Critique 198: A study of alcohol and cardiovascular disease based on “Big Data,” linked electronic medical records from the population — 24 April 2017
Bell S, Daskalopoulou M, Rapsomaniki E, George J, Britton A, Bobak M, Casas JP, Dale CE, Denaxas S, Shah AD, Hemingway H.
Association between clinically recorded alcohol consumption and initial presentation of 12 cardiovascular diseases: population based cohort study using linked health records. BMJ 2017;356:j909. http://dx.doi.org/10.1136/bmj.j909
Objectives To investigate the association between alcohol consumption and cardiovascular disease at higher resolution by examining the initial lifetime presentation of 12 cardiac, cerebrovascular, abdominal, or peripheral vascular diseases among five categories of consumption.
Design Population based cohort study of linked electronic health records covering primary care, hospital admissions, and mortality in 1997-2010 (median follow-up six years).
Setting CALIBER (ClinicAl research using LInked Bespoke studies and Electronic health Records).
Participants 1 937 360 adults (51% women), aged ≥30 who were free from cardiovascular disease at baseline.
Main Outcome Measures 12 common symptomatic manifestations of cardiovascular disease, including chronic stable angina, unstable angina, acute myocardial infarction, unheralded coronary heart disease death, heart failure, sudden coronary death/cardiac arrest, transient ischaemic attack, ischaemic stroke, intracerebral and subarachnoid haemorrhage, peripheral arterial disease, and abdominal aortic aneurysm.
Results 114 859 individuals received an incident cardiovascular diagnosis during follow-up. Non-drinking was associated with an increased risk of unstable angina (hazard ratio 1.33, 95% confidence interval 1.21 to 1.45), myocardial infarction (1.32, 1.24 to1.41), unheralded coronary death (1.56, 1.38 to 1.76), heart failure (1.24, 1.11 to 1.38), ischaemic stroke (1.12, 1.01 to 1.24), peripheral arterial disease (1.22, 1.13 to 1.32), and abdominal aortic aneurysm (1.32, 1.17 to 1.49) compared with moderate drinking (consumption within contemporaneous UK weekly/daily guidelines of 21/3 and 14/2 units for men and women, respectively). Heavy drinking (exceeding guidelines) conferred an increased risk of presenting with unheralded coronary death (1.21, 1.08 to 1.35), heart failure (1.22, 1.08 to 1.37), cardiac arrest (1.50, 1.26 to 1.77), transient ischaemic attack (1.11, 1.02 to 1.37), ischaemic stroke (1.33, 1.09 to 1.63), intracerebral haemorrhage (1.37, 1.16 to 1.62), and peripheral arterial disease (1.35; 1.23 to 1.48), but a lower risk of myocardial infarction (0.88, 0.79 to 1.00) or stable angina (0.93, 0.86 to 1.00).
Conclusions Heterogeneous associations exist between level of alcohol consumption and the initial presentation of cardiovascular diseases. This has implications for counselling patients, public health communication, and clinical research, suggesting a more nuanced approach to the role of alcohol in prevention of cardiovascular disease is necessary.
Essentially all observational cohort epidemiologic studies have shown a “J-shaped” curve for the association of moderate alcohol consumption with coronary heart disease and myocardial infarction, in that non-drinkers and heavy drinkers have higher risk than moderate drinkers. Numerous mechanisms of both alcohol in all alcoholic beverages and polyphenols in wine, including effects on lipids, coagulation, fibrinolysis, inflammation, etc., have been shown to help explain the epidemiologic results. The present study presents a new approach for evaluating this relation by using clinical medical records for a very large number of patients (“big data”) to judge the relation between alcohol consumption and a variety of cardiovascular disease (CVD) outcomes.
Forum members considered this to be an innovative and well-done study using a very large dataset (almost 2 million subjects; about 115,000 events) based on electronic medical records from the population in the United Kingdom. The authors used a number of measures in attempting to be more precise in their estimates of alcohol intake and in the definition of outcomes. Still, in most cases, the exposures and outcomes came solely from the medical diagnoses given by individual physicians, and such diagnoses are known to often be inaccurate. For example, the most common cause of death in a community (usually CVD) is often coded as the cause of death when detailed evaluation fails to support such a diagnosis. Similarly, for individual subjects, coding of alcohol use or misuse in the medical record surely varies according to the physician making the diagnosis.
