Cancer awareness and socio-economic position: results from a population-based study in Denmark

Data on cancer awareness among the general population in Denmark were collected as part of the International Cancer Benchmarking Partnership (ICBP), Module 2 [17]. The survey consisted of a 20-minute computer-assisted telephone interview undertaken by trained native-language interviewers from the market research company Ipsos MORI using the Awareness and Beliefs about Cancer (ABC) measure [18].

In Denmark, a target study population of 1,000 respondents aged 30–49 years and 2,000 respondents aged 50 years and older was initially defined. Using the Danish CRS [16], a random study base was selected consisting of 20,000 persons aged 30–49 years and 40,000 persons aged 50 years and older. Among these 60,000 persons, a total of 6,570 persons (11.0%) were excluded because of research protection (i.e. publicly recorded rejection to be contacted for research purposes). For the remaining 53,430 persons, full name and complete address were obtained from the CRS and used by the Danish market research and consulting firm NN Markedsdata to obtain the phone number (landline and/or mobile) belonging to the identified person. Phone numbers could not be obtained for 6,309 (11.8%) persons, who were therefore excluded. Lastly, another 55 (0.1%) persons were excluded just before the data collection began, either because of a newly established research protection status, emigration from Denmark, or because the person had passed away. In total, 47,066 persons (78.4% of the study base) were thus eligible for being contacted to answer the ABC measure (Figure 1).

The Danish survey was conducted from 31 May to 4 July 2011, and each person was contacted on up to seven occasions at different weekdays and times. Interviews were not performed if the person was unable to speak or understand Danish.

The ABC measure was applied; the development and the validation of the ABC measure is described elsewhere [18]. It explores awareness of cancer symptoms by using recognition as well as recall [19]; awareness of risk factors for cancer; awareness of growing risk of cancer with age; awareness of perceived 5-year survival from cancer; access to a doctor; anticipated healthcare seeking for cancer symptoms; anticipated barriers to healthcare seeking; beliefs about cancer in general; beliefs about cancer screening and actual screening behaviour. In addition, the measure explores smoking status, self-rated health and personal experience of cancer (own or close relatives, if any). The English version of the ABC measure was translated into Danish in accordance with the WHO guidelines for translation procedures, which included forward and backward translations [20].

Data reported here include awareness of cancer symptoms using the recognition method, awareness of risk factors for cancer, awareness of growing risk of cancer with age and awareness of 5-year survival from four different types of cancer.

Information on SEP indicators was obtained by data linkage to Statistics Denmark [23]. For each person in the study population, we obtained information on seven SEP indicators: gender (female, male); age (30‐49, 50–69 and 70+ years); marital status (married/cohabiting, living alone); ethnicity (ethnic Dane, immigrant/descendant); level of education (low: ≤10 years, middle: >10 ≤ 15 years and high: >15 years) according to UNESCO’s International Standard Classification of Education (ISCED) groups [24]; occupation (in the labour force: employed and students, outside the labour force: unemployed, early retirement pensioner, disability retirement pensioner, personal or sick leave and retired: voluntarily retired person (special or old-age pensioner); and, lastly, OECD-modified disposable household income adjusted for number of adults and children in the household [25]. To level out yearly variation, this SEP indicator was calculated as an average for the preceding three years and categorised as low, middle and high income (low: ≤16,536 £/year, middle: >16,536 ≤ 33,095 £/year and high: >33,095 £/year) based on the 20%, 60% and 20% income distribution among the 60,000 persons in the study base. To examine the association between previous personal experiences with cancer and cancer awareness, we retrieved data from the Danish Cancer Registry on all registered cancer diagnoses (yes/no) for each person within the past 10 years [26]. For these register-based SEP indicators, missing data ranged from 0% for age, gender and information on cancer diagnosis to 3.9% for information about educational level for the study base of 60,000 persons. Data on close relatives with cancer (yes/no) and self-rated health (very good, good, fair, poor and very poor dichotomised into good and fair/poor) were obtained from the ABC survey.

Analyses were carried out using Stata version 13.1. Statistical analyses were performed with unweighted data, because weighting may introduce additional bias. Prevalence ratios (PRs) with 95% confidence intervals (95% CIs) were used to determine the association between SEP indicators and awareness of cancer symptoms, risk factors for cancer, growing risk of cancer with age and 5-year survival for four different types of cancer. Unadjusted analyses were carried out with each of the independent variables, and an adjusted model was used to control for possible confounding. PRs were chosen over odds ratios (ORs) as ORs may overestimate the associations when there is a high prevalence of the dependent variables [27].

The study was approved by the Danish Data Protection Agency (J. no. 2011-41-6237) and the Danish Health and Medicines Authority. In accordance with the Central Denmark Region Committees on Biomedical Research Ethics, the study needed no further approval (Report no. 128/2010).

