Family Practice Vol. 21, No. 1, 33-38
© Oxford University Press 2004, all rights reserved.
Article |
Trends in smoking habits
A longitudinal population study
General Practice Research Unit, North West Lung Research Centre and a Department of Medical Statistics, South Manchester University Hospitals NHS Trust, Wythenshawe Hospital, Manchester M23 9LT, UK
E-mail: gpresearchunit{at}yahoo.co.uk
Received 25 February 2003; Accepted 8 September 2003.
Frank PI, Morris JA, Frank TL, Hazell ML and Hirsch S. Trends in smoking habits. A longitudinal population study. Family Practice 2004; 21: 33–38.
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionConclusionsAppendixReferences |
|---|
Background. In 1998, the UK government published a White Paper in which it set long-term targets for reducing smoking in the population. This longitudinal study aimed to examine whether progress has been made in achieving these in two adult general practice populations over an 8-year period.
Methods. Postal respiratory questionnaires, based on the European Community Respiratory Health Questionnaire, were sent to all patients registered with two practices in North West England on four occasions between 1993 and 2001. Two analyses were carried out. The first (smaller cohort) included only those subjects answering the question concerning current smoking on all four occasions, the second (larger cohort) those answering at least twice.
Results. The smaller cohort included 2403 subjects (19.6% of all respondents). Almost one-quarter reported in all four surveys that they smoked, the highest proportion being in those aged 35–44 years. The proportion of smokers decreased from 34.2% (1993) to 30.3% (2001) (P < 0.001 for trend) and the prevalence of heavy smokers fell from 15.9 to 13.3% (P < 0.001 for trend) over the same period. There were, however, no reductions in those aged <45 years in 1993. These changes were confirmed in the larger cohort which included 7274 subjects (59.3% of respondents).
Conclusions. If smoking-related disease is to be reduced, it is important that adults are targeted for smoking prevention and cessation before they reach middle age. Only when such initiatives show success will smoking prevention among their children become a practical proposition.
Keywords. Longitudinal study, questionnaire, secular trend, smoking.
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionsAppendixReferences |
|---|
Smoking is a major cause of ill health1 and, over the past decade, in the UK, there has been much publicity and expenditure concerned with the dangers of smoking and with smoking cessation strategies. Most have been directed at groups already suffering from or at specific risk of smoking-related illnesses, although recent publications have also discussed the problem of smoking among children.2–4 In 1998, a White Paper ‘Smoking Kills’5 outlined the government's policy on smoking, and gave some long-term targets, including reducing overall adult smoking prevalence and also the prevalence in children. Despite this, smoking patterns in England, based principally on cross-sectional household surveys, appeared to show little change between 1992 and 2000.6–8
The present longitudinal study examines changes in current smoking prevalence in adults from two general practices in North West England, observed on four occasions between 1993 and 2001. It forms part of the Wythenshawe Community Asthma Project (WYCAP), a long-term study of the natural history of respiratory symptoms in this population.9
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionsAppendixReferences |
|---|
Postal questionnaires were sent to all patients aged 16 years and over, registered with two neighbouring general practices, in four separate surveys (1993, 1995, 1999 and 2001).
Questionnaires (Appendix) were based on the European Community Respiratory Health Questionnaire,10 with some minor modifications including an additional question concerning current smoking ‘How many cigarettes do you smoke each day?’
For all four surveys, questionnaires were mailed with an accompanying letter on practice-headed paper, and non-responders were sent reminders 4 and 8 weeks after the first mailing. Subjects who had still not responded comprised those no longer living at the mailing address (‘ghosts’) and those still at the address who failed to reply (‘true non-responders’). The number of ghost patients was estimated in the 1993 survey by identifying, from a 5% random sample of all non-respondents, those who were neither on the 1993 Electoral Register, nor in the telephone directory, and who had not consulted at the practice in the previous 12 months. The proportion of ghost patients in the whole population was found to be 5.4%, and this was used to calculate an adjusted response rate. The exercise was not repeated in the other surveys, but a similar proportion of ghosts was assumed. Each practice had between 5000 and 5500 patients registered during the period of the study.
Two main analyses were carried out. The first included all subjects who had answered the smoking question in all four surveys (referred to as the smaller cohort). The second included all who had replied to the smoking question on two or more occasions (larger cohort).
