Primary care practitioner diagnostic action when the patient may have cancer: a vignette survey in 20 European countries

Background Cancer survival rates vary widely between European countries, with differences in timeliness of diagnosis thought to be one key reason. There is little evidence on the way in which different healthcare systems influence Primary Care Practitioners’ (PCPs’) referral decisions in patients that could have cancer, and how this links with cancer survival. This study aimed to explore PCPs’ diagnostic actions in patients with symptoms that could be due to cancer, how they vary across European countries with marked differences in socio-economic development, healthcare investment and organisation, and how they relate to cancer survival rates. Methods A primary care study, with centres in twenty European countries with widely varying cancer survival rates. The on-line survey of PCPs used vignettes describing four patients with symptoms that could indicate cancer (lung, ovary, breast and colorectal). PCPs were asked how they would manage these patients. Correlations between the likelihood of taking immediate diagnostic action and physician characteristics were calculated. The likelihood of taking immediate diagnostic action in the different participating countries was also analysed, as well as the correlation with national 1-year relative cancer survival rates. Results A total of 2,086 PCPs answered the survey question, with a median of 72 PCPs per country. PCPs’ likelihood of organising a diagnostic test and/or referring to a specialist at the first consultation varied from 50 to 82% between countries PCPs who were more experienced, were working in more remote areas, or worked alone or in smaller practices, were more likely to take immediate diagnostic action than their peers. There was a significant negative correlation between national

healthcare expenditure levels and likelihood of immediate diagnostic action (r=-0.55, P=0.012). However, there was no significant correlation between the likelihood of taking immediate diagnostic action and cancer survival (r=-0.27, P=0.278).
Conclusions Europe shows large between-country variations in PCPs' diagnostic action rates for patients who could have cancer. These are linked with differences in healthcare organisation and levels of healthcare investment.

Background
Cancer survival varies vary widely across Europe (1). The fifth cycle of the European Cancer Registry (EUROCARE)-based Study on Survival and Care of Cancer Patients shows that national 1-year relative survival rates for all cancer sites vary from 58.2% to 81.1% (2). Comparison of European 1-year relative cancer survival (2)(3)(4)(5) shows that some countries have higher survival from most cancers (including Belgium, France, Sweden and Switzerland), while others have consistently lower survival (including Bulgaria, Croatia, Poland and Scotland). This suggests that an improvement in cancer awareness and early detection in relatively poorly performing countries could reduce the survival gap (4). While recent cancer survival rates show improvement in most countries (5), the between-country differences remain (6). However, this is not inevitable: Denmark's considerable efforts to improve early detection rates (7) have resulted in a narrowing of the gap between its own relatively poor cancer survival rates and those of its better performing Nordic neighbours (8). There has been a call for studies which compare cancer diagnostic pathways between well and poorly performing countries, to help gain an understanding of how these disparities may be remedied (5).
Although 1-year relative survival can be affected by differences in registration, as well as by overdiagnosis and lead-time biases (9,10), poorer 1-year survival in some countries is thought to be rooted in diagnostic delay (11,12) and more advanced disease at diagnosis (13,14). The more advanced a cancer is, the more difficult it is to treat it successfully (15) and, for many cancers, disease stage at diagnosis is associated with survival (16,17). There is considerable evidence that longer time to diagnosis and treatment increases cancer mortality (18)(19)(20)(21)(22)(23)(24). Timely diagnosis of cancer is, therefore, a cornerstone of health policy throughout Europe (25).
However, there is a substantial challenge in deciding where and how to achieve this (26), as it is uncertain whether late diagnosis is due to cancer patients presenting later, not being referred quickly enough from primary care, or whether they are inefficiently investigated, diagnosed and treated in secondary care (15). This may be a particular issue where cancer patients present without 'red-flag' symptoms, as how the Primary Care Practitioner (PCP) acts will depend to a large extent on local health service organisation (27).
There is little evidence on how different healthcare systems influence PCPs' referral decisions (26). However, a large variety of non-clinical factors affect these referral decisions (27). These include the extent of gatekeeping, funding systems, access to special investigations, concerns over litigation, and barriers to accessing specialist advice. The International Cancer Benchmarking Partnership (ICBP) (28) examined differences in cancer awareness and beliefs between six countries with comparable wealth in an attempt to explain differences in cancer survival (29). It found a positive association between national cancer survival rates and the readiness of PCPs in those countries to investigate potential cancer symptoms (30). However, there has not yet been an investigation of how PCPs' diagnostic actions with respect to potential cancer symptoms vary across Europe, amongst countries with a wide range of socio-economic development, healthcare systems and healthcare spending.
We therefore aimed to explore the diagnostic action rates of PCPs for patients with symptoms that could be due to cancer, how they compare across European countries, and how they relate to cancer survival. Translation and validation by backtranslation were done in a standardised way (31) and are described elsewhere (32).

