| Cardiovascular Disease 98 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Chapter 3 Risk factors for
Richard Boreham, Bob Erns, Emanuela Falaschetti,
This chapter reviews the current prevalence among adults of a number of risk factors for cardiovascular disease (CVD), and, for those where comparative data are available, examines changes in their prevalence since 1994. The risk factors reviewed are:
Another risk factor, physical activity level, was treated in considerable depth in a module of the questionnaire, and is reported on in a separate chapter of this report (Chapter 5). 3.2.1 Background Alcohol is a significant component of the diet, in this as in other countries. Epidemiological studies have suggested that heavy drinking constitutes a severe risk for cardiovascular disease, but that low levels of consumption can have a protective effect against coronary heart disease (CHD) mortality. A U- or J-shaped association between alcohol consumption and various types of ischaemic illnesses, including myocardial infarction and stroke, was demonstrated in several studies, with heavy drinkers and abstainers most at risk, while light and moderate drinkers showed the lowest risk.2,3 Since its inception in 1991, the Health Survey has carried a set of questions on alcohol consumption (the same questions as in the General Household Survey (GHS)). These questions were designed to provide an estimate of average weekly consumption. Until the end of 1995, advice about sensible drinking was given in terms of weekly amounts, men being advised not to exceed 21 units per week and women 14 units. In late 1995, an inter-departmental Working Group recommended that advice on sensible drinking should be expressed in terms of daily, rather than weekly, consumption, and that it should reflect evidence that moderate consumption can be beneficial for certain groups of the population.4 Advice about sensible drinking was revised as follows:
The main series of questions in the Health Survey and in the GHS was designed to estimate average weekly consumption rather than daily drinking patterns. To provide continuity with earlier reports, the results given in this chapter include weekly consumption estimates and show the proportions of men and women exceeding the weekly levels advised before 1996. But the chapter also briefly reports some findings from new questions intended to throw more light on daily consumption. The 1997 survey had already included some questions with similar aims,5 but these were revised for the 1998 survey. 3.2.2 Weekly consumption patterns Previous reports have stressed the well-known finding that surveys tend to underestimate alcohol consumption, and have suggested that the results should be used primarily to compare the consumption of different population groups and to monitor change over time, rather than as absolute estimates of actual consumption. They have also described in some detail the way in which estimates of weekly consumption were derived from the responses.6 Several changes were made to the module of questions on drinking in the 1998 survey. The biggest innovation was the distinction made for the first time between normal (alcoholic strength less than 6%) and strong (6% or more) beer, lager and cider.7 The separate question on shandy, which had been asked in all previous Health Surveys, was dropped, and shandy was included with normal strength beer. Also, a question on alcoholic lemonades, colas, and fruit drinks was added to the adult interview for the first time. (Previously it had only been included in the self-completion booklets for those aged 13-17, beginning with the 1997 Health Survey.) Finally, a new series of questions on drinking behaviour over the seven days before the interview was added (and is described in section 3.2.4). The weekly consumption tables annexed to this chapter show patterns of variation by age and social class in 1998 that were similar to those discussed in previous Health Survey reports. There are two convenient summary measures: the proportion drinking more than the weekly levels advised before 1996 (21 units for men, 14 units for women), and estimated mean weekly units. In 1998, the proportion was 31% for men and 18% for women; the mean was 18.0 units for men and 7.2 units for women. Variations in estimated weekly consumption by age Among both men and women, the general pattern emerging from the surveys since 1994 has shown mean weekly consumption decreasing slightly from age 16-24 until age 45-54, and thereafter decreasing rapidly with increasing age. For men in 1998, mean weekly units was estimated at 23.9 units at age 16-24, decreasing to 9.0 units at age 75 and over. Comparable figures for women were 10.8 decreasing to 3.4. The proportion whose weekly consumption was above the levels advised
before 1996 showed a broadly similar pattern to the mean. For men, it
decreased from 41% in the youngest group to 13% in the oldest, and for
women from 27% to 7%. Variations in estimated weekly consumption by social class of head of household Among women, estimated alcohol consumption in the Health Survey series
has generally been found to increase from Social Classes IV and V to
Social Classes I and II. Among men the social class pattern is different.
In the surveys since 1994, there has been a persistent tendency for
men in Social Class II to have the highest proportion drinking above
the pre-1996 advised levels, with some evidence of a below-average proportion
in Social Class IV (though this latter feature is not seen in the 1998
results). Estimated mean weekly units showed a similar pattern, but
with the difference that mean consumption among men was high in Social
Class V as well as in Social Class II. Social Class V thus does not
contain a particularly high proportion of men drinking above the pre-1996
advised levels, but those who are drinkers in Social Class V have heavier
consumption that increases the overall mean for this group. Variations in estimated weekly consumption by equivalised household income The tables annexed to this chapter present analyses of weekly consumption not featured in earlier reports, by (equivalised) household income and Health Authority area type. Regional tables will also be found among those annexed, but are not commented on. Among both men and women, the (age-standardised) proportion drinking above the pre-1996 advised levels increased markedly with income from the lowest two income quintiles to the top quintile. The increase for men in the 1998 sample was from 26% in both the lowest quintiles to 38% in the highest, and for women from 13% in both the lowest quintiles to 26% in the highest. (Age-standardised) estimated mean weekly consumption followed the
same pattern, rising among men from the two lowest income quintiles
(16.5 and 17.3 units) to the highest (21.1 units), with the corresponding
