ARTICLE

Teenaged Girls, Carbonated Beverage Consumption,

Objective: To determine the possible association be- Results: In the total sample, carbonated beverage con-
tween carbonated beverage consumption and bone frac- sumption and bone fractures are associated: odds ra-
tures among teenaged girls given the awareness of the tio = 3.14 (95% confidence limit, 1.45, 6.78), P = .004.
concern about the impact of carbonated beverage con- Among physically active girls, the cola beverages, in par-
sumption on children’s health. ticular, are highly associated with bone fractures: odds
ratio=4.94 (95% confidence limit, 1.79, 13.62), P=.002.
Setting: An urban high school.
Conclusions: The results reported confirm previous find-
Methods: A cross-sectional (retrospective) study. ings, but the mechanism by which cola drinks are asso-
Four hundred sixty 9th- and 10th-grade girls attend- ciated with bone fractures in physically active girls has
ing the high school participated in this study by com- neither been fully explored nor determined. Neverthe-
pleting a self-administered questionnaire relating to less, national concern and alarm about the health im-
their physical activities and personal and behavioral pact of carbonated beverage consumption on teenaged
practices. The school system and the Harvard School girls is supported by the findings of this study. The re-
of Public Health Institutional Review Boards approved sults have policy implications for improving the dietary
the study. The girls’ self-reports on physical activity, practices and health of children.
carbonated beverage consumption, and bone fractures
are analyzed. Arch Pediatr Adolesc Med. 2000;154:610-613

A
LARMED BY the fact that cal activity program, the regular high
“teens drink breathtaking school sports’ teams, or both programs.
quantities of soda,”1 the With respect to activity level, 10.9% re-
Center for Science in the ported themselves as being “inactive”;
Public Interest in a letter to 22.0% as engaging in “light” activity;
the Secretary of Health and Human Ser- 28.7%, in “moderate” activity; and 17.2%,
vices urged that department to commis- in “high-level” activity; and 18.5%, in “vig-
sion a study on the health impact of soda orous” activity; 2.8% did not specify ac-
pop. This article herein confirms an earlier tivity level participation. Nearly 80% drink
finding of an association between carbon- carbonated beverages at present; 49.8%,
ated beverage consumption and bone frac- cola beverages only, 11.5%, noncola bev-
tures among teenaged girls, and particu- erages only; and 15.0%, both cola and non-
larly the association of cola drinks and bone cola beverages. Thus, nearly two thirds of
fractures among physically active teen- the 460 high school girls consumed cola
aged girls.2
For editorial comment
RESULTS see page 542
The mean (SD) age of the girls was 15 years drinks. Of those who consumed carbon-
From the Department of
Psychiatry, Harvard Medical
8 months (10 months). One girl was in the ated beverages, most drank regular (sugar)
School, and Departments of 11th grade; 2, in the 12th grade; and the in contrast to diet drinks; only 20% of the
Biostatistics and Population and rest were either ninth or 10th graders. Of participants drank diet drinks, either diet
International Health, Harvard the total sample of 460, 55.65% did not cola or other diet drinks. One fifth of the
School of Public Health, Boston, participate in any sports activity; the rest girls reported a history of a bone fracture
Mass (Dr Wyshak). participated in either the special physi- at any age; 9 girls reported having had frac-

