Men Who Do Not Drink: A Report from the British Regional Heart Study

(1)

©International Epidamiologlcal Association 1388 Printed in Great Britain

Men Who Do Not Drink: A Report from the British Regional Heart Study

GOYA WANNAMETHEE AND A G SHAPER

Wannamethee G (Department of Clinical Epidemiology and General Practice, Royal Free Hospital School of Medicine, London NW3 2PF, UK) and Shaper A G. Men who do not drink: a report from the British Regional Heart Study.

International Journal of Epidemiology 1988,17: 307-316.

Men who do not drink are frequently used as a baseline against which the effects of alcohol consumption are measured. The characteristics of such men have been examined in a large-scale prospective study of cardiovascular disease Involving 7735 middle-aged men drawn from general practices in 24 British towns. Non-drinkers include lifelong teetotallers and ex-drinkers, both long-term and recent Long-term ex-drinkers have many characteristics likely to increase their morbidity and mortality; recent ex-drinkers have similar characteristics but to a less marked degree. Ex-drinkers are older than the other groups and Include an Increased proportion of unmarried mervand men in manual occupations. They have the same high percentage of current cigarette smokers as moderate/heavy drinkers and a prevalence of hypertension and obesity similar to moderate/heavy drinkers and higher than lifelong teetotallers or occasional/light drinkers.

Ex-drinkers have the highest percentage of men with multiple doctor-diagnosed disorders. In particular, they have the highest prevalence rates of angina and possible myocardlal infarction on standardized questionnaire, of myocardial infarction on electrocardiogram and of recall of a doctor-diagnosis of Ischaemlc heart disease. They also have high prevalence rates of recall of high blood pressure, peptic ulcer, diabetes, gall bladder disease and bronchitis. They have the highest rates for regular medical treatment and the highest proportion of men who consider their hearth to be poor.

It is abundantly clear that the general category of non-drinkers, which Includes a large proportion of ex-drinkers, should not be used as a baseline against which to measure the effects of alcohol consumption. Overall, it would appear that the occasional/light drinking category (<15 drinks/week) provides a large and satisfactory baseline group for comparative purposes in the study of cardiovascular and other organic disorders.

In the majority of studies which examine the relationship between alcohol intake and disease, men who do not drink are used as a baseline against which to measure the effects of alcohol consumption. Ever since Pearl in 1926 described the 'U-shaped curve' of all- cause mortality related to alcohol intake,¹ it has been a common finding that non-drinkers have a higher mortality for all causes and sometimes for specific diseases, eg cardiovascular disease, than men who are light or moderate drinkers.24 This finding has been interpreted as showing that light or moderate drinking is beneficial in general or that it protects against specific diseases, eg cardiovascular disease/ischaemic heart disease.4-5

Clearly, this interpretation depends on the definition and characteristics of those designated as non-drinkers. This study of middle-aged men drawn from all social classes in all the major regions of Great Britain (The British Regional Heart Study) examines the characteristics of men who do not drink, in order to

Department of Clinical Epidemiology and Genera] Pnctice, Royal Free Hospital School of Medicine, London NW3 2PF, UK.

evaluate the use of the non-drinkers as a baseline group for comparative purposes.

SUBJECTS AND METHODS

In 1978-80, 7735 men aged 40-59 years were selected at random from general practices in 24 towns in England, Wales and Scotland for a prospective study of cardiovascular disease. The criteria for selecting the town, the general practice and the subjects as well as the methods of data collection, have been reported.⁶ The social class distribution of the practice population was required to represent that of the town. Research nurses administered to each man a standard questionnaire (Ql) which included questions on alcohol consumption, smoking habits, chest pain and medical history. A number of physical measurements were made, an electrocardiogram recorded and blood taken for biochemistry and hacmatology. Following examination, all men were followed for morbidity from cardiovascular events (stroke, myocardial infarction) and for mortality from all causes.7

Five years after the initial screening a similar ques- 307

at Pennsylvania State University on May 9, 2016http://ije.oxfordjournals.org/Downloaded from

(2)

tionnaire (Q5) was mailed to all the surviving men.

Detailed information was obtained on present and past drinking habits, changes in smoking behaviour, history of chest pain, medical history and employment status.

A total of 7275 men, 98% of available survivors, completed the questionnaires satisfactorily. The available survivors exclude those men who had died (n «= 297) or emigrated (n = 42) prior to the fifth-year questionnaire. The 2% non-response comprises those who had moved and not yet registered with a new doctor and those who did not wish to complete the questionnaire.

This report is concerned with the 7275 men who completed the fifth-year questionnaire.

Drinking Behaviour

Alcohol consumption was recorded at initial screening using questions on frequency, quantity and type, similar to those used in the 1978 General Household Survey.* The men were asked to describe their current alcohol intake as none, on special occasions only or once/twice a month, at weekends, or daily/most days.