increase for women being 5.7 to 9.8 units. Variations in estimated weekly consumption by Health Authority area type The (age-standardised) proportion drinking above the pre-1996 advised levels was highest for men in the two Health Authority area types labelled Mining and Industrial (34%) and Urban (35%), and lowest in Inner London (26%). A similar pattern was found in estimated mean weekly units, which was 19.7 both in Mining and Industrial and in Urban areas, compared with 14.2 in Inner London. Among women there was less variation both in the proportion drinking
above pre-1996 advised levels (the higher figure for Inner London is
not statistically significant), and in estimated mean weekly units. 3.2.3 Trends in weekly consumption There is evidence of an upward trend in alcohol consumption among women. Women's estimated mean weekly units were 6.3 in 1994 and 6.2 in 1995, increasing to 6.6 in 1996, 6.7 in 1997 and 7.2 in 1998. In the same five years, the proportion of women drinking more than the levels advised prior to 1996 were respectively 14%, 14% 15%, 16% and 18%. The increase appears to have occurred mainly among younger adults (those aged 16-24). Among men, figures for 1998 were not very different from those for
1994, and there is no evidence of a consistent trend. Questions asked in 1998 were changed in some respects from those in earlier Health Surveys, and this could affect comparability. For the first time, a distinction was made between normal and strong beer, and a separate question was added on alcoholic lemonades. Strong beer and alcoholic lemonades are most commonly drunk by the youngest age group, so the change in methodology would be likely to affect this group more than others. 3.2.4 The heaviest day's drinking in the previous week Informants were asked whether they had drunk alcohol in the past seven days; if so, on how many days and, if on more than one, whether they had drunk the same amount on each such day or more on one day than others. If they had drunk more on one day than others, they were asked how much they drank on that day. If they had drunk the same on several days, they were asked how much they drank on the most recent of those days. If they had drunk on only one day, they were asked how much they had drunk on that day. In each case, the questioning obtained details of amounts drunk of each type of drink (similar to those obtained for establishing average weekly consumption), rather than a direct estimate of units consumed. The proportions claiming to have drunk alcohol in the previous seven days were 78% for men and 62% for women. Of men who had drunk alcohol in the past seven days, 21% had drunk on only one day and 20% on two days, with decreasing proportions on three (15%), four (10%), five (7%) and six days (6%), and then increasing again to 21% on all seven days. The distribution (which had a mean of 3.6 days) was thus to some extent polarised between every day drinking and drinking on only a few days. This was also true of women, though with a lower mean (3.0 days) the distribution was more skewed towards the lower end: 33% of women who had drunk alcohol in the past seven days drank on only one day, 22% on two days and 15% on all seven days. The proportion of past seven day drinkers who drank on all seven days
increased with age. Figures for the youngest and oldest groups respectively
were 9% and 38% for men, 4% and 32% for women. This is consistent with
patterns commented on (using evidence from other questions) in earlier
Health Survey reports.8 As noted above, those who drank on more than one day were asked whether they had drunk the same on each day or different amounts. The results showed very marked differences by age, with younger people tending to drink different amounts and older people the same amounts. Figures for the youngest and oldest groups were as follows:
Thus the tendency is for drinking habits to become more regular with increasing age: as noted above, daily drinking is common among older people, and the amounts consumed per occasion are more uniform than among younger people. Where amounts differed between days within the last seven, consumption estimates were obtained for the heaviest day. Amalgamating these with estimates where there had been a uniform sequence of days and also with estimates where there had been drinking on only one day, it is possible to derive an estimate of the heaviest day's consumption out of the previous seven, the base being all drinking in the past seven days. Amounts consumed on the heaviest (or only) day are shown in Table
3.8. The table does not give sufficient information to allow a full
assessment of whether current advice on sensible drinking, quoted earlier
in this chapter, is being followed, since the advice refers to regular
consumption rather than consumption on a single day. 'Binge' drinking
- drinking an excessive amount on a single occasion - is thought to
be less healthy than drinking moderate amounts more regularly.9
The table throws some light on this by indicating whether 'binge-level'
amounts are being consumed, though there is no medically-specified criterion
for such amounts. The level chosen for the analysis that follows is
over 8 units for men and over 6 for women - that is, double the daily
amounts that people are advised not to exceed on a regular basis. It was found that 33% of men who had drunk alcohol in the past week had drunk more than 8 units on their heaviest drinking day. This proportion decreased with age, from 58% of men aged 16-24 to 6% of men aged 75 and over. A similar pattern was seen among women, though at a lower overall level of consumption, with 38% of women aged 16-24 drinking more than 6 units, compared with 2% of women aged 75 and over. The heaviest day's amount was highly correlated with estimated weekly consumption: mean weekly units varied from 8.5 among men whose heaviest day's consumption was under 2 units to 37.0 among men with a heaviest day's consumption of 8 or more units. Comparable figures for women were 5.4 units rising to 24.2 units. (Table not shown.) Variations in heaviest day's consumption by social class of head of household The social class pattern was similar for men and women, with drinkers in Social Classes IV and V more likely to exceed these amounts than drinkers in Social Classes I and II. These figures contrast with drinking prevalence and estimated weekly
mean consumption for the sample as a whole. For women, the prevalence
of drinking more than the pre-1996 advised amounts, and overall mean
consumption, was higher in Social Classes I and II than in IV and V,
but drinkers in Social Classes IV and V had heavier drinking days than
those in Social Classes I and II.10 Variations in heaviest day's consumption by equivalised household income Among women, the drinkers most likely to exceed the specified levels
were those in the lowest income quintile. There was very little difference
among the other four quintiles. No clear pattern was seen among men. Variations in heaviest day's consumption by Health Authority area type The pattern by Health Authority area type was similar to that described
for weekly consumption, with the two area types of Mining and Industrial
and Urban showing the highest proportions of men and women drinkers
likely to exceed the specified levels (8 units for men, 6 for women)