(REPRINTED) ARCH PEDIATR ADOLESC MED/ VOL 154, JUNE 2000 WWW.ARCHPEDIATRICS.COM
610
Downloaded from www.archpediatrics.com on April 17, 2009
©2000 American Medical Association. All rights reserved.
 SUBJECTS AND METHODS and carbonated beverage consumption, behavioral prac-
tices were asked about. The areas of inquiry were as fol-
lows: (1) current height and weight; satisfaction with pres-
SETTING ent body weight, weight gained or lost (if .2.3 kg) in the
past year; being on a special diet for weight loss or gain,
The study was conducted in an urban high school. Per- vegetarian concerns, low cholesterol, low salt, low fat, or
mission to carry out the study was granted by the school other special dietary reasons; (2) smoking history; (3)
system’s authorities and also approved by the Harvard School medical history, including regular care by a physician for
of Public Health’s Human Subjects Committee, Boston, Mass. medical conditions, whether they had heart disease or dia-
Data were obtained in the course of a project designed betes, whether they ever had an operation, and medica-
to reduce the incidence of teenaged pregnancy by enrolling tions taken; (4) personal matters in life, eg, sexual inter-
girls attending an urban high school in group physical ac- course, pregnancy; (5) menstrual history, ie, age at
tivity programs. Permission from the school was obtained to menarche, regularity of periods; (6) current life (related to
distribute a questionnaire to 9th- and 10th-grade girls dur- self-esteem and self-efficacy); (7) work history outside of
ing the school day; approval was granted by the Human Sub- school or the home; and (8) thoughts about the future, ie,
jects Committees of the school authorities and of the Har- college, job, any plans.
vard School of Public Health. Four hundred sixty teenaged
girls completed the questionnaire. STATISTICAL METHODS
The questionnaires were self-administered in a class-
room setting and monitored by classroom personnel. Since Statistical analyses were done on a personal computer
the responses were confidential, they were not examined using SAS Statistical software procedures that included
by school personnel. However the girls were instructed about descriptive statistics and logistic regression analysis.3
the need to answer as best they could and to provide com- Participation in any physical activity, either the spe-
plete answers. The information provided had internal con- cial program or a high school athletic team, was dichoto-
sistency and face validity, although as in any question- mized as high level or vigorous, coded 1/0. (Table 1).
naire, a few questions were unanswered or incompletely Carbonated beverage consumption variables were
answered. dichotomized as yes or no and coded as 1/0. Bone frac-
The questions relevant to this article relate to the tures were dichotomized as yes or no and coded as 1/0.
intensity of the girl’s activity level, based on a 5-point Table 2 gives the results for all fractures. Analyses were
scale: inactive, light, moderate, high-level, and vigorous; also done in which bone fractures that happened when
their medical history; and their carbonated beverage the girl was younger than 8 years old were excluded
consumption. In particular, participants were asked because it was felt fractures that early in life were not
whether they had ever been told by a physician that they likely to be associated with carbonated beverage con-
had a bone fracture, and if so, the type or site of the frac- sumption. Table 3 gives the results when fractures that
ture and the year of its occurrence. Questions on car- occurred before the age of 8 were coded as 0 in the
bonated beverage consumption were: “Do you drink car- analyses relating carbonated beverage consumption to
bonated beverages at present? yes/no”; “What type(s) of bone fractures. Results based on analyses of the data on
these beverages do you most commonly consume, ie, all fractures irrespective of time of occurrence did not
diet cola, cola with sugar, other diet carbonated drinks, differ significantly from the analyses that excluded those
other carbonated drinks with sugar?” that occurred when the girl was younger than 8 years.
In addition to questions about physical activity, bone Therefore, the results in Table 3 will not be discussed
fractures and other questions relating to medical history, further.

tures when younger than 8 years old. The activity level OR=4.94 (95% CL, 1.79, 13.62), P =.002, while for less
was associated with neither the drinking of carbonated active girls, the OR=1.16 (96% CL, 0.60, 2.24), P=.66.
beverages nor the type of beverage they consumed, ie, Table 2, C, gives the ORs for 3 comparisons of non-
cola or noncola. Similarly, participation in either the carbonated beverages with noncola beverages only; cola
special program or the high school teams activities beverages only; and both cola and noncola beverages. For
was not associated with drinking carbonated beverages active girls, the ORs are, respectively, for consumers of
(Table 1). noncola beverages only, 0.43(95% CL, 0.05, 4.22), P=.47;
Table 2 gives the results for the association of car- cola beverages only, 2.83 (95% CL, 0.87, 9.23), P=.09;
bonated beverage consumption and bone fractures among and both cola and noncola beverages, = 7.00 (95% CL,
all girls, active girls, and less active girls, controlling for 2.00, 24.46), P =.002. The x21 trend is statistically sig-
participation in physical activity programs and for the nificant, (P =.001), indicating a dose-response relation-
level of activity, ie, high or low. For all girls, the odds ship. For less active girls the results are marginally sig-
ratio (OR) for the association of bone fractures and drink- nificant for cola drinkers; but, for the noncola drinkers:
ing carbonated beverages is 3.14 (95% confidence limit OR=4.27 (95% CL, 1.40, 13.07), P=.01.
[CL], .45, 6.78), P = .004. In summary, from Table 2, B and C, we observed
Table 2, B, gives the results comparing the group that for less active girls, the association between carbon-
who drank either no carbonated beverages or noncola bev- ated cola beverages and bone fractures is marginal; how-
erages with the group who drank either cola drinks only ever, for active girls who consume both cola and non-
or both cola and noncola drinks. For the active girls the cola drinks, the risk of bone fractures was highest.