For each survey, subjects were categorized into one of four groups according to the number of cigarettes they smoked each day. Those smoking 20 or more were labelled ‘heavy smokers’ for the purposes of this study. Longitudinal smoking behaviour in those who had answered on all four occasions was investigated by classifying as ‘persistent smokers’ [smoker (S) in each survey, SSSS], ‘never smokers’ [non-smoker (N) in all surveys, NNNN], ‘ex-smokers’ (SSSN, SSNN or SNNN) or ‘new smokers’ (NNNS, NNSS or NSSS). Those who did not fit into one of these categories were categorized as ‘other’. No information was available for cigarette smoking prior to the first survey.
Ethical approval was obtained from South Manchester Local Research Ethics Committee.
Statistical analysis
The data were analysed using the Generalized Estimating Equations (GEE) approach, implemented using STATA 6.0.11 This method takes account of the repeated measures nature of the prevalence data. An exchangeable correlation structure was assumed, i.e. there is a constant correlation between any pair of time points (i.e. it is independent of the year of measurement), and robust standard errors were calculated. Thus, for the second analysis, it was possible to include all patients who answered two or more of the four surveys, regardless of which surveys.
For the analysis of subjects replying on all four occasions (smaller cohort), the proportions of the various levels of smoking are presented for each survey. Since the patients included were the same for each survey year, it was possible to calculate a significance value for trend based on these proportions. For the analysis of those subjects who answered two surveys or more, the group of subjects at each time point was different, and the cross-sectional proportions did not accurately reflect within-subject change over time; these proportions are, therefore, not presented herein. It was, however, possible to estimate within-subject trends, and the statistical significance of these changes over time, expressed as P-values, are shown in the Results. Because of the relatively large number of statistical tests carried out, the threshold of statistical significance for this study was made more stringent (P < 0.01).
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionsAppendixReferences |
|---|
Between 10 000 and 11 200 questionnaires were sent out in each of the four surveys. The response rates after adjustment for ‘ghost’ patients were 76.3% (1993), 76.8% (1995), 71.0% (1999) and 68.9% (2001).
A total of 12 262 subjects answered the smoking questionnaire in at least one survey, of whom 2403 (19.6%) replied in each of the four surveys and were included in the first analysis; 7274 (59.3%) individuals responded on two or more occasions and were included in the second analysis.
Among subjects responding to all four surveys, 23.0% reported smoking on each occasion. The highest proportion of these persistent smokers, in both sexes, was in those aged 35–44 years at the time of the first survey (Table 1). In all, 61.1% reported ‘no smoking’ in any of the studies. This was most frequent in those aged >65 years. Subjects who stopped smoking during the period of the study and were still non-smokers at the time of the last survey (ex-smokers) were most frequent in the age group 45–64 years in both males and females. ‘New smokers’(those who started to smoke during the study and continued up to the time of the last survey) tended to be in the younger age groups.
|
When overall changes in smoking were considered (Table 2), there was an identifiable decrease in the proportion of smokers over the 8-year period in both the smaller and larger cohorts; in the first, the prevalence fell from 34.2% in 1993 to 30.3% in 2001 (P for trend <0.001). This trend remained statistically significant when those answering two or more surveys were included. The prevalence of heavy smoking decreased from 15.9% in 1993 to 13.3% in 2001 (P for trend <0.001) for the smaller cohort, and again remained statistically significant for the larger cohort (P = 0.002).
|
When respondents were categorized by their age (in 1993) and gender (Table 3), there were no significant decreases (i.e. P < 0.01) in smoking prevalence in either sex in subjects <45 years old. There were, however, significant decreases in females aged 45–54 years (smaller cohort) and 45–64 (larger cohort), and in males aged 55–64 years (smaller cohort) and 45–74 years (larger cohort). In subjects between 16 and 24 years, there was an increase in the proportion of smokers over the period of the study in both the smaller and larger cohorts, although statistical significance for trend was only reached in the larger cohort for male respondents (P = 0.004).
|
The prevalence of heavy smokers (Table 4) was higher in males in each survey, and was highest, for both genders, in those aged 35–44 years. In males, there were decreases in prevalence in those aged 45–64 in both cohorts. In females, the only significant decrease was in those aged 75 years or more, and this was significant only when all who replied on two or more occasions were included (larger cohort). In those aged under 45 years, there were no significant reductions in the prevalence of heavy smoking in either sex. In fact, in the youngest age group, there was an increased prevalence in both sexes in both cohorts, although significance was only reached when those answering two or more surveys were included (larger cohort, P < 0.001).