in a single day. 3.3.1 Introduction Government White Papers The importance attached by the government to reductions in levels of smoking in all social classes is emphasised in the White Paper 'Smoking Kills'.11 One of the targets is: To reduce adult smoking in all social classes so that the overall rate falls from 28% to 24% or less by the year 2010; with a fall to 26% by the year 2005. The White Paper 'Our Healthier Nation'12 identifies smoking as a major risk factor for deaths from cancer and coronary heart disease and stroke - two of the four health targets set. Questions asked about cigarette smoking in the Health Surveys Since its inception in 1991, the Health Survey series has collected information about smoking. It uses essentially the same questions as the General Household Survey (GHS), which is the source for the Smoking Kills prevalence targets. An analysis of levels of smoking reported by informants in the GHS and the Health Survey for England has shown that the measurements in the two surveys are comparable in spite of their different contexts.13 In 1998, as in previous surveys, information about cigarette smoking was collected from those aged 16 and 17 by means of a self-completion questionnaire, while for those aged 18 or over14 it was collected as part of the main interview. Questions about the main brand of cigarette smoked were added in the 1998 survey. Cotinine Before 1998, cotinine levels in the Health Survey were measured in serum in adults, but in 1998 were measured in saliva, primarily to increase the number of people being measured as more people refuse to give a blood sample than a saliva sample. Cotinine is a metabolite of nicotine. It is one of several biological markers that are indicators of smoking (others include carbon monoxide and thiocyanate), and is generally considered the most useful. It can be measured in, among other things, saliva or serum. Cotinine has a half-life in the body of between 16 and 20 hours, which means that it will detect regular smoking but will not detect occasional smoking if the last occasion was several days ago. Tar, nicotine and carbon monoxide content of cigarettes smoked New information is provided in the present report, which utilises analyses of cigarette content by the Laboratory of Government Chemists for the Department of Health. This is examined in Section 3.3.5 . 3.3.2 Cigarette smoking prevalence Cigarette smoking prevalence is measured in two ways in the Health Survey for England. Informants are asked directly whether they smoke cigarettes nowadays, and cotinine levels in saliva are measured for those providing a saliva sample at the nurse interview. A saliva cotinine level of 15 ng/ml and over is taken as an indication that the informant currently smokes (those who use other nicotine products are excluded). The measurement of cotinine levels in the Health Survey series provides an objective cross-check on self-reports of smoking behaviour, which are known not always to be accurate. Inaccuracies in reporting arise in part from difficulties informants may experience in providing quantitative summaries of variable behaviour patterns, but in some cases arise from a desire to conceal the truth from other people, such as household members who may be present during the interview. However, previous Health Survey reports have shown a very high level of agreement between self-report and cotinine levels. Systematic differences are mostly minor, and are due to under-reporting for the reasons given above, or to alternative sources of nicotine, notably pipe or cigar smoking or the use of other nicotine products, or to passive smoking, though nicotine levels due to passive smoking are normally not high enough to result in the informant's misclassification as a smoker. Cotinine-based estimates of prevalence were more or less the same as self-report for women (27% overall in either case), but for men they were higher (32%) than self-report (28%), due to some extent to pipe and cigar smoking. Given the close resemblance between patterns shown by self-report and those shown by cotinine analysis, the present report deals mainly with self-report. But Table 3.20 presents saliva cotinine levels by age and sex, and Table 3.21 presents a logistic regression in which a saliva cotinine level of 15 ng/ml is the dependent variable, while Table 3.28 explores the determinants of particularly high cotinine levels among smokers. Analysis of self-reported cigarette smoking prevalence was conducted using cross-tabulations (age-standardised where appropriate) and logistic regression models. Separate logistic regressions were run for men and women in SPSS with the following independent variables: age, social class of head of household, equivalised household income, Health Authority area type and highest level of educational qualification. The odds ratios shown in the tables are relative to average. An odds ratio of less than one means that the group was less likely than average to smoke cigarettes currently, and an odds ratio greater than one indicates a greater than average likelihood of smoking. Cigarette smoking prevalence by sex and age As in previous reports, the 1998 survey showed that similar proportions of men (28%) and women (27%) reported smoking cigarettes. Cigarette smoking prevalence was highest among those aged 16-24 (41%
among men aged 16-24, 38% among women aged 16-24), and declined with
increasing age to levels of 9% among men aged 75 and over and 10% among
women aged 75 and over. The decrease in prevalence with age is likely
to be due in part to higher death rates among smokers than non-smokers.15 Cigarette smoking prevalence by social class of head of household The 1998 survey results again showed the well-documented social class
gradient in cigarette smoking. The (age-standardised) prevalence of
cigarette smoking by men in Social Class I was 15%, rising to 42% in
Social Class V. Corresponding proportions of women were 14% to 37%. Cigarette smoking prevalence by equivalised household income Cigarette smoking prevalence increased as equivalised household income
decreased. The age-standardised proportion of men and women who were
current smokers rose consistently from 21% of men and 18% of women in
the highest income quintile to 42% of men and 37% of women in the lowest
income quintile. Cigarette smoking prevalence by Health Authority area type There was much less variation between area types than between age,
class and income groups, and no clear overall pattern. Logistic regression models predicting cigarette smoking and cotinine levels After adjusting for the other variables in the model, age was still
a strong predictor of cigarette smoking, and of saliva cotinine levels
of 15 ng/ml or more. The odds of cigarette smoking decreased as age
increased, being nine times as great at age 16-24 as at age 75 and over
