Abstract
Background: We examined the ability of adult Canadians to recall cardiovascular disease risk factors to determine the associations between their ability to recall risk factors for cardiovascular disease and their socioeconomic status.
Methods: This study used the database assembled by the Canadian Heart Health Surveys Research Group between 1986 and 1992 - a stratified representative sample comprising 23 129 Canadian residents aged 18 to 74. Nurses administered a standard questionnaire asking respondents to list the major risk factors for cardiovascular disease: fat in food, smoking, lack of exercise, excess weight, elevated blood cholesterol and high blood pressure. Six logistic regressions examined the multivariate associations between ability to recall each risk factor with education, income adequacy, occupation, sex, age, marital status and province of residence.
Results: More people knew about the behaviour-related risk factors for cardiovascular disease than about the physiologic risk factors: 60% recalled fat in food, 52% smoking and 41% lack of exercise, but only 32% identified weight, 27% cholesterol and 22% high blood pressure. Education was the socioeconomic status indicator most strongly and consistently associated with the ability to recall risk factors for cardiovascular disease. The odds ratios of reporting an association of the risks between people with elementary education and those with university degrees varied between 0.16 (95% confidence interval 0.12 to 0.22) for lack of exercise to 0.55 (95% confidence interval 0.39 to 0.77) for smoking.
Interpretation: People in categories at greater risk of cardiovascular disease, such as those aged 65 or more or those with only elementary education, are less able to recall important cardiovascular disease risk factors.
In Canada, as in many Western countries, the death rate from cardiovascular disease (CVD) for men aged 45 to 64 years has been declining since the beginning of the 1970s.1 A corresponding decrease in the prevalence of the main CVD risk factors is thought to have contributed to this trend.2 Prevention trials and information campaigns encouraging people to reduce their risk of developing CVD have been launched during the past 30 years. These programs have focussed primarily on disseminating information about CVD risk factors such as smoking, eating too much fat, not getting enough exercise, being overweight, having high blood pressure and having elevated blood cholesterol levels.3 We examined the ability of a representative sample of Canadian adults to identify the major CVD risk factors.
Although knowledge alone is insufficient, it is thought to be a prerequisite for making sound decisions about health.[4, 5] Indeed, many theories of behaviour modification rely on a person's knowledge or their access to information, such as a person's perceptions of risk and severity in the Health Belief Model,[6, 7] self-efficacy beliefs and outcome expectations,8 and behavioural, normative, control beliefs.[9–11] More specifically, knowledge that a particular condition is a CVD risk factor has been identified as a prerequisite for change and is often targeted by prevention programs.[3, 12] Monitoring the population's knowledge of risk factors can help guide public health programs.
Of the few studies that measured how much the general population knows about CVD risk factors, most were done in the context of CVD primary-prevention community trials,[12–20] and only a few used national population samples.[21–26] Although most studies are merely descriptive accounts of what a population knows, some have tried to examine trends in the degree of knowledge people have about CVD risk factors,[17, 22–24] and others to identify the correlates of that knowledge. Only 5 have used multivariate analysis techniques.[14–16, 18, 21, 27] The main predictors of whether a person will know the CVD risk factors have been identified in multivariate studies. They are ethnicity,[14, 16, 18, 21] education,[14, 16, 18, 21, 27] age,[14, 16, 21] sex,[14, 21, 27] income,21 marital status,21 source of medical care21 and geographical region of residence.21
One of 2 approaches is generally used to measure a person's knowledge of CVD risk factors. The first uses probes; respondents are asked whether each of a series of actions would place them at risk for CVD. The second approach asks respondents to list everything they can think of that would reduce their risk of CVD or what they think are CVD risk factors. The former method leads to consistently greater estimated levels of knowledge.[22, 28]
This study had 2 objectives: to identify segments, particularly socioeconomic segments, of the Canadian population that are not aware of the specific CVD risk factors and to estimate the independent associations between knowing about CVD risk factors and indicators of socioeconomic status (SES) - education, income level, and occupation or main activity. A 1992 study showed that the level of education completed is a stronger predictor of the prevalence of CVD risk factors than either income or occupation.29
Method
Subjects for this study included 23 129 men and women who were interviewed for the Canadian heart health surveys that took place in each Canadian province between 1986 and 1992. Each provincial survey targeted people between the ages of 18 and 74 years who did not live in an institution or on a military base and (except in Manitoba) were not aboriginal and living on a reserve. Each provincial research team designed a stratified 2-stage replicated probability sample selected from the provincial health insurance registry. In each province the targeted sample, with equal numbers of men and women, included 1200 respondents aged 18 to 34 years, 600 respondents aged 35 to 64 years and 400 respondents aged 65 to 74 years.