|
20 per day) in each survey by age |
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionsAppendixReferences |
|---|
This longitudinal study enabled the collection of smoking data from two general practice populations on four separate occasions over 8 years. It was carried out in the North West region of England, which has the highest prevalence of smoking and smoking-related disease in the country.6 Further, the population studied was from an area with high levels of deprivation, where cigarette smoking is even more prevalent than in the remainder of the region. The results between the two practices were, however, consistently similar, and although they may not be nationally representative, they are likely to reflect the situation of deprived populations in other parts of the country.
There was a significant improvement in smoking patterns over the 8-year period to 2001, with a decreased prevalence of smokers, including heavy smokers. These findings contrasted with National Household Survey figures,6 which reported no overall changes since 1994. Despite these positive changes, the overall prevalence of smoking in these deprived populations remained above the latest available national figures, and even above those reported for the whole North West region.
In addition, the improvements found were confined to the middle and older age groups, those most likely to be already suffering from, or at known risk of smoking-related disease. The variation from national figures may be a reflection of the deprived nature of these populations and their consequent increased prevalence of smoking-related disease. Targeting this group for smoking cessation advice may have been a factor in reducing their smoking.
In subjects <45 years at the start of the study, there was little change in smoking prevalence. In fact, in the youngest group, those between 16 and 24 years, the proportion of smokers (and heavy smokers) showed increases for both males and females. The longitudinal design of the study enabled comparison of the same individual at different time points, i.e. within-individual analyses of change.12 In addition, in the first analysis (smaller cohort), it was possible to include four observation points for all subjects, which provides inherently more reliable results than when those with fewer available time points are included.13
The four surveys had adjusted response rates of between 69 and 77%. The estimate of 5.4% ‘ghost’ patients used to calculate this adjusted rate is likely to be conservative, as a proportion of telephone owners were ex-directory and not all subjects were on the Electoral Register. Two postal surveys from Newcastle14 which used the Health Authority Register as the source of subjects calculated an adjusted response rate after taking account of concordance between this register and the electoral roll. The adjusted rates were 33 and 22% above the crude response figures for the first and second studies, respectively.
Loss of subjects over time has been reported as a potential bias in studies concerning smoking, because smokers are more likely to drop out.15 In this study, <20% of those who completed at least one questionnaire answered all four. To address this problem, a second analysis was undertaken which used information from all who had replied to at least two questionnaires (larger cohort). In this way, it was possible to include 60% of all respondents. Both analyses showed broadly similar trends. It seems unlikely, therefore that there was any important bias, although the populations included in the two analyses were older and contained fewer smokers than those answering only one survey, a finding similar to previous surveys.16 It is possible that this may have caused some over-estimation of the improvements in smoking prevalence over the study period.
Although a number of national household surveys in England have published cross-sectional data concerning smoking prevalence and change over time,6 a search of the peer-reviewed literature revealed no critical analyses of these findings, or other comparable studies in England, conducted over the past decade. No studies have been reported with a ‘within-subject’ longitudinal design.
Although the government's White Paper ‘Smoking Kills’5 gave some long-term targets for reducing overall adult smoking prevalence and also the prevalence in children and pregnant women, no differentiation was made between various adult age groups. The smoking behaviour of parents has been found to influence smoking behaviour and health care of their children. A recent study2 reported that in children age 9–15 years, the proportion who had ever tried smoking more than doubled in the group where at least one parent smoked. A study from Scotland17 reported that children whose parents smoke heavily are less likely to be taken to the doctor for asthma symptoms than those of non-smoking parents.
The results of the present study reflect the smoking control policies of these practices, where specific targeting was made at health promotion and later chronic disease management clinics, of patients already suffering from or at specific risk of smoking-related disease. Healthy young adults were not given special attention. This may have been a factor in the lack of improvement in smoking habits of this age group. In fact, the highest rate of persistent smokers was in the age group 35–44 years at the start of the study. This is the group most likely to have children of an age at the greatest risk of starting to smoke. Educating these children in the dangers of smoking,3 and reducing the glamour of smoking2 may have little impact on their smoking habits if their parents continue to smoke.
|
Conclusions |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionsAppendixReferences |
|---|
These findings have implications for the achievement of government smoking targets concerning younger adults. Education and smoking cessation programmes appear to have had some success in the older age groups. In younger adults, however, new strategies are needed in order to prevent the onset of smoking-related disease. Until beneficial changes are achieved in young parents, it seems unlikely that smoking education of their children will be successful.
|
Appendix |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionsAppendixReferences |
|---|
Please tick the appropriate box
- What is your date of birth?