for both men and women. Socio-economic variables were also strong predictors of cigarette smoking. The odds of smoking showed social class, income and educational qualification gradients. After adjustment for the other factors in the model, men in Social Class V were 21/2 times as likely to smoke as men in Social Class I, men in the lowest income quintile were 58% more likely to smoke than men in the top income quintile, and men with no qualifications were more than twice as likely to smoke as men with a degree. Similar results were found for women. 3.3.3 Trends over time in self-reported cigarette smoking There has been no consistent overall trend in cigarette smoking prevalence
between 1994 and 1998. The proportion of men reporting cigarette smoking
varied between 28% and 30%, while the proportion of women reporting
cigarette smoking remained constant at 27%. Early indicators of an increase
in smoking prevalence among young adults16 were not confirmed
in the 1997 survey, but prevalence in 1998 among those aged 16-24 supports
the hypothesis of an upward trend among young people. The proportions
of men aged 16-24 who reported current cigarette smoking in each of
the five years from 1994 to 1998 were 35%, 36%, 38%, 36%, 41%. Corresponding
proportions of women were 34%, 37%, 35%, 38% 38%. 3.3.4 Number of cigarettes smoked by smokers Those who smoked cigarettes were asked how many cigarettes they smoked during the week and at the weekend. Analysis was conducted using cross-tabulations and logistic regression. Number of cigarettes smoked per day by sex and age Reported daily consumption was higher among men smokers (15.7 cigarettes a day) than among women smokers (13.6 per day). Among men smokers, the number smoked per day was highest among those
aged 45-54 and 55-64 (18.2 cigarettes per day). Among women smokers,
it was highest among those aged 45-54 (16.0 cigarettes per day). Number of cigarettes smoked per day by social class of head of household There was a clear social class gradient in number of cigarettes smoked
per day, with smokers in Social Classes I and II smoking least and smokers
in Social Classes IV and V smoking most, among both men and women. However,
there was no social class gradient once the effects of age, educational
qualifications and income were taken into account using logistic regression. Number of cigarettes smoked per day by equivalised household income Among those who smoked cigarettes, the mean number of cigarettes per
day tended to increase as income decreased, although the gradient was
not consistent. Among men smokers, the highest average reported daily
consumption figure (16.8 cigarettes a day) was found in the second lowest
income quintile, and the lowest average consumption figure in the highest
income quintile (14.3 cigarettes/day). The same pattern was found for
women smokers (those in the second lowest income quintile smoked 14.3
cigarettes per day, and those in the highest income quintile smoked
11.6 cigarettes per day). Logistic regression model predicting heavy cigarette smoking The dependent variable for the logistic regression model was heavy smokers, those who smoked 20 or more cigarettes per day. As well as being a main predictor of cigarette smoking prevalence, age was also a main predictor, among current cigarette smokers, of heavy smoking (20 or more cigarettes per day), but the relationship between age and heavy smoking was different from that between age and cigarette smoking prevalence. For men who smoked, the odds of being a heavy smoker increased with age to reach a maximum among those aged 45-54 (odds ratio 2.06 relative to average), but then decreased with age, the odds among men aged 65 and over of being heavy smokers not being significantly different from those of men aged 16-24 (odds ratio relative to average 0.54 for men aged 16-24, odds ratio relative to average 0.53 for men aged 75 and over). For women the pattern was the same, but the differences were not as pronounced. Among women smokers, those aged 45-54 had an odds ratio of 1.76 (relative to average) of being heavy smokers, those aged 16-24 an odds ratio of 0.59 and those aged 75 and over an odds ratio of 0.67. The contrasting patterns of cigarette smoking prevalence and of heavy smoking among smokers can be seen by comparing Figure 3D with Figure 3C. Smoking prevalence decreases with age, and smoking cessation increases.
Those who stop smoking are likely to be lighter smokers, so that people
who remain smokers tend to be heavier smokers. This probably accounts,
at least in part, for the increase into middle age in the odds of smokers
being heavy smokers. In addition, heavy smokers have higher death rates,
and this may explain why the odds of being a heavy smoker decrease with
increasing age from the peak at 45-54. Social class was not a significant predictor of heavy smoking once adjusted for other factors in the model. For men smokers, there was no clear relationship between household income and the odds of being a heavy smoker. In contrast, for women smokers, the odds of being heavy smokers increased as income decreased, so that women in the lowest income quintile were nearly twice as likely to be heavy smokers as smokers in the top income quintile (odds ratios relative to average: top quintile 0.74 bottom quintile 1.41). The highest educational qualification achieved was a significant predictor
of heavy smoking among men and women smokers. Smokers with no educational
qualifications were more likely to be heavy smokers (odds ratio relative
to average 1.37 for men, 1.34 for women). 3.3.5 Type of cigarettes smoked Tar, nicotine and carbon monoxide yields The Laboratory of Government Chemists (LGC) conducts an annual survey for the Department of Health to determine the tar, nicotine and carbon monoxide yields of brands of cigarettes available in the UK. This data was linked to the Health Survey data using the information collected about the main brand of cigarette smoked by informants. Yields of tar, nicotine and carbon monoxide are highly correlated, so to avoid repetition, this section, after an initial discussion of cigarette content, considers tar yields only. Those who roll their own cigarettes, for whom information about content was not available, were included as a separate category. Tar describes the particulate matter inhaled when the smoker draws on a lighted cigarette, and tar levels have been linked to prevalence of CVD.17 Nicotine, an alkaloid, is a powerful drug which stimulates the central nervous system, increasing the heart rate and blood pressure, leading to the heart needing more oxygen. Its effects are related to mode of delivery. Cigarette smoking is the optimal delivery system, producing effects on the brain within seconds and peak blood levels not achieved by tobacco dependence products.18 In a report published in March 1998, the Government's Scientific Committee on Tobacco and Health said: 'Over the past decade there has been increasing recognition that underlying smoking behaviour and its remarkable intractability to change is addiction to the drug nicotine. Nicotine has been shown to have effects on brain dopamine systems similar to those of drugs such as heroin and cocaine.'19 Carbon monoxide, the main poisonous gas in car exhausts, is present in all cigarette smoke. It binds to haemoglobin much