The authors did a good job of attempting to take these problems into consideration, and deal with them in the Discussion. However, such problems would be expected to decrease the precision of exposure and of outcomes in any study that is based only on medical codes available in public health records.
The authors conclude that there is a higher risk among abstainers (versus moderate drinkers) for many CVD outcomes; non-drinkers had significantly higher risk for unstable angina, myocardial infarction (MI), unheralded coronary death, heart failure, ischemic stroke, peripheral arterial disease, and abdominal aortic aneurysm. For most of these outcomes, heavy drinkers tended to have an increased risk (a J-shaped curve), but the risk for myocardial infarction and for stable angina remained decreased even for the heavy drinking category. For other manifestations of CVD, especially those related to cerebral artery disease (other than ischemic stroke), there was less of a clear pattern relating alcohol to disease.
Specific comments by Forum members: Forum member Ellison had a number of favorable comments about this paper: “I think that among the strengths of this paper was an attempt to adjust for ‘socioeconomic deprivation,’ which is a strong predictor of adverse health outcomes from alcohol consumption, as described recently by Towers et al (a review of this paper by the Forum is available as Critique 195 under Recent Reviews at www.alcoholresearchforum.org). It is reassuring that the authors tested not only for each separate manifestation of CVD but for the effects of alcohol on the risk of ‘aggregated coronary heart disease,’ as this makes the overall implications of their findings more relevant.
“In some of their analyses, they adjusted for HDL-cholesterol, which is a key mechanism for the effect of alcohol on CVD; this may have led to over-adjustment in some of their analyses. Among key weaknesses of the paper were that there were no data on the pattern of alcohol (most previous studies have shown large differences between regular moderate drinkers and binge drinkers) or the type of beverage (wine, especially with meals, often shows greater protective and fewer adverse effects than other beverages). Overall, the results of this paper strongly support a J-shaped (or inverse) relation of alcohol with most manifestations of CVD. Both the main outcomes and most sensitivity analyses showed a similar pattern. In their Discussion, the authors present an excellent overview of the strengths and weaknesses of their approach.”
Ellison added: “A priori, ‘stable angina’ and ‘TIA’ would be expected to be the softest end-points, and this is supported by the data in this study. Further, ‘cardiac arrest’ is a non-specific outcome and is often coded for many other diseases, even for deaths from cancer, as it, obviously, occurs as the final event. Again, the data from this study show weaker relations with alcohol for this diagnosis than for MI and other specific causes of death. For the ‘hard’ outcomes such as MI, cardiac death, abdominal aneurysm, and even unstable angina, the data support a protective effect of alcohol against these outcomes. Cerebral artery diseases (hemorrhagic stroke, intracranial hemorrhage, etc.) show less of an effect than outcomes related to coronary artery disease, as shown in many previous studies. Effects on peripheral arterial disease and aortic aneurysm were similar to those for coronary disease. Protection against CVD was greater among smokers than non-smokers and among the obese versus the non-obese.”
Forum member Van Velden wrote: “I find this paper very interesting; it underlines what we already know, but also stresses that alcohol consumption should not be seen in isolation from other lifestyle factors such as responsible exercise habits, not smoking, and weight management through healthy eating habits. Health is a summary of several variables working in concert with each other, also interacting with the genetic profile of the individual. Not everybody will react the same to a diet intervention; alcohol in moderation is merely a diet component, and not a drug.”
Reviewer Estruch stated: “I also agree with other Forum members that it is an impressive study on the association between alcohol consumption and incidence of several types of cardiovascular disease; it re-confirms once more the validity of the J-shape curve observed for the relationship between alcohol consumption and incidence of cardiovascular events. However, three criticisms should be considered:
- Alcohol consumption was recorded only at baseline and it would be better to investigate the shape of the dose-response association using continuous measures of alcohol consumption.
- The effects of alcohol consumption may differ according to the dietary pattern followed and the proportion of foods and nutrients consumed. It would have been preferable if the authors were able to include these variables in the multivariate analysis as confounding factors.
- Physical activity is another key determinant of health status and it should be included in the analysis.
Despite these problems, I consider that it is an excellent work and it will be a landmark study when analyzing alcohol consumption and incidence of cardiovascular disease.”
