Alcohol And the Gastrointestinal Tract: Special Issue: Digestive Diseases 2005

152 Pages·2006·2.2 MB·English

by  M. V. Singer

Alcohol and the
Gastrointestinal
Tract
Editors
Manfred V. Singer, Mannheim
David Brenner, New York, N.Y.
36 fi gures, 15 tables, 2005
Basel • Freiburg • Paris • London • New York •
Bangalore • Bangkok • Singapore • Tokyo • Sydney
S. Karger Disclaimer All rights reserved.
Medical and Scientifi c Publishers The statements, options and data contained in this publication No part of this publication may be translated into other
Basel • Freiburg • Paris • London are solely those of the individual authors and contributors languages, reproduced or utilized in any form or by any means,
and not of the publisher and the editor(s). The appearance of electronic or mechanical, including photocopying, recording,
New York • Bangalore • Bangkok
advertisements in the journal is not a warranty, endorsement, microcopying, or by any information storage and retrieval
Singapore • Tokyo • Sydney or approval of the products or services advertised or of their system, without permission in writing from the publisher or, in
effectiveness, quality or safety. The publisher and the editor(s) the case of photocopying, direct payment of a specifi ed fee to
disclaim responsibility for any injury to persons or property the Copyright Clearance Center (see ‘General Information’).
resulting from any ideas, methods, instructions or products
referred to in the content or advertisements. © Copyright 2005 by S. Karger AG,
P.O. Box, CH–4009 Basel (Switzerland)
Drug Dosage Printed in Switzerland on acid-free paper by
The authors and the publisher have exerted every effort to en- Reinhardt Druck, Basel
sure that drug selection and dosage set forth in this text are in ISBN 3–8055–8030–4
accord with current recommendations and practice at the time
of publication. However, in view of ongoing research, changes
in government regulations, and the constant fl ow of informa-
tion relating to drug therapy and drug reactions, the reader is
urged to check the package insert for each drug for any change
in indications and dosage and for added warnings and precau-
tions. This is particularly important when the recommended
agent is a new and/or infrequently employed drug.
Fax [phone]
E-Mail [email protected]
www.karger.com
Vol. 23, No. 3–4, 2005
Contents
161 Editorial 232 Molecular Mechanisms of Alcoholic Pancreatitis
Singer, M.V. (Mannheim); Brenner, D.A. (New York, N.Y.) Apte, M.V.; Pirola, R.C.; Wilson, J.S. (Sydney)
162 Measuring the Health Consequences of Alcohol 241 Treatment of Alcoholic Pancreatitis
Consumption: Current Needs and Methodological Pfützer, R.H.; Schneider, A. (Mannheim)
Challenges
247 Genetic Polymorphisms in Alcoholic Pancreatitis
Bloss, G. (Bethesda, Md.)
Whitcomb, D.C. (Pittsburgh, Pa.)
170 Moderate Alcohol Consumption and the
255 Clinical Syndromes of Alcoholic Liver Disease
Gastrointestinal Tract
Adachi, M.; Brenner, D.A. (New York, N.Y.)
Taylor, B. (Toronto); Rehm, J. (Toronto/Zürich); Gmel, G.
(Toronto/Lausanne) 264 Molecular Mechanisms of Alcohol-Induced Hepatic
Fibrosis
177 Moderate Alcohol Consumption and Diseases of
the Gastrointestinal System: A Review of Siegmund, S.V. (New York, N.Y./Heidelberg); Dooley, S.
Pathophysiological Processes (Heidelberg); Brenner, D.A. (New York, N.Y.)
Taylor, B.; Rehm, J. (Zürich) 275 Treatment of Alcoholic Liver Disease
181 Animal Models and Their Results in Gastrointestinal Bergheim, I.; McClain, C.J.; Arteel, G.E. (Louisville, Ky.)
Alcohol Research 285 Alcohol and Hepatitis C
Siegmund, S.V. (Heidelberg/New York, N.Y.); Haas, S.; Jamal, M.M.; Saadi, Z.; Morgan, T.R. (Long Beach, Calif.)