more readily than oxygen does, thus raising the blood carboxyhaemoglobin levels, particularly in heavy smokers. Reliability and validity of tar, nicotine and carbon monoxide yield data The LGC data was derived from tests of packs sampled between January and December 1997,20 the year preceding the 1998 Health Survey for England fieldwork. At the end of 1997 new legislation affecting the permitted tar levels of cigarettes came into effect. From January 1st 1998 the maximum tar yield for new production of cigarettes was reduced to 12 mg/cigarette. Between January 1st 1998 and December 31st 1998 existing production stock with declared tar yields in the range 12-15 mg/cigarette was allowed to be retailed. From January 1st 1999 it was illegal to sell any cigarettes with declared tar yield over 12 mg/cigarette. The analysis of tar yields may thus slightly overestimate the actual yields of tar in the cigarettes that informants were smoking at the time they were interviewed. There are concerns whether measured tar, nicotine and carbon monoxide yields have much relevance to actual smokers' exposures. The yields used in the analysis were those determined by smoking machines.21 It is known that there is variation in the way people smoke cigarettes, and that smokers may compensate for low nicotine content in cigarettes by inhaling more, or by blocking ventilation holes in the filter with fingers, saliva or lips. Thus people may receive higher tar and nicotine levels than those indicated by the machine test and there may also be considerable variation between smokers. Cigarette type smoked by sex and age Men were more likely than women to smoke roll-ups (26% of men, 7% of women), and less likely to smoke cigarettes with a tar yield of under 10 mg/cigarette (20% of men, 37% of women). Among men smokers, the proportion smoking roll-ups increased with age, from 15% of those aged 16-24 to 31% of those aged 45 and over. For women, the proportion smoking roll-ups did not vary greatly by age, but the proportion smoking cigarettes
with a tar content of 10 mg/cigarette or more decreased from 52% of
those aged 16-24 to 33% of those aged 55 and over. Cigarette type smoked by social class Compared to men in non-manual classes, men in manual classes were
more likely to smoke roll-ups (age-standardised proportion 35% in Social
Classes IV & V, 16% in Social Classes I & II), and less likely
to smoke cigarettes with a tar yield of under 10 mg/cigarette (12% in
Social Classes IV & V, 32% in Social Classes I & II). Among
women, those in Social Classes I & II were more likely to smoke
cigarettes with a tar content of under 10 mg/cigarette than those in
manual classes (age-standardised proportion 53% in Social Classes I
& II, 30% in Social Classes IV & V). Cigarette type smoked by income Among men smokers, the proportion smoking roll-ups decreased as income
increased, from 34% (age-standardised) among the bottom income quintile
to 12% among the top income quintile. The reverse was true for cigarettes
with a tar yield of under 10 mg/cigarette, where prevalence increased
with increasing income from 13% in the bottom income quintile to 38%
in the top income quintile. Men in higher income groups were thus more
likely to smoke branded cigarettes and to smoke those with lower tar
levels. Women smokers in the bottom income quintile were more likely to smoke
cigarettes with a higher tar yield (51% with tar yield of 10 mg/cigarette
or more, 28% with tar yield of under 10 mg/cigarette), while the reverse
was true for women in the top income quintile (28% with tar yield of
10 mg/cigarette or more, 60% with tar yield of under 10 mg/cigarette). 3.3.6 Levels of saliva cotinine among smokers Previous research22 has shown that people with high scores on a measure of deprivation23 tend to have higher cotinine levels, even when controlling for the number of cigarettes smoked, but that analysis could not control for the brand of cigarette smoked. The addition of cigarette brand questions to the 1998 survey allows a further investigation of the relationship between deprivation or socio-economic factors and high cotinine levels when both the number of cigarettes smoked and the nicotine yield are taken into consideration. A logistic regression was undertaken among smokers in which the dependent variable was a cotinine value in the top quartile (at or above the 75th percentile) of smokers' cotinine values (431.5 ng/ml for men and 369.0 for women). Values in the top quartile are referred to below as 'high' cotinine levels. The logistic regression was conducted in the same way as the others reported above, with additional variables for number of cigarettes smoked per day, and the type (tar yield) of cigarette smoked. Not surprisingly, the odds of a high cotinine level increased with the number of cigarettes smoked per day. The odds of high cotinine were only one-third of average among those smoking under ten per day, but 21/2 times average among those smoking 20 or more a day, these results being similar for both men and women. Age was also an important predictor of high cotinine levels. Men smokers aged 16-24 were the least likely to be in the top cotinine quartile (odds ratio 0.43 relative to average), and men smokers aged between 35 and 54 were the most likely to be in the top quartile (odds ratio relative to average 1.53 for those aged 35-44, 1.63 for those aged 45-54). Similar results were found for women. Level of educational qualification was an important predictor of high cotinine levels among women smokers. Women smokers with no qualifications or whose highest qualification was 'O' level standard were more likely than average to be in the top cotinine quartile (odds ratios relative to average 1.36 for no qualifications, 1.56 for 'O' levels), whilst women with a degree had an odds ratio of 0.55 relative to average. There was no significant relationship between educational qualifications and high saliva cotinine levels among men. The type of cigarette smoked was a predictor of high cotinine levels for both men and women. Men smokers who smoked roll-ups were 48% more likely than average to be in the top cotinine quartile. Chapter 7 of the Scientific Committee report already referred to19 commented that hand-rolled cigarettes have on average higher yields of nicotine than manufactured cigarettes. For men (in cases where the tar level of the cigarette smoked was known), the odds of high cotinine levels were not significantly different between those smoking branded cigarettes with tar yields of 10 mg/cigarette or more and those smoking cigarettes with under 10 mg/cigarette. The odds ratio for women who smoked branded cigarettes with a tar yield of 10 mg/cigarette or more was 1.34, compared to an odds ratio of 0.79 for women who smoked cigarettes with a tar yield of under 10 mg/cigarette. There were no additional effects due to income or social class once
other variables were taken into consideration, but it should be borne
in mind that all the variables identified as significant predictors
are themselves highly correlated with both income and social class.
