You are on page 1of 4

Public Health Briefs

and its major subfractions: the Caerphilly 6. Prevention ofCardiovascularDiseaseRecom- 11. Hunter SM, Webber LS, Berenson GS.
and Speedwell Collaborative Heart Dis- mendationsforNational Governments. Tech- Cigarette smoking and tobacco usage be-
ease Studies. J Epidemiol Community nical reports of the World Health Organiza- havior in children and adolescents: Boga-
Health. 1988;42:220-225. tion, Coronary Prevention Group. London, lusa Heart Study. Prev Med. 1980;9:701-
2. Kivela S-L, Nissinen A, Punsar S, Puska P, England: World Health Organization; 1992. 712.
Karvonen MJ. Determinants and predic- 7. Schoenbom CA, Cohen BH. Trends in 12. Bresnahan JL, Shapiro MM. A general
tors of heavy alcohol consumption among smoking, alcohol consumption, and other equation and technique for the exact
aging Finnish men. Compr Gerontol [B]. health practices among US adults, 1977 partitioning of chi-square contingency
1988;2:103-109. and 1983. Adv Data Vital Health Stat. June tables. Psychol Bull. 1966;66:252-262.
3. Johnson CC, Hunter SM, Amos CI, Elder 30, 1986; no. 118. DHHS publication PHS 13. Johnston LD, O'Malley PM, Bachman JG.
ST, Berenson GS. Cigarette smoking, 86-1250. National Survey Results on Drug Use from
alcohol, and oral contraceptive use by type 8. Healthy People 2000: National Health Promo- Monitoring the Future Study, 1975-1992:
A adolescent[s}-the Bogalusa Heart Study. tion and Disease Prevention Objectives, Con- Volume I. Secondary School Students. Rock-
JBehav Med. 1989;12:13-24. ference Edition. Washington, DC: US Dept ville, Md: National Institutes of Health,
4. The 1988 Joint National Committee. The of Health and Human Services; 1990. National Institute on Drug Abuse; 1993.
9. Berenson GS, McMahan CA, Voors AW, 14. Brooks SD, Williams GD, Stinson FS,
1988 Report of the Joint National Commit- et al. Cardiovascular Risk Factors in Chil- Noble J. Apparent per Capita Alcohol
tee on Detection, Evaluation, and Treat- dren: The EarlyNatural History ofAtheroscle- Consumption: National, State and Regional
ment of High Blood Pressure. Arch Intem rosis and Essential Hypertension. New York, Trends, 1977-1987. Rockville, Md: Na-
Med. 1988;148:1023-1038. NY: Oxford University Press; 1980. tional Institute on Alcohol Abuse and
5. LaPorte R, Valvo-Gerard L, Kuller L, et 10. Croft JB, Freedman DS, Cresanta JL, et al. Alcoholism; 1989. Surveillance Report 13.
al. The relationship between alcohol con- Adverse influences of alcohol, tobacco, and 15. Barber JG, Bradshaw R, Walsh C. Reduc-
sumption, liver enzymes, and high density oral contraceptive use on cardiovascular ing alcohol consumption through television
lipoprotein cholesterol. Circulation. 1981; risk factors during transition to adulthood. advertising. J Consult Cin Psychol. 1989;57:
64(III):67-72. Am JEpidemiot 1987;126:202-213. 613-618.

A Gender-Specific Measure of Binge


IE.Z.::U.S.: 111111
~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ . . . .. Drinking among College Students
Henry Wechsler, PhD, George W Dowdall, PhD, Andrea Davenport, MPH, and
Eric B. Rimm, ScD

Introduction This paper contrasts the use of the


currently accepted definition of binge
*. relatepoblems.amon malen..d Binge drinking, or the consumption drinking to the use of a gender-specific
*fe a . . . . .o. stde .t deveo a of large amounts of alcohol on a single
occasion, has been linked to an increased
standard among college students.
...... ..

