Health Education Research, Vol. 14, No. 2, 209-223,
April 1999
© 1999 Oxford University Press
Predictors of the prevalence of tobacco use among Francophones and Anglophones in the province of Ontario
Social Evaluation and Research Department, Addiction Research Foundation, 100 Collip Circle, Suite 200, London, Ontario N6G 4X8, Canada
 |
Abstract |
|---|
 Top Abstract IntroductionMethodResultsDiscussionConclusionReferences |
|---|
This study examines Francophone/Anglophone differences in levels and patterns of tobacco use and associated risk factors in the province of Ontario, Canada. Estimates are derived from the self-administered portion of the 1990 Ontario Health Survey, a random probability survey of Ontario residents. The sample consists of 1127 Francophones and a random subset of 4023 Anglophones. Evidence, unique to Francophones, indicates a steady age-related decline in the median age of onset of daily cigarette consumption. Unlike Anglophones, multivariate results reveal that Francophones age 35–44 are significantly more likely than all other age groups to smoke cigarettes daily and to smoke a pack or more daily. Sub-groups within Ontario's Francophone community may be experiencing health-related risks associated with frequent and heavy consumption of cigarettes, and thus may be in need of addiction-related services.
|
Introduction |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionConclusionReferences |
|---|
Recent population surveys on the health of Canadians have shown a greater propensity to smoke among Francophones (Letourneau, 1988
; Health and Welfare Canada, 1992a; Millar, 1992). Much of this research, however, has been largely descriptive in nature and has occurred at the national level or in the province of Quebec. A concerted effort has been lacking to investigate the prevalence and patterns of smoking behaviour and associated risk factors among Francophone Ontarians (Franco-Echanges, 1991). A greater understanding of smoking patterns and behaviour among Francophone Ontarians is important for a variety of reasons. Outside of Quebec, Ontario has the largest population of French-speaking persons in the country. In 1991 the number of individuals in Ontario with a French mother tongue stood at approximately 500 000, representing just over 5% of the total population. In total, 23 areas in the province were designated for French language services, areas where French-speaking persons comprise at least 10% of the local population or in urban areas where they number at least 5000.
Statistics indicate that a large percentage of Francophones in Ontario are steadily becoming assimilated into the dominant Anglophone society. An increase in the proportion of mixed marriages, the declining influence of the Catholic church in French language instruction and family life, and a reluctance on the part of French-speaking couples to send their children to French language schools have contributed to an ever increasing trend toward reduced language retention rates (Mougeon, 1984a,b; Mougeon and Heller, 1986). In some ethnic-linguistic minority groups, `acculturation stress' resulting from feelings of alienation, marginalization and identity confusion has been linked to lower mental heath (Berry et al., 1987; Kaplan and Marks, 1990) and elevated levels of substance use (Padilla, 1980; Burnam et al., 1987). Interestingly, a recent analysis of 261 Canadian census divisions found that `proportion Francophone' ranked third in a list of 10 major predictors of suicide (Hasselback et al., 1991).
Social isolation from one's relatives and friends is also an important predictor of substance use, and may be particularly salient in cultural groups which place a lot of emphasis on close family relations. Evidence suggests that Francophones maintain closer family and kinship ties than Anglophones. Burch's analysis of the 1985 General Social Survey found that respondents with a French ethnic background were significantly more likely than those with an English background (as well as other major ethnic groups) to see their parents and siblings at least once a week (Burch, 1988). Only among Italians was face-to-face contact higher. This relationship held after controlling for demographic variables (including geographic mobility) as well as indicators of socioeconomic status.
Finally, Francophones differ from their Anglophone counterparts on a number of key demographic and socioeconomic indicators (Gilbert and Langlois, 1994; Ontario Office of Francophone Affairs, 1996) which bear a strong association with tobacco use (Health and Welfare Canada, 1992a; Flewelling et al., 1993). Francophone Ontarians have higher divorce rates, are more likely than Anglophones to be geographically mobile and rank well behind most other ethno-cultural groups in Ontario in post-secondary education.
The range of adverse social, economic and physiological effects of tobacco consumption are well known. Studies have linked cigarette smoking (including the presence of second hand smoke) to various forms of cancer, respiratory disease, cardiovascular disease, spontaneous abortions and still births, and low birth weight among newborn children (Gordon and Kannel, 1982; Stellman and Garfinkel, 1986; Stachenko et al., 1992; Health Canada, 1995). Compared to non-smokers, mortality rates among male and female smokers are 70 and 30% higher, respectively. In 1991 it was estimated that 13 575 deaths in Ontario were indirectly attributed to tobacco use (Addiction Research Foundation, 1995). Smoking also results in lost productivity in the workplace due to smoking-related illnesses (Douville, 1990).