Forum member Finkel stated: “This analysis reconfirmed much of what we have learned during the last 20 or 30 tears. I hope this is part of a wedge that will expand understanding by elucidating some of the intricacies of what are multifaceted diseases or, in some cases, separate diseases. How many times have we, in reviewing a paper, wished for more detail about this or that: beverage choice, pattern of drinking, criteria for quantity of drinking, whether drinking occurred while eating, how diagnoses were established, what was included under the diagnosis being studied, and so on? And have some of us wondered out loud whether the confusion about the disease being related positively or negatively or both (!) to alcohol isn’t really a set of diseases. [Breast cancer is the prime example of this.] So, this large and well-done study does take us a step further into finer-sectioning of the relationship between alcohol and cardiovascular disease.”
Use of “Big Data” for research: Reviewer de Gaetano wrote: “The most original and important contribution of this paper is to have used data from general practice. Obviously, case-control or prospective observational studies and especially randomized trials offer a more accurate setting for studies on humans than general practice does. They try to reproduce in some way the strictly controlled conditions obtained in animal experiments. However, any kind of experimental design in epidemiological research introduces and generates more or less important artificial situations that may be quite far from everyday life. Indeed we do not live in a cage and do not drink the same water nor eat the same laboratory chow. An important question that is not frequently asked in clinical research is the transferability of the results obtained in controlled conditions to the general practice and real life. Now, in this paper, data derive directly from the setting to which they are addressed.”
De Gaetano continued: “This is not simply data, but big data! And this is a second important merit of this paper. Big data is not only a large amount of data on the specific object of that particular study, but also a large, possibly integrated amount of information contributing to a final result adjusted for many confounding or contributing variables. In these real conditions alcohol in moderation is clearly confirmed to be a lifestyle habit protecting against CVD. This is the important, though confirmatory, message of this innovative paper. All the rest is literature.”
The advantages and disadvantages of using mega-data sets of electronic medical records for epidemiologic research are well discussed in an editorial by Mukamal and Lazo that accompanies this paper. These authors point out that studies using this new approach make it possible to evaluate extremely large numbers of subjects. However, because the sources of the data are from medical diagnoses recorded by a large number of clinicians, it is not possible to judge the accuracy of diagnostic data as well as in well-done cohort studies. For example, for many outcomes, studies frequently show poor correlation between the cause of death on the death certificate and that determined by autopsy; the correlation was higher for cancer deaths but less than 50% for cardiac disease and less than 10% for pneumonia in a study by Mieno et al. Lund et al found a good correlation between physician-recorded cause of death and the initial diagnosis obtained from a national surveillance system for cancer, while Smith Sehdev et al found poor correlation for most outcomes even when the physician’s cause of death was recorded just prior to autopsy.
Further, it is more difficult to control for systematic error in studies based on clinical data, and validation of exposure data is, to a large extent, not possible. However, as Mukamal and Lazo state, “Bell and colleagues examined a diverse, if quizzical, set of endpoints that would be difficult to study with precision in smaller cohorts. They largely reproduce findings from previous large cohort studies, such as from the Kaiser-Permanente group (Klatsky et al, 1974, 1986, 2001, 2002, 2005), but add risk estimates for subtypes of myocardial infarction, stable and unstable angina, and abdominal aortic aneurysm, among others. Clearly, this new approach is particularly attractive for rarer health outcomes.”
Reviewer Skovenborg stated: “As a family physician, I am very pleased to see data from general practitioners used as ‘big data’ to refine the epidemiological investigation. Many results were as expected and the weakness of the study is (as always) the alcohol exposure data and the real risk of information bias – though not worse here in this study than what we are used to see elsewhere.” Forum member Mattivi noted: “This looks to me as a really key paper, for the impressive size of the information collected, and the appropriate use of many important and useful tools for the accurate study of complex health conditions and their several interconnected components.”
Forum member Thelle wrote: “Bell et al have done a very good job in my opinion. Their study is impressive, first and foremost because the magnitude and the design enables a valid ascertainment of the end-points, with probably less misclassification than in most other cohort studies. The alcohol exposure variable is well described and likely to represent a valid assessment. The statistical analysis is up to par, and probably more than that. The paper is as close as we can get to a public health situation and represents a valid picture of what is going on in the UK.
“The authors underline that they haven’t assessed transition between consumption categories, and can therefore not encourage people to take up drinking alcohol as a health measure. I would support their attitude on that and also claim that most of us do not drink alcoholic beverages for preventive reasons. Their very last paragraph stating that ‘finding that moderate drinking is not universally associated with a lower risk …, etc’ is more difficult to accept as their own results show that moderate drinkers are doing better than any other for the cardiac disease categories. But again, a very (or very, very, very as present POTUS would have said) important study.”