risk of negative health outcomes.1 It has Methods and Statisical Analysis


become common practice in research on
* a 'heath
puli cetc alcohol to define heavy or binge drinking The data for this research were
in terms of episodes involving five or more gathered as part of a representative
drinks in a row for both men and survey of 17 592 students at 140 colleges.
sampleofsuntat104er women.2-7 Yet blood alcohol level tables A self-administered 20-page question-
that determine the legal definition of naire received by 25 627 students in early
driving while intoxicated8 are based on sex 1993 yielded an overall response rate of
as well as on weight. Recent research
suggests that the gender differences are Henry Wechsler, George W. Dowdall, and
owing to women's lower rates of gastric Andrea Davenport are with the Department of
metabolism of alcohol (initially only 80% Health and Social Behavior, and Eric B. Rimm
of men's) and, therefore, to their higher is with the Department of Epidemiology and
Nutrition, Harvard School of Public Health,
blood alcohol levels for a fixed amount of Boston, Mass. Dr Dowdall is also with St.
alcohol, even after accounting for differ- Joseph's University, Philadelphia, Pa.
ences in body weight or lean body mass.9 Requests for reprints should be sent to
Psychiatric epidemiologists have sug- Henry Wechsler, PhD, Department of Health
gested that clinical criteria should there- and Social Behavior, Harvard School of Public
Health, 677 Huntington Ave, Boston, MA
fore be defined differently for men and 02115.
women in the diagnosis of alcohol depen- This paper was accepted January 11,
dency and alcoholism.10 1995.
;. .....

.;........ ....

:;...
.
July 1995, Vol. 85, No. 7
Public Health Briefs

69%. Details of the study design have


been published elsewhere.' TABLE 1-Odds Ratios (95% Confidence Intervals) for Risk of Alcohol-Related
This analysis is based on the 12 243 Problems with Gender-Neutral and Gender-Specific Definitions
respondents who reported drinking alco- among 12 243 US College Students
hol in the 30 days prior to the survey and
could be classified as current drinkers, Gender-Neutral Gender-Specific
nonbinge or binge. To quantify differ- Definition: OR of Definition: OR of
Women Consuming Women Consuming
ences by sex and to control for other Alcohol-Related % Reporting 5 Drinks vs Men 4 Drinks vs Men
potential confounders, multiple logistic Problem Problem Consuming 5 Drinks Consuming 5 Drinks
regression was used to compare the
likelihood of an alcohol-related problem Hangover 64 1.78*** (1.36-2.34) 1.30* (1.02-1.67)
Miss a class 30 1.47*** (1.19-1.81) 0.99 (0.81-1.21)
among men with that among women for a Fall behind 23 1.33* (1.06-1.66) 1.04 (0.83-1.29)
typical drinking level (i.e., the usual Cause regret 36 1.41** (1.15-1.73) 1.17(0.96-1.43)
number of drinks per occasion in the past Forget 27 1.31* (1.06-1.62) 0.92 (0.74-1.13)
Argument 22 1.08(0.86-1.35) 0.90 (0.72-1.12)
30 days). The dependent variable was Unplanned sex 21 0.89 (0.71-1.12) 0.89 (0.72-1.11)
whether a student reported experiencing Unsafe sex 11 0.94 (0.70-1.26) 0.87 (0.66-1.16)
1 of 12 outcomes as a result of drinking. Injury 10 1.23 (0.92-1.65) 0.81 (0.60-1.09)
Each student was asked: "Since the
beginning of the school year, how often has *P < .05; **P < .01; ***P < .001.
your drinking caused you to [experience
each of twelve problems]." The nine
alcohol-related problems that were expe-
rienced by at least 5% of students of each
sex are presented in Table 1. The three
problems that fell below this cutoff-
vandalism, trouble with the police, and
alcohol overdose-were excluded from
the rest of the analysis.
The logistic model was as follows: &$I'fl I141U' 4 4 " ^ ;

In1 P
= 1O + P1 Xgender
n

+ E r3iXi + i3intrxXXgenderq ij ' si l 0


~> ~~~~~~~~~~~~~ X_it';j
2i~~~~~~~~~' -

i=2
where Xgender is 1 if male and 0 if female ffi:.;i <it > 3 @_2 gs <J 2r }8
9! |
|
i W ~~~~~~~~~~~~~~~~~~~~~~7i
+ . ;w g ,. r, -te,
and the xi are a series of indicator f1 24 ;studnt wh8o i.reported misin a
FIUR 1-umlaive prcentag
variables specifying the level of a student's aRinresult
clssaoccasion ofsdrJinkn lchl by usual nube of drink pr;
typical drinking during the previous month. the past month~~~~~~~~~~~~~~11k
For example, X3 takes on the value 1 for
students who reported typically drinking
three drinks per occasion andx3 is 0 for all
other students. Students typically drinking
one drink were considered the reference
category. The antilogs of the beta coeffi- FIGURE 1 -umulative percentage of 12 243 students who reported missing a
cients can be directly interpreted as the class as a result of drinking alcohol, by usual number of drinks per
occasion in the past month.
odds of a particular outcome between the
"exposed" and the referent populations.
The P,I is the estimate for the effect of sex,
the P3i's are the estimates of effect for each
individual level of alcohol consumption, drinks five drinks can be estimated from vals (CIs) using the variance estimates of
and the 3inx's are the estimates for the the ratio of fitted logits: each coefficient plus two times the covari-
interaction between sex and level of ance of each possible pair of terms."
drinking. Odds of woman who drinks 4 Self-reported items on height and weight
OR Odds of man who drinks 5 were transformed into metric measure-
To determine if female and male
students had similar odds of an alcohol- ments. The body-mass index was created
related outcome for a given level of using the conventional definition of weight
alcohol consumption, the odds ratio (OR) 5+5 inoi,
el+culaende 9+ (in kilograms) divided by the square of
height (in meters). In this sample, body
=