They were also asked how much they usually took per day when they drank—one to two drinks, three to six drinks, or more than six drinks. A drink is defined as i pint of beer, a single of spirits or a glass of wine (approximately 8-10 g alcohol). No history of previous drinking was requested on this occasion and it was therefore not possible to separate non-drinkers into lifelong teetotallers and ex-drinkers. In the initial analysis of data, eight drinking categories are used:

non-drinkers, occasional drinkers (special occasions or 1-2 per month), weekend drinkers (1-2, 3-6 or more than 6 drinks per day) and men drinking daily or on most days (1-2, 3-6, or more than 6 drinks per day).9

Five years later (1983-85), in addition to the questions on frequency and quantity of alcohol consumed, the men were asked about their past drinking habits.

Those who said they were non-drinkers at Q5 were asked whether they bad been drinkers in the past and if so, what their past alcohol consumption had been.

There was inadequate information at Ql or Q5 in 33 men and they were excluded from the analysis, leaving 7242 men.

Classification of Drinking Behaviour

Classification of drinking behaviour in this paper is based on the combined information obtained at the initial screening (Ql) and five years later (Q5).

1. Teetotallers. Men who were non-drinkers at Ql and Q5, and who said at Q5 that they had never been drinkers (n = 204).

2. Ex-drinkers (Long-term). Non-drinkers at Ql and Q5 who reported previous drinking (n = 108).

3. Ex-drinkers (Recent). Non-drinkers at Q5 who were occasional or regular drinkers at Ql (n = 388).

4. Occasional/Light drinkers (n = 3914). (a) Occa- sional drinkers at Ql who were drinking at any level at Q5 (n = 1492). (b) Non-drinkers at Ql who were drinking at Q5 (n =119); 80% of this group claimed to be occasional drinkers at Q5. (c) Light drinkers include weekend 1-2 and 3-6/day and daily 1-2 drinks/day (n = 2303) based on Ql.

5. Moderate!Heavy drinkers (n = 2628) based on Ql.

Moderate drinkers comprise weekend >6/day and daily 3-6 drinks/day (n = 1876). Heavy drinkers are those taking >6 drinks/day on a daily basis (n =» 752).

Smoking

Smoking habits were derived from the initial questionnaire (Ql) and the men were classified as current smokers, ex-cigarette smokers and those who had never smoked. Ex-cigarette smokers who currently smoked pipe/cigars are included in ex-cigarette smokers. Those men who only ever smoked pipe/cigars are included in 'never smoked'. Smoking data were missing for 14 men.

Social class

The longest held occupation of each man was recorded at Ql and the men were grouped into one of the six social classes defined by the Registrar General: I, II, HI non-manual, III manual, IV and V. Those whose longest-held occupation was in the Armed Forces form a separate group. For some comparisons, social classes I, II and III non-manual were combined as 'non- manual' workers, and social classes III manual, IV and V as 'manual' workers. Social class data were missing for 13 men.

Prevalence of Ischaemic Heart Disease

The prevalence of ischaemic heart disease was measured in three ways at initial examination (Ql).

1. A standardized chest pain questionnaire for angina or possible myocardial infarction.¹⁰

2. A three-lead orthogonal electrocardiogram for myocardial ischaemia or infarction.¹⁰

3. Recall of a doctor diagnosis of angina or myocardial infarction (heart attack, coronary thrombosis).¹¹ , Doctor Diagnoses and Medical Treatment

Men were asked to recall whether they had ever been told by a doctor that they had any of the conditions listed on the questionnaire (Ql). A total disease score was obtained by giving one point for each affirmative

(3)

TABLE 1 Distribution of 7242 middle-aged British men in five drinking categories

Alcohol group No

1 Teetotallers (lifelong) 2 Ex-drinken (long-term) 3 Ex-drinken (recent) 4 Occasional/light 5 Moderate/heavy

204 108 388 3914 2628

2.8 1.5 5.4 54.0 36.3

answer. A single point for ischaemic heart disease was given for any one of the following: angina, heart attack, coronary thrombosis, myocardial infarction. A single point was given for each of the following conditions: 'other heart trouble', high blood pressure, stroke, gout, gall bladder disease, thyroid disease, arthritis, bronchitis, asthma, peptic ulcer, and diabetes. In addition, an affirmative answer to an open question on 'Other conditions' was also given a single point, irrespective of the number or type of such conditions. The maximum score was 13 points. The disease score was grouped into four classes: 0, 1, 2 and >3 disease conditions. The men were also asked details of any regular medical treatment.

RESULTS

Table 1 shows the distribution of men in the five drinking categories. Of the original 466 non-drinkers at Ql, 431 provided information at Q5. About half turned out to be lifelong teetotallers (n = 204), about one-quarter were long-term ex-drinkers (n = 108) and about one- quarter were drinking at Q5 (n = 119), mainly occa- sionally. Men in this last group are now classified as occasional drinkers. The ex-drinkers (recent), those who were drinking at Ql but not at Q5, constitute a group of particular interest, as their characteristics at initial screening relate to a group who give up drinking in the following five years.