Reviewer McEvoy noted: “This is a very important study demonstrating the use of a very large clinical database to address public health issues. That the findings confirm cardioprotective effects of moderate alcohol consumption despite the limitations in exposure and outcomes as described in the paper, and by forum members, reflects the robustness of that association. This is an impressive study that is likely to have great impact.”
Forum members tended to agree with the conclusions of Mukamal and Lazo: “The new study does not offer a materially new view of the associations between alcohol consumed within recommended limits and risk of cardiovascular disease. They report lower rates of essentially every meaningful cardiovascular outcome except hemorrhagic stroke among moderate drinkers than among abstainers. Four decades of epidemiological studies have largely found the same. This work, however, sets the stage for ever larger and more sophisticated studies that will attempt to harness the flood of big data into a stream of useful, reliable, and unbiased findings that can inform public health, clinical care, and the direction of future research.”
Consistency of clinical data (as used in the present paper) with previous epidemiologic and experimental data: Reviewer Stockley provided an overview of how the results of the present paper relate to those from observational epidemiologic studies and experimental studies. “The results from this analysis are generally consistent with those of other studies attempting to delineate the beneficial versus detrimental effects of different levels of alcohol consumption on the various cardiovascular diseases. For example, in addition to reducing the risk of atherosclerosis-related cardiovascular events, such as coronary heart disease and myocardial infarction, studies have also considered the relationship between alcohol and the specific cardiovascular diseases of atrial fibrillation, heart failure, hypertension and peripheral vascular disease as well as to non-coronary CVD diseases such as stroke. Light-to-moderate drinking was observed as generally beneficial in minimising the risk of these cardiovascular events, even after accounting for incident MI, except for atrial fibrillation and hypertension (Elkind et al; Samokhvalov et al).
“Heart failure, for example, is a complex syndrome with multiple causes including coronary heart disease, endomyocardial fibrosis, hypertension, myocardial infarction, obesity and type 2 diabetes (Lloyd-Jones; Wilhelmsen et al). Accordingly, the relationship between alcohol and the risk of heart failure appears to be a summation of the relationships between alcohol and the risk of the contributing cardiovascular events (Klatsky et al, 2005; Padilla et al). From the US Physician’s Health Study I (1982 to 2008), moderate alcohol consumption independently and jointly with four other healthy lifestyle habits, reduced the lifetime risk of heart failure (Djoussé et al). The relationship between atrial fibrillation, a common chronic cardiac arrhythmia, and alcohol appears to be a casual, linear and dose-response relationship, which may reflect alcohol-induced electrophysiological changes in atrial cells (Habuchi et al; Steinbigler et al; Elkind et al; Marcus et al; Samokhvalov et al). An excellent review of the alcohol-atrial fibrillation relation was published by Voskoboinik et al, which concluded that even moderate drinking may increase the risk of atrial fibrillation. Atrial fibrillation is closely associated with both heart failure and hypertension.
“In contrast to moderate alcohol consumption, heavy alcohol consumption is an established risk factor for hypertension and the development and progression of atherosclerosis, and is associated with lower fibrinolytic activity, pro-coagulation events, and higher blood viscosity (Toth et al). Alcohol-induced oxidative stress (production of reaction oxygen species), accumulation of fatty acid ethyl esters, modification of lipoproteins, and increased expression of pro-inflammatory cytokines and vascular cellular adhesion molecules, all contribute to and promote the formation of atherosclerotic plaques (Hannuksela et al). In addition, the alcohol-induced increase in blood pressure may counteract direct athero-protective mechanisms. Recently, in vitro models of atherosclerosis have shown that the primary breakdown product of alcohol, acetaldehyde, may also affect pro-inflammatory cytokines and vascular cellular adhesion molecules (Redmond et al).
“Concerning all-cause mortality, the 1996 Australian meta-analysis by Holman et al of 16 cohort studies undertaken between 1980 and 1993 was the first to suggest that the J-shaped relationship between alcohol and CVD could be extended to total or all-cause mortality. A 16% and 12% reduction in risk of death from all-causes was calculated for men and women, respectively, at 10-19 and 1-9 g alcohol/day. Ten years later, a larger meta-analysis by Di Castelnuovo et al of 34 studies undertaken between 1980 and 2005 corroborated this initial observation. It suggested that there was now a 17 and 18% reduced risk for men and women, respectively, which could be extended out to approximately 40 g/day for men and 20 g/day for women. Overall, the results of the present study are quite consistent with this previous research.”