Those selected (29 855) were telephoned and asked to make an appointment for a 40- to 60-minute interview at home; 23 129 (77%) agreed to participate. Fewer men (76%) than women (79%), and fewer people aged 65 years and older (75%) than people aged 18 to 64 years (78%) agreed to participate. A detailed description of the survey method has been published elsewhere.30
The questionnaire used in the provincial surveys was developed by a core group of researchers using validated questions from other surveys.30 A French translation was available. The risk factors used in this study were all derived from respondents' answers to the following question: "Can you tell me the major causes of heart disease or heart problems?" All elements spontaneously listed by the respondents were classified into 15 categories. Our analysis focuses on the ability to identify 6 modifiable risk factors: fat in food (including poor diet, too much fat and too much cholesterol); smoking; lack of exercise; excess weight; elevated blood cholesterol level; and high blood pressure (including hardening of the arteries and arteriosclerosis). Not mentioning a risk factor was interpreted as not knowing its association with heart disease. As noted above, this procedure leads to an underestimation of the level of knowledge. All respondents were asked the question so there are no missing values. The reliability of the answers to this question was not estimated by a test-retest procedure.
Three socioeconomic variables (education, income level and occupation) and 4 sociodemographic variables (sex, age, marital status and place of residence) were used as covariates. There were 3 categories of income level: high, middle and low. (High income is considered to be one person with an income of $25 000 or more or 2 or more people with an income of $50 000 or more; middle income is considered to be one person with an income between $12 000 and $24 999, 2 people with an income between $12 000 and $49 999, or 3 or more people with an income between $25 000 and $49 999; low income is considered to be 1 or 2 people with an income of less than $12 000 or 3 or more people with an income of less than $25 000.)
Because of the complex sampling design of the surveys, standard errors of estimates provided by standard statistical software are all biased. To address this issue, all bivariate and multivariate estimates were jackknifed in an SAS routine to provide exact standard errors (SE) for all estimates.[31–33] Hypothesis testing and computation of confidence intervals (CI) were performed with α-level sets at 0.05. All analyses were performed using weighted data. A socioeconomic or sociodemographic characteristic was judged to be associated with knowing a risk factor when the confidence intervals for some categories of that characteristic did not include one. Six multivariate models identifying the individual characteristics associated with knowing the CVD risk factors are presented, one logistic regression for each risk factor. All equations were computed using the complete sample of 23 129 respondents, including a category for covariates with missing values. The odds ratios (OR) associated with missing categories are not reported in the table. Because the bivariate relationships between education, income level and occupation were moderate (all tau-b coefficients [tau-b is a rank correlation coefficient] were between 0.30 and 0.40), there was no multicollinearity when the 3 were included in the same multivariate equation.
Results
When respondents were asked to name CVD risk factors (Table 1), they mentioned fat in food (60%) most often, followed by smoking (52%), lack of exercise (41%), excess weight (32%), elevated blood cholesterol (27%) and high blood pressure (22%).
There was a bivariate association among 3 variables - education, occupation and region of residence - and knowing each of the 6 risk factors. Income level was not associated with knowing that either excess weight or high blood pressure are risk factors. The sex of the respondents was associated only with knowing that excess weight is a risk factor. Marital status was not associated with knowing that smoking, elevated blood cholesterol or high blood pressure are CVD risk factors. Age group was not associated with knowing that high blood pressure is a risk factor.
Analysis of the multivariate associations (Table 2) between sociodemographic and socioeconomic variables and identification of CVD risk factors revealed that controlling for the other variables women were more likely than men to know that fat in food (OR 1.2, 95% CI 1.07 to 1.36) and excess weight (OR 1.48, 95% CI 1.20 to 1.83) are CVD risk factors. There was an association between age and knowing each of the risk factors, except high blood pressure. In general, people aged 65 to 74 years were less likely to mention a given risk factor than people aged 18 to 24 years.
