lism offers a biologically plausible mechanism for the apparent protective association o f alcohol.679 M oderate alcohol intake raises HDL cholesterol,684,686,691,692 and this inverse association between HDL cholesterol and alcohol intake has a dose-response relation.684 A full review of this association has previously been published.679 Serum cholesterol does not show a consistent relationship to alcohol intake,684,692 but serum triglyceride is positively associated.679 The inverse association between alcohol and CAD does not differ by age or smoking status.686
5.5.2 Methods
In the interview with the case and control subjects, each was asked the approximate year that alcohol use commenced, the year ceased if this was the case, the number of days per w eek alcohol was taken, and the usual num ber o f portions per day when used. The self-reported number o f portions were the num ber o f standard glasses of beers, standard glasses o f wines and o f spirits. The standard glass of beer was de fined in the customary terms of those that drink beer, that is, as the equivalent to one
10 oz glass (or one "middie" of beer). From this information, the duration of alcohol intake for those using or having used alcohol, the number o f portions per week, the
percent ceasing alcohol use and when, and the percent of alcohol users was docu mented.
Similarly, in the dietary diary, the subjects were requested to document daily the equivalent to the number of standard glasses of alcohol used for each type of alcohol as described above. The mean number of portions of alcohol used by the groups and the number of individuals using alcohol at the time of the assessment was thus documented.
The estimated amount of alcohol from the questionnaire and the dietary diary, the intake for the two groups and the percent and number of former and current drinkers were compared.
5.5.3 Results
5.5.3.1 Group differences. The difference in the estimated number of portions of alcohol previously used by both the case and control groups was not significant (Table 5.10). In the case group, the estimate was slightly higher than in the controls, but because of the wide variance the difference was not significant (Table 5.10). The documented number of portions of alcohol in the dietary diary was slightly greater in the control group, but, again, the difference was not significant given the large stan dard deviation (Table 5.10). There was no difference in the percent of subjects in the two groups who previously used alcohol (Table 5.11). The length of time that indi viduals used alcohol in the two groups was also the same. The mean number of years of drinking for the control and case groups was 22.4+/-10.3 and 21.5+/-11.3 years respectively (95% confidence interval for the mean difference was -5.2 to 3.4, p=0.66).
The number of subjects not using alcohol within 24 hr of the platelet function tests was similar in both groups (35 in the case group and 31 in the control group). The amount of alcohol used by the remaining subjects was also the same (Table 5.10). However, significantly fewer case subjects admitted to the use of alcohol after the diagnosis of their CHD (Tables 5.11 and 5.12). Approximately 18% of subjects using alcohol in the case group stopped at the time of their diagnosis (Table 5.12). There was a borderline significant difference between the number previously using alcohol and the number at the time of the interview for the case group (Table 5.12). The number not using alcohol was also confirmed by the dietary diary (Table 5.13). In addition, the number of controls using alcohol at the time of the dietary diary was less than the number who stated that they used alcohol at the time of the interview (Table 5.12).
5.5.3.2 Correlation of weekly alcohol intake and other variables
(i) Coronary heart disease risk factors. The amount of alcohol used per week correlated positively with the triglyceride level in both the case and control group (Table 5.14). The total cholesterol level was significantly and positively asso ciated with the alcohol intake in the control group (Table 5.14). HDL cholesterol values did not correlate with this measure of alcohol intake. The alcohol intake
correlated with the level of cholesterol and LDL cholesterol in the control group (Table 5.14).
The control group also demonstrated a relationship between alcohol intake and previous cigarette intake (Table 5.14). However, there were very few current sm okers in both groups (n=23 past smokers in the control group, n=40 past smokers in the case group, n=4 currently smoking in the case group, and n=7 currently smoking in the control group). None of the other measurable risk factors, including the BP, BMI, and the fibrinogen level, was associated with alcohol intake.