Singer, M.V. (Heidelberg)
297 Alcohol Consumption and Cancer of the
195 Alcohol-Related Diseases of the Mouth and Throat Gastrointestinal Tract
Riedel, F.; Goessler, U.R.; Hörmann, K. (Mannheim) Seitz, H.K.; Maurer, B.; Stickel, F. (Heidelberg)
204 Alcohol-Related Diseases of the Esophagus and 304 Therapy and Supportive Care of Alcoholics:
Stomach Guidelines for Practitioners
Franke, A. (Mannheim); Teyssen, S. (Bremen); Singer, M.V. Kienast, T.; Heinz, A. (Berlin)
(Mannheim)
214 Effect of Alcohol Consumption on the Gut
Rajendram, R. (London/Oxford); Preedy, V.R. (London) 310 Author Index Vol. 23, 2005
311 Subject Index Vol. 23, 2005
222 Alcoholic Pancreatitis
after 312 Contents Vol. 23, 2005
Schneider, A.; Singer, M.V. (Mannheim)

© 2005 S. Karger AG, Basel
Fax [phone] Access to full text and tables of contents,
E-Mail [email protected] including tentative ones for forthcoming issues:
www.karger.com www.karger.com/ddi_issues
Dig Dis 2005;23:161
DOI: 10.1159/000090161
Editorial

Alcohol-related disorders account for an enormous tal morbidity and mortality. Moreover, recent evidence
part of the global mortality and for an even greater part from epidemiology studies suggests that moderate alco-
of the life years lost by disabilities due to alcohol con- hol consumption might have some health benefi t, mainly
sumption. In Europe and the USA, more than 20% of men by lowering the risk of coronary heart disease in a certain
and approximately 9% of women hospitalized in varying subset of the population. Therefore, we present critical
medical departments in general hospitals feature alcohol- evaluations of the impact of moderate alcohol consump-
related disorders. In Germany, alcohol-induced diseases tion on the risk and the pathophysiological mechanisms
caused direct or indirect costs of about 20.6 billion EUR of gastrointestinal diseases.
in the year 2000. Strikingly, most of the patients with al- An appraisal of the available animal models used for
cohol-induced organic disorders are being treated in gas- the study of alcohol-related diseases explains the latest
troenterology. fi ndings in basic alcohol research. The effects of alcohol
Thus, our aim in this issue of Digestive Diseases was on the various parts of the gastrointestinal system are dis-
to provide state-of-the-art reviews by a team of interna- cussed in separate chapters with special emphasis on the
tionally well-renowned experts on alcohol-related epide- pancreas and liver.
miology as well as diseases of the gastrointestinal tract, A review of the well-known association between alco-
liver and pancreas. hol consumption and increased risk of cancer is followed
In addition, we also wanted to provide strategies on by a discussion on how best to care for alcoholics in view
how to guide alcoholic patients psychologically, because of the advances presented in alcohol research.
not only do alcohol use and abuse contribute to a variety We would like to thank the international experts who
of medical disorders, but they can strongly affect the so- contributed well-organized and well-written summaries
cial, socioeconomic and personal situation of these pa- of the current knowledge in this fi eld, and have success-
tients such as family interactions or worker productivity. fully presented up-to-date and solid scientifi c data on the
Alcohol abuse can kill, both directly and indirectly; it re- subject. We would also like to thank Dr. Peter Feick for
sults in an increase in the number of injuries, automobile his reliable assistance in the preparation of this issue. It
collisions and violence-related deaths. Therefore, any has also been a great pleasure to correspond with the au-
physician who wants to successfully treat patients who thors.
suffer alcohol-related diseases should not only treat the We do hope that readers fi nd this issue of Digestive
medical conditions. Diseases fulfi ls their expectations.
This issue starts with an overview of the epidemio- Manfred V. Singer, Mannheim
logical data on the impact of alcohol consumption on to- David A. Brenner, N ew York
© 2005 S. Karger AG, Basel
0257–2753/05/0234–0161$22.00/0
Fax [phone]
E-Mail [email protected] Accessible online at:
www.karger.com www.karger.com/ddi
Dig Dis 2005;23:162–169
DOI: 10.1159/000090162
Measuring the Health Consequences of
Alcohol Consumption: Current Needs
and Methodological Challenges
Gregory Bloss
National Institute on Alcohol Abuse and Alcoholism, National Institutes of Health, US Department of
Health and Human Services, B ethesda, Md., USA
Key Words sumption and individual characteristics. Comparisons
Mortality, alcohol-attributable (cid:1) Health consequences, over time are needed to identify areas where improve-
alcohol (cid:1) Alcohol abuse ments in public health may be occurring or are most
needed, to support evaluation of specifi c interventions,
and to encourage the public awareness of alcohol prob-
Abstract lems that is necessary to change attitudes and behaviors
Background/Aims: Extensive research has shown that involving alcohol consumption.