To obtain inclusion of 1,000 respondents aged 30–49 years and 2,000 respondents aged 50 years or older, we approached a random sample of 11,297 persons. A response rate of 36.7% was achieved (Table 1); this was estimated as the number of completed interviews divided by the number of eligible persons made contact to. The SEP of the respondents and of the study base are shown in Table 2. A higher proportion of the respondents than of the entire study base of 60,000 persons were females, younger, married/cohabiting, ethnic Danes, had a high-level education, a high household income and were in the labour force. The differences were statistically significant at the 0.01 level.

The median number of cancer symptoms recognised by the respondents was nine out of 11. The associations between SEP and recognition of less than nine symptoms of cancer are presented in Table 3. In both the unadjusted and adjusted analyses, several of the SEP indicators were statistically significantly associated with awareness of cancer symptoms. Especially men, immigrant/descendants, people with low-level education, people outside the labour force, people with a low household income and people with no close relatives with cancer were more likely to recognise less than nine symptoms of cancer than women, ethnic Danes, people with a high-level education, people in the labour force, people with a high household income and people with close relatives with cancer, respectively.

The highest awareness of risk factors for cancer was found for smoking (96.5%) and using a sunbed (95.5%), while the lowest awareness was found for infection with HPV (23.6%) and intake of less than five daily portions of fruit and vegetables (41.0%).

The median number of risk factors recognised by respondents was nine out of 13. Certain characteristics were strongly associated with recognising less than nine risk factors for cancer; these include being a man, older, having a low-level education, or having a low household income (Table 4). Again, sensitivity analyses showed a stronger social gradient (data not shown).

In total, 24% of the respondents were aware that 70-year-olds are more likely to be diagnosed with cancer than 30-year-olds, 50-year-olds and people of any age. However, the majority (42%) responded that people of any age are equally likely to be diagnosed with cancer or that 50-year-olds are more likely (30%). Being a woman, having a low level of education, and a low income were associated with non-recognition of growing risk of cancer with age (Table 5).

The 5-year survival from bowel, breast, ovarian, and lung cancer was correctly identified by 42%, 49%, 9% and 19% of the respondents, respectively. For ovarian and lung cancer, a large majority (86 and 78%, respectively) of the respondents overestimated the 5-year survival, whereas almost half of the respondents underestimated survival from breast cancer. The distributions of the respondents’ estimations (underestimating, correctly estimating and overestimating) are shown in Figure 2.

Table 6 shows the associations between the SEP indicators and underestimation/overestimation of the 5-year cancer survival. People outside the labour force were more likely than people within the labour force to wrongly estimate the 5-year survival from breast cancer, and men were more likely to wrongly estimate the 5-year survival from lung and ovarian cancer than women. Furthermore, people with a low and middle income and people with no close relatives with cancer were less aware of the 5-year survival from bowel cancer than people with a high income and with close relatives with cancer.

A key strength of the present study was the use of the Danish CRS. All Danish residents are registered in the CRS which contains complete information on any Danish resident’s date of birth, gender, migration, etc. Owing to our use of the CRS, we were able to define a study base of 60,000 persons, a representative sample of the entire Danish population aged 30 years and older. Furthermore, the use of the CRS and the data linkage to a range of register-based SEP indicators provided us with precise and valid insight into variables that may be related to cancer awareness. Naturally, the SEP indicators capture correlated aspects. Still, since the correlation is not a hundred percent, each indicator contributes with unique information about the association with cancer awareness.

To analyse associations between SEP and cancer awareness of symptoms and risk factors, cancer awareness was categorised into low/high using the median split procedure. One of the shortcomings of this procedure is that the median is contingent upon the particular sample on which it is based [50, 51]. Thus, respondents categorised as having a low cancer awareness in this sample may be categorised as having a high cancer awareness in another sample. However, sensitivity analyses using both awareness of less than five and less than seven cancer symptoms and risk factors showed a similar, but intensified social gradient in cancer awareness.

A limitation of the study was the modest response rate. Only 36.7% of the persons whom we made contact to agreed to participate in the study. Unfortunately, response rates have been declining over the past decades and telephone surveys have been particularly affected by this decline [52]. However, by collecting data using a telephone interview, the respondent did not have the possibility to look for information elsewhere. This advantage could not have been achieved with paper-based or web-based surveys. The respondents completing the ABC measure were more often females, younger, married/cohabiting, had a high-level education and a high household income than people in the study base. As a consequence, selection bias may in some way affect the generalisability of the findings since women and persons with a high-level education and a high household income were generally more aware of cancer symptoms and risk factors than men and persons with a low educational level and a low household income. Consequently, the actual awareness level in the population is most probably lower than estimated here.