When the other individual characteristics were controlled for, the strongest and most consistent association was between education and knowing CVD risk factors. However, it was only for lack of exercise, fat in food and, to a lesser extent, elevated cholesterol and high blood pressure that the increment in knowledge increased steadily as education increased. For smoking, the most significant contrast was between those with only elementary school education (OR 0.55, 95% CI 0.39 to 0.77) and those with a university degree. Those who had completed secondary school were most likely to identify excess weight as a risk factor (OR 1.36, 95% CI 1.16 to 1.60) followed by those with a university degree and those with some high school; those with only elementary school education were least likely to identify excess weight (OR 0.52, 95% CI 0.39 to 0.69) as a risk factor. Income level was not associated with knowing that excess weight or elevated blood cholesterol are CVD risk factors. For the other factors, the most significant contrast was between those with high and low income levels. Finally, among the 3 socioeconomic variables, the weakest association was between occupation and knowing about CVD risk factors when controlling for other individual characteristics. Occupation was not associated with knowing that elevated blood cholesterol or high blood pressure are risk factors and was marginally associated with knowing that fat in food and smoking are risk factors. For excess weight, there was a significant contrast between professionals and all other categories of occupation, whereas homemakers (OR 0.92, 95% CI 0.69 to 1.22) and professionals (OR 1) were more likely than other groups to know that lack of exercise is a risk factor.
Discussion
Two features of these results are most important. First, all segments of the Canadian population are missing some information about the different CVD risk factors. Second, the results of the multivariate analyses clearly identified particular groups of Canadians who still do not know about specific CVD risk factors.
The results presented here clearly show that knowing about the main modifiable CVD risk factors is strongly related to an individual's SES. This observation is supported by previous findings.[14, 16, 18, 21, 27] It also corroborates the findings of Choinière and colleagues (page S13) that behavioural CVD risk factors are more prevalent among Canadians of low SES.
After more than a decade of mass-media campaigns, slightly more than half of our sample spontaneously identified eating habits and smoking as CVD risk factors and about 40% mentioned that lack of exercise and CVD are related. In Canadian and American surveys on national samples respondents were usually asked whether or not they thought given behaviours or physiologic states could affect their risk of CVD.[21, 22, 24] Not surprisingly, people who were asked this question could identify more risk factors than could those who were asked merely to name CVD risk factors. The measures of knowledge used in this study were probably biased by the prominence of the risk factors in the respondents' minds. It should be kept in mind, however, that both probed and unprobed measures of knowledge are widely used and considered to be valid.
Studies that have used unprobed questions about CVD risk factors have been conducted on community samples as part of the evaluation of CVD-prevention community trials. In a sample from the comparison city of the Pawtucket Heart Health Program, a survey done in 1987 and 1988 found rates of knowledge similar to those reported here.17 Folsom and associates20 also reported similar results for the 1985 to 1986 baseline survey of the Minnesota Heart Health Program in the Minneapolis metropolitan area. Finally, Avis, McKinlay and Smith27 reported that many more people knew that exercise and fat in food are risk factors; however, their sample comprised mostly white people from the Boston area. The results of our study clearly indicate that although the situation in Canada is probably comparable to that in other Western countries, large segments of the population are still only aware of some of the CVD risk factors.
Of course, a limitation of this finding is that knowledge was estimated using a general question that treated not mentioning a specific risk factor the same as not knowing it. This approach certainly underestimates the level of knowledge in the population. However, it does allow us to identify the most commonly known risk factors.
In an attempt to create a summary index of knowledge, the intercorrelations among knowledge items were checked. Most of these correlations were too low to group knowing the risk factors into an overall knowledge index. However, analysing them as different items enabled us to identify gaps in the population's knowledge.
Results presented here clearly identify 2 types of risk factor. The behaviour-related risk factors - fat consumption, smoking and exercise - were mentioned more often than the physiologic ones - high blood pressure and elevated blood cholesterol. The proportion of people that knew that excess weight was a risk factor was mid-way between the proportion that knew that the behaviour-related variables are risk factors and the proportion that knew that the physiologic ones are.
