The Association of Caffeinated Beverages with Blood Pressure in Adolescents

ARTICLE The Association of Caffeinated Beverages With Blood Pressure in Adolescents

Margaret R. Savoca, PhD; Conner D. Evans, BA; Martha E. Wilson, MA; Gregory A. Harshfield, PhD; David A. Ludwig, PhD

Objective: To assess the association between the con- feine had higher systolic blood pressure readings than the sumption of caffeinated beverages and blood pressure in groups consuming 0 to 50 mg/d (mean difference, 6.0 African American and white adolescents. mm Hg; 95% confidence interval [CI], 2.3 to 9.7) or more than 50 to 100 mg/d (mean difference, 7.1 mm Hg; 95% Design: This study was part of ongoing research exam- CI, 3.4 to 10.7). The effect on diastolic blood pressure was ining stress-induced hemodynamic responses in adoles- less pronounced (P=.08). The diastolic blood pressure of cents. African American and white adolescents (n=159) the group consuming more than 100 mg/d was 3.7 mm selected foods and beverages for a 3-day sodium- Hg (95% CI, 0.41 to 7.0) higher than the group consum- controlled diet. Caffeine in these foods was used to stratify ing more than 50 to 100 mg/d and was not statistically dif- participants into 3 categories (0-50 mg/d, Ͼ50-100 mg/d, ferent from the group consuming 0 to 50 mg/d (mean dif- and Ͼ100 mg/d). Before menu selection, blood pres- ference, 2.4 mm Hg; 95% CI, −0.9 to 5.8). There was no sure readings were obtained. evidence that the association between diastolic blood pressure and caffeine intake varied by race (P=.80). Statistical Analysis: A general linear model (multiple regression with both categorical and continuous vari- Conclusions: For adolescents, especially African Ameri- ables) was developed to assess the effects of race, cat- can adolescents, caffeine intake may increase blood pres- egory of caffeine intake, and interaction of race and sure and thereby increase the risk of hypertension. Al- caffeine intake on systolic and diastolic blood pressure ternatively, caffeinated drink consumption may be a controlling for sex and body mass index (calculated as marker for dietary and lifestyle practices that together in- weight in kilograms divided by height in meters squared). fluence blood pressure. Additional research is needed owing to rising rates of adolescent hypertension and soft Results: The association between systolic blood pres- drink consumption. sure and caffeine category varied by race (P=.001). Af- rican Americans consuming more than 100 mg/d of caf- Arch Pediatr Adolesc Med. 2004;158:473-477

HE PREVALENCE OF HYPER- aged 12 to 17 years drink 1 or more soft tension among youth is ris- drinks daily and that 21% of boys and 22% ing.1 By adolescence, Afri- of girls consume coffee or tea on a daily can American girls and basis.5 The percentage of adolescents boys have higher systolic drinking caffeinated soft drinks is more blood pressure than white individuals.2 To than double the percentage of those who T 6 reduce the risk of hypertension among this consume noncaffeinated soft drinks. The vulnerable group, a better understanding question of whether caffeine in the quan- is needed of the environmental (includ- tities consumed by many adolescents can ing dietary) and genetic factors that con- affect blood pressure has not been exten- tribute to the blood pressure differences sively evaluated. between African American and white adolescents.3 One such dietary factor is For editorial comment caffeine consumption. Caffeine is con- see page 418 sidered a preventable risk factor for hypertension and cardiovascular disease.4 In In a preliminary evaluation of the caf- From the Georgia Prevention adults the primary source of caffeine is cof- feinated beverage consumption of a sample Institute, Department of fee, whereas in adolescents the major of African American adolescents, we found Pediatrics, Medical College source is caffeinated soft drinks. It is es- a higher increase in diastolic blood pres- of Georgia, Augusta. timated that 68% of boys and 62% of girls sure in response to competitive stress