between the sexes was calculated. For e N- 3gender- P - intS .


mass index averaged 24 for males and 22
=

example, the odds ratio for a woman who for females.


drinks four drinks versus a man who We calculated 95% confidence inter-

American Journal of Public Health 983


July 1995, Vol. 85, No. 7
Public Health Briefs

Results sons are not across the two sets of odds due to under- or overreporting or untruth-
ratios (1.47 and 0.99) but are embedded fulness. However, investigators have stud-
The overall sample is representative within the confidence limits of each. The ied the validity of self-reported question-
of full-time undergraduates currently only meaningful comparison across col- naires and have used this method with
studying at 4-year American colleges and umns may be that one odds ratio is success.1',2"-7 There is also no evidence of
universities. The proportions of men and significantly different from 1.00 while the differences between men and women in
women were similar in such variables as other is not. For most of the problems self-report inconsistencies or inaccura-
age, race, marital status, self-rating of examined, the evidence (presented in the cies.
health, having a parent who was a college column labeled "Gender-Neutral Defini- What are the practical consequences
graduate, and living with a roommate. tion") suggests that women who drank of using this cutoff to understanding
Figure 1 illustrates the differences five drinks were significantly more likely college binge drinking? Using the old
between the sexes for one drinking- to experience an adverse outcome than definition of binge drinking (five or more
related problem. Plotted along the verti- men who drank at a similar level. By drinks) would identify 33% of the women
cal axis are the usual numbers of drinks contrast, for eight of the nine problems as binge drinkers; using the new definition
consumed per occasion during the previ- reported (the column labeled "Gender- increases that figure to 39%. This increase
ous 30 days. The horizontal axis presents Specific Definition"), women who typi- of 6% (a relative increase of 18%) in the
the cumulative percentage of students cally drank four drinks had a similar number of women binge drinkers is in line
who reported missing a class since the likelihood of each alcohol-related prob- with the growing emphasis on alcohol
beginning of the school year as the result lem as men who had five drinks. problems in women.1820 Clinicians and
of drinking. Other figures (not presented At least some of the association administrators might draw additional im-
for reasons of space) illustrating indi- between usual drinking and the occur- plications from the findings. Women
vidual alcohol-related problems show simi- rence of alcohol-related problems might should be advised that they cannot drink
lar results. be owing not to sex but to differences in at the same level as men without risking
If men and women missed a class body mass. However, controlling for body greater health and behavioral conse-
because of drinking and this outcome was mass index only somewhat attenuated the quences. College alcohol educators should
due to the same usual number of drinks odds ratios. For example, the odds of any consider programs to alert women to the
per episode for both sexes, Figure 1 would of the alcohol-related problems (exclud- heightened risk they run in matching male
show two overlapping lines, with the same ing unplanned or unsafe sex) for a woman drinking patterns. While "blaming the
cumulative proportion of men and women consuming five drinks versus a man victim" is poor social policy, it is entirely
at each level of usual drinking. Instead, consuming five drinks dropped from 1.66 appropriate to educate women to protect
Figure 1 shows a clear difference, with a (95% CI = 1.18, 2.34) to 1.53 (95% their own health in an environment in
gap between the two sexes indicating that CI = 1.08, 2.16) after controlling for body which gender-neutral drinking norms ac-
women reported missing a class at a mass index. When women who consumed tually put them at higher risk than men.
significantly lower level of usual drinking four drinks were compared with men who Clinicians should suspect the presence of
than men. consumed five drinks, the odds also drinking problems in their female patients
Table 1 gives the odds of each of nine diminished from 1.36 (95% CI = 1.00, at lower levels of alcohol use than is seen
drinking-related problems for a woman 1.87) to 1.25 (95% CI = 0.90, 1.71). This in male patients.
who usually drinks four or five drinks supports the argument that women have Although many women drink as
compared with a man who usually drinks lower rates of gastric metabolism of heavily as men, their recognition of
five drinks. The problems are presented in alcohol and therefore higher blood alco- alcohol problems lags. Among drinkers
the same order as in the original question- hol levels than men for a fixed amount of who binged three or more times in the
naire. The crucial numbers in this table alcohol, even after accounting for differ- past 2 weeks, 22% of the men described
are not the odds ratios of the risk of ences in body weight and lean body mass.9 themselves as heavy or problem drinkers
disease; rather, they are the comparisons. Further control for year in school, type of compared with only 8% of the women. It
For example, the second row reports on school, or living arrangements did not is important to correct the underestima-
the results from one question, "Since the appreciably alter these results. tion of the extent and seriousness of
beginning of the school year, how often drinking problems in women,20 contrib-
has your drinking caused you to miss a uted to in part by the use of a single
Discussion standard for heavy alcohol consumption.
class?" Thirty percent of the students who
drank during the last year reported The methodological implications of A gender-specific definition of binge
missing a class because of drinking. this research are clear: a lower standard drinking should be used in future research
Women who usually consumed five drinks defining heavy or binge drinking needs to and clinical practice. O
were 1.47 times (95% CI = 1.19, 1.81) be used for women than for men. For
more likely to miss a class than were men eight of the alcohol-related problems
who consumed five drinks, strong and examined in this study of college students, Acknowledgments
women who typically drank four drinks This research was supported by the Robert
statistically significant evidence that using Wood Johnson Foundation.
the same cutoff for binge drinking for men had a similar likelihood of experiencing We wish to thank the following persons
and women is misleading. By contrast, that problem as men who usually drank who assisted with the project: Andrew Brodsky,
women who usually consumed four drinks five drinks. The findings did not apply, BS; Sonia Castillo, PhD; Gary Gregg, PhD;
however, to the most frequently experi- Jeffrey Hansen, BS; Lloyd Johnston, PhD;
were almost equally likely to miss a class Avtar Khalsa, MSW; Marianne Lee, MPA; and
compared with men who usually con- enced problem, hangover. Barbara Moeykens, MS. We also wish to thank
sumed five drinks (OR = 0.99; 95% The study relies on self-reports, Scott Chasan-Taber, PhD, for assistance in
CI = 0.81, 1.21). The important compari- which introduce the possibility of error developing the statistical models.