(REPRINTED) ARCH PEDIATR ADOLESC MED/ VOL 154, JUNE 2000 WWW.ARCHPEDIATRICS.COM
611
Downloaded from www.archpediatrics.com on April 17, 2009
©2000 American Medical Association. All rights reserved.
 COMMENT supplementation have been found to increase bone min-
eral density.5-7 Osteoporotic fractures may be affected by
One health impact of carbonated beverage consump- diet and activity among young women.8 Adolescence could
tion among teenaged girls, the risk of bone fractures, is be a critical period for bone mass formation; adolescent
reported herein. Among all the girls, the OR for the as- females at the time of puberty may be the optimal popu-
sociation between carbonated beverages and bone frac- lation for early prevention of osteoporosis with calcium
tures is 3.14 (95% CL, 1.45, 6.78), P=.004. Among physi- intake.9,10 Laboratory investigators have reported pos-
cally active girls, the risk was highest for those who sible bone resorption from high levels of phosphorus
consumed both cola and noncola beverages, as com- intake.11,12 Others have found that there may be a del-
pared with those who did not drink carbonated bever- eterious effect on bone owing to the change in the calcium-
ages, the OR being 7.00 (95%CL, 2.00, 24.45), P =.002. phosphorus ratio in the diet.13,14 The association ob-
This association between cola drinks and bone fractures served may be because of the phosphoric acid content
among physically active adolescents confirms results from in cola drinks.2,15 As long ago as the 19th century, an in-
previous work.2 crease in bone fractures from occupational exposure to
Teens have doubled or tripled their consumption of phosphorus was reported.16,17 Research to elucidate and
soft drinks and they have cut their consumption of milk understand the biological mechanism by which cola drinks
by more than 40%.1 Low bone mineral density is due to may lead to bone fractures among physically active girls
genetic, hormonal, or environmental factors, eg, diet may is needed.
be causally related to fractures.4 Calcium and calcium Our present results confirm earlier findings,2 but the
study has some of the following possible limitations. The
design is cross-sectional and causality cannot be in-
Table 1. Characteristics of Participants ferred from the data; causality could be inferred from a
in Teenaged Girl Survey* longitudinal study. Further, a study design with min-
eral density measurements would have been preferable
All Girls Active Less Active for the assessment of the relationship between carbon-
Variable (N = 460) (n = 164) (n = 296) ated beverage consumption and fractures; such measure-
Activity level ments were not part of the study design. The question-
Inactive 50 (10.9) ... 50 (16.9) naire did not include questions on calcium consumption,
Light activity 101 (22.0) ... 101 (34.1) so the effect of calcium consumption could not be as-
Moderate activity 132 (28.7) ... 132 (44.6)
High-level activity 79 (17.2) 79 (48.2) ...
sessed. The teenaged subjects studied provided informa-
Vigorous activity 85 (18.5) 85 (51.8) tion on whether they consumed carbonated beverages and
Not specified 13 (2.8) ... 13 (4.4) the types of these beverages, ie, cola or noncola, regular
Participants in or diet, but not on the amount. No difference between
No organized activity 256 (55.6) 44 (26.8) 212 (71.6) regular and diet drinks was found; but no questions were
Special program only 39 (8.5) 11 (6.7) 28 (9.5) asked regarding caffeinated vs noncaffeinated drinks.
High school sports only 102 (22.2) 66 (40.3) 36 (12.1)
Both special program 63 (13.7) 43 (26.2) 20 (6.8)
The data were self-reports of adolescent girls via self-
and high school team administered questionnaires. The girls were informed in
Drink carbonated beverages their classrooms of the study’s importance, and they were
None 96 (20.9) 34 (20.7) 62 (20.9) asked to respond as accurately and as completely as they
Noncolas 53 (11.5) 18 (11.0) 35 (11.8) could. On the whole, the responses were complete, in-
Colas 229 (49.8) 74 (45.1) 155 (52.4) ternally consistent, and had face validity. The results ob-
Both noncolas and colas 69 (15.0) 35 (21.3) 34 (11.5)
Yes, type not specified 13 (2.8) 3 (1.8) 10 (3.4)
tained in this study confirm the relationship between cola
Bone fractures consumption and bone fractures among physically ac-
Yes 90 (19.6) 49 (25.0) 41 (16.5) tive girls found in a smaller sample of adolescent girls.2
In the earlier study, calcium intake and the calcium-
*Data are given as number (percentage). Ellipses indicate not applicable. phosphorus ratio were assessed and controlled for in the