(REPRINTED) ARCH PEDIATR ADOLESC MED/ VOL 158, MAY 2004 WWW.ARCHPEDIATRICS.COM 473

©2004 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 10/01/2021 among the participants who consumed the largest quan- The total amount of foods and beverages selected minus tity of caffeinated drinks.7 The purpose of the present study the amount returned was used to determine subjects’ dietary was to extend our earlier finding and examine the asso- intake during the 3 days. Caffeine content for all foods and beverages on the menu was determined using the University of Min- ciation between the consumption of caffeinated bever- 12 ages and blood pressure in a sample of both African Ameri- nesota nutrient database. The amount of caffeine consumed during the 3 days was calculated for each subject based on the can and white adolescents. amount of caffeine in each food and the quantity of each food consumed by that subject. Examination of the distribution of METHODS caffeine intake showed a skewed distribution due to the large Data for this investigation were obtained in the course of an numbers of subjects consuming less than 50 mg/d of caffeine. ongoing research program to examine hemodynamic re- For this reason, caffeine intake was treated as a categorical vari- sponses to competitive stress in healthy adolescents.8 Subjects able by stratifying subjects into 3 caffeine-intake categories based completed the study protocol between February 2, 2002, and on their mean daily caffeine intake. May 8, 2003. The protocol was approved by the Human As- A general linear model (multiple regression including both surance Committee of the Medical College of Georgia (Augusta). continuous and categorical variables) was used to assess the Subjects were volunteers throughout the public and private high effects of race and category of caffeine intake on systolic and schools in Richmond County and Columbia County, Georgia. diastolic blood pressure and determine if the association be- Subjects were recruited according to their interest in partici- tween caffeine intake and blood pressure varied by race. Sex pating via school announcements, flyers, and handouts to 1 high and body mass index were controlled for in the model. The 2 school (900 students) and through word of mouth from sub- amount of variance explained (R ) by the model and the unique jects who had already participated in similar research projects. contribution of each variable in the model are provided. We Interested subjects contacted research assistants and were used t tests to compute individual pairwise comparisons of least screened by telephone. Inclusion criteria were African Ameri- squares regression means in the model. Differences between can or white ethnicity, aged 15 to 19 years, not taking any medi- least squares regression means were reported using mean dif- cations including contraceptives, and having no food allergies ferences and 95% confidence intervals (CIs). by parental and self-report as well as the ability to meet the dietary and testing requirements of the protocol. Written in- RESULTS formed parental consent and subject assent were obtained from 194 subjects prior to participation. Of these, 35 were excluded The sample (n=159) included 32 African American boys, before testing as a result of exceeding the age requirement between consent and their availability for testing, rescheduling 49 African American girls, 56 white boys, and 22 white owing to conflicts, relocation out of the area, pregnancy as de- girls with a mean±SD age of 16.4±1.0 years. Girls and termined by urine pregnancy testing, illness on test day, fail- boys were similar in regard to age, with a mean differ- ure to comply with the dietary protocol (determined by daily ence of 0.14 years (95% CI, −0.2 to 0.5). African Ameri- overnight urine samples), or withdrawal of parental consent cans were younger than white adolescents, with a mean or subject assent. difference of −0.58 year (95% CI, −0.88 to −0.27). Afri- Measurements, which included systolic and diastolic blood can American girls had a higher body mass index com- pressure readings as well as height and weight, were obtained pared with white girls (25.9 vs 22.4; mean difference, 3.5; during an initial orientation session. Trained research assis- 95% CI, 0.7 to 6.2). African American boys and white tants measured blood pressure levels using a mercury manom- boys had a similar body mass index (23.6 vs 23.4) with eter. Normotensive status and values used in subsequent statistical analyses were determined using the mean of 3 successive a mean difference of 0.19 (95% CI, −2.1 to 2.4). blood pressure readings. Subjects were considered normoten- Participants were stratified into 3 caffeine-intake cat- sive if their mean blood pressure reading was lower than the egories (0-50 mg/d, Ͼ50-100 mg/d, and Ͼ100 mg/d). The 95th percentile based on weight, height, and age norms.9,10 A Table provides a summary by race of the 3 caffeine- digital eye-level scale with a height rod (Detecto 6439; Cardi- intake categories according to sex, age, body mass in- nal Scale Manufacturing Company, Webb City, Mo) was used dex, and caffeine intake. Based on the amount of caf- to measure weight in kilograms and standing height in centi- feine in a 12-oz regular cola (37.2 mg of caffeine), the meters. Body mass index was calculated as weight in kilo- mean caffeine intake in each successive category was grams divided by height in meters squared. Race was based on equivalent to 0.75 can, 2 cans, and 3.9 cans per day. Sev- self-report. enteen subjects (11%) did not consume any soft drinks The subjects received a sodium-controlled diet for 3 days.11 After the research assistants obtained blood pressure readings or iced tea, all of whom were in the lowest intake group. and weight and height measurements, subjects planned their Of these 17, 1 participant had no caffeine, and the oth- meals by selecting foods and beverages from a menu that in- ers had a mean±SD consumption of 5.1±3.9 mg/d from cluded a wide array of breakfast, lunch, dinner, snack, and bev- foods containing chocolate. Of the African American sub- erage items. Of the 15 beverage choices, 6 contained caffeine. jects, 14% were in the highest caffeine-intake category There were also 10 chocolate-containing items with small and 47% were in the lowest intake category. This was in amounts of caffeine. Controlling mean±SD sodium intake to contrast to the white subjects, of whom 46% were in the 4000±200 mg/d was the only selection criterion. The pre- highest group and 26% were in the lowest group. Afri- selected meals were packed into coolers, and subjects or par- can Americans in the highest intake group consumed less ents picked up the coolers from the Georgia Prevention Insti- caffeine than white individuals in the same group, with tute (Department of Pediatrics, Medical College of Georgia) along with a container for an overnight urine collection, which was a mean difference in caffeine intake of −34.4 mg/d (95% used to estimate subjects’ compliance with the diet. Subjects CI, −16.7 to −52.2). were required to return any uneaten food as well as the pack- The model using systolic blood pressure as the de- aging from the foods and beverages that were eaten. The amount pendent variable accounted for 33% (PϽ.001) of the vari- of each food returned was recorded. ance in systolic pressure. There were unique effects for