984 American Journal of Public Health July 1995, Vol. 85, No. 7
Public Health Briefs

US Dept of Health and Human Services; 13. Midanik L. Validity of self-reported alco-
References 1993. hol use: a literature review and assessment.
1. Wechsler H, Davenport A, Dowdall G, 7. Hingson RW, Strunin L, Berlin BM, BntJAddict. 1988;83:1019-1030.
Moeykens B, Castillo S. Health and behav- Heeren T. Beliefs about AIDS, use of 14. Reinisch OJ, Bell RM, Ellickson PL. How
ioral consequences of binge drinking in alcohol and drugs, and unprotected sex Accurate Are Adolescent Repois of Drug
college: a national survey of students at 140 among Massachusetts adolescents. Am J Use? Santa Monica, Calif: RAND; 1991.
campuses.JAMA. 1994;272:1672-1677. Public Health. 1990;80:295-299. RAND publication N-3189-CHF.
2. Wechsler H, Isaac N. "Binge" drinkers at 8. O'Brien R, Chafetz M. The Encyclopedia of 15. Kupetz K, Klagsbrun M, Wisoff D, LaRosa
Massachusetts colleges: prevalence, drink- Akoholism. New York, NY: Facts on File; J, Davis DI. The acceptance and validity of
ing styles, time trends, and associated 1982:49-51. the Substance Use and Abuse Survey
problems.JAMA. 1992;267:2929-2931. 9. Frezza M, diPadova C, Pozzato G, Terpin (SUAS).JDrugEduc. 1979;9:163-188.
3. Hanson DJ, Engs RC. College students' M, Baraona E, Lieber CS. High blood 16. Rachel JV, Guess LL, Hubbard RL, et al.
drinking problems: a national study, 1982- alcohol levels in women: the role of Adolescent Drinking Behavior. Vol 1. Re-
1991. Psychol Rep. 1992;71:39-42. decreased gastric alcohol dehydrogenase search Triangle Park, NC: Research Tri-
4. Presley CA, Meilman PW, Lyerla R. activity and first-pass metabolism. N Engl J angle Institute; 1980.
Alcohol and Drugs on American College Med. 1990;322:95-99. 17. Room R. Survey vs sales data for the US.
Campuses: Use, Consequence, and Percep- 10. Hetzler JE, Burnham A. Alcohol abuse Drink DrugPract Sunr. 1971;3:15-16.
tions of the Campus Environment. Vol 1. and dependency. In: Robins LN, Regier
1989-1991. Carbondale, Ill: Core Institute; DA, eds. Psychiatnic Disorders in America. 18. Celis W. Drinking by college women raises
1993. New York, NY: Free Press; 1991. new concern. New York Times. February
5. Johnston LD, O'Malley PM, Bachman JG. 11. Hosmer DW, Lemeshow S. Applied Logis- 16, 1994:A18.
National Survey Results on Drug Use from tic Regression. New York, NY: John Wiley 19. Ettore E. Women and Substance Use. New
the Monitonng the Future Study, 1975-1993. & Sons; 1989. Brunswick, NJ: Rutgers University Press;
Vol.1. Secondary School Students. Washing- 12. Freier MC, Bell RM, Ellickson PL. Do 1992.
ton, DC: US Government Printing Office; Teens Tell the Truth? The Validity of 20. Rodin J, Ickovics JR. Womens' health:
1994. NIH publication 94-3809. Self-Reported Tobacco Use by Adolescents. review and research agenda as we ap-
6. National Household Survey on Drug Abuse: Santa Monica, Calif: RAND; 1991. RAND proach the 21st century. Am Psychol.
Population Estimates, 1992. Rockville, Md: publication N-3291-CHF. 1990;45:1018-1034.