Canada has one of the highest tobacco consumption rates in the world (USDHHS, 1988) and is committed to developing a national program to achieve a non-smoking environment that will assist in producing a generation of non-smoking Canadians by the year 2000 (Millar, 1992). To achieve this goal, increased knowledge about possible risk factors that are associated with different types of smoking behaviour are required. A greater understanding of the differences between Francophones and Anglophones in tobacco use will guide the development of smoking prevention and smoking cessation programs sensitive to the unique situation of each group.
The primary objective of this study is to explore within a multivariate framework the extent to which traditional predictors of cigarette consumption among Anglophone Ontarians serve to adequately predict patterns of use among the province's Francophone population. Meeting this objective will help to identify sub-groups within the Francophone population possibly in need of specialized prevention and intervention programs as a result of their unique levels and patterns of use.
|
Method |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionConclusionReferences |
|---|
Sample and weighting
Data for this study are obtained from the Ontario Health Survey (OHS) conducted in 1990 by the Ontario Ministry of Health (Ontario Ministry of Health, 1992
). The OHS is based on a multi-stage stratified cluster sampling design. Sampling was carried out to provide reliable estimates of the population at the level of the Public Health Unit. A sample of 35 650 households was selected from a sample of geographic clusters. Within each household, interviews were conducted with a household member who would provide information (e.g. age, sex, marital status and relationship to household head) for all household members for the interviewer-completed portion of the survey. A self-report form was left for all household members age 12 and over to complete. Questions pertaining to alcohol and other drug use appeared on this form only. In total, over 63 000 individuals completed the interview portion of the survey (a response rate of 87%) and just over 49 000 completed the self-report portion (a response rate of 77%). This study focuses on the smoking behaviour of a sample of Francophones and a sample of Anglophones age 16 and over who completed the self-report portion of the OHS. Certain questions pertaining to general lifestyle and health were not asked of those age 12–15.
According to the French Language Services Act, a Francophone is defined as someone who reports first learning French as child and who still understands or speaks the language. Unfortunately, the OHS did not ask this question. Instead, the survey asked respondents to list up to four ethnic groups with whom they closely identify and to report the language spoken most often in the home.
To isolate a sample of individuals who closely conform to the definition under the Act, survey respondents were designated as Francophone if they reported French as their first or second most important ethnic group and if they reported French as the language spoken most often in the home. Anglophones (defined in the dictionary as English-speaking persons) consisted of those who identified with an ethnic group other than French and who reported speaking English most often in the home.
Our rationale for choosing a combination of ethnic identity and language use as criteria for selecting our sub-samples was based on several considerations. First, used in isolation, language spoken most often in the home is a biased and vague indicator of linguistic identity because it precludes the possibility that two languages can be used with equal frequency in the home. Moreover, such a measure does not tell with whom a respondent is speaking French or English (e.g. the spouse, children or other relatives). Second, many households in Ontario consist of French–English couples. Within these households, spouses learning a language other than French as a child and belonging to non-French ethnic groups would be incorrectly classified as `Francophone' (as defined under the 1986 Act) if, as a result of marriage to a Francophone, French was the dominant language spoken in the household. By combining French as the dominant household language with French ethnic identity, we move in the direction of obtaining a `cleaner' sample of respondents. Excluded from our derived sample were persons who reported speaking French most often in the home but who reported an ethnic identity other than French. In total this number represented 70 cases.
All multivariate analyses in this study were performed using sample weights that were rescaled so that they sum up to the actual number of cases in the sample. These are called analytic or average weights and are designed to yield more meaningful tests of statistical significance. For those age 16 years and older, analytic weights resulted in 1265 Francophones and 37 525 Anglophones.
Analytic weights were adjusted for design effects (Lee et al., 1986; Goel, 1993). These adjustments are necessary to correct for the problem of inflated type 1 error rates. In complex survey designs such as the one adopted in the OHS, clustering of sampling units results in reduced variance and hence reduced standard errors. The reduced variance may lead to inflated estimates of statistical significance (increased type 1 error rates). For each ethno-linguistic group age 16 and over, the calculated analytic weight was divided by the square root of the design effect. In the OHS the design effect was 2. Application of the adjusted weights to the data reduced the sample size for the Francophone group to 1127. To facilitate a comparison of the multivariate results across both groups, a random subset of 5691 Anglophones was selected from the Anglophone sample. The adjusted weights reduced this number to 4023.