alcohol consumption leads to poor health and premature Copyright © 2005 S. Karger AG, Basel
death through its causal or contributing roles in numer-
ous chronic health conditions and acute health out-
comes, including various cancers, liver disease, and in- Introduction
juries. Paradoxically, advances in understanding of the
causal associations between alcohol consumption and There is clear scientifi c consensus that alcohol con-
various conditions have complicated our ability to dis- sumption leads to a variety of harms to health, and sub-
cern trends in the health consequences of alcohol con- stantial agreement regarding many specifi c health conse-
sumption over time. Methods: Four distinct needs for quences that result from excessive alcohol consumption.
information on alcohol’s role in causing adverse health Several recent reviews and meta-analyses provide surveys
outcomes are identifi ed. Estimates of alcohol-attribut- of the epidemiological evidence regarding alcohol effects
able mortality from two US studies are compared and for various health outcomes [ 1–5] . Nevertheless, further
differences identifi ed. Results: Differences in the condi- research effort devoted to discovering and documenting
tions included and alcohol-attributable fractions em- the nature and extent of those harms is clearly needed.
ployed accounted for large differences in the estimated Four distinct needs for this information – clinical, indi-
alcohol-attributable mortality for several health out- vidual, scientifi c, and social – justify further work to as-
comes. Conclusion: Despite the broad consensus on sess the health consequences of alcohol consumption.
many health consequences of alcohol consumption, fur- First, advances in knowledge of the effects of alcohol on
ther research is needed to clarify the conditions that are specifi c health outcomes support good clinical practice,
caused by alcohol consumption, magnitudes of causal including appropriate identifi cation of causal factors un-
relationships, and effects of different patterns of con- derlying patients’ health conditions and guidance to clini-
© 2005 S. Karger AG, Basel Gregory Bloss
NIAAA Division of Epidemiology and Prevention Research
Fax [phone] Fishers Lane, Room 2075
E-Mail [email protected] Accessible online at: Bethesda, MD 20892-9304 (USA)
www.karger.com www.karger.com/ddi Tel. [phone], Fax [phone], E-Mail [email protected]
cians regarding other conditions that may threaten the specifi c patterns of consumption interact with individual
health of individual patients. The variety of health condi- characteristics to affect overall health risks and benefi ts.
tions associated with alcohol consumption, and the pos- This represents a large and important research agenda
sibility that individuals may deny the existence of an al- whose scope encompasses the many known conditions for
cohol problem, make it important for physicians to be which risk may be affected multiplied by the many per-
well-informed about these effects. Awareness of the po- mutations of individual characteristics and consumption
tential health consequences of alcohol consumption also patterns, as well as additional conditions that may yet be
supports appropriate screening and intervention by pri- linked to alcohol consumption. Given the great breadth
mary care clinicians [6, 7] . The limits to current knowl- of this research agenda, a key priority is to strengthen
edge regarding the relationships between consumption at measurement approaches that apply across the full range
various levels and the risks for specifi c outcomes, as well of conditions. Particular needs include further refi nement
as the subtler issues involving co-variation in risks across in measurement of alcohol consumption levels and pat-
different population groups and different levels and pat- terns as well as development, testing, and validation of
terns of alcohol consumption, leave important areas of metrics and assessment tools that support comparisons
uncertainty that underscore the need for further re- of health outcomes at both the individual and aggregate
search. levels. Advances in these areas will support further prog-
Second, clearer information than that currently avail- ress in the broader scientifi c goal of developing a rich and
able will assist in evaluating the risks and benefi ts associ- detailed understanding of the full spectrum of alcohol ef-
ated with alcohol-related behaviors that may be contem- fects on health.