Table 2. Odds Ratios for the Association of Carbonated Beverage Consumption and Bone Fractures in Teenaged Girls*

All Girls, Odds Ratio Active, Odds Ratio Less Active, Odds Ratio P
(95% CL) P (95% CL) P (95% CL) Value
A: Drink carbonated beverages
Some/None 3.14 (1.45, 6.78) .004 2.78 (0.90, 8.62) .08 3.28 (1.13, 9.54) .03
B: Drink colas
Some/None 2.01 (1.17, 3.43) .011 4.94 (1.79, 13.62) .002 1.16 (0.60, 2.24) .66
C: Drink
None 1.00 (. . .) 1.00 (. . .) 1.00 (. . .)
Noncolas 2.48 (0.97, 6.34) .06 0.43 (0.05, 4.22) .47 4.27 (1.40, 13.07) .01
Colas 2.70 (1.30, 5.60) .008 2.83 (0.87, 9.23) .08 2.46 (0.97, 6.24) .06
Both 3.68 (1.58, 8.53) .002 7.00 (2.00, 24.45) .002 1.31 (0.34, 5.04) .69

*Note: for all girls, adjusted for the sports program and activity level. For less active and active girls adjusted only for the sports program. CL indicates
confidence limit; ellipses, not applicable.

(REPRINTED) ARCH PEDIATR ADOLESC MED/ VOL 154, JUNE 2000 WWW.ARCHPEDIATRICS.COM
612
Downloaded from www.archpediatrics.com on April 17, 2009
©2000 American Medical Association. All rights reserved.
 Table 3. Odds Ratios for the Association of Carbonated Beverage Consumption and Bone Fractures
at Age 8 Years or Older in Teenaged Girls*

All Girls, Odds Ratio P Active, Odds Ratio P Less Active, Odds Ratio P
(95% CL) Value (95% CL) Value (95% CL) Value
A: Drink carbonated beverages
Some/None 3.16 (1.39, 7.16) .006 3.44 (0.97, 12.20) .06 2.75 (0.94, 8.05) .07
B: Drink colas
Some/None 2.02 (1.12, 3.62) .02 5.28 (1.74, 16.01) .003 1.07 (0.53, 2.19) .08
C: Drink
None 1.00 1.00 1.00
Noncolas 2.84 (1.00, 8.04) .05 0.52 (0.05, 5.45) .58 4.85 (1.36, 17.31) .02
Colas 3.13 (1.34, 7.28) .008 3.12 (0.83, 11.72) .09 2.80 (0.93, 8.45) .07
Both 3.71 (1.43, 9.66) .002 7.49 (1.88, 29.76) .004 0.80 (0.14, 4.63) .80

*Fractures that occurred when girls were younger than age 8 years are coded 0. For all girls, adjusted for the sports program and activity level. For less active
and active girls, adjusted only for the sports program.