Measurement
Variables of interest for the multivariate analyses are listed and defined in Table I. Because of severely skewed distributions, dependent measures are collapsed into dichotomous categories. All independent variables are treated as categorical. Nominal-level predictor variables with more than two categories are transformed into dummy variables and assigned reference or comparison groups. Although categorization of continuous covariates entails some loss in efficiency in parameter estimation (Morgan and Elashoff, 1986), there are certain advantages. Categorical predictors yield odds ratios that are more easily interpretable and allow the investigator to check for non-linear patterns in the data.
|
Dependent variables include daily cigarette consumption and consumption of a pack or more of cigarettes daily.
Independent variables were selected on the basis of correlates of substance use examined previously within a general population setting (Adlaf, 1991; Flewelling et al., 1993; Adlaf et al., 1994) and grouped under five major headings: regional/geographic (includes place of residence and geographic location), demographic (includes age, sex and marital status), socioeconomic (includes, education, occupation—based on 1980 Standard Occupational classifications, dwelling status—rent/own, number of rooms in dwelling, and need repairs for dwelling not including remodelling or renovations), health and social support (includes perceived stress level, perceived health status, presence of close confidant—can confide in at least one person about problems, and voluntary membership in clubs and organizations—church and school groups, labour unions or social, civic and fraternal clubs), risk-taking behaviour (includes heightened sexual activity—sexual intercourse before age 15 or two or more sexual partners in the past 12 months and serious suicidal thoughts at least once in lifetime), social influence variables related to substance use (includes number of friends who smoke) and substance use variables (include drinks alcohol daily). Based on the work on measurement of social class by Liberatos (Liberatos, 1988), we chose not to create a single index of socioeconomic status.
Because of small n sizes, we collapsed certain variable categories. To contrast currently married respondents with all others, the separated, divorced and widowed groups were combined with the never married. Respondent education was collapsed from five categories to three (primary or some secondary, completed secondary and at least some post-secondary) and occupation from six categories to three (professional, non-professional and not employed). Central and southwest regions of the province were combined. For detailed information on the location of specific Public Health Units within each of the broad geographic regions, interested readers should refer to the Ontario Health Survey User's Guide Documentation Manual (Ontario Ministry of Health, 1992). Perceived stress level was reduced from four to two categories.
Statistical analysis
The standard 
2 statistic with 
set at 0.05 was used to test for significant differences between Francophones and Anglophones on a number of measures of smoking behaviour. We also use life table survival methods to examine the timing of onset and cessation of daily cigarette use for Francophones and Anglophones by broad age groupings at the survey date. Statistically significant differences in survival distributions between age groups is assessed using the Lee–Desu statistic (Lee and Desu, 1972). This statistic indicates whether two or more groups are likely to come from the same survival distribution and it compares all groups at all age intervals simultaneously. The Lee–Desu statistic is asymptotically distributed as 2 with g – 1 degrees of freedom, where g represents the number of groups being compared.
Logistic regression is used to model the effects of the hypothesized predictor variables on each of the dependent variables. This method is most appropriate when the dependent variable(s) is a skewed dichotomous (binary) outcome (Morgan and Teachman, 1988). In logistic regression, the dependent variable is the logarithm of the odds of falling into one category versus another or the odds of experiencing an event versus not experiencing an event. To facilitate interpretation, estimated effects of the independent variables may be expressed in terms of odd ratios calculated by exponentiating the unstandardized raw ß coefficients. Odds ratios may be interpreted as the relative likelihood of experiencing the event of interest with each unit increase in the independent variable, controlling for the other variables in the model. Odds ratios of less than 1 indicate a negative relationship (i.e. one that is less likely to occur) and ratios above 1 indicate a positive relationship (i.e. one that is more likely to occur). Significance is calculated by dividing the unstandardized ß coefficient by its standard error. The result is a t value.
Caution must be exercised in making direct comparisons of results across the two groups given the different sample sizes involved. However, it is possible to make general comparisons. Statistical significance of the difference in magnitude between raw coefficients from each group can be assessed with the following Z statistic:
|
|
is the group sample size.
Measures of model fit are assessed using the Pearson 2 goodness of fit (GOF) statistic and a pseudo-R2 somewhat analogous to R2 in ordinary least squares regression. The Pearson 2 statistic compares the expected distribution of cases based on our model with the actual distribution of cases. This value should be non-significant. Because the Pearson 2 statistic has no upper boundary, we also include a pseudo-R2 measure of GOF (Walsh, 1987). This measure is based on the Pearson 2 statistic, and provides a standardized measure of fit bounded by 0 and 1 with values approaching 1 indicating a good fitting model.