plated. Rational individual decisions about alcohol con- Fourth, continued research on the health conse-
sumption may appear diffi cult in the context of an quences of alcohol consumption is needed for social pur-
extensive list of associated health conditions and an un- poses: to provide up-to-date information on the range and
certain schedule of risks and benefi ts that vary in different magnitude of specifi c risks associated with particular con-
ways across conditions by age and sex, as well as by levels, sumption behaviors. Credible and reliable data on the
patterns, and contexts of consumption. In addition, the health effects of alcohol consumption can serve to focus
evidence of signifi cant benefi cial effects for at least some public attention on a major public health issue, provide
population groups may reduce the attraction of absten- valid metrics for evaluating progress (or the lack thereof)
tion even among those who are highly averse to health in reducing adverse consequences through health-promo-
risks. Individuals need information to help address not tion programs and policies, and justify public and private
just the global questions of ‘how much, how often?’ but interest and investment in a range of research, preven-
the more immediate issue of ‘Should I have a(nother) tion, and treatment efforts. Addressing this need requires
drink now?’ Although full-blown risk-benefi t calculations more than just measurement of the individual and aggre-
cannot be part of every interaction with a bartender or gate health effects of alcohol consumption, but also track-
social host, clearer understanding of the scope, severity, ing of how those effects change over time, i.e., reliable
and specifi c risks of potential health consequences can and comparable data on trends in alcohol-related health
contribute to better planning and ultimately better out- consequences. Recent work by the US Centers for Disease
comes. Control and Prevention to provide public web-based ac-
Third, there are continuing scientifi c opportunities to cess to Alcohol-Related Disease Impact software that al-
be explored in terms of additional conditions that may be lows consistent calculation of alcohol-attributable frac-
caused (or prevented) by alcohol consumption and revi- tions (AAFs) represents an important and constructive
sions to previous understanding of the relationships be- step in this direction [8] .
tween consumption and conditions that have already
been identifi ed as alcohol-related. The major areas of sci-
entifi c focus continue to be on identifying, characterizing, Lessons from Recent Studies of Alcohol-
and quantifying associations between alcohol consump- Related Mortality in the United States
tion and various disease and other health outcomes and
in describing the causal mechanisms that underlie such To see the potential social importance of advanc-
associations. To address the clinical, individual, and so- ing measurement of alcohol-attributable health conse-
cial needs identifi ed here, the most important scientifi c quences, it is instructive to consider how the collection,
opportunities involve advancing understanding of how validation, reporting, and analysis of data on alcohol-at-
Measuring Health Consequences of Dig Dis 2005;23:162–169 163
Alcohol Consumption
Fig. 1. US alcohol- and non-alcohol-related
traffi c fatalities per 100,000 population,
1982–2004. Source: Fatality Analysis Re-
porting System, National Highway Traffi c
Safety Administration and US Census Bu-
reau. Courtesy of R. Hingson, NIAAA.
tributable traffi c crash fatalities in the US over the past The situation is considerably different with respect to
25 years has contributed to successful efforts to reduce many other alcohol-attributable health outcomes, espe-
alcohol-related traffi c crash deaths. Since the 1970s, the cially for chronic conditions associated with long-term
Fatality Analysis Reporting System (FARS; previously alcohol consumption. Although there is general agree-
known as the Fatal Accident Reporting System) of the US ment on many specifi c conditions that may be wholly or
National Highway Traffi c Safety Administration has col- partially attributed to alcohol consumption, there are
lected and reported data on traffi c crash fatalities in a some clinically and epidemiologically important condi-
systematic framework. Detailed information on alcohol tions for which scientifi c agreement is very recent and not
involvement, based on physical testing of drivers in- yet fi rmly established. Within the past 10–20 years there
volved in fatal crashes, has been part of the FARS data have been signifi cant revisions in our understanding of
from the early 1980s, and this has provided a rich source the specifi c conditions that may be caused by alcohol con-
of data to serve as input to many studies of efforts to re- sumption and the magnitude of alcohol’s causal effects,
duce alcohol-related crash fatalities. The FARS data also including the identifi cation of conditions not previously
have provided a consistent and reliable source of infor- associated with alcohol (e.g., breast cancer) and deletion
mation for use in publicizing both the enormous burden of conditions once thought to be partially caused by alco-
of alcohol-related crashes and the potential value of suc- hol consumption (e.g., pneumonia and infl uenza). In ad-
cessful interventions. The long-term success of the dual dition, there is growing consensus that moderate alcohol
strategy of scientifi c analysis and publicity may be seen consumption confers a degree of protection against coro-
in the overall trends in alcohol-related traffi c crash fa- nary heart disease for at least some population groups;
talities: from 1982 to 2004, alcohol-related traffi c crash this important effect on the leading cause of death re-
death rates in the US have declined by 50% even as traf- mained controversial until quite recently. Similarly, evi-
fi c fatalities not involving alcohol have increased by 15% dence is now growing that moderate alcohol consumption
over the same period (fi g. 1 ). Our ability to assess and may confer a protective effect for some against type-2
celebrate this success, as well as a portion of the success diabetes, a qualitative reversal from the received scien-
itself, may be traced to the development and consistent tifi c wisdom of just a decade ago [ 9, 10] .