Implications of this study for setting public policy: The authors use the standard, politically acceptable statements about not advising people to drink for their health because you can protect better against CVD by increasing exercise or stopping smoking; however, they do not comment on how extremely difficult it is for physicians to actually get people to begin to exercise or to stop smoking. The authors of this paper state that the latter changes in behavior “do not incur increased risks of alcohol related harm . . .” without stating that these adverse outcomes occur essentially only with immoderate, rather than moderate, drinking. Their comments suggest further that they believe that patients who decide to drink moderately after discussing the topic with their physician are at increased risk of becoming alcohol abusers, but we are unaware of scientific data supporting this supposition.
Forum member Keil also had some questions about the importance of this study: “I agree with the editorial by Mukamal and Lazo, which said that ‘The new study does not offer a materially new view of the associations between alcohol consumed within recommended limits and risk of cardiovascular disease. They report lower rates of essentially every meaningful cardiovascular outcome except hemorrhagic stroke among moderate drinkers than among abstainers. Four (!) decades of epidemiological studies have largely found the same . . . ‘
“With this big data set I have, however, my doubts concerning the quality of the data on alcohol consumption assessed by physicians or medical personnel in a clinical setting. Doing a study on alcohol in a region where participants respond to the question ‘Do you drink alcohol?’ with “NO” but later on in the questionnaire answer “YES” when asked ‘Do you drink beer?’ seems more reasonable. What do I want to say? In a field study (‘free living population’) in Bavaria there seems to be no taboo to admit to drinking alcohol, while in a medical setting patients may be intimidated to admit to alcohol consumption. The same is true also for occupational settings.”
Keil continued: “My resumee: The many new studies on alcohol and health do not contribute much to our knowledge, but some studies like the one by Bell suggest that the authors misinterpret their own results when proposing implications of their study (because of moral or ideological reasons?). I think it was Lewis Kuller who coined the phrase ‘circular reasoning’ — always the same again and again.”
Forum member Finkel added: “I want to add that maybe scientific papers should stick to science, and leave the editorializing for editorials. Setting public policy is a highly specialized field, and involves scientific data, cultural and lifestyle factors of the population for which it is intended, and many other factors. When scientists stray outside their field they chance embarrassing themselves and irritating their readers. Our Forum should be in the business of editorializing, at least about the subject and execution of the papers we review, but not necessarily the implications for public policy.”
References from Forum Critique
Di Castelnuovo A, Costanzo S, Bagnardi V, Donati MB, Iacoviello L, de Gaetano G. Alcohol dosing and total mortality in men and women: an updated meta-analysis of 34 prospective studies. Arch Intern Med 2006;166:2437-2445.
Djoussé L, Driver JA, Gaziano JM. Relation between modifiable lifestyle factors and lifetime risk of heart failure. JAMA 2009;302:394-400.
Elkind MS, Sciacca R, Boden-Albala B, Rundek T, Paik MC, Sacco RL. Moderate alcohol consumption reduces risk of ischemic stroke: the Northern Manhattan Study. Stroke 2006;37:13-19.
Habuchi Y, Furukawa T, Tanaka H, Lu LL, Morikawa J, Yoshimura M. Ethanol inhibition of Ca2+ and Na+ currents in the guinea-pig heart. Eur J Pharmacol 1995;292:143-149.
Hannuksela ML, Liisanantti MK, Savolainen MJ. Effect of alcohol on lipids and lipoproteins in relation to atherosclerosis. Crit Rev Clin Lab Sci 2002;39:225-283.
Holman CDJ, English DR, Milne E, Winter MG. Meta-analysis of alcohol and all-cause mortality: a validation of NHMRC recommendations. Med J Aust 1996;164:141-145.
Klatsky AL, Friedman GD, Siegelaub AB. Alcohol consumption before myocardial infarction. Results from the Kaiser-Permanente epidemiologic study of myocardial infarction. Ann Intern Med 1974;356:294-301. doi:10.7326/0003-4819-81-3-294
Klatsky AL, Armstrong MA, Friedman GD. Relations of alcoholic beverage use to subsequent coronary artery disease hospitalization. Am J Cardiol 1986;356:710-714. doi:10.1016/0002-9149(86)90342-5.