Prior to the multivariate analysis, Pearson zero-order correlation coefficients were generated for all combinations of hypothesized predictors. The vast majority of coefficients fell below a value of 0.70, considered to be problematic for efficient parameter estimation. Number of persons living in the household and place of birth were not included in the logistic regression analyses. Preliminary work found that these variables contributed very little to overall model fit. Respondent income was also excluded. In the OHS a large percentage of respondents refused to report their level of income. Estimates of income for this group were derived using related variables in the data set containing more complete information (i.e. a high percentage of valid cases). Missing cases were handled using the list-wise deletion option.
|
Results |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionConclusionReferences |
|---|
Results from Table II
reveal higher life time use of cigarettes on the part of Francophones; roughly 35% of Francophones report never having smoked compared to 44% of Anglophones. Results also show a higher proportion of current daily smokers (31.3%) among Francophones compared to Anglophones (26.7%). If we consider just smokers, Francophones are more likely than Anglophones to start smoking daily before the age of 16 (39.5 versus 33.7) and are slightly more likely to smoke in excess of 20 years (46.9 versus 44.1%). Interestingly, both groups do not differ in terms of the quantity of cigarettes smoked daily, with 59% of Francophones smoking a pack or more of cigarettes a day compared to 60% of Anglophones. Results from Table II should be interpreted with caution. The 2 statistic is heavily reliant on sample size. In large samples statistical significance may be achieved even when the magnitude of the difference (between two proportions) is quite small.
|
Ethno-linguistic group differences in smoking behaviour are most pronounced for the age group 35–44 (See Table III
). Among Francophones, more than one in every three respondents in the age group 35–44 (36.5%) reports smoking at least a pack of cigarettes daily compared to one in every five Anglophones (20.6%). Roughly 46% of Francophones age 35–44 report smoking cigarettes on a daily basis compared to 30% of Anglophones. Differences also emerge for those age 65 and over. Results show that 12.4% of Francophones age 65 and over smoke at least a pack of cigarettes daily compared to 6.3% of elderly Anglophones. Roughly 18% of the Francophone elderly report daily consumption compared to a value of 13.2% for Anglophones.
|
Table IV
presents cumulative proportions of Francophone and Anglophone respondents not having started smoking cigarettes on a daily basis to selected ages by broad age groupings at the time of the survey. Results demonstrate a greater propensity among Francophones to begin smoking at an early age; however, this finding is not uniform across all age groups. Anglophone/Francophone differences in smoking onset rates are very small for the age groups 45–54 and 55–64. The largest contrast in rates occurs for the age groups 25–34, 35–44 and 65 and over where Francophones are more likely to have reported started smoking daily at all ages. For example, by 16 years of age only 56.9% of Francophones age 35–44 have still not started smoking daily compared to 70.4% of Anglophones age 35–44. Interestingly, the median age at onset of daily cigarette smoking among the Francophone group shows a steady age-related decline falling from 21.7 years of age for those age 65 and over to 16.8 years for those age 25–34. Because of truncation, it is not possible to calculate a median survival value for those under age 25. Among Anglophones, the median falls from an unusually high value of 45 plus for those age 65 and over to 21.61 for those age 55–64 and then remains fairly constant across the younger age groups.
|
Table V
presents cumulative proportions of Francophone and Anglophone smokers not having stopped smoking cigarettes on a daily basis to selected years since age at first use by broad age groupings at the time of the survey. Results here should be interpreted with caution. In the OHS, respondents classified as current daily smokers did not report prior episodes of quitting for periods of at least 1 year. Quitting behaviour was only recorded for former daily smokers. As a result, some amount of underestimation in quitting behaviour should be expected.
|
With this caution in mind, strong age-related differences are found in terms of the propensity to quit smoking on a daily basis. Among Francophones, results show a greater reluctance to quit smoking daily for those age 35–44, 45–54 and 65 and over. For example, at 12 years duration, 85.9% of Francophone smokers age 35–44 had still not reported quitting their habit compared to 75.6 of Anglophone smokers. Interestingly, our results provide some indication of a reversal in smoking patterns for the younger age groups. Francophone smokers age 25–34 appear more inclined to quit smoking daily compared to their Anglophone counterparts.
Table VI presents logistic regression analyses for two different measures of tobacco consumption by selected demographic, economic and psycho-social characteristics of the respondent. Odds ratios are listed first followed by the standard errors associated with the unstandardized ß coefficients. To conserve space, the latter are not shown. Levels of significance are indicated using asterisks.
|
For both ethno-linguistic groups, region of residence and place of residence (urban/rural) do not appear to be important correlates of daily cigarette consumption and quantity of cigarettes consumed daily.