reporting of outcome data on the role of alcohol in traffi c A clear illustration of the effects of changing scientifi c
crash deaths. assessments of the role of alcohol in mortality from vari-
164 Dig Dis 2005;23:162–169 Bloss
ous conditions is apparent in a comparison between two Table 1. Alcohol-attributed mortality from two US studies
recent estimates of alcohol-attributable deaths in the US.
Table 1 shows estimated alcohol-attributable mortality in Cause of death McGinnis Midanik
and Foege et al.
the US from various causes reported for 1995 by McGin-
[11] [12]
nis and Foege [11] and for 2001 by Midanik et al. [12] .
The totals at the bottom of the table suggest a decrease of Chronic conditions
approximately 28% in alcohol-attributable mortality over Cancers
Stomach cancer 3,000 0
the period. However, examination of the numbers of
Liver cancer 1,500 690
deaths estimated within various disease categories shows
Oropharyngeal cancer 3,000 360
so much greater variation that it seems improbable that
Esophageal cancer 4,500 447
the health consequences of alcohol could have changed Laryngeal cancer 800 233
so much in such a short time. There is reasonable consis- Breast cancer 0 352
Prostate cancer 0 233
tency in the components of the unintentional injuries and
Circulatory diseases
intentional injuries categories, but major discrepancies in
Stroke 11,000 2,401
deaths attributed to toxicity/overdose and to a number of
Hypertension 3,000 1,221
chronic health conditions. Alcoholic cardiomyopathy 0 499
An obvious source of disparity in the chronic condi- Ischemic heart disease 0 908
Supraventricular cardiac dysrhythmia 0 165
tions category is that each study reports substantial num-
Respiratory diseases
bers of deaths from conditions for which the other study
Pneumonia/infl uenza 4,000 0
reports zero. For example, McGinnis and Foege [11] re-
Tuberculosis 300 0
ported 3,000 alcohol-attributable deaths from stomach Digestive diseases
cancer for 1995, but Midanik et al. [ 12] reported none in Ulcers/digestive diseases 900 107
Chronic liver disease 6,000 18,926
2001. On the other hand, Midanik et al. reported nearly
Pancreatitis 1,000 1,261
7,000 deaths from mental, nervous system, and neuro-
Cholelithiasis 0 0
muscular disorders, while McGinnis and Foege reported
Diabetes 3,000 0
none. The earlier study included 4,000 deaths from pneu- Mental, nervous system, and neuromuscular disorders
monia and infl uenza and 3,000 deaths from diabetes, Alcohol abuse and dependence 0 5,841
while the more recent report had no alcohol-attributable Alcoholic psychosis 0 742
Alcoholic polyneuropathy 0 3
deaths in those categories. The absence of ICD codes
Degeneration of nervous system 0 114
identifying specifi c conditions in these studies suggests
Epilepsy 0 177
that some of these differences may refl ect category assign- Alcoholic myopathy 0 2
ment discrepancies (e.g., alcohol use disorders, toxicity/ Fetal and newborn effects 0 152
overdose), but others more clearly refl ect differences in Psoriasis 0 0
Total, chronic conditions 42,000 34,833
the conditions included as causally related to alcohol con-
sumption. Further scrutiny reveals important differences Acute conditions
Injuries – unintentional
in the numbers of deaths reported in the categories in
Motor vehicle 17,000 14,109
which both reports count some deaths. Of particular note
Fire injuries 1,000 1,167
is the difference in deaths from digestive diseases, which
Fall injuries 5,000 4,766
are reported at 20,223 by Midanik et al. for 2001 but only Water, aviation injuries 1,500 1,071
7,900 by McGinnis and Foege for 1995, with the discrep- Other unintentional injuries 0 570
Injuries – intentional
ancy refl ecting a threefold difference in the chronic liver
Homicide 10,000 7,957
disease category (almost 19,000 in one study vs. 6,000 in
Suicide 8,500 6,995
the other).
