, our approach to interpreting the results was to focus on absolute differences between the countries that were likely to be of public health significance.
There was some variation in estimated response rates between countries, with lower rates in the Scandinavian countries and the highest rate in Australia. It is difficult to know whether the response rates achieved in Australia, Canada and the UK are comparable to other surveys, because few reports of random digit dialling surveys in these countries have reported response rates according to AAPOR conventions. In Scandinavian countries, where there are population registers, telephone surveys often achieve higher response rates than this, but it is more usual for the researchers to write to people before telephoning them (e.g., Feveile et al, 2007
). We chose not to write to potential participants in advance in the Scandinavian countries, because it would have increased the variation in survey methods. Despite the variation in response rates, the age distribution of the samples were broadly similar to the national populations except for some differences in gender balance (more women) and higher levels of education (which often happens in health surveys). However, weighting made little difference to the estimates of cancer awareness or beliefs that suggests that under- or over-representation of particular groups did not significantly influence the pattern of cancer awareness and beliefs.
We calculated aggregate scores for two sets of items (beliefs about cancer outcomes and about barriers to symptomatic presentation), because factor analysis provided strong evidence that responses to each set of items were driven by common underlying factors: broad perceptions about either beliefs about cancer outcomes or barriers to symptomatic presentation (Simon et al, 2012
). The mean differences between countries in the aggregate scores are more likely to represent real differences and are more precise (with narrower confidence intervals) than the responses to the individual items.
We monitored the three most widely distributed national newspapers in each of the participating countries for 2 weeks before and during fieldwork, focusing on stories or campaigns that would have affected all or most of the geographical area covered reported in the first three pages. In Denmark and Norway, there were no relevant news stories. In Sweden, there were stories about breast screening, HPV vaccination and the quality of cancer care. In the UK, there was a story about a blood test to identify ovarian cancer early, and a negative story about the quality of cancer care in the NHS. In Canada, there was a story about a celebrity with oral cancer but no details about symptoms, and a story about the quality of care in cervical screening. In Australia, there were stories about men's health checks for a range of health issues (not just cancer), dietary risk factors for cancer and mammography. Although we cannot rule out an effect of these stories on our results, none appeared sufficiently specific to have influenced population levels of cancer awareness or beliefs to the extent that they would explain any observed international differences.
The biggest differences in cancer awareness and beliefs between the participating countries were in beliefs about barriers to symptomatic presentation and awareness of age-related risk, with the UK having the highest level of barriers and the lowest awareness of age-related risk; this is consistent with previous UK surveys (Robb et al, 2009
; Forbes et al, 2011
). Our study was not designed to address the underlying reasons for international differences; that would be for future studies. We speculate that people's worry about wasting the doctor's time in the UK could be shaped by a belief that they might be told they were wasting the doctor's time if they presented, that others might judge that they should not waste the doctor's time or that they should not waste public resources. British people could be more concerned about embarrassment because of the traditional ‘stiff upper lip'. Barriers to symptomatic presentation in Britain warrant further research to inform interventions to promote early presentation.
The pattern of differences in cancer awareness and beliefs between the participating countries did not follow the pattern of differences in survival, but there was some evidence that it followed cultural/language demarcations: Scandinavian people had lower levels of barriers to symptomatic presentation and better awareness of age-related risk than people in the Commonwealth countries, but further studies are needed to identify the origins of these differences.
Although we found no evidence that the international pattern of cancer awareness and beliefs followed the 1-year survival pattern, our results do not rule out an association between cancer awareness and beliefs and cancer survival at an individual level, because it was an ecological analysis looking at average levels of cancer awareness and beliefs.
Other factors, for example, delay in primary care referral or delay in diagnosis in secondary care, may contribute to delay in diagnosis. It has been hypothesised that poor cancer survival in Denmark is related to gate-keeping by primary care doctors (Vedsted and Olesen, 2011
), although a Danish study found that a large proportion of delay in diagnosis was attributable to delay between the GP initiating investigations for cancer and the start of treatment (Hansen et al, 2011
). Furthermore, other analyses from the ICBP exploring stage and survival data for ovarian (Maringe et al, 2012
) and breast cancer (Michel Coleman, personal communication), suggest that differences in access to treatments or tumour biology may also be having a role in differences in survival rates, although differences in staging procedures make comparisons difficult. All these issues – beliefs and behaviours in primary care, diagnostic pathways and availability of treatments – are now being studied across the ICBP countries. This will help to work out the relative importance of each factor in determining the 1-year cancer survival rates in each country.
Our findings have some specific implications for individual countries. In Denmark, poor 1-year cancer survival rates are not likely to be due to poor cancer awareness and negative beliefs in the population, and the causes must be sought elsewhere. In the UK, interventions to promote early presentation might usefully focus on addressing awareness of the age-related risk and increasing the public's confidence to approach the GP with possible cancer symptoms.