Results do show interesting differences by sex. For example, Anglophone males appear to be significantly less likely than Anglophone females to smoke cigarettes daily. For Francophones, sex differences in daily consumption do not emerge. Results also show that Francophone males are twice as likely as Francophone females to smoke a pack or more of cigarettes daily. For Anglophones, sex differences in the quantity of cigarettes consumed do not emerge.
Age emerges as a fairly strong correlate of smoking behaviour. For Anglophones and Francophones, respondents less than age 25 are considerably less likely than those in the reference group (age 45–54) to smoke daily and to smoke a pack or more of cigarettes a day. Unique to the Francophone sub-sample are the elevated levels of smoking found among those age 35–44. Adjusting for other factors, Francophones age 35–44 are roughly 1.7 times more likely than the reference group to smoke on a daily basis and 2.6 times more likely to smoke at least a pack of cigarettes a day.
In general, indicators of socioeconomic status fair poorly as correlates of smoking behaviour. One exception worth noting is number of rooms in the respondent's household. Interestingly, results show that Francophones living in households with four or fewer rooms are much more likely to smoke daily and to smoke a pack of cigarettes or more per day than those in larger households. Among Anglophones, the opposite relationship occurs.
Perceived health status emerges as an important correlate of consumption but only among Anglophones. Controlling for other factors, Anglophones reporting their health status as fair to poor are significantly more likely than those reporting good to excellent health to smoke daily and to smoke a pack or more of cigarettes a day. No relationship appears for Francophones.
Table VI
shows that Francophones reporting membership in clubs or organizations are significantly less likely than non-members to smoke daily or to smoke a pack or more of cigarettes per day. Similarly, Anglophone club members also show a lesser propensity to consume cigarettes, but the relationships are only marginal in terms of statistical significance.
Risk-taking measures appear to be fairly strong correlates of smoking behaviour. Unique to Anglophones, findings show that respondents reporting having thought seriously about suicide are much more likely than those not to consume cigarettes daily and to smoke at least a pack a day. For both groups, being sexually active bears a significant positive relationship with daily consumption. Thus, Francophones who are sexually active are at least twice as likely as those who are non-active to smoke cigarettes daily. Anglophones who are sexually active are 1.65 times more likely to smoke daily than those who are non-active.
For both ethno-linguistic groups, the number of respondent friends who smoke cigarettes is a strong correlate of cigarette consumption. Holding constant other influences, Francophones who report that most or all of their friends smoke are over 4 times more likely than others to smoke daily; for Anglophones the calculated odds ratio is 7.26. Similarly, Francophones reporting that most or all of their friends smoke are nearly 5.5 times more likely than others to smoke at least a pack of cigarettes daily; for Anglophones the calculated odds ratio is 5.89.
Finally, our results show that drinking alcohol daily is an important risk factor for elevated levels of smoking. For both Anglophones and Francophones, respondents who reported drinking alcohol daily were well over twice as likely to smoke cigarettes daily and to smoke heavy quantities as those not drinking daily.
|
Discussion |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionConclusionReferences |
|---|
This report presents strong evidence to support the claim that Francophone Ontarians demonstrate levels and patterns of tobacco use which depart significantly from those of the larger Anglophone population. Francophones in this study were significantly more likely than Anglophones to smoke cigarettes in their lifetime, to smoke on a daily basis, to initiate daily smoking before the age of 16 and to smoke daily for lengthier periods of time. Both groups were equal in terms of the quantity of cigarettes smoked daily.
Our descriptive analysis revealed interesting group differences in the patterns of smoking behaviour by respondent age. Francophones aged 35–44 far exceeded all other age groups in both samples in terms of reported prevalence of daily cigarette consumption and quantity consumed daily.
Results also showed that certain age groups within the Francophone sample (i.e. those age 25–34, 35–44 and 65 and over) were more likely than similar aged members of the Anglophone sample to begin smoking daily at an early age. For example, by 16 years of age, only 56.9% of Francophones age 35–44 had not started smoking daily compared to 70.4% of Anglophones in this age group. We found evidence of a steady age-related decline in age of onset of smoking daily unique to Francophones. At age 65 and over, the median age of onset stood at 21.7 years and declined to 16.8 years for those age 25–34.
Certain age groups within the Francophone sample were more strongly reluctant than their Anglophone age counterparts to quit smoking daily after starting to smoke on a daily basis. Most reluctant to quit smoking were those age 35–44, 45–54 and 65 and over. In contrast, results indicated a greater willingness on the part of younger Francophones, particularly those age 25–34, to quit their habit.
Our multivariate results revealed important Francophone/Anglophone differences in smoking behaviour on a number of respondent background characteristics. Consistent with the descriptive analyses above, Francophones age 35–44 were much more likely than other age groups to smoke cigarettes daily and to consume heavy quantities. Francophone males were more than twice as likely as females to smoke a pack or more of cigarettes daily. Females were just as likely as males, however, to smoke on a daily basis.