Table 2 shows the mean age for each drinking category and the percentage of men from each drinking category within each of the four age groups. Non- drinkers (teetotallers and ex-drinkers) were significantly older than drinkers (51.6 versus 49.9 years;

p<0.001). The oldest group was long-term ex-drinkers who had the highest proportion of men aged 55-59 years and the lowest proportion of men aged 40-44 years. A similar but less marked pattern was seen among recent ex-drinkers and teetotallers in whom the proportion of men in the oldest age group was nearly twice that seen in the youngest age group. Drinkers showed approximately the same proportion of men in each age group.

Marital Status

The ex-drinkers (long-term and recent) had a higher proportion of single men than the drinkers or teetotallers (Table 3). The long-term ex-drinkers had by far the highest proportion of divorced men.

Social Class

Table 4 shows the percentage distribution of men in each drinking category by social class. In the total sample, 41% were non-manual and 59% manual workers. Ex-drinkers had the highest proportion of manual workers and occasional/light drinkers had the highest proportion of non-manual workers.

Smoking

Table 5 shows the percentage distribution of men in each drinking category by their cigarette smoking status (Ql). Teetotallers had by far the highest proportion of men who had never smoked and by far the lowest proportion of current smokers. Ex-drinkers, both long-term and recent, had patterns of cigarette smoking very similar to those seen in moderate/heavy drinkers. Occasional/light drinkers contained a much

TABLE 2 Number of men and mean age in each drinking category and percentage distribution of men from each drinking category by ihefour age groups

Number Mean age (yn) Age group 40-44 45-49 50-54 55-59 All

No 1761 1820 1846 1815

7242

TeetotaDen 204 51.5

% 21.6 14.7 29.4 34.3 100

Ex-drinken Long-term

108 52.5

% 13.9 14.8 32.4 38.9 100

Recent 388 51.3

% 16.2 23.7 30.9 29.1 100

Occ/light 3914 50.1

% 23.8 25.7 25.3 25.2 100

Drinkera

Mod/heavy 2628 49.7

% 26.9 25.7 24.4 23.0 100

(4)

INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

TABLE 3 Distribution (%) ofmen from each drinking category by marital status

Married Single Widowed Divorced All

No 6572

336 90 244 7242

TeetotaBen 89.7

5.4 2.5 2.5 100

fit-drink en Long-term

83.3 6J 1.9 8.3 100

Recent 88.7

7.0 1.0 3.4 100

Occ/light 93.3

3.5 1.2 2.3 100

Drinken

Mod/heavy 87.7

5.9 1.6 4.8 100

higher proportion of men who had never smoked and a lower proportion of current smokers than ex-drinkers or moderat^eavy drinkers. The relationships between drinking and smoking patterns were similar in both manual and non-manual workers (data not displayed).

Other Risk Factors

Table 6 shows the percentage of men from each drinking category who fall into the highest fifth of the distribution of systolic blood pressure, body mass index and serum total cholesterol and the lowest fifth of the HDL-cholesterol distribution. All of these based on the distribution at Ql.

Blood pressure. Ex-drinkers and moderate/heavy drinkers have similar high rates of systolic hypertension. Teetotallers had a slightly higher rate of hypertension than occasional/light drinkers.

Body mass index. The highest rate of obesity was seen in recent ex-drinkers and moderate/heavy drinkers and the lowest rates in occasional/light drinkers and teetotallers.

Serum total cholesterol. Teetotallers had the lowest rate of hypercholesterolaemic men and drinkers had higher rates of raised total cholesterol than ex-drinkers. None of these differences were statistically significant.

HDL-cholesterol. The highest proportion of low HDL-cholesterol concentrations were observed in the ex-drinkers (long-term and recent), followed by teetotallers. The lowest proportion of low concentrations was seen in the moderate/heavy drinkers.

Doctor-Diagnoses

Table 7 shows the age-adjusted percentage of each drinking group with recall of specific diagnoses made by a doctor. Those diseases for which less than 100 cases (total) were recorded were omitted including stroke (41 cases) and thyroid disease (42 cases). The prevalence of doctor-diagnosed disease among teetotallers was in general similar to those among occasional/light drinkers. Teetotallers had the lowest prevalence of high blood pressure, peptic ulcer and gall bladder disease.

The prevalence of high blood pressure was significantly higher among ex-drinkers (long-term and recent) than among drinkers, and among the drinkers the prevalence was slightly higher in the moderate/

heavy drinkers. Peptic ulcer was very common among long-term but not recent ex-drinkers, and like high blood pressure was slightly more frequent in the moderate/heavy drinkers. Gout was most common in the moderate/heavy drinkers, followed by long-term ex-drinkers. Diabetes was more frequent among non-

TABLE 4 Distribution (%) of men from the drinking categories by social class.