Klatsky AL, Armstrong MA, Friedman GD, Sidney S. Alcohol drinking and risk of hospitalization for ischemic stroke. Am J Cardiol 2001;356:703-706. doi:10.1016/S0002-9149(01)01824-0.
Klatsky AL, Armstrong MA, Friedman GD, Sidney S. Alcohol drinking and risk of hemorrhagic stroke. Neuroepidemiology 2002;356:115-122. doi:10.1159/000054808.
Klatsky AL, Chartier D, Udaltsova N, Gronningen S, Brar S, Friedman GD, Lundstrom RJ. Alcohol drinking and risk of hospitalization for heart failure with and without associated coronary artery disease. Am J Cardiol 2005;96:346-351.
Lloyd-Jones DM. The risk of congestive heart failure: sobering lessons from the Framingham Heart Study. Curr Cardiol Rep 2001;3:184-190.
Lund JL, Harlan LC, Yabroff KR, Warren JL. Should Cause of Death From the Death Certificate Be Used to Examine Cancer-Specific Survival? A Study of Patients With Distant Stage Disease. Cancer Invest 2010;28:758–764. doi: 10.3109/07357901003630959
Marcus GM, Smith LM, Whiteman D, Tseng ZH, Badhwar N, Lee BK, Lee RJ, Scheinman MM, Olgin JE. Alcohol intake is significantly associated with atrial flutter in patients under 60 years of age and a shorter right atrial effective refractory period. Pacing Clin Electrophysiol 2008;31:266-272.
Mieno MN, Tanaka N, Arai T, et al. Accuracy of Death Certificates and Assessment of Factors for Misclassification of Underlying Cause of Death. J Epidemiol 2016;26:191–198. doi: 10.2188/jea.JE20150010.
Mukamal K, Lazo M. Alcohol and cardiovascular disease. Big data puts the link between moderate drinking and lower risk under the microscope. BMJ 2017;356:j1340 doi: 10.1136/bmj.j1340.
Padilla H, Michael Gaziano J, Djoussé L. Alcohol consumption and risk of heart failure: a meta-analysis. Phys Sportsmed 2010;38:84-89.
Redmond EM, Morrow D, Kundimi S, Miller-Graziano CL, Cullen JP. Acetaldehyde stimulates monocyte adhesion in a P-selectin- and TNFalpha-dependent manner. Atherosclerosis 2009;204:372–380.
Samokhvalov AV, Irving HM, Rehm J. Alcohol consumption as a risk factor for atrial fibrillation: a systematic review and meta-analysis. Eur J Cardiovasc Prev Rehabil 2010;17:706-712.
Smith Sehdev AE, Hutchins GM. Problems With Proper Completion and Accuracy of the Cause-of-Death Statement. Arch Intern Med 2001;161:277-284. doi:10.1001/archinte.161.2.277
Steinbigler P, Haberl R, König B, Steinbeck G. P-wave signal averaging identifies patients prone to alcohol-induced paroxysmal atrial fibrillation. Am J Cardiol 2003;91:491-494.
Toth A, Sandor B, Papp J, Rabai M, Botor D, Horvath Z, Kenyeres P, Juricskay I, Toth K, Czopf L. Moderate red wine consumption improves hemorheological parameters in healthy volunteers. Clin Hemorheol Microcirc 2014;56:13-23.
Towers A, Philipp M, Dulin P, Allen J. The “Health Benefits” of Moderate Drinking in Older Adults may be Better Explained by Socioeconomic Status. Pre-publication: J Gerontol B Psychol Sci Soc Sci 2016. doi:10.1093/geronb/gbw152
Voskoboinik A, Prabhu S. Ling L-H, Kalman J-M, Kistler PM. Alcohol and Atrial Fibrillation. A Sobering Review. J Am Coll Cardiol 2016;68:2567-2576. http://dx.doi.org/10.1016/j.jacc.2016.08.074.
Wilhelmsen L, Rosengren A, Eriksson H, Lappas G. Heart failure in the general population of men–morbidity, risk factors and prognosis. J Intern Med 2001;249:253-261.