Sample size
We aimed for a total sample size of at least 1000 PCPs, with at least 50 responses from each of the participating countries.

Recruitment of participants
Each ÖRG local lead was asked to email an invitation to take part in the survey to the PCPs in their local health district or jurisdiction, and to recruit at least 50 participants. In six countries (Denmark, Norway, Portugal, Romania, Slovenia, Sweden), the invitation was distributed to a national sample. The recruitment email stated that the research aimed to identify which health system factors affect PCPs' decisions to refer patients for further investigation. The possibility of cancer as being a cause of the vignette symptoms was not mentioned in either the recruitment email or the survey. As low survey response rates are common in primary care (38) and can vary between jurisdictions, any local leads who had difficulty in achieving the required sample sizes were asked to increase the number of responses by using snowballing, a recognised technique for recruiting hard-toreach populations in health studies (39,40).

Distribution of the questionnaire
The questionnaire was designed using SurveyMonkey (41). Because of the study's wide geographical coverage, on-line delivery of the questionnaire was used; this methodology has previously been successfully used in research involving cancer care professionals (42).

Statistical analysis
Demographic questions and those relating to vignette diagnostic actions were analysed using descriptive statistics for decision to arrange a diagnostic test, and to refer to a specialist. As it was considered that some PCPs would not investigate because they were referring to a specialist, and conversely some PCPs would not refer to a specialist because they were investigating, we also used a composite measure of a decision to either arrange a diagnostic test and/or refer to a specialist, i.e. the likelihood of taking diagnostic action for cancer. For each individual PCP, mean diagnostic action rates were calculated from the four individual vignette responses. From those, mean diagnostic action rates were calculated for each country. For comparisons between countries, medians and ranges were calculated.
For each country, the mean 1-year and 5-year relative cancer survival rates for the four cancers of interest (lung, ovary, breast and colorectal) were calculated from EUROCARE-5 data (2). These are shown in Table 1. Linear correlations and 2-tailed significance levels were estimated for the proportions of PCPs opting to take diagnostic action and these national 1-year relative cancer survival rates, and also between PCPs' likelihood of organising a diagnostic test and their likelihood of referral to a specialist. As one-year relative survival rates for cancer can be affected by lead-time bias, we made a sensitivity analysis using 5-year survival.

Results
A total of 2,086 PCPs completed the questionnaire. There was a median of 72 respondents per country, range 59-446 (Table 1).
Participants' demographic distributions are shown in Table 2.
Organising a diagnostic test The range of PCPs who stated that they would organise a diagnostic test at this first consultation varied from 35.6%-80.1%, median 53.0% (Figure 1).
Referring the patient to a specialist Correlation of diagnostic action with PCP demographics There was no significant gender difference for taking immediate diagnostic action: for female PCPs, the likelihood was 62.7%, for male GPs it was 61.3%, P = 0.346.
There was, however, a link between the number of years since graduation and the likelihood of taking immediate diagnostic action (Table 3): a mean of 55.5% of PCPs who had graduated less than 10 years ago were likely to take diagnostic action, compared with a mean of 63.4% for those who had graduated 10 or more years ago,

P<0.001.
In the twelve countries with respondents who self-identified as working in remote or island practices, those PCPs were significantly more likely to take immediate diagnostic action than their colleagues (71.4% vs. 60.7%, P = 0.021). There was a trend towards lower likelihood of investigation and/or referral with larger practices,

Principal findings
When faced with vignettes of patients with symptoms that could be due to cancer, there was a marked variation between different European countries in PCPs' stated actions. In all the participating countries at least half of PCPs would have taken an immediate diagnostic action (i.e. either organised a diagnostic test or referred the patients to a specialist, or both). There was a significant negative correlation between national healthcare expenditure levels and likelihood of PCPs' immediate diagnostic action, but no significant correlation between the likelihood of taking diagnostic action and cancer survival. PCPs who were more likely to arrange a diagnostic test were also more likely to refer their patients to a specialist at the same time.
PCPs who had graduated more recently were less likely to take diagnostic action than their more experienced peers, but PCPs working in more remote locations were more likely to take diagnostic action than their colleagues in other localities. PCPs working alone or in smaller practices were more likely to take diagnostic action than those in larger practices.