For both ethno-linguistic groups, certain variables emerged as strong correlates of consumption. These included drinking alcohol on a daily basis, having a large number of friends who smoke and being sexually active. Unique to Francophones was the influence of number of household rooms. Francophone respondents reporting four or fewer rooms were far more likely to smoke daily and to smoke heavily each day than those from larger households. No difference was observed for Anglophones. Unique to Anglophones was the influence of perceived health status and serious thoughts about suicide. Interestingly, Anglophones who reported their health status as fair to poor were much more likely to smoke daily and to smoke heavily than those reporting good to excellent health. No relationship was observed for Francophones.
|
Conclusion |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionConclusionReferences |
|---|
Caution should be exercised in using our findings as a guide for program and policy decisions. This study is the first in Ontario to investigate in detail Francophone/Anglophone differences in tobacco prevalence and patterns of behaviour. To validate our findings, other province-wide general population surveys must be undertaken including special surveys of Francophones at the local level. Secondly, this study has assessed the addiction-related service needs of Francophone Ontarians using prevalence estimation techniques and models. Prevalence estimates, however, are very crude indicators of need. At best they may be used to imply a level of need, and should not be used in isolation from methods or approaches which gather more direct and detailed qualitative information at the community level.
With these cautions in mind, this study does provide a clear indication that certain sub-groups within the Francophone population may be experiencing health-related risks associated with heavy and frequent tobacco use, and thus may be in need of addiction-related services. Most notable are the elevated levels of consumption among the age group 35–44. Studies indicate that people who quit smoking before the age of 50 reduce their risk of death from cancer and heart disease by 50% over a period of 15 years (USDHHS, 1990). Smoking cessation programs, therefore, should be targeted with greater intensity at this group.
Our results also indicate that the median age at which Francophones start smoking cigarettes on a daily basis may be on the decline. Although it is generally understood that a certain amount of cigarette smoking in adolescence constitutes healthy psychological experimentation, health risk behaviours initiated at an early age have been linked to negative consequences later in life (Chassin et al., 1990). The trend observed here suggests the need for smoking prevention (and cessation) programs aimed at Francophone children and adolescents.
Because of a number of methodological limitations, the results presented in this study should be interpreted with caution. First of all, we did not have a precise definition of a Francophone Ontarian on which to base our analyses. To validate our findings, future work is needed which examines the correlates of substance use among Francophones selected on the basis of more refined and accepted measures such as mother tongue or language first learned as a child and still understood or spoken.
Other measurement issues pertain to a number of theoretically important variables not captured in our regression models. For example, we did not have questions pertaining to the geographic mobility of the respondent as well as patterns of face-to-face contact and exchange of instrumental and emotional supports between older adult respondents and their teenage or adult children. Geographic mobility has been found to be a strong predictor of substance use (including use of tobacco) in both adolescent and adult populations (Flewelling et al., 1993). Social isolation is also important either in terms of its direct impact on substance use or indirectly via the level of involvement of close kin and family networks in dealing with family members' substance-related problems. Also excluded were important social influence variables such as number of family members who smoke (parental and sibling smoking behaviour), knowledge of the health risks of smoking and smoking attitudes (Pederson and Lefcoe, 1986). Although our `region' variable did not seem to elicit significant Francophone/Anglophone differences in smoking behaviour, differences might have emerged had we been able to utilize a more refined measure. Interestingly, a study just completed by the authors comparing the alcohol and other drug-related service needs of over 840 Francophones in two Ontario towns (one in Eastern Ontario and the other in Northern Ontario) found significant differences in drug use prevalence including tobacco consumption (DeWit et al., 1996).
Although it is the most widely used method of collecting information from a general population, the sample survey is not without its limitations. Survey estimates often fall short of the true population figure. One reason for the underestimation is that tobacco consumption is increasingly becoming a socially undesirable activity. Surveys also miss certain hard-to-reach elements in the population such as street youth or the homeless who may exhibit higher than average rates of consumption.
Third, the sample size for the Francophone sub-group was somewhat small, raising some concern with respect to the issue of statistical power. In the case of rare events or occurrences, large samples are required to detect an effect which actually exists in the population of interest (Hsieh, 1989). Because of the relatively small number of respondents in the Francophone group, we were forced to collapse the number of categories for certain background variables such as martial status, education and occupation prior to conducting the multivariate analysis. Future surveys must pay attention to the issue of statistical power to ensure that sufficient numbers of respondents are sampled.