analyses.2 Despite our study’s limitations, the results re- 3. SAS Statistical Analysis System. Cary, NC. Statistical Institute; 1987.
4. Goulding A, Cannon R, Williams SM, Gold EJ, Tayor RW, Lewis-Barned NJ. Bone
ported herein warrant attention and confirmation in lon-
mineral density in girls with forearm fractures. J Bone Miner Res. 1998;13:143-
gitudinal and biomedical studies. 148.
In conclusion, nationally, there is great concern 5. Johnston CC Jr, Miller JZ, Slemenda CW, et al. Calcium supplementation and
about the effects of carbonated beverage consumption increases in bone mineral density in children. N Engl J Med. 1992;327:82-87.
on obesity, tooth decay, osteoporosis, and other health 6. Lloyd T, Andon MB, Rollings N, et al. Calcium supplementation and bone min-
eral density in adolescent girls. JAMA. 1993; 270:841-844.
problems.1 Concern about the health impact of carbon-
7. Ruiz JC, Mandel C, Garabedian M. Influence of spontaneous calcium intake and
ated beverage consumption, in particular, the associa- physical exercise on the vertebral femoral bone mineral density of children and
tion with bone fractures in adolescent girls, is validated adolescents. J Bone Miner Res. 1995;10:675-682.
by our findings. Our findings have implications both for 8. Kanders B, Dempster DW, Lindsay R. Interaction of calcium nutrition and
the health of teenagers and for the health of women at physical activity on bone mass in young women. J Bone Miner Res. 1988;2:
145-149.
later ages. In addition to research, health and social poli- 9. Matkovic V, Fontana D, Tominac C, et al. Factors that influence peak bone mass
cies to promote better behavioral practices, including formation: a study of calcium balance and the inheritance of bone mass in ado-
one’s diet, among American women of all ages are lescents. Am J Clin Nutr. 1990;52:878-88.
needed. 10. Matkovic V. Skeletal development and bone turnover revisited [editorial]. J Clin
Endocrinol Metab. 1996;81:2013-2016.
Accepted for publication November 10, 1999. 11. Draper HH, Scythes CA. Calcium phosphorus and osteoporosis. Fed Proc. 1981;
I thank Rose Frisch, PhD, who was responsible for the 40:2434-2438.
12. Portale AA, Halloran BP, Murphy MM, Morris RC Jr. Oral intake of phosphorus
administration of the project and for its development. can determine the serum concentration of 1,25-dihydroxyvitamin D by deter-
Reprints: Grace Wyshak, PhD, Department of Popula- mining its production rate in humans. J Clin Invest. 1986;77:7-12.
tion and International Health, Harvard School of Public Health, 13. Jowsey J. Osteoporosis, its nature and the role of diet. Postgrad Med. 1976;60:
665 Huntington Ave, Boston, MA 02115 (e-mail: wyshak 75-79.
14. Lutwak L. Current concepts in bone metabolism. Ann Intern Med. 1974;80:630-
@hsph.harvard.edu).
644.
15. Wyshak G, Frisch RE, Albright TE, Albright NL, Schiff I, Witschi J. Nonalcoholic
REFERENCES carbonated beverage consumption among women former college athletes. J Or-
thopaedic Res. 1989;7:91-99.
1. Jacobson MF. Liquid Candy: How Soft Drinks Are Harming Americans’ Health. 16. Dearden WF. Fragilitas ossium amongst workers in Lucifer match factories. BMJ.
Washington, DC: Center for Sciences in the Public Interest; 1998. 1899; July29:270-271.
2. Wyshak G, Frisch RE. Carbonated beverages, dietary calcium, the dietary calcium/ 17. Magitot M. Des accidents du phosphore et en particuliere du phosphorisme. Bull
phosphorus ratio, and bone fractures in girls and boys. J Adolesc Health. 1994; Acad Med (Paris). Vol 1895;XXXIII(series 3, March 12th session):267-289, 321-
15:210-215. 327.

(REPRINTED) ARCH PEDIATR ADOLESC MED/ VOL 154, JUNE 2000 WWW.ARCHPEDIATRICS.COM
613
Downloaded from www.archpediatrics.com on April 17, 2009
©2000 American Medical Association. All rights reserved.