(ii) Platelet function. No platelet function measurement correlated with alcohol intake in this study in the control group. There was an equivocol relationship with platelet measures in the case group. The estimated intake of alcohol negatively correlated with the MPV and positively with the PCR (Table 5.14). No other meas ure of platelet fuction was related.
(iii) Alcohol intake and other parameters. In the control group only, the amount of alcohol intake correlated positively with the MCV, the pack years of cigarettes and the blood urate level (Table 5.14).
(iv) Alcohol intake and angiographic CAD. The alcohol intake did not correlate with any angiographic measure of CAD in the case group.
5.5.4 Discussion
These observations demonstrate that the use of alcohol was frequent in both the case and control groups, but there was no major difference between the two groups. There was, however, an increased likelihood that the subjects with CHD stopped drinking alcohol after their diagnosis. Importantly, there was not a significant dif ference in the percent of subjects using alcohol at the time of the evaluation and when the blood was collected for the platelet function tests and measurement of risk factors for CHD.
The results from this small case-control study confirm that the amount of alcohol use can be estimated reasonably well by a single self-report of intake.693 In the control group, the mean weekly intake estimated by the dietary record was slightly but not significantly higher than estimated in the questionnaire, an observation also reported by ohers.693 However, the lower number of control group subjects using alcohol during the week of the dietary diary indicates that a single measure of alcohol may not reflect the pattern of drinking for a group. Alternatively, this reduction may arise from intentional under-reporting, although under-reporting would be expected to be less likely for moderate consumption.693 As in other aspects of the dietary diary and self-reported measures discussed above, an important bias that has not been con trolled for in this evaluation is response bias.471,677,678
The pattern of alcohol intake by the CHD case group in this study supports the concept that, at least in part, the increased CHD mortality differences related to alcohol intake are due to the downward shift of intake by individuals already with disease symptoms.689 As a cross-sectional study, the results cannot negate the ob-
servations from a large prospective study, that, even taking the downward shift into account, there remains an inverse association between alcohol intake and CHD.686
However, there are a number of other important observations arising from this latter study. Firstly, despite an increase in dietary risk factors for CHD in the groups with higher alcohol intake, the CHD mortality was less, implying that alcohol can protect against the adverse effects of other dietary risk factors. Those drinking more alcohol on the average also smoked more and had a more frequent diagnosis of hyperten sion, and despite both of these being very powerful risk factors for CHD, there was less cardiovascular events in the follow up period.686 If the observations are correct,
alcohol would have a very powerful protective effect against CHD. Secondly, the results confirm that there is no increased risk of CHD in abstainers, although the total mortality data were not reported.686 This observation does support the prospect
that movement of higher risk individuals into the group of abstainers from alcohol, but not the light to moderate drinking group, may have been a cause of the U-shaped curve for cardiovascular mortality in previous studies. 689 The exclusion of non
drinkers and men with disorders potentially related to CHD which may have lead to a prior reduction in alcohol intake, did not affect the relative risk associated with alcohol. Thirdly, it would appear that the more an individual drinks alcohol and the more often, then the lower the risk of CHD. Obviously, this probably would be offset by the total mortality increase well known with high levels of alcohol intake, although the data was not presented. From this study, and looking at alcohol intake alone, alcohol appears to impart a strong protective effect against cardiovascular disease.686
As in all such epidemiological studies, potential areas of bias exist. Firstly, the group consisted of upper socioeconomic and educated individuals, a population different to that in previous studies.689 This is pertinent since both the risk and in
cidence of CHD is greater in the lower socioeconomic groups,646 the prevalence of
risk factors are greater459,460,646 and the prevalence of increased alcohol intake is
greater.646 Secondly, the results appear to differ from most other prospective studies.