Finally, we based our analyses on cross-sectional data. Although less costly and time intensive than other survey methods, cross-sectional surveys do not generally provide valid data on the causal sequencing or temporal ordering of events increasing the chance of spurious results. Even retrospective data must be interpreted with caution. Evidence suggests that respondents often tend to forget when historical events have occurred, to report distant events in the past as happening much more recently than is actually the case, a phenomenon known as telescoping, and to round dates or ages indicating the starting or ending time of an event to the nearest 0 or 5. Cross-sectional surveys are also ineffective in measuring past attitudes or beliefs related to the use of substances. Future surveys of Francophone substance use must be longitudinal in nature in order to map with greater precision the correlates and patterns of tobacco use over the life course.
In summary, this study identified important differences between Anglophones and Francophones on various measures of smoking behaviour and associated risk factors. The results are expected to play a key role in the development of smoking prevention programs for the Francophone minority population in Ontario. In addition to between-group differences, we also found substantial heterogeneity within the Francophone sample particularly with respect to age-related smoking patterns, a finding that challenges the common stereotype that all Francophones are frequent or heavy smokers. As noted in previous reviews of ethnic variations in drug use (Collins, 1992; Trimble, 1990/91), heterogeneity within racial/ethnic groups does not receive widespread acknowledgement among researchers and consequently is rarely incorporated into most research designs. The results of the present study call for the development of programs targeted at specific sub-groups. This approach has the potential to be more effective both in terms of cost efficiency and in bringing about measurable change in smoking behaviour.
|
References |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionConclusionReferences |
|---|
Addiction Research Foundation (1995) Ontario Profile: Alcohol and other Drugs. Addiction Research Foundation, Toronto, Ontario.
Adlaf, E. M. (1991) Alcohol and other drug use. In 1990 Health Promotion Survey: Technical Report. Statistics Canada, Ottawa, pp. 104–123.
Adlaf, E. M., Ivis, F. J. and Smart, R. G. (1994) Alcohol and other Drug Use among Ontario Adults in 1994: Changes since 1977. Addiction Research Foundation, Toronto, Ontario.
Berry, J. W., Kim, U., Minde, T. and Mok, D. (1987) Comparative studies of acculturative stress. International Migration Review, xxi, 491–511.
Burch, T. K. (1988) Family Structure and Ethnicity in Canada: An Exploration of the 1985 General Social Survey. Population Studies Center, The University of Western Ontario, London, Ontario.
Burnam, M. A., Hough, R. L., Karno, M., Escobar, J. I. and Telles, C.A. (1987) Acculturation and lifetime prevalence of psychiatric disorders among Mexican Americans in Los Angeles. Journal of Health and Social Behaviour, 28, 80–102.
Chassin, L., Presson, C. C., Sherman, S. J. and Edwards, D. A. (1990) The natural history of cigarette smoking: predicting young adult smoking outcomes from adolescent smoking patterns. Health Psychology, 9, 701–716.[Web of Science][Medline]
Collins, R. L. (1992) Methodological issues in conducting substance abuse research on ethnic minority populations. Drugs and Society, 6, 59–77.
DeWit, D., Beneteau, B., Vuksanovic, V. and Newton-Taylor, B. (1996) A Profile of Francophone Health Promotion Needs related to Alcohol and Other Drugs in Two Ontario Communities. Addiction Research Foundation, Toronto, Ontario.
Douville, J. A. (1990) Active and Passive Smoking Hazards in the Workplace. Van Nostrand Reinhold, New York.
Flewelling, R. L., Ennett, S. T., Rachal, J. V. and Theisen, A. C. (1993) National Household Survey on Drug Abuse: Race/Ethnicity, Socioeconomic Status, and Drug Abuse: 1991. Research Triangle Institute, Durham, NC.
Franco-Echanges (1991) Franco-Echanges 91. Internal Report. Addiction Research Foundation, Toronto, Ontario.
Gilbert, A. and Langlois, A. (1994) Regard Sur Les Nouvelles Realites Franco-Ontariennes. Publication de l'Association Canadienne-francaise de l'Ontario, Toronto, Ontario.
Goel, V. (1993) Analysis of Complex Surveys. Institute for Clinical Evaluative Sciences, Toronto, Ontario.
Gordon, T. and Kannel, W. B. (1982) Multiple risk functions for predicting coronary heart disease: the concept, accuracy, and application. American Heart Journal, 103, 1031–1039.[Web of Science][Medline]
Hasselback, P., Lee, K. I., Mao, Y., Nichol, R. and Wigle, D. (1991) The relationship of suicide rates to sociodemographic factors in Canadian census divisions. Canadian Journal of Psychiatry, 36, 655–659.[Web of Science][Medline]
Health and Welfare Canada (1992a) Alcohol and other drug use by Canadians: A National Alcohol and Other Drugs Survey. Technical Report, Minister of Supply and Services, Ottawa, Ontario.