Fewer people knew that physiologic factors are associated with an increased risk of heart disease than knew that behavioural risk factors are associated with it. About 1 in 5 Canadians reported that high blood pressure is associated with an increased risk of CVD, slightly more than 1 in 4 knew that elevated blood cholesterol increases the risk of CVD, and slightly less than 1 in 3 knew that excess weight is associated with CVD. Again these figures are quite similar to those found in the recent literature.27
There is no easy way to explain why behaviour-related risk factors are more likely to be identified than physiologic ones. One can speculate that the behavioural risk factors of CVD are much simpler to disseminate through mass media than the more complex physiologic risk factors. The latter may be more effectively delivered by health care professionals in a clinical setting.34 Further studies are needed to determine whether the generally low number of people who knew about the physiologic risk factors can be attributed to physicians transmitting that information only to people who actually have one of those risk factors, or whether only a few physicians are systematically informing all their patients about CVD risk factors. Another explanation may be that behaviour-related risk factors are more easily understood than physiologic ones. Previous research has shown that people are not likely to conduct an extensive search for information when making health-related decisions;5 they are more likely to select the most accessible or familiar option.5 It is also possible that people learn the easily accessible behavioural information and make no effort to process other information. Or, it might be attributed to the emphasis put on making the public aware of the behavioural risk factors since the middle of the 1970s.
The second important feature of these results is that they identify segments of the population that are less likely to know about CVD prevention. People of low SES and older people were less likely than younger people or people of high SES to identify CVD risk factors. Women were more likely than men to identify association between either weight or fat in food and CVD. Clearly, the segments of the population that are at the greatest risk of developing CVD are those who have received or retained the least amount of information about its prevention.
Results of research on communicating information[5, 35, 36] are useful in the study of associations between sociodemographic characteristics and knowledge. Individual characteristics, particularly those related to age, sex and SES, have been consistently shown to influence the way a person seeks information as well as their level of knowledge. Other factors, such as having a special interest in health because of a previous illness, could also play important roles.37
Another interesting aspect of our results is that the relationship between SES and risk factor is not the same for each of the risk factors. Adler and colleagues38 have suggested that even if education, income and occupation are interrelated indicators of SES, they do not completely overlap. One might speculate that education is a reflection of living conditions and access to resources during the early part of a person's life, whereas income reflects actual conditions and opportunities. As well, occupation provides cultural environments and access to information that are different from the family and neighbourhood culture and access to information. Our results suggest that all 3 aspects influence a person's knowledge of CVD risk factors. However, because education is a reflection of living conditions during the early part of a person's life, and because education is the indicator most consistently associated with knowing CVD risk factors, then conditions during early life are likely to be most predictive of access to and retrieval of health information.
One finding evident from our work is that health-promotion campaigns should consider individual differences and include distinct messages for subgroups of the population, at least those defined by age and education level. Other variables identified as determinants in the communication process (e.g., method of disseminating information or source of information) should also be considered. It seems that CVD awareness programs have been successful in reaching some segments of the population, but programs need to be developed for the most disadvantaged sectors of the population. Ultimately, we should also recognize the limit of an approach based strictly on persuasive communication and acknowledge that other strategies aiming, for example, at modifying the social or political components of a person's environment should be an integral part of interventions targeting disadvantaged sectors of the population.[39–42]
We thank the members of the Canadian Heart Health Surveys Research Group for access to the data of the Canadian heart health surveys: C. Balram, P. Connelly, A. Edwards, D. Gelskey, K. Hogan, M. Joffres, R. Lessard, S. MacDonald, D. MacLean, E. Macleod, M. Nargundkar, B. O'Connor, G. Paradis, A. Petrasovits, B. Reeder, R. Schabas, S. Stachenko, T. Young. We also acknowledge the helpful comments made on previous drafts by the members of this monograph group, especially Doreen Neville, the monograph coordinator.
Funding has been provided in part by the National Health Research and Development Program, Health Canada; provincial ministries of health and the Heart and Stroke Foundation of Canada.