Non-manual = I, II and III non-manual. Manual = III manual, IV and V

Social class

I Profesxkmal II Managerial III NM Skilled HIM Skilled IV Semi-fkilled V Unskilled

AH (*)

24 U l

oj

4 5

10} 59

1

*) 100

Teetotaller!

(*)

18 [35 10 J

4 1

18 V 66

1 A

100

Ex-drinkeri Long'tciiu

(*)

M

_{8 [24}

sj

521 19 V77

100

Recent (%)

16 [27

501 14 [72

«J

100

Occ/light (*)

11 1

28V50 11 J 391

8>50

3j

100

Drinken

Mod/heavy (*)

20V33 8j 501 11 >67

6}

100

(5)

TABLE 5 Distribution (%) of men from each drinking category by smoking status

Never Ex-jmoken Current

Totil 1753 2559 2916 7228

TeetouQen 44.6 28.9 26.5 100

Ex-drinkers Long-term

15.7 34.3 50.0 100

Recent 20.9 29.1 50.0 100

Occ/tight 29.0 36.8 34.2 100

Drinkers

Mod/heavy 16.4 34.6 49.0 100

drinkers than drinkers, particularly among recent ex-drinkers.

Long-term ex-drinkers had a significantly higher prevalence of gall bladder disease than any other category. The prevalence of bronchitis was high among moderate/heavy drinkers which presumably reflects the smoking habit in this group. Ex-drinkers also have a high prevalence of bronchitis, possibly indicating their historic relationship to the moderate/heavy drinkers with their high rate of cigarette smoking.

Mean disease score. Table 8 shows the age-adjusted mean disease score and the percentage of men with one, two, three or more disorders in each alcohol group. Long-term ex-drinkers had the highest mean disease score, followed closely by recent ex-drinkers.

The long-term ex-drinkers also had the lowest percentage of men free of disease and the highest percentage of men recalling three or more conditions. The recent ex-drinkers also had a low percentage of men with no disease and a higher percentage of men with three or more disorders than teetotallers or drinkers. Tee- totallers and light/occasional drinkers had very similar findings.

Regular medical treatment. Table 9 shows the age- adjusted percentage of men in each drinking category on regular medical treatment. Almost half (44.1%) of the long-term ex-drinkers were on regular treatment, which is far higher than any of the other groups.

Recent ex-drinkers also had a significantly higher frequency of medical treatment than drinkers. The long- term ex-drinkers showed the highest prevalence of treatment with tranquillisers, pain-killers and anti-

hypertensive drugs. Recent ex-drinkers show higher rates of regular treatment than drinkers and a frequency of anti-hypertensive treatment similar to long- term ex-drinkers. Teetotallers had a significantly higher percentage of men taking regular medical treatment than drinkers but otherwise seem to have similar rates of drug usage (tranquillisers, pain-killers, anti- hypertensives) to drinkers. It is interesting to note that the prevalence of tranquilliser use was lowest among moderate/heavy drinkers, possibly because alcohol may act as a tranquilliser.

Men who do not drink (teetotallers and ex-drinkers) are more likely to be on regular medical treatment than drinkers. The higher prevalence of treatment among ex-drinkers is to be expected as people taking drugs or medication may be advised to abstain.

Prevalence oflHD at Screening (Table 10)

Chest pain questionnaire. Long-term ex-drinkers had by far the highest prevalence of angina, more than twice the prevalence among drinkers. The prevalence among teetotallers was similar to that seen in occasional/light drinkers. Long-term ex-drinkers had a higher rate of possible myocardial infarction (MI) than any other group. Ex-drinkers, particularly long-term, had the highest prevalence rate of possible Ml plus angina and teetotallers had a slightly higher prevalence rate than drinkers. Overall, it was clear that ex-drinkers, long-term or recent, had a higher prevalence of EHD on questionnaire than drinkers or teetotallers.

Age-adjusted rates showed that teetotallers had a similar rate to occasional/light drinkers.

TABLE 6 Percentage of men from each drinking caugory in the highest fifth of the distribution of systolic blood pressure (>160 mmHg), body mass Index (*28 kglm2) and total cholesterol f»7.2 mmol/I), and in the lowest fifth of the HDL-cholesierol distribution (<0.93 mmoUl). •£ test for

difference between percentages: '" " p<0.001

Systolic BP Body miss index Total cholesterol HDL-chotesterol

Teetotallers 21.6 16.7 14.2 25.5

25.0 20.4 15.7 26.7

Recent 24.5 22.2 17.0 26.8

Occ/light 18.5 16.6 19.1 20.5

Drinkers Mod/heavy

24.3 22.3 20.4 12.9

Significance

• ••

NS

• ••

(6)

TABLE7 Age-adjusted prevalence (%) of recall ofdoctor diagnoses in the drinking categories. Overallx? test for differences between percentages:

NS » not significant, ' - p<0.05, " - p<0.01, " * = p<0.001

Doctor diagnosis High blood pressure (887) Peptic ulcer (774) Gout (183) Diabetes (104)

Gall bladder disease (119) Bronchitis (1264) Arthritis (715) Asthma (267) Other conditions (2164)

TeetouDen ( n - 2 0 4 )

10.1 8.0 1.9 1.9 0.5 15.5 9.6 4.5 26.6

(n = 108) 14.3 20.9 2.6 1.7 4.3 18.2 12.4 3.9 45.4

Recent ( n - 3 8 8 )

15.3 10.9 1.7 2.5 2.0 21.0 11.2 5.8 36.2

Drinkers Occ/Kght (n - 3914)

11.1 9.7 1.7 1.5 2.1 15.2 9.0 3.4 30.3

Mod/heavy ( n - 2 6 2 8 )

13.6 11.9 3.9 1.2 0.9 20.5 10.9 3.7 28.0

Significance

• • • NS

• ••

• ««

NS NS

*•*

Electrocardiogram. Long-term ex-drinkers had by far the highest prevalence of definite myocardial infarction on electrocardiogram. Recent ex-drinkers were similar to moderate/heavy drinkers. Teetotallers had a higher prevalence of definite myocardial infarction than drinkers. When definite myocardial ischaemia (including possible myocardial infarction) was considered, ex-drinkers (long-term and recent) had a higher prevalence than drinkers, and teetotallers had the lowest prevalence. Overall, ex-drinkers, particularly long-term ex-drinkers, had a higher prevalence of electrocardiograph^ evidence of ischaemic heart disease than drinkers and teetotallers. Age-adjust- ment confirmed the high rate in ex-drinkers (long- term); teetotallers and occasional/light drinkers had similar low rates.

Recall oflHD and 'other heart troubles'. Long-term ex-drinkers had more than twice the rate of recall of LHD recorded in drinkers. Recent ex-drinkers also had a higher recall rate than drinkers. Teetotallers had a substantially higher rate of LHD than drinkers. The trends among 'other heart trouble' were not statistically significant, although ex-drinkers (long-term) had the highest rates.

Health Status

On Q5 men were asked to record their current health status as excellent, good, fair, or poor and Table 11 shows the distribution of this self-assessment within each drinking category. Long-term and recent ex-drinkers had by far the highest proportion of men who regarded their health status as poor. Only half of the long-term ex-drinkers considered their health to be excellent or good compared with 81% among occasional/light drinkers. Overall, drinkers considered themselves to be in better health than teetotallers and the occasional/light drinkers had by far the highest opinion of their health status.

Prevalence of Disease After Five Years

Thus far we have used the information reported on Ql to examine the prevalence of disease within the various drinking categories. In the five years following initial examination some of these men were likely to develop further diseases. On Q5 men were asked similar questions on recall of doctor-diagnosed disease and regular treatment. This information was again used to examine the prevalence of disease by the alcohol groups. The pattern of disease at Q5 followed a similar trend to that

TABLE 8 Mean total disease score (age-ad/usied) for each drinking category and percentage of each drinking category recalling one, two, three or more disorders. Analysis of variance test for difference between means: '" - p<0.001.

Number Mean score Group 0 1 2

Teetotallers n = 204

0.96

38.7 33.8 20.1 7.4

n - 108 1.48

22.2 31.5 25.0 21.3

Recent

n-388 1.21

28.4 34.8 24.2 12.6

Drinkers Occ/Hght n - 3 9 1 4 0.96

40.4 34.4 16.9 8.4

Mod/heavy Significance n - 2628

1.06

36.1 35.2 18.9 9.8

(7)

TABLE 9 Age-adjusted prevalence (%) of regular medical treatment and specific treatments for each drinking category. Overall -^ test for differences between percentages: •• -p<0.01, ••• = p<0.001.

Medical treatment On regular treatment Tranquillisers Pain-killers Antihypertensive

Total 1975 524 457 320

Teetotallers 31.6

9.1 7.4 3.1

44.1 18.8 10.8 7.0

Recent 34.8

7.8 7.8 6.4

Drinkers Ox/light

25.9 7.4 5.9 4.3

Mod/heavy 27.1

6.3 6.5 42

Significance

• ••

NS NS

recorded at initial screening (data not displayed). The prevalence of recall of IHD and high blood pressure was higher amongst long-term ex-drinkers than at initial screening (20% versus 11% for IHD and 30%

versus 16% for high blood pressure). The prevalence of men on regular medical treatment had increased for all groups. Overall, the pattern of disease among the drinking categories was similar to that observed at screening, with long-term ex-drinkers having by far the higher prevalence of recall of disease, followed closely by recent ex-drinkers.