Health and Welfare Canada (1992b) Smoking Behaviour of Canadians: A National Alcohol and Other Drugs Survey (1989) Report. Minister of Supply and Services Canada, Ottawa, Ontario.
Health Canada (1995) Effects of Tobacco Smoke and Second Hand Smoke on Prenatal and Postpartum Periods: A Summary of the Literature. Minister of Supply and Services, Ottawa, Ontario.
Hsieh, F. Y. (1989) Sample size tables for logistic regression. Statistics in Medicine, 8, 795–802.[Web of Science][Medline]
Kaplan, M. S. and Marks, G. (1990) Adverse effects of acculturation: psychological distress among Mexican American young adults. Social Science and Medicine, 31, 1312–1319.
Lee, E. S., Forthofer, R. N. and Lorimor, R. J. (1986) Analysis of complex sample survey data: problems and strategies. Sociological Methods and Research, 15, 69–100.
Lee, E. T. and Desu, M. M. (1972) A computer program for comparing 
samples with right censored data. Computer Methods Programs Biomedical, 2, 315–321.
Letourneau, G. (1988) Francophones and Smoking. Report Prepared Under Contract for Health and Welfare Canada, Ottawa, Ontario.
Liberatos, P., Link, B. and Kelsey, J. (1988) The measurement of social class in epidemiology. Epidemiologic Reviews, 10, 87–121.
Millar, W. J. (1992) Place of birth and ethnic status: factors associated with smoking prevalence among Canadians. Health Reports, cat. 82-003, vol. 4, no. 1, Statistics Canada, pp. 7–24, Ottawa.
Morgan, T. M. and Elashoff, R. M. (1986) Effect of categorizing a continuous covariate on the comparison of survival time. Journal of the American Statistical Association, 81, 917–921.
Morgan, S. P. and Teachman, J. D. (1988) Logistic regression: description, examples and comparisons. Journal of Marriage and the Family, 50, 929–936.
Mougeon, R. (1984a) Mixed marriages and the assimilation of French Canadians into Canada's English-speaking majority. Multiculturalism, 8, 5–10.
Mougeon, R. (1984b) Retention of French among young Franco-Ontarians. Language and Society, 13, 17–20.
Mougeon, R. and Heller, M. (1986) The social and historical context of minority French language education in Ontario. Journal of Multilingual and Multicultural Development, 7, 199–227.
Ontario Office of Francophone Affairs (1996) Francophones in Ontario: A Statistical Profile. Queen's Park, Toronto, Ontario.
Ontario Ministry of Health. (1992) 1990 Ontario Health Survey. User's Guide. Vol. 1. Documentation. Toronto, Ontario.
Padilla, A. A. (1980) Acculturation: Theory, Models and Some New Findings. Westview Press, Boulder, CO.
Pederson, L. L. and Lefcoe, N. M. (1986) Change in smoking status among a late cohort of adolescents: prediction and explanation of initiation, maintenance and cessation. International Journal of Epidemiology, 15, 519–526.
Stachenko, S. J., Reeder, B. A., Lindsay, E., Donovan, C., Lessar, R. and Balram, C. (1992) Smoking prevalence and associated risk factors in Canadian adults. Journal of the Canadian Medical Association, 146, 1989–1996.[Abstract]
Stellman, S. D. and Garfinkel, L. (1986) Smoking habits and tar levels in a new American cancer prospective study of 1. 2 million men and women. Journal of the National Cancer Institute, 6, 1057–1063.
Trimble, J. (1990/91) Ethnic specification, validation prospects, and the future of drug use research. International Journal of the Addictions, 25, 149–170.[Web of Science][Medline]
US Department of Health and Human Services (1988) Adult Health Practices in the United States and Canada. US Government Printing Office, Washington, DC.
US Department of Health and Human Services (1990) The Health Benefits of Smoking Cessation: A Report of the Surgeon General. USDHHS, Rockville, MD.
Walsh, A. (1987) Teaching understanding and interpretation of Logit regression. Teaching Sociology, 15, 178–183.
Received on September 30, 1997; accepted on February 11, 1998
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  J. Angus, E. Rukholm, R. St. Onge, I. Michel, R. P. Nolan, J. Lapum, and S. Evans Habitus, Stress, and the Body: The Everyday Production of Health and Cardiovascular Risk Qual Health Res, October 1, 2007; 17(8): 1088 - 1102. [Abstract] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||