Essentially all observational cohort epidemiologic studies have shown a “J-shaped” curve for the association of moderate alcohol consumption with coronary heart disease and myocardial infarction, in that non-drinkers and heavy drinkers have higher risk than moderate drinkers. The present study presents a new approach for evaluating this relation by using electronic medical records (“big data”) from a very large number of patients in the UK to judge the relation between alcohol consumption and a variety of cardiovascular disease (CVD) outcomes. The study evaluated clinical data recorded by physicians from almost 2 million subjects, with about 115,000 CVD events recorded in public records. The exposures and outcomes came solely from the medical diagnoses given by individual physicians, and such diagnoses are known to often be inaccurate. For individual subjects, the exposure was based on codes for alcohol use or misuse in their medical records, and the cause of death was based on that recorded by the physician (which is not always consistent with diagnoses from autopsies or medical record reviews). Such problems would be expected to decrease the precision of the exposure and of outcome diagnoses.
Forum reviewers of this paper considered that the authors made a valiant effort in attempting to take into account the limiting factors of using big data for epidemiologic research. It is recognized that studies such as these cannot evaluate the pattern of drinking (regular versus binge) and the investigators were unable to evaluate different effects according to the type of beverage consumed (beer, wine, or spirits). It would be expected that both of these factors may have affected their results.
Nevertheless, their conclusions strongly support the extensive observational data from cohort studies over many decades that have shown a higher risk among abstainers (versus moderate drinkers) for many CVD outcomes. In the present study, non-drinkers had significantly higher risk for unstable angina, myocardial infarction (MI), unheralded coronary death, heart failure, ischemic stroke, peripheral arterial disease, and abdominal aortic aneurysm. For most of these outcomes, heavy drinkers tended to have an increased risk (a J-shaped curve), but the risk for myocardial infarction and for stable angina remained decreased even for the heavy drinking category. For other manifestations of CVD, especially those related to cerebral artery disease (other than ischemic stroke), there was less of a clear pattern relating alcohol to disease. For their aggregate, overall outcome based on all manifestations of CVD, their analyses add further support for a J-shaped curve between alcohol intake and CVD.
* * * * * * * *
Comments on this critique of the International Scientific Forum on Alcohol Research were provided by the following members:
Elizabeth Barrett-Connor, MD, Distinguished Professor, Division of Epidemiology, Department of Family Medicine and Public Health and Department of Medicine, University of California, San Diego, La Jolla, CA USA
Giovanni de Gaetano, MD, PhD, Department of Epidemiology and Prevention, IRCCS Istituto Neurologico Mediterraneo NEUROMED, Pozzilli, Italy
R. Curtis Ellison, MD, Professor of Medicine & Public Health, Boston University School of Medicine, Boston, MA, USA
Ramon Estruch, MD, PhD, Hospital Clinic, IDIBAPS, Associate Professor of Medicine, University of Barcelona, Spain
Harvey Finkel, MD, Hematology/Oncology, Boston University Medical Center, Boston, MA, USA
Ulrich Keil, MD, PhD, Professor Emeritus, Institute of Epidemiology & Social Medicine, University of Muenster, Germany
Arthur Klatsky, MD, Dept. of Cardiology, Kaiser Permanente Medical Center, Oakland, CA, USA
Fulvio Mattivi, MSc, Head of the Department of Food Quality and Nutrition, Research and Innovation Centre, Fondazione Edmund Mach, in San Michele all’Adige, Italy
Linda McEvoy, PhD, Department of Radiology, University of California at San Diego (UCSD), La Jolla, CA, USA
Professor Andrzej Pajak, Epidemiology and Population Studies, Jagiellonian University Medical College, Kraków, Poland
Erik Skovenborg, MD, specialized in family medicine, member of the Scandinavian Medical Alcohol Board, Aarhus, Denmark
Creina Stockley, PhD, MSc Clinical Pharmacology, MBA; Health and Regulatory Information Manager, Australian Wine Research Institute, Glen Osmond, South Australia, Australia
Arne Svilaas, MD, PhD, general practice and lipidology, Oslo University Hospital, Oslo, Norway
Pierre-Louis Teissedre, PhD, Faculty of Oenology–ISVV, University Victor Segalen Bordeaux 2, Bordeaux, France
Dag S. Thelle, MD, PhD, Department of Biostatistics, Institute of Basic Medical Sciences, University of Oslo, Norway; Section for Epidemiology and Social Medicine, Sahlgrenska Academy, University of Gothenburg, Sweden
Fulvio Ursini, MD, Dept. of Biological Chemistry, University of Padova, Padova, Italy
David Van Velden, MD, Dept. of Pathology, Stellenbosch University, Stellenbosch, South Africa
Andrew L. Waterhouse, PhD, Department of Viticulture and Enology, University of California, Davis, USA