DISCUSSION

In many of the studies showing that light or moderate drinking is associated with a low IHD mortality rate, the interpretation of the findings depends heavily on the high rates of IHD observed in the non-drinkers.

The definition of non-drinkers varies considerably between studies and often it is not defined. The reasons for a person being a non-drinker are rarely explored and the health characteristics of such people are infre-

quently presented. Drinking behaviour, like many other social behaviours is a dynamic process and may change over time, from heavy drinking to more moderate drinking or even to abstinence. It may also go the other way, and those claiming to be non-drinkers may on later inquiry be found to be drinkers. Some of this change may be genuine; often it will reflect the weak- ness of the classification systems based on administered or self-completed questionnaires. Validation of alcohol intake by biomedical or haematological methods is possible on a group basis,9 but because of the striking individual variation in response to alcohol intake, it is not possible to completely validate individual histories of alcohol intake.

In our present study, separating life-long teetotallers from ex-drinkers, we have observed that ex-drinkers have characteristics which are likely to increase their morbidity and mortality rates from a number of conditions. They are older, more likely to be manual work- eTS and current smokers and they have the highest rates of recall of doctor-diagnosed disease, particularly high

TABLE 10 Prevalence (%) ofischaemic heart disease on chest pain questionnaire, electrocardiography and recall of doctor-diagnosis in each drinking category, t Total prevalence rates adjusted for age appear In brackets. Overall H test for differences between percentages: NS - not significant;

• - p<0.05, " °p<0.01

Chest pain questionnaire Angina only

Possible MI only Both

Totarf

Electrocardiogram

Definite myocardial infarction Definite iscfaaemia/possible MI

Totarf

Recall of doctor diagnosis bcfaaemic heart dtseact

age-adjusted 'Other heart troubk't

No

342 450 184 976

196 305 501

352 427

Teetotallers

3.9 6.4 3.4 13.7 (12.9)

3.9 2.5 6.4 (6.0)

9.3 (8.2)

6.8 (6.7)

10.2 8.3 7.4 25.9 (23.8)

6.5 7.0 13.5 (11.6)

11.2 (9.8) 9.3 (9.0)

Recent

7.2 6.4 4.4 18.0 (17.2)

3.1 5.4 8.5 (8.0)

7.0 (6.2)

4.9 (4.8)

Drinkers Occ/light

4.3 6.0 2.4 12.7 (12.7)

2.4 3.8 6.2 (6.2)

5.0 (5.0)

5.6 (5.6)

Mod/heavy

4.8 6.5 22 13.5 (13.8)

2.8 4.8 7.6 (7.7)

3.8 (4.0)

6.3 (63)

Significance

•

• • NS

(8)

TABLE 11 Distribution <%) of men in each drinking category according to their self-assessed health status

Health Excellent Good Fair Poor

Teetotaller!

. 22.9 48.8 24.9 3.5

15.4 34.6 35.6 14.4

Recent 14.1 48.6 28.2 9.1

Occ/Iigbt 23.5 57.6 17.4 1.5

Drinken

Mod/heavy 18.7 55.6 22.7 3.0

blood pressure, peptic ulcer, and cardiovascular disease. It is well-recognized that heavy drinking ele- vates blood pressure,13 so that men with a diagnosis of high blood pressure may be advised to abstain. Long- term drinkers have an increased prevalence of all the indicators of IHD and are far more likely to be on regular treatment than any other group. Recent ex-drinkers appear to follow a similar pattern. In many ways the ex-drinkers in the present study had characteristics similar to moderate/heavy drinkers, in particular with regard to many of the known risk factors for ischaemic heart disease such as blood pressure and cigarette smoking. Though the majority of ex-drinkers claimed they had been occasional/light drinkers five years earlier, there is suggestive evidence that some may have been heavy drinkers who had gradually reduced their drinking, possibly due to ill health and/or medical advice. An investigation into the reasons why the non-drinkers did not drink revealed that 82% did not drink out of 'personal choice'. No further details were given and only 12% cited health reasons or medical advice. However, among regular drinkers (daily/weekend) who had given up since initial examination (Ql), about 30% gave health reasons compared to only 9% among occasional drinkers who had become non-drinkers. The Tecumseh study also found that 20-25% of ex-drinkers gave up because of health reasons.3

In general, teetotallers have a similar prevalence of disorders to occasional/light drinkers, including evidence of IHD on ECG and chest pain questionnaire.

However, they have a significantly higher rate of recall of doctor-diagnosis of IHD and a higher percentage of them are on regular treatment compared with drinkers. Overall, the teetotallers did not appear to be healthier than occasional/light drinkers.

Ex-drinkers in Other Studies

In the Tecumseh (Michigan, USA) Study, men aged 45-59 years were separated into teetotallers (never drank), ex-drinkers and drinkers.³ The ex-drinkers had about three times the incidence rate of myocardial infarction over 8-10 years as the teetotallers or drinkers, who both had the same lower rate. The ex-drinkers

did not have different levels of the standard risk factors. It was suggested that men with symptoms or diagnoses of heart disease are more likely to reduce or stop drinking.