(REPRINTED) ARCH PEDIATR ADOLESC MED/ VOL 158, MAY 2004 WWW.ARCHPEDIATRICS.COM 474

©2004 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 10/01/2021 Characteristics of African American and White Participants in Each Caffeine-Intake Category*

0-50 mg/d Ͼ50-100 mg/d Ͼ100 mg/d

African American White African American White African American White Variable (n = 38) (n = 20) (n = 32) (n = 22) (n = 11) (n = 36) Girls, % 55.3 20.0 65.6 27.3 63.6 33.3 Caffeine, mg 26.1 ± 15.5 25.0 ± 18.0 74.8 ± 12.7 74.0 ± 13.4 119.2 ± 10.5 153.7 ± 47.6 Age, y 16.2 ± 1.0 16.8 ± 1.0 15.9 ± 1.0 16.8 ± 1.0 16.0 ± 0.6 16.5 ± 1.0 Body mass index, kg/m2 24.0 ± 5.8 22.0 ± 2.9 25.6 ± 6.2 22.1 ± 2.9 26.7 ± 6.1 24.3 ± 6.0 Dietary sodium, mg/d 3023 ± 885.0 3261 ± 773.0 3040 ± 781.0 3514 ± 680.0 3146 ± 949.0 3156 ± 714.0 Systolic blood pressure, mm Hg 105.4 ± 10.5 105.2 ± 6.9 105.5 ± 10.2 103.1 ± 7.6 119.3 ± 9.0 105.6 ± 9.8 Diastolic blood pressure, mm Hg 68.1 ± 7.9 63.3 ± 9.2 67.4 ± 7.5 62.0 ± 7.5 72.7 ± 9.1 65.8 ± 7.9

*Data are presented as mean ± SD unless otherwise indicated.

130 100 African American African American White White 120 90

110 80

100 70

90 60 Systolic Blood Pressure, mm Hg Diastolic Blood Pressure, mm Hg

80 50 0-50 >50-100 >100 0-50 >50-100 >100 Category of Caffeine Intake, mg/d Category of Caffeine Intake, mg/d

Figure 1. Systolic blood pressure by category of caffeine intake in African Figure 2. Diastolic blood pressure by category of caffeine intake in African American and white adolescents. Data points were means derived from least American and white adolescents. Data points were means derived from least squares regression adjusted for sex and body mass index, and error bars squares regression adjusted for sex and body mass index, and error bars represent standard error of the means. represent standard error of the means.