Almost all prospective studies in heavy drinkers have consistently shown an associa tion between increased mortality from CAD and increasing alcohol intake, in con trast to moderate alcohol intake.679 Again, there are potential confounding issues
leading to such spurious results,679 although the observations appear reproducible in
other studies.689 Thirdly, the groups with higher alcohol intake also used more aspir
in on a regular basis, a potentially powerful treatment bias considering acute coro nary events are included in the outcome analysis.100,103’105,694"698
Using simple linear correlation without adjustment for other risk factors, there was no association between HDL cholesterol values and prior total weekly alcohol intake in both study groups. However, with greater numbers there may be an association, as has been demonstrated in large studies. 693 There was an apparent association
between alcohol intake and triglyceride levels and measures related to LDL cholest erol. Others have demonstrated a decrease in LDL cholesterol with alcohol intake,699
while other studies have shown no relationship.680 It is clear that those who consume
more alcohol have a diet richer in fat.686 There is evidence that the case group in this
study changed both the fat content and alcohol content of their diet, which would have major confounding influence on any associations that may have been present pre-morbidly. This is also supported by the consistent relationship in the controls
between variables that are known to be influenced by alcohol intake. Those that used m ore alcohol in the control group had a higher MCV, sm oked m ore cigarettes and had a higher blood urate level. The increased use o f tobacco in those using m ore alcohol is well documented.686
A num ber o f causes o f this possible protective effect of alcohol have been postulat ed,679,699 in particular an elevation of the HDL cholesterol level.604 If increased HDL cholesterol does provide a mechanism by which alcohol use can protect against the developm ent of CHD, this would only explain about 50% o f the protective effect.699 It has been postulated that factors influencing haem ostasis are potential candidates for the added protective e ffe c t.164,699,700 P latelet aggregation in response to ADP appears to be inhibited by alcohol intake,700 and by the same level of alcohol use that protects against C H D .164 There w as no m easurable asso ciatio n betw een alcohol intake and platelet function measures, including measures o f platelet aggregation in the control group in this study. This may again be due to a type II statistical error, but even if so, the clinical im portance o f such an association w ould only be small. However, in the control group, increasing alcohol intake was associated with smaller platelet volum e and also less platelet aggregate form ation during blood collection. This equivocol association in the case group would in fact increase the possibility of a false negative result because o f alcohol use, that is a dilution bias. N evertheless, a difference in these parameters was docum ented and moreover, there was no m easur able difference between the alcohol intake of the two groups.
5.5.5 Summary
1. Any associations, or lack o f association, betw een alcohol in tak e and other param eters needs to be interpreted with caution given the m ajor confounding influence resulting from the change in diet, including alcohol use, in the case group since their diagnosis was first established.
2. The d iffe re n c e s in p la te le t fu n c tio n d em o n strated in this stu d y are unlikely to be biased by alcohol use resulting in increased platelet reactivity in the case group.
Table 5.1. Lipoprotein measures for the case group patients. Mean+/-SD Mean+/-SD Cholesterol(Pr) Cholesterol(Sr) Triglyceride(St) HDL(St) Apoprotein B 6.08+/- 1.41 5.64+/- 1.11 2.21+/- 0.23 0.86+ /- 0.21 0.95+/-0.17 Cholesterol(St) 5.18+/-0.74 Cholesterol (Sm) 5.46+/-0.90 Triglyceride(Sr) 2.24+/-0.17 HDL(Sr) 0.95+/-0.23 Apoprotein A l 1.21+/-0.28 Pr=previous (n=27),
St=reference laboratory (serum), Sr=service laboratory (plasma), Sm=service laboratory (serum), HDL=HDL cholesterol.
All measures in mmoles/1.
Table 5.2. Analysis of variance of the comparison between all the cholesterol levels.
Comparison mean d iffe r e n c e uncorrected d v a lu e Bonferroni p v a lu e A versus B 0.90 0.0006 p<0.01 ** A " C 0.44 0.0825 p>0.05 A " D 0.62 0.0154 p>0.05 B " C -0.46 0.0199 p>0.05 B " D -0.28 0.0153 p>0.05 C ” D 0.20 0.3568 p>0.05 A = previous cholesterol (Pr, n=27),
B = study serum cholesterol (St, from reference laboratory in serum), C = service laboratory plasma cholesterol (Sr),
Table 5.3. Correlations between lipoproteins and angiographically defined athero matous disease.