In a study of 7705 Japanese men aged 45-65 years- living in Hawaii, both ex-drinkers and teetotallers had significantly higher rates of coronary heart disease than drinkers, after six years follow-up.^M This is a somewhat unusual community with a very low prevalence rate of coronary heart disease and a very high proportion of teetotaller (36%) and ex-drinkers (11%). The ex-drinkers had higher rates than the teetotallers, but the difference was not significant. Because of this lack of significant difference between these two groups, the authors consider that the increased incidence of disease in non-drinkers cannot be attributed to the presence of underlying disease in this group.

Studying a general population, Day found that lifelong teetotallers had the lowest mortality ratio, about half that of ex-drinkers.¹⁵ Heavy drinkers had a mortality ratio four times that of lifelong teetotallers.

While emphasizing that the increased mortality risk in non-drinkers is a consequence of the increased risk in ex-drinkers, she provides no detailed information on the health status of the ex-drinkers.

In the Chicago Western Electric Co. Study of men aged 40-55 years initially free of definite coronary heart disease, the 17-year mortality rates for all causes, cardiovascular disease and coronary heart disease were all highest in the ex-drinkers, followed closely by the heavy drinkers.16 In the heavy drinkers much of the excess cardiovascular disease and coronary heart disease mortality is explained by the established risk factors (blood pressure, serum total cholesterol, cigarette smoking); no data were presented to explain the excess mortality in ex-drinkers.

A Swedish study of middle-aged men provided material for a case-control analysis of 82 deaths compared with 115 living controls.2 The non-drinkers (current abstainers) had the highest observed:

expected mortality ratio, and current drinkers with no drink problems (on a nine-item questionnaire) had the lowest mortality ratio. It is suggested that there may be

(9)

increased ill-health amongst the non-drinkers but no data were presented in support of this suggestion.

A Japanese study of male physicians aged 25-70 years followed for 13 years, found that ex-drinkers had the highest overall mortality followed by heavy drinkers.17 Non-drinkers (little or no drink over 20 years) did not differ significantly in mortality rate from occasional or light drinkers. After 19 years follow-up,18 the situ- ation was unchanged and the excess mortality in ex-drinkers was found to be due mainly to cardiovascular disease and in particular to coronary heart disease other than acute myocardial infarction. It is suggested that ex-drinkers have given up drinking because of ill-health and the authors emphasize the need to separate ex-drinkers and lifelong non-drinkers in analysis.

Other British Studies

The results of The British Regional Heart Study (BRHS) are largely in agreement with data from a DHSS survey of 2000 men and women aged 18 upwards carried out by Wilson in England and Wales, despite differences in the methodology of classifying drinking behaviour.¹⁹ Wilson observed 4% of non- drinkers in men aged 45-54 years and 10% in men aged 55-64 years. In the BRHS, about 6% of men aged 40-59 years were non-drinkers at initial screening and 10% were non-drinkers five years later. The BRHS recorded 11 % of men as heavy drinkers compared with 4% in a roughly similar age-band (45-64 years) recorded by Wilson. Both studies showed similar social class differences in drinking patterns, with semi-skilled and unskilled manual workers having the highest percentage of non-drinkers and professional men the lowest. In a broader survey which included data from Scotland and Northern Ireland, Wilson observed little difference between average alcohol consumption between manual and non-manual workers in England and Wales, but higher levels in manual workers in Scotland.20 Overall, in men aged 20 years or more, he reported 6% non-drinkers in England and Wales and 7% in Scotland. In neither of these surveys is there any detailed discussion of non-drinkers.

The General Household Survey has examined the drinking patterns in men and women aged 18-65 years.21 The findings for men aged 45-64 are similar to the BRHS observations, with an increasing percentage of non-drinkers with increasing age. They also observed a higher percentage of non-drinkers and heavier drinkers in the manual workers. A recent DHSS study has examined drinking patterns in 3000 men and women aged 18-54 years living in private households in two different regions, with stratified

sampling from three types of neighbourhood and a high response rate (82%).²² Merseyside and Northern Regions were chosen to represent a 'high-risk' region while Trent and East Anglia represented a 'low-risk' region. 'Risk' was based on three indicators of alcohol abuse: death rates for cirrhosis of the liver, admission rates to mental illness hospitals and conviction rates for drunkenness other than motoring offences. Drinking categories were based on reported alcohol consumption in the week prior to interview, and only a 'tiny minority' (about 2%) of men in both regions were abstainers. All the analyses are for those who drink (light, medium, heavy) and there is no specific discussion of subjects who do not drink.