Toxicity/overdose 20,000 4,297
As noted, it seems unlikely that both sets of fi ndings Total, acute conditions 63,000 40,933
could be correct, given the magnitude of the differences,
Total 105,000 75,766
the relatively short time period separating the two stud-
ies, and the fact that per capita alcohol consumption
changed by less than 2% during that time [ 13] . The more
plausible conclusion is that differences in data and meth-
odology account for the differences in the composition of
Measuring Health Consequences of Dig Dis 2005;23:162–169 165
Alcohol Consumption
estimated alcohol-attributable mortality between these tions should be included as causally linked to alcohol con-
two studies. Both of these studies have been conducted sumption, the methods used in studies seeking to quan-
by highly respected researchers using broadly similar tify relationships between alcohol consumption and health
methods and relying on published fi ndings and offi cial outcomes can have signifi cant effects on estimated effect
vital statistics. The methods involve application of popu- sizes (e.g., magnitudes of estimated relative risk ratios).
lation-attributable fractions for alcohol (i.e., AAFs) to Several recent studies have identifi ed key methodological
cause-specifi c data on deaths by age and sex. The studies concerns that must be addressed in epidemiologic re-
differ in the sources and values of the AAFs employed. search into the health effects of alcohol consumption [ 1–3,
Differences in the conditions included (i.e., for which 14–16] . Among the most important issues for improving
only one of the two studies employed an AAF of 10 ) and measurement of the relationships between alcohol con-
differences in the AAF values assigned to conditions that sumption and health outcomes are the following.
are included in both studies are clearly the primary source
of the differences in reported mortality. An important ef- Selection of Appropriate Cohorts and Comparison
fect of these differences is to leave uncertainty as to Groups
whether and how overall alcohol-attributable mortality Cohorts must be reasonably representative and allow
may have changed over the period between the two stud- suffi cient variation to avoid omitting relevant aspects of
ies. the relationship between alcohol consumption and health
Signifi cant variation in included conditions across outcomes. Comparison groups must be chosen with care
studies is more the norm than the exception. A recent as well, so that the counterfactual drinking scenario (e.g.,
study found substantial differences in the included condi- abstinence, or drinking at a ‘safe’ level) is well specifi ed
tions across three major studies from the US, Canada, and does not incorporate unmeasured heterogeneity. In
and Australia and, based on an extensive assessment of particular, the distinction between lifetime abstainers
the epidemiologic literature in fi ve languages published and former drinkers should be observed in selecting a
in 1995 or later, recommended a list of 47 conditions comparison group [ 1, 2].
(with associated ICD-9 codes) for inclusion in studies of
the social costs of alcohol (and, by extension, studies of Measurement of Consumption Levels and Patterns
the health-related burden of alcohol) [9]. It appears that Epidemiologic analysis requires accurate measure-
the potential contribution of this effort to assess the full ment of the exposure in question, i.e., the relevant aspects
scope of conditions for which there is evidence of a caus- of alcohol consumption. Different aspects of the con-
al association with alcohol consumption may have been sumption pattern (i.e., the temporal distribution of con-
underappreciated; efforts to quantify the health effects of sumption of a given volume of ethanol) and drinking con-
alcohol consumption would benefi t from reference to (pe- text (setting and surroundings, presence or absence of
riodically updated) assessments of this sort. food, etc.) may exert critical infl uence on the risks for dif-
The foregoing comparison between two specifi c stud- ferent health outcomes, and these infl uences may vary
ies illustrates that scientifi c progress in understanding the across population groups. An additional complicating
health effects of alcohol consumption has provided our factor is the timing of consumption relative to reported
best public health researchers – and hence, the public – health outcomes; the appropriate time lag between ob-
with changing information about which conditions are served consumption and health outcome assessment may
partially caused by alcohol consumption, the strength of be expected to vary across health conditions [1, 17] .
the causal relationships involved, and the extent of result-
ing overall damage to health at the individual and popu- Correcting for Confounding Effects
lation levels. The interactions among alcohol consumption levels
and patterns, other health behaviors, environmental fac-
tors, and individual and demographic characteristics cre-
Methodological Issues ate complex confounding effects. Analysts must seek to
correct for the full range of relevant confounders while
The lack of stable consensus regarding the health effects avoiding over-correcting (which may occur when stan-
of alcohol consumption ultimately refl ects the scientifi c dard correction methods are applied to factors that fi gure
challenges involved in identifying and reliably estimating in the causal chain leading from alcohol consumption to
such effects. Beyond gaining agreement on which condi- health outcomes) [2] .