Strengths and limitations of the study
One of the strengths of our study is the wide spectrum of participating centres, with four countries from each of the Central, Eastern, Northern, Southern and Western European geographical areas, providing variation in geography, socioeconomic and health systems, and levels of healthcare spending. It included the views of PCPs who are not usually involved in research. The questionnaire was carefully developed and piloted by GPs and other PCPs, and therefore grounded in their clinical experience. While low survey response rates are common in primary care (38) and are known to vary between countries, those in our study compared favourably with those of a recent ICBP survey, in which response rates varied from 5.5% to 45.6% (30). There is evidence that responses to vignettes in surveys correspond well to clinical practice (44), and such surveys have previously been used to study primary care investigation preferences in patients who could have cancer (30,45).
Our study examined groups of PCPs as the unit of observation. Measuring the possible link between these groups' stated actions and national cancer survival creates the risk of an ecological fallacy (46), as any identified correlation may be an indicator of the effect of unmeasured system factors. While the demographic data that we collected included the gender of participants and the number of years that they had been in practice, we have found no equivalent data on national PCP populations that would allow us to assess how representative our samples were.
Most samples were taken from each local lead's own locality, and these may not have been representative of their nations as a whole (47). The recruitment method used in this study resulted in variable response rates, leading to a risk of nonresponse bias (38). However, the goal of 50 survey participants per country and more than 1000 respondents in total was achieved. We have no data on nonresponders as the survey was anonymous, however this anonymity might have reduced the risk of social desirability bias. It is possible that the PCPs with the most interest in diagnostic decision-making were the most likely to respond.

Interpretation of the results
Diagnostic testing and referral rates differed widely between participating countries. Whereas we might expect that referring a patient would result in the PCP being less likely to investigate the patient her-or himself, and vice versa, this was not confirmed in the survey: the more likely PCPs were to refer a patient, the more likely they were to organise a diagnostic test at the same time.
While we found that PCPs working in remote locations were more likely to take diagnostic action, this may be due to confounding, as doctors with young families have been found to be less likely to work rurally (48), and we found that more experienced PCPs were more likely to take diagnostic action.

Comparison with existing literature
In contrast to our study, an ICBP study showed a positive correlation between jurisdictional cancer survival rates and readiness to investigate or refer to secondary care in 4 out of 5 of its vignettes (30). However, it only studied five, relatively wealthy, countries (Australia, Canada, Denmark, Norway, Sweden and the United Kingdom). While the ICBP research found no health system characteristics that explained their findings, our study shows a link between PCP diagnostic actions and practice location, size of practice and PCP experience. Our finding of a lower likelihood of diagnostic action for PCPs working in larger practices corresponds with other study findings (49). Our evidence that PCPs working in more remote locations were more likely to take diagnostic action links across to evidence that such remote living is associated with more rapid cancer diagnosis and treatment following GP referral (50). While our data show higher diagnostic action rates in PCPs with more years since graduation, the opposite was found in a Finnish study (51), and no difference was found in a United Kingdom study (52). However, those two studies did not specifically study referrals for suspected cancer: it may be that experienced PCPs are more likely to recognise symptoms that suggest a possibility of cancer, even in the absence of 'red-flag' symptoms.
The extent to which respondents were gatekeepers, and needed to authorise their patients' access to specialist care and diagnostic tests (53)

Implications for research and practice
The poorer survival in some of our surveyed countries appears not to be due to delayed referral from primary care, suggesting that the cause may be related to cancer patients presenting later, or being less efficiently investigated, diagnosed and treated in secondary care. It may be that in those countries a stronger gatekeeper role for primary care, allowing secondary care to focus on patients that have a higher risk of cancer, would be more effective.
While it might be expected that PCPs who are more likely to arranging a diagnostic test would be less likely to refer their patient in the same consultation, we found the opposite was the case, and research is needed to explain this. We cannot explain our finding that PCPs in countries with a higher per capita healthcare expenditure were less likely to take immediate diagnostic action, and the reasons for this need investigation.

Conclusions
When given vignettes of patients with a low but significant possibility of cancer, more than half of PCPs across Europe would take diagnostic action, most often by ordering diagnostic tests. However, there are substantial between-country differences, which are linked with the healthcare organisation in those countries, as well as their levels of investment in healthcare. In countries with higher healthcare expenditure, immediate diagnostic action is less likely.

Availability of data and material
To avoid the risk of identification of individual participants, the datasets generated and analysed during the current study are not publicly available. However, they are available from the corresponding author on reasonable request. (BRC).

Figure 2
Percentage of PCPs in each country who would refer the patients to a specialist.

Figure 3
Percentage of PCPs in each country who would organise an investigation and/or refer the pat Figure 4 Percentage of PCPs in each country who would organise a diagnostic test.

Figure 5
Percentage of PCPs in each country who would refer the patients to a specialist.