3.4.1 Introduction The White Paper 'Our Healthier Nation' states that a good diet is an important way of protecting health.24 Unhealthy diets have been linked to cardiovascular disease (CVD), cancers and dental decay. One of the targets of 'Our Healthier Nation' is to improve the diet of the population by educating and providing information about diet and health to groups at risk, and to ensure that there is adequate access to, and availability of, a wide range of healthy foods. An increase in the intake of fruit and vegetables and a reduction in the consumption of fats and salt can have a beneficial influence on health. Dietary modifications which reduce fat intake,25 increase fibre intake from fruits and vegetables (in particular from cereals and grains26) and reduce salt intake27 can aid in reducing the risk of developing CVD.28,29,30 Increased fat intake is directly related to obesity,31,32 a major risk factor for CVD. There are many studies which suggest that antioxidant vitamins from dietary sources such as fruits and vegetables have a preventive role in the development of atherosclerosis (a condition where fatty plaques are deposited on the walls of arteries causing hardening and narrowing) which is associated with CVD.33,34 There is evidence to suggest that there are significant associations between greater sodium intake and high blood pressure.30,35 A diet rich in fruits, vegetables, and low-fat dairy foods and with reduced saturated and total fat can substantially lower blood pressure, a major risk factor for CVD.36 It is therefore important to develop strategies aimed at offering an additional nutritional approach to preventing and treating hypertension alongside the more traditional pharmacological treatment. The Health Survey for England included questions on eating habits in the years 1994 and 1997 and focused on the behavioural patterns relating to a few 'healthy eating' messages. The interview included questions about the types and frequency of categories of foods eaten. This was a simplified version of a food frequency questionnaire aimed at getting a broad view on the general eating habits of the population in England. Detailed information on the British diet is collected in other surveys, the National Food Survey37 and the National Diet and Nutrition Surveys38 carried out among different age cohorts. 3.4.2 Methods In 1998 the eating habits questionnaire underwent substantial changes. The modified version was based on the Dietary Instrument for Nutrition Education (DINE) questionnaire, developed by the Imperial Cancer Research Fund's General Practice Research Group to assess dietary fat and fibre intake.39,40 The DINE consists of a weighted food frequency questionnaire of 19 groups of food which together accounted for 70% of the fat and fibre in the typical UK diet according to the National Food Survey, together with measures of the types of spread, frying and cooking fat used.39 Scores were assigned to food groups proportionally to the fat and fibre content of a standard portion size. The DINE provides a quick assessment of an individual's diet by adding the scores relevant to the frequency of consumption of the groups of foods to give a total fat and a total fibre score. For both fat and fibre, three categories are then derived grouping the scores: low intake (less than 30), medium intake (30-40) and high intake (more than 40). A total fat score of 30 or less on the DINE is estimated to represent a fat intake of 83g/day or less, which corresponds to about 35% of the energy recommended dietary allowance (RDA) for adults in the UK.41 A score of 40 or more indicates a fat intake greater than 122g/day or about 40% of energy RDA. Fibre intake was assessed from sub-scores for fruit and vegetable intake, breakfast cereal, and bread. A total fibre score of 30 or less is estimated to correspond to a dietary fibre intake of 20g/day or less, which is about the national average, and the high fibre score of 40 or more represents more than 30g/day. The DINE scores for fat and fibre consumption measure absolute intakes, not intakes relative to recommended daily allowances, or to individual differences in size. This should be taken into account when interpreting the results. The DINE questionnaire was adapted for use on the Health Survey (See Appendix A for a copy of the questionnaire used). Some food categories, such as pasta or rice and potatoes, which were separated categories in the original DINE within the vegetables sub-section, were combined in the Health Survey questionnaire into one question covering all three foods. The same applied to the meat and meat products sub-sections, where questions on beefburgers or sausages, beef, pork or lamb, bacon, meat pies and processed meat were combined into one question covering all these foods. The scores for the combined foods question were assigned by comparison with results obtained from the Oxford and Collaborators Health Check (OXCHECK) data.42 Moreover, the frequency of consuming a serving of particular foods was kept the same as DINE but a 'Rarely or never' category was added for most food groups. This category was scored the same as the lowest frequency group in the original DINE questionnaire. Cases where people did not consume the 'usual types of food' specified were excluded from the analysis. Every attempt was made to ensure that the Health Survey yielded results comparable to those from DINE, but given the differences in the questionnaires, the correspondence may not be exact. It has been assumed in this chapter that it is close enough to justify imputing to the Health Survey categories the same actual intakes as the equivalent DINE categories, but it should be noted that the correspondence is an assumption that has not been verified. The questionnaire was administered by the interviewer only to informants aged 16 and over. Informants were also asked whether salt was added to prepared foods at the table. Fat, fibre and salt consumption and socio-economic variations (by social class, equivalised household income, Health Authority area type) in fat and fibre consumption are reported in this chapter. Tables of fat and fibre scores by region are also appended, but are not commented on. 3.4.3 Fat and fibre consumption, and addition of salt to food, by age Fat consumption Fat consumption was notably higher among men than women. The mean fat score was 34.0 for men, 28.3 for women. The prevalence of high fat scores (over 40) was 26% among men and 11% among women. In men, the youngest age group (16-24) had the highest mean fat score (38.7). This was also the age group with the highest prevalence of high fat consumption (38%): more than one in three men in this age group were in the high fat intake category, while in the middle age groups (25-64) this proportion was less than 25%. Prevalence of high fat consumption increased to 27% in those aged 65-74 and 28% in those aged 75 and over. In women too, fat consumption was higher in the extreme age groups than in the middle age groups, nevertheless a difference from the pattern for men was seen; the highest mean fat score (31.8) was in those aged 75 and over, amongst whom about a fifth were in the high fat consumption category (18%). Young women aged 16-24 were in the second highest fat consumption category (16%).The prevalence of high fat was much lower for women than for men in all age groups. Fibre consumption There was less difference between the sexes in fibre consumption than in fat consumption. 53% of men and 60% of women had low fibre scores. In both sexes, there was a steady increase in fibre intake with age. Fibre intake was thus lowest among the youngest age group. About two thirds of young men aged 16-24 (65%) and more than two thirds of women in the same age group (72%) had low fibre intake. It is therefore the youngest group who appear at most risk, having
diets that are relatively high in fat and low in fibre. Salt consumption Slightly over a third of men (35%) stated they added salt to food
without tasting it first. Men aged 55 and over were more likely than
those aged 16-54 to 'add salt to food without tasting it first'. The
proportion of women adding salt without tasting was lower, at 24%. Contrary
to what was observed in men, this eating pattern was more likely to
be present in those aged 16-44 years than in the older age groups. 3.4.4 Socio-economic variations in fat and fibre consumption The report of the Independent Inquiry into Inequalities in Health raised important issues about the link between diet and inequalities in health.43 There is increasing evidence suggesting that nutrition-related diseases cluster in disadvantaged groups. People in lower socio-economic groups buy more foods that are high in fat, tend to eat less fruit and vegetables, and less food which is high in fibre. The report highlighted the differences in dietary intake with social class, income, and areas of residence (disadvantaged vs affluent). These issues are investigated below by examining differences in the prevalence of high fat and low fibre intake by social class, equivalised income and Health Authority area type. All prevalences quoted are age standardised. Social class of head of household There was a clear social class gradient in age-standardised high fat intakes in both sexes; the prevalence of high fat intake increased steadily from Social Class I to Social Class V in both sexes. In men, the increase was from 19% in Social Class I to 38% in Social Class V and in women from 7% in Social Class I to 17% in Social Class V. In all social classes, the prevalence of high fat intake was much higher in men than in women. There was a strong social class gradient in the prevalence of low