166 Dig Dis 2005;23:162–169 Bloss
Separate Reporting of Benefi cial and Adverse viduals as a result of drinking by others; such effects need
Consequences to be quantifi ed and catalogued along with other health
Much recent attention has focused on potential benefi - consequences of alcohol consumption. For example, in
cial effects of alcohol consumption. There is now substan- the US, 44% of decedents in crashes involving drinking
tial evidence of a U- or J-shaped relationship between drivers were persons other than the drinking drivers [23] ,
alcohol consumption and coronary heart disease for at and alcohol has been estimated to be involved in one
least some population groups [ 3, 5, 18, 19], and suggestive fourth of all violent crimes [24] . Most such ‘third-party’
evidence that qualitatively similar associations may exist effects may be expected to fall in the categories of unin-
for type-2 diabetes and cholelithiasis [9, 20, 21]. When tentional injuries, intentional injuries, and fetal alcohol
there is evidence of both adverse and benefi cial effects of effects. Identifying and measuring the effects of alcohol
alcohol consumption – either within or across outcome that result from someone else’s drinking poses special
categories – it is important to document both kinds of problems, both at the individual level in terms of prima-
consequences, and not simply the net balance of two com- ry observation and causal attribution and at the popula-
peting effects. The study by Corrao et al. [ 16] sets a useful tion level in the estimation of aggregate effects. Neverthe-
standard in this area by distinguishing different effects less, such effects represent an important component of
and associating each outcome with specifi c consumption the health consequences of alcohol consumption.
ranges. This disaggregation provides much more infor-
mation about the health effects of alcohol than if the Measurement of Health Outcomes
roughly equal numbers of deaths reported as caused and Studies of the health effects of alcohol consumption
prevented by low-level consumption in that study were most commonly assess mortality outcomes, although al-
netted against one another and reported as no overall ef- cohol also has many non-fatal effects on health. Mortal-
fect on mortality. ity is relatively easy to measure because of the dichoto-
mous character of life/death outcomes, and the severity
Appropriate Construction, Application, and of these outcomes makes them obviously important. Even
Interpretation of AAFs in the relatively straightforward case of injury outcomes,
Attributable fractions may be misused in a variety of however, signifi cant uncertainties remain about the mag-
ways, and the results of such misuse could lead to signifi - nitude of alcohol involvement in injury mortality. To
cant misstatements of the health consequences of alcohol build on the successful model from the traffi c crash area
consumption [22] . Even for well-constructed attributable and improve understanding of the role of alcohol in in-
fractions, problems may arise when previously published jury mortality from various causes, Hingson et al. [25]
fractions are applied to current vital statistics on cause- have called for comprehensive testing of all injury deaths
specifi c deaths without consideration of the possibility for alcohol involvement. For mortality from all causes,
that the fractions may change over time. Such changes the need for additional research to clarify the risk rela-
could result from changes in population prevalence of tionships associated with various consumption behaviors
drinking at particular levels or from changes in the un- and individual characteristics is manifest.
derlying relative risks. For some outcomes, such changes Beyond mortality alone, a more complete view of the
may result from technological, legal, or other factors not health effects of alcohol must incorporate non-fatal health
directly related to alcohol, such as widespread use of air consequences as well. There is credible evidence from
bags, enforcement of speed limits, or changes in health various countries that non-fatal effects account for a sig-
care practices. The possibility that changes in alcohol-at- nifi cant share of the overall health burden associated with
tributable mortality could result from changes in AAFs alcohol [3, 26–28]. Full accounting of morbidity out-
should encourage the use of up-to-date information in comes must confront a range of qualitative consider-
estimating aggregate health effects of alcohol consump- ations, encompassing illness, pain, and disability, al-
tion. though most studies have used indicators of health care
utilization (e.g., hospitalizations, hospital days, health
Recognition of Health Consequences of Drinking by care expenditures). Advances in measurement of health-
Others related quality of life represent a particularly promising
Most of the health consequences of alcohol consump- development in the measurement of non-fatal outcomes
tion accrue to the drinker. However, some important al- with two signifi cant potential benefi ts. First, cost-effec-
cohol-attributable health outcomes may accrue to indi- tiveness and cost-utility analyses of interventions de-
Measuring Health Consequences of Dig Dis 2005;23:162–169 167
Alcohol Consumption