- Are you FEMALE MALE
- Have you had wheezing or whistling in your chest at any time in the last 12 months? NO YES
IF ‘NO’ GO TO QUESTION 4,
IF ‘YES’:
3.1 Have you been at all breathless when the wheezing noise was present? NO YES
3.2 Have you had this wheezing or whistling when you did not have a cold? NO YES
- Have you woken up with a feeling of tightness in your chest in the last 12 months? NO YES
- Have you been woken by an attack of shortness of breath at any time in the last 12 months? NO YES
- Have you been woken by an attack of coughing at any time in the last 12 months? NO YES
- Have you had an attack of asthma in the last 12 months? NO YES
- Are you currently taking any medicine for asthma? (including inhalers, aerosols or tablets) NO YES
- Has any person in your family (parents, grandparents, sisters or brothers, or your children) had asthma? NO YES
- Have you ever had hay fever or eczema? NO YES
- How many cigarettes do you smoke each day?.........per day
- How many other adults live in your house?
How many of these adults smoke?
|
Acknowledgments |
|---|
The authors wish to thank the staff at Bowland Medical Practice and Tregenna Group Practice for their co-operation, and the patients who replied to the questionnaires. Funding was provided by the NHS Executive North West.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionsAppendixReferences |
|---|
1 Callum C. The Smoking Epidemic: Deaths in 1995. London: Health Education Authority; 1998.
2 Sargent JD, Beach ML, Dalton MA et al. Effect of seeing tobacco use in films on trying smoking among adolescents: cross-sectional study. Br Med J 2001; 323: 1394–1397.
3 Tønnesen P. How to reduce smoking among teenagers. Eur Respir J 2002; 19: 1–3.
4 Friend JAR. Do smoking parents seek the best advice for their asthmatic children? Thorax 2001; 56: 1.
5 Department of Health. Smoking Kills—a White Paper on Tobacco (Cm 4177). London: Stationery Office; 1998.
6 Walker A, Maher J, Coulthard M, Goddard C, Thomas M. Living in Britain. Results from the 2000–01 General Household Survey. London: The Stationery Office; 2001.
7 Lader D, Meltzer H. Smoking Related Behaviour and Attitudes, 2000. National Statistics. London: The Stationery Office; 2001.
8 Erens B, Primatesta P (eds). Health Survey for England: Cardiovascular Disease 1998. London: The Stationery Office; 1999.
9 Frank P, Ferry S, Moorhead T, Hannaford P. Use of a postal questionnaire to estimate the likely underdiagnosis of asthma like illness in adults. Br J Gen Pract 1996; 46: 295–297.[ISI][Medline]
10 Burney PG, Luczynska C, Chinn S, Jarvis D. The European community respiratory health survey. Eur Respir J 1994; 7: 954–960.[Abstract]
11 Stata. Version 6.0 for Windows. College Station (TX): Stata Corporation; 1999.
12 Siddiqui O, Flay B, Hu FB. Factors affecting attrition in a longitudinal smoking prevalence study. Prev Med 1996; 25: 554–560.[CrossRef][ISI][Medline]
13 Burney P. The changing prevalence of asthma? Thorax 2002; 57 (Suppl II): ii36–ii39.[Medline]
14 Barraclough R, Devereux G, Hendrick DJ, Stenton SC. Apparent but not real increase in asthma prevalence during the 1990s. Eur Respir J 2002; 20: 826–833.
15 Flay BR, Koepke D, Thomson SJ, Santi S, Best JA, Brown KS. Six year follow up of the first Waterloo smoking prevention trial. Am J Public Health 1989; 79: 1371–1376.
16 Hanssen WB, Collins LM, Malotte CK, Johnson CA, Fielding JE. Attrition in prevention research. J Behav Med 1985; 8: 261–275.[CrossRef][ISI][Medline]
17 Crombie IK, Wright A, Irvine L, Clarke RA, Slane PW. Does passive smoking increase the frequency of health service contacts in children with asthma? Thorax 2001; 56: 9–12.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||