fibre intake, increasing steadily in both men and women from Social
Class I (45%, 47%) to Social Class V (59%, 68%) although the increase
was greater in women. Equivalised household income In men, the prevalence of high fat intake was highest (33%) in the lowest income quintile and lowest in those in the highest income quintile (19%). In women the prevalence of high fat intake also increased from the highest income quintile (6%) to the lowest income quintile (18%). Unlike the prevalence of high fat intake, the prevalence of low fibre
intake did not show a clear overall relationship to income in either
sex, but both men and women in the lowest income quintile had the highest
prevalence of low fibre intake (59% and 67% respectively). Health Authority area type High fat intakes were highest in men from Rural areas (30%) and lowest
in men living in Inner London (19%). This did not appear to be true
for women, among whom the highest prevalence of high fat intake (14%)
was in Inner London. High prevalence was observed among women living
in Mining and Industrial (13%) and Rural areas (13%). 3.5.1 Introduction Obesity, a major risk factor for cardiovascular disease, diabetes, hypertension and premature death,44,45,46,47,48,49,50 is increasing amongst adults in England ,51,52,53 and in other western populations.48,54 In particular, the abdominal or android type of obesity (see section on waist-hip ratio in Section 3.5.2) has been generally recognised as a risk factor in relation to these chronic diseases.50,54,55 The anthropometric measures presented in this chapter for adults (aged 16 and over) focus on measurements relevant to obesity. Height and weight data used to calculate body mass index (BMI) were collected in each year of the Health Survey series. Waist and hip data, used to calculate waist-hip ratio (WHR), were collected in 1994, 1997 and 1998. Firstly, the methods and definitions of these measurements are described. The distributions and trends over time of these measurements are then reported. Finally, the associations of some socio-economic variables (social class, equivalised household income, Health Authority area type) with raised BMI, obesity and WHR are examined. Tables for height, weight and demi-span are appended, but not commented on. 3.5.2 Methods and definitions of measurement Full details of the protocols for carrying out the measurements are contained in Volume II, Appendix B and are briefly summarised here. Height and weight were measured during the interview visit while waist and hip circumferences and demi-span were measured during the nurse visit. Height Height was measured using a portable stadiometer with a sliding head plate, a base plate and three connecting rods marked with a metric measuring scale. Informants were asked to remove shoes. One measurement was taken, with the informant stretching to the maximum height and the head positioned in the Frankfort plane. The reading was recorded to the nearest millimetre. Weight Weight was measured using a Soehnle electronic scale with a digital display. Informants were asked to remove shoes and any bulky clothing. A single measurement was recorded to the nearest 100g. Informants who were pregnant, chairbound, or unsteady on their feet were not weighed. Informants who weighed more than 130 kg were asked for their estimated weights because the scales are inaccurate above this level: these estimated weights were included in the analysis. In the analysis of height and weight, data from those who were considered by the interviewer to have unreliable measurements, for example those who had excessive clothing on, were excluded from the analysis. Body Mass Index (BMI) In order to define overweight or obesity, a measurement is required which allows for differences in weight due to height. A widely accepted measure of weight for height, the Body Mass Index (BMI), defined as weight (kg)/height (m2), has been used for this purpose in the Health Survey series. However BMI does not distinguish between mass due to body fat and mass due to muscular physique. It also does not take account of the distribution of fat. BMI was calculated for all those informants for whom a valid height and weight measurement was recorded. Adult informants were classified into the following BMI groups:
In the 1998 report the obese category has been split further into 30-40 and 40+; the latter category defined as morbid obesity. Morbid obesity is recognised as a serious illness which is associated with a poor quality of life and with co-morbidities, and has an economic impact on health care. Weight loss treatments such as behavioural, diet, exercise and drug treatments have shown limited success. Surgical methods have been shown to be more effective for reducing weight in those who are morbidly obese.56,57 Previous Health Surveys have indicated that the prevalence of morbid obesity in England, particularly in women, is increasing.58 Waist and hip Waist was defined as the midpoint between the lower rib and the upper margin of the iliac crest. Waist was measured using a tape with an insertion buckle at one end. Hip was defined as the widest circumference around the buttocks below the iliac crest. Both measurements were taken twice, using the same tape, and were recorded to the nearest even millimetre. Those whose two waist or hip measurements differed by more than 3 cm had a third measurement taken. The mean of the two valid measurements was used in the analysis. For waist and hip measurements all those who reported that they had a colostomy or ileostomy, or were chairbound or pregnant, were excluded from the measurement. All those with measurements considered unreliable by the nurse, for example due to excessive clothing or movement, were excluded from the analysis. Waist-hip ratio Waist-hip ratio (WHR) was defined as the waist circumference divided by the hip circumference, ie waist girth (m)/hip girth (m). WHR is a measure of deposition of abdominal fat, ie central obesity. Unlike BMI there is no consensus about appropriate WHR criterion levels.59 For consistency, the same cut-off values as in the 1994 report have been used. A raised WHR has been taken to be 0.95 or more in men and 0.85 or more in women. WHR was calculated for all informants who agreed to a nurse visit and for whom a valid waist and hip circumference measurement was recorded. Demi-span Demi-span is defined as the distance between the mid-point of the sternal notch and the finger roots with the arm outstretched laterally. It is an alternative to height as a measure of skeletal size, especially useful in elderly people in whom a certain height loss occurs with age. Measurements were made with the right arm outstretched using a metal retractable tape. Two measurements were taken to the nearest even millimetre with the informant in light clothing and with bulky jewellery removed. If there was a difference between the two measurements of more than 3 cm, a third measurement was taken. The mean of the two valid measurements was used in the analysis. All informants aged 65 and over were eligible for the measurement. Measurements considered unreliable, for example, due to excessive clothing, or if there was partial response (where only one measurement was obtained or if the difference between the two measurements was greater than 3 cm and a third measurement was not taken) were excluded from the analysis. Demi-span measurements are not commented on in this chapter. Response to anthropometric measurements Valid height (94%) and weight (93%) measurements were obtained from a majority of informants. Weight and height measurements allowed BMI to be computed for 91% of people aged 16 years and over in both sexes (including 37 informants who gave their estimated weights because they weighed more than 130 kg). Valid WHR measurements were obtained for 99% of informants aged 16
years and over who were visited by a nurse. Valid demi-span measurements
were obtained for 96% of informants aged 65 years and over who were
visited by a nurse. 3.5.3 Trends in body mass over time, by age Trends over time in BMI In 1998, mean BMI was 26.5 kg/m2 in men and 26.4 kg/m2 in women. Overall, obesity was more prevalent in women (21.2%) than in men (17.3%), and the prevalence of morbid obesity among women (1.9%) was also higher than in men (0.6%). Both mean BMI and the prevalence of obesity increased: up to a maximum of 27.8 kg/m2 in men aged 55-64 (prevalence of obesity 23.3%) and to 27.8 kg/m2 (prevalence of obesity 29.0%) in women aged 65-74. BMI and obesity then decreased in older age groups. During the period 1994-1998 mean BMI gradually increased among men and women in most age groups. The overall age-standardised increase of 0.44 kg/m2 in men (95% CI 0.31-0.57) and 0.57 kg/m2 in women (95% CI 0.41-0.72) was statistically significant in both sexes (p<0.001). The prevalence of obesity also increased gradually in most age groups
in both sexes from 1994 to 1998. The increase was greater in women than
men and in those aged 45 years and over than those aged 16-44 years.