CONCLUSIONS

Our data support the hypothesis that non-drinkers include a considerable proportion of ex-drinkers who have a high prevalence of ill health. Some ex-drinkers may indeed have been advised to stop drinking because of their ill health. It seems likely that a considerable proportion of the excess mortality found in non-drinkers in many studies may be explained by the mortality experience of ex-drinkers. There are also indications that ex-drinkers as a group share many of the characteristics of moderate/heavy drinkers, and that a substantial proportion of ex-drinkers may derive from this group.

In some studies the lifelong teetotallers have a particularly low mortality rate but in most studies they have not been separated clearly from the ex-drinkers.

Except in the Japanese studies, lifelong teetotallers tend to be a relatively small group, and they do not provide a reliable baseline in mortality studies. It is suggested that occasional/light drinkers, ie those drinking 15 or fewer drinks per week, could provide an acceptable baseline group for comparative purposes in studies of the effects of alcohol intake. One thing is clear, non-drinkers are a heterogenous group usually containing a large proportion of ex-drinkers who have an increased morbidity and mortality from a wide range of disorders. They should not be used as a baseline for comparative purposes. The conclusions drawn from studies which have used them in this way should be regarded with caution.

ACKNOWLEDGEMENTS

The British Regional Heart Study is a British Heart Foundation Research Group. It also receives support from the MRC, DHSS, the Chest, Heart and Stroke Association and the Scottish Endowment Research Trust. The alcohol-related studies are supported by the Institute of Alcohol Studies.

(10)

REFERENCES

1 Pearl R. Alcohol and longevity. New York: Alfred A Knopf, 1926.

2 Petersen B, Trel) E , Kristenson H. Alcohol abstention ind prema- ture mortality in middle-aged men. Brit Med J 1982; 285:

1457-9.

3 US Dept Health Education and Welfare. Alcohol and the heart.

Second special report on alcohol and health US Govt Printing Office 1974; pp 68-72.

'Turner T B, Bennett V L, Hernandez H. The beneficial side of moderate alcohol use. The Johns Hopkins Medical Journal 1981; 148:53-63.

9 Marmot M G. Alcohol and coronary heart disease. ImJ Epidemiol 1984; 13: 160-7.

• Shaper A O, Pocock SJ, Walker M, Cohen N M, Wale CJ, Thomson A G. British Regional Heart Study: cardiovascular risk factors in middle-aged men in 24 towns. Brit MedJ 1981; 282:179-86.

7 Walker M, Shaper A G. Follow-up of subjects in prospective studies in general practice. J R Coll Gen Pract 1984; 34: 365-70.

I Office of Population Censuses and Surveys. Social Survey Division.

General Household Survey; 1978. London, HMSO; 1980.

•Shaper A G, Pocock S J, Ashby D, Walker M, Whitehead T P.

Biochemical and hatmatologictl response to alcohol intake.

Ann din Biochem 1985; 22: 50-61.

a Shaper A G, Cook D G, Walker M, Macfarlane P W. Prevalence of ischacmic heart disease in middle-aged British men. Br Heart J 1984; 51: 595-605.

II Shaper A G , Cook D G, Walker M, Macfarlane P W. Recall of diagnosis by m e n with jschitfmic heart disease. Br Heart J 1984; 51: 6 0 6 - 1 1 .

12 Cummins R O , Cook D G , H u m e R C, Shaper A G. Tranquillizer

use in middle-aged British men. / Roy CoU Gen Pract 1982; 32:

745-52.

° Klatsky A L, Friedman G D , Sigelaub A B , Gerard M J. Alcohol consumption and Wood pressure. N Engl J Med 1977; 290:

1194.

" Y a n o , R h o a d s G G , Kagan A . Coffee, alcohol and risk of coronary heart disease among Japanese men living in Hawaii. N Engl J of Med 1977; 297: 4 0 5 - 9 .

u D a y W L. Alcohol mortality: separating the drink from the drinkers.

Abstract of P h D thesis in 1978, University of Berkeley, California.

14 D y e r A R , Stamler J, Paul O , ef a/. Alcohol consumption, cardio- vascular risk factors and mortality: The Chicago experience.

Circulation 1981; 64 (Suppl III): pp 20-27.

17 Kcmo S, Ikeda M, Ogata M.etal. Relationship between alcohol and mortality among Japanese Physicians. Ins 1 Epidemiol 1983;

12:437-44.

"Kcmo S, Ikeda M, Tokadome S, a al Alcohol and mortality: A cohort study of male Japanese physicians. Ins J Epidemiol 1986; 15: 527-32.

" Wilson P. Drinking in England and Wales. OPCS Soda] Survey Division London: HMSO, 1980.

" Wilson P. Drinking Habits in the UK. Population Trends 1980; 22:

14-8.

21 Office of Population Censuses and Surveys. Social Survey Division.

General Household Survey; 1982. London, HMSO, 1984.

°Brceze E. Differences in Drinking Patterns between Selected Regions. London, HMSO, 1985.

{Revised version received October 1987)