CAS mCAS nCAS dist 50/75% 25/50% 10/25% DOB (year) -0.27* -0.29* -0.20 -0.19 -0.31* -0.10 -0.12 NYHA FC 0.20 0.23 0.40# 0.13 -0.09 0.19 0.03 Choi (Pr) 0.46# 0.44# 0.48* 0.51# 0.35* 0.45# 0.23 Choi (St) 0.23* 0.27* 0.15 0.40* 0.28* 0.25* 0.29* Choi (Sr) 0.29* 0.33# 0.29* 0.08 0.36# 0.30* 0.33# Choi (Sm) 0.26* 0.28* 0.15 0.16 0.28* 0.27* 0.29* Trig (St) -0.17 -0.20 -0.18 -0.19 -0.26* 0.07 0.02 Trig (Sr) -0.19 -0.21 -0.19 -0.23 -0.15 0.01 -0.11 HDL (St) 0.01 0.03 -0.10 0.01 0.03 -0.03 -0.04 HDL (Sr) -0.05 -0.02 -0.16 0.08 0.09 -0.05 -0.01 Apo B 0.35# 0.35* 0.23 0.26* 0.30* 0.33* 0.36# Apo A1 0.09 0.12 0.02 0.01 0.13 0.09 0.08
*p<0.05, #p<0.01, All values are the Pearson correlation coefficient.
CSS=coronary stenosis score, mCSS=mean CSS, pCSS=proximal CSS, CAS=coro- nary atheromatous score, mCAS=mean CAS, pCAS=proximal CAS, Jenkins=Jenkins coronary score, and LVS=left ventricular score. Data are shown as mean +/- SD. See the Methods section for an explanation of the scores.
Chol=cholesterol, trig=triglyceride, HDL=HDL cholesterol, Apo B=apoprotein B, Apo Al=apoprotein A l, NYHA FC=New York Heart Association functional class for angina. See Table 5.1 for other abbreviations.
Table 5.4. Correlations between lipoproteins and angiographically defined measures of disease severity. CSS mCSS pCSS 100% 90/99% 75/90% Jenkins DOB (year) -0.25* -0.25* -0.22 0.01 -0.34# -0.17 0.28* NYHA FC 0.31* 0.34# 0.55** 0.41# 0.13 0.03 0.21 Choi (Pr) 0.28 0.27 0.45* 0.09 0.30 0.45# 0.38* Choi (St) 0.16 0.17 0.18 0.03 0.15 0.25* 0.21 Choi (Sr) 0.21 0.21 0.28* -0.01 0.16 0.30* 0.27* Choi (Sm) 0.10 0.10 0.07 -0.16 0.20 0.27* 0.22 Trig (St) -0.21 -0.23* -0.13 -0.19 -0.01 0.07 -0.24* Trig (Sr) -0.19 -0.21 -0.13 -0.20 -0.03 0.01 -0.19 HDL (St) 0.16 0.19 0.07 0.14 0.09 -0.03 0.13 HDL (Sr) 0.16 0.07 -0.12 -0.04 0.03 -0.05 0.07 Apo B 0.17 0.15 0.19 -0.06 0.17 0.33* 0.27* Apo A1 0.20 0.12 0.09 0.05 0.10 0.09 0.21 *p<0.05, #p<0.01, **p<0.001
Table 5.5. Correlations between lipoproteins, functional class for angina, age and number of infarcts and normal segments, LV function and collateral supply.
Norm LVS CS Coll DOB (years) 0.23* -0.11 -0.10 0.08 NYHAFC -0.15 0.05 0.39# 0.33# No AMI -0.18 0.42# -0.03 0.13 Choi (Pr) -0.46* 0.09 0.36* 0.13 Choi (St) -0.33# -0.05 0.09 0.05 Choi (Sr) -0.38# -0.03 0.11 0.08 Choi (Sm) -0.30* 0.04 0.03 -0.09 Trig (St) 0.25* 0.01 0.05 0.02 Trig (Sr) 0.29* 0.11 -0.08 -0.06 HDL (St) -0.09 0.10 0.02 0.04 HDL (Sr) -0.01 0.10 -0.12 -0.11 Apo B 0.39# -0.06 0.07 0.02 Apo Al 0.24 0.05 0.02 0.06 *p<0.05, #p<0.01, **p<0.001
See Tables 5.1 and 5.3 for abbreviations.