Tuberculosis Knowledge among New


York City Injection Drug Users

.~ ~ . ;
*. .;~ ~ ~. ~ . ~ ~ .; Hannah Wolfe, MS, Michael Marnor, PhD, Robert Maslansky, MD,
Stuart Nichols, MD, Michael SimberkoffJ MD, Don Des Jarlais, PhD,
and Andrew Moss, PhD, MPH
..:: .........k...
;.::.. ... ad
.e....
:.; ;. . minority communities most affected by
.. itee to 49 New YokCt Introdution tuberculosis about its increasing inci-
In the past decade, New York City dence, and to provide knowledge and
has witnessed a dramatic increase in other resources needed to influence tuber-
* s re'ie-.wof 'a.. .-
': pulmonary tuberculosis. Increases in hu- culosis programs directed toward those
man immunodeficiency virus (HIV) infec- communities.8 In 1994, the CDC added
tion, homelessness, and poverty are re- questions on tuberculosis transmission to
sponsible for much of this resurgence; all
of these are prevalent among New York Hannah Wolfe and Michael Marmor are with
City injection drug users, who are among the Division of Epidemiology and Biostatistics,
. . .tacti.e. i. :.
Mosrdetsjidertood
4ata those at highest risk for tuberculosis.16 New York University Medical Center; Robert
The present study sought to assess tuber- Maslansky is with the Department of Psychia-
try, Bellevue Hospital; Stuart Nichols and Don
culosis-related knowledge in this popula- Des Jarlais are with the Beth Israel Medical
tion. Center; and Michael Simberkoff is with the
Virtually no public health education Manhattan Veterans Affairs Medical Center,
regarding tuberculosis has been done in all in New York, NY. Andrew Moss is with the
.otnoiaceo ..n........ Department of Epidemiology and Biostatistics,
...... h .d.s....
p ...n...
: ;.- Y ~~~~~~~~~~~~~~~~~.

s*::: ; :.
e; Pinedito
......... :. :- -

totbe' fe.r
:--.- -. -.
the past 3 decades. A recent survey of
injection drug users in Brooklyn, NY,
University of California, San Francisco.
Requests for reprints should be sent to
Michael Marmor, PhD, Department of Envi-
edcain4a.0tibtt suggests the presence of a high level of
w.'r\.ith''
and..coAfusio tha inte..,;>;:rfere misinformation and fear about tuberculo- ronmental Medicine, New York University
e.'. rts towconreltubrelom (
..n Medical Center, 341 E 25th St, New York, NY
,
' ..
SiS.7 The Centers for Disease Control and 10010.
Prevention (CDC) recently called for This paper was accepted on October 4,
.~hifuW ,95~ ,8 .88 public awareness campaigns to alert the 1994.
:ul '$95:.VoL.......
8 N"'7

American Journal of Public Health 985

You might also like