Looking at all ages, the prevalence of obesity in men rose from 13.8%
in 1994 to 17.3% in 1998. In women, the increase was more marked, from
17.3% in 1994 to 21.2% in 1998. Trends over time in WHR In 1998, mean WHR was greater in men than in women, and, in general, tended to increase with age. This pattern was more marked in women than men. For all ages combined, mean WHR was 0.91 in men and 0.80 in women. In both sexes the changes in mean WHR from 1994 to 1997 and 1998 were
very small and no clear pattern emerged. 3.5.4 Body mass differences by socioeconomic status BMI and social class of head of household Age-standardised mean BMI did not show a clear pattern by social class in men, being highest in Social Class IIIM (26.7 kg/m2) and lowest in Social Class I (25.9 kg/m2). In women, there was a steady increase from Social Class I (25.4 kg/m2) to Social Class V (27.2 kg/m2). In men, the age-standardised prevalence of being overweight was highest in Social Class II (46.8%) and was generally higher in non-manual than manual social classes. The reverse was true for the prevalence of obesity and of morbid obesity which was higher in manual than non-manual social classes. Men from Social Class IIIM showed the highest prevalence (19.6%) of obesity while the prevalence of morbid obesity was highest in Social Class V (1.6%). In women, the age-standardised prevalence of being overweight was
highest in Social Class II (32.5%). The prevalence of obesity showed
a strong social class gradient, increasing progressively from Social
Class I (14.4%) to Social Class V, where more than a quarter (28.1%)
of women were obese. The prevalence of morbid obesity also increased
steadily from Social Class I (0.7%) to Social Class V (3.3%). WHR and social class of head of household Mean WHR did not vary consistently from Social Classes I to V in men. The highest mean WHR in men was 0.91 in Social Class IIIM and the lowest was 0.89 in Social Class I and IIINM. In women there was a gradual increase in mean WHR from Social Class I (0.78) to V (0.81). In both sexes, the age-standardised prevalence of raised WHR (>0.95
in men and >0.85 in women) was higher in informants from manual
Social Classes (IIIM, IV, V) than from non-manual Social Classes (I,
II, IIINM). Again, the pattern was more evident in women, where the
prevalence of raised WHR generally increased from Social Class I (18.0%)
to Social Class V (26.6%). BMI and equivalised household income There was no great difference in age-standardised mean BMI in men from the highest to the lowest quintile of equivalised income. In women however, there was a gradual increase in mean BMI from the highest income quintile (25.4 kg/m2) to the lowest income quintile (27.1 kg/m2). In men, the prevalence of being overweight increased from the lowest income quintile to the highest, where it was 46.8%. No clear pattern emerged in women. The prevalence of obesity increased from the highest to the lowest income quintile in both men (14.5%, 20.3%) and women (15.9%, 26.3%) respectively. The prevalence of morbid obesity also showed a strong gradient for
equivalised income, increasing for both sexes. In men it increased from
0.6% in the highest income quintile to 1.2% in the lowest income quintile
and in women it increased from 0.9% to 3.7% respectively. WHR and equivalised household income In men, observed mean WHR showed only small differences by income from the highest to lowest quintile. In women, however, there was a stronger gradient for income, increasing from the highest (0.77) to the lowest (0.81) quintile. In both sexes, the prevalence of age-standardised raised WHR was higher
in informants from the lower income quintiles than in those from the
higher income quintiles. BMI and Health Authority area type Age-standardised mean BMI was highest in men (26.6 kg/m2) and women (26.7 kg/m2) living in Rural areas. The lowest mean BMI was in men living in Inner London (26.0 kg/m2) and in women living in Prosperous areas (26.0 kg/m2). In men, no clear pattern of variation in the prevalence of raised BMI was seen by area type. The age-standardised prevalence of being overweight was higher in men in Mature areas (46.1%). The prevalence of being obese was higher in Urban areas (19.5%) and the prevalence of being morbidly obese was higher in Inner London (2.2%). In women, the age-standardised prevalence of being overweight and
of being obese was higher in Rural areas (33.3%, 23.0%), however the
prevalence of morbid obesity was higher in Mining and Industrial areas
(2.1%). WHR and Health Authority area type In both sexes, mean WHR showed only small differences between Health Authority area types . In men, mean WHR was highest in Mining and Industrial areas (0.91). In women, mean WHR was highest in Inner London (0.81) and lowest in both Urban and Prosperous areas (0.79). A similar pattern to mean WHR was seen for the prevalence of age-standardised
raised WHR in both sexes. In men, raised WHR was highest in Mining and
Industrial areas (30.4%) and for women raised WHR was higher in Inner
London (28.8%) than in other areas.
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||