Table 5.6. Multiple regression entering apoprotein A l, apoprotein B, the study cho lesterol level, triglyceride level, HDL cholesterol level and age as independent varia bles and the coronary artery scores as the dependent variables.
Dependent Independent
Variable Variable B SEB Beta Rsq Si?
NORMSEG Apo B -0.0854 0.0260 -0.4469 0.1547 0.0022 CAS Apo B 0.2220 0.0940 0.3497 0.1224 0.0231 mCAS Apo B 0.1808 0.0756 0.3535 0.1250 0.0216 mCSS DOB -0.9300 0.4535 -0.3083 0.0951 0.0469 JENKINS DOB -0.4264 0.2070 -0.3097 0.0959 0.0459 LVS DOB -0.1461 0.0575 -0.3726 0.1388 0.0151
B=regression coefficient, Rsq=coefficient of determination.
SE B = stan d ard erro r of B, D O B =year of birth . See T ables 5.1 and 5.3 for other abbreviations.
Table 5.7. Main Dietary Constituents Measured in Portions. CONTROLS CASES
Mean SD Mean SD Diff 95% Cl p
DAIRY PRODUCTS Milk - full 2.9/0.5 1.7/2.5 1.1 -2.4 to 0.01 0.07 - low 2.4/3.9 3.5/5.1 1.1 -0.9 to 3.2 0.28 Addedmilk-high 8.2/12.7 4.0/7.5 4.3 -8.6 to 0.1 0.05* Addedmilk-low 4.1/9.1 7.3/11.7 3.2 -1.6 to 7.9 0.19 Addedsugar 8.3/12.9 9.3/12.7 1.0 -4.6 to 6.7 0.71 Cream 0.3/0.6 0.2/0.6 0.1 -0.4 to 0.2 0.53 Yoghurt -full 0.4/1.3 0.4/1.1 0.0 -0.6 to 0.5 0.92 -low 0.4/1.2 0.2/0.9 0.1 -0.6 to 0.3 0.53 Cheese -full 3.1/4.0 2.0/3.4 1.0 -2.6 to 0.6 0.20 -low 0.3/1.1 0.4/1.1 0.01 -0.4 to 0.6 0.76 Icecream -full 1.1/1.4 1.0/1.9 0.01 -0.8 to 0.7 0.81 Butter 3.9/8.7 2.5/6.8 1.4 -4.8 to 2.0 0.41 Margarine 20.4/15.2 19.7/11.6 0.6 -6.4 to 5.2 0.83 Eggs - total 1.2/1.7 1.0/1.8 0.2 -1.0 to 0.6 0.63 MEAT Total Meat 11.3/5.5 9.4/4.1 1.9 -4.0 to 0.2 0.07 Fried Meat 1.7/2.7 1.6/1.8 0.1 -1.1 to 0.9 0.86 Beef 5.1/3.5 4.0/2.7 1.1 -0.2 to 2.5 0.10 Chicken 2.4/2.4 2.1/1.9 0.2 -1.2 to 0.7 0.65 FISH Total Fish 3.0/1.6 2 1/2.6 0.3 -1.5 to 0.9 0.60 GBC Fish 1.4/1.6 1.9/2.2 0.5 -0.4 to 1.4 0.26 Shell Fish 0.4/0.9 0.3/0.7 0.01 -0.4 to 0.3 0.61
G B C = g rilled + b ak ed + cassero led , h ig h = fu ll cream , low = reduced fat
Table 5.8. Main Dietary Constituents.
CONTROLS CASES Mean SD