body mass index (R2 =13%; PϽ.001), sex (R2 =7%; sure readings. The mean differences in caffeine intake be- PϽ.001), caffeine-intake group (R2=7%; P=.006), and the tween the highest group and the lower 2 groups were not association between race and caffeine-intake group statistically significant, with differences of −0.5 mm Hg (R2=6%; P=.001). There was no statistically significant (95% CI, −5.1 to 4.2) for those consuming 0 to 50 mg/d effect for race (R2Ͻ0.001; P=.70). For the covariates of and 1.3 mm Hg (95% CI, −3.2 to 5.7) for subjects con- body mass index and sex, there was a positive relation- suming 50 to 100 mg/d. ship between systolic blood pressure and body mass in- The model with diastolic blood pressure as the de- dex. Boys had higher systolic blood pressure, with a mean pendent variable accounted for 23% (PϽ.001) of the vari- difference of 5.8 mm Hg (95% CI, 3.0 to 8.6). The high- ance in diastolic pressure with unique effects for body est caffeine-intake group had a higher systolic blood pres- mass index (R2=11%; PϽ.001), race (R2=3%; P=.04), and sure when compared with the other 2 groups, with dif- caffeine-intake category (R2=3%; P=.08). There was no ferences of 6.0 mm Hg (95% CI, 2.3 to 9.7) for subjects evidence that the association between diastolic pressure consuming 0 to 50 mg/d and 7.1 mm Hg (95% CI, 3.4 to and caffeine-intake category varied by race (R2Ͻ0.001; 10.7) for those consuming more than 50 to 100 mg/d. P=.80). For the covariates of body mass index and sex, The difference in systolic blood pressure between the 2 there was a positive relationship between diastolic blood racial groups was not statistically significant (0.9 mm Hg; pressure and body mass index and similar diastolic blood 95% CI, −5.6 to 3.8). As shown in Figure 1, the effects pressure readings for both sexes (mean difference, 1.77 of race and caffeine intake were best understood by con- mm Hg; 95% CI, −0.8 to 4.3). African Americans had a sidering how systolic blood pressure for each racial group higher diastolic blood pressure, with a mean difference varied by caffeine-intake category. For African Ameri- of 4.4 mm Hg (95% CI, 0.2 to 8.6). There was a differ- cans, the systolic blood pressure was higher in the high- ence in diastolic blood pressure between the highest caf- est caffeine intake group compared with the 2 lower in- feine-intake group and those consuming more than 50 take groups, with differences of 12.5 mm Hg (95% CI, to 100 mg/d (3.7 mm Hg; 95% CI, 0.41 to 7.0), but the 6.8 to 18.2) for those consuming 0 to 50 mg/d and 12.9 mean difference between the highest group and sub- mm Hg (95% CI, 7.1 to 18.2) for subjects consuming more jects consuming 0 to 50 mg/d (2.4 mm Hg; 95% CI, −0.9 than 50 to 100 mg/d. In the case of white subjects, all 3 to 5.8) was not statistically significant. Figure 2 shows caffeine-intake categories had similar systolic blood pres- how diastolic blood pressure readings were similar within

(REPRINTED) ARCH PEDIATR ADOLESC MED/ VOL 158, MAY 2004 WWW.ARCHPEDIATRICS.COM 475

©2004 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 10/01/2021 What This Study Adds amounts for participants who were unable to consume all of the foods and beverages they had selected. We compared the soft drink intake found in this Soft drink consumption represents the major source of study with the levels reported by adolescents in nation- caffeine for adolescents. The association of adults’ caf- wide dietary surveys. Harnack et al14 reported 4 catego- feine consumption from coffee with blood pressure has been widely reported. To our knowledge, no studies have ries of soft drink consumption from the Continuing Sur- explored the relationship between caffeine consump- vey of Food Intakes by Individuals (1994) for a sample tion from soft drinks and blood pressure in adolescents. containing 423 adolescents aged 13 to 18 years. The mean Exploring this question, particularly in African Ameri- soft drink consumption for each category was none, 0.1 cans, is important as the rates of soft drink consump- to 12.9 oz/d, 13.0 to 25.9 oz/d, and 26 oz/d or more. In tion and hypertension continue to rise. our study, the amounts of soft drinks per day in each cat- Data from this observational study indicate that the egory were slightly higher than those reported by Har- choice of large amounts of caffeinated soft drinks by Af- nack and colleagues. Their sample combined intakes from rican American adolescents may be associated with el- adolescents aged 13 to 18 years. Our subjects were aged evated blood pressure readings. These results suggest the 15 to 19 years, possibly reflecting the increased con- need for further research on the direct effect of caffeine sumption pattern of older adolescents and youths living on the blood pressure of African American youths and 15 the high consumption of soft drinks as an indicator of in the Southeast. The Bogalusa Heart Study reported caf- lifestyle behaviors that increase the risk of hypertension. feine intake among children and adolescents in 1988. The mean daily caffeine intakes in our study were similar to levels obtained from the 15- and 17-year-old participants in the Bogalusa Heart Study. Consistent with our each racial group across caffeine-intake categories. For findings, that study reported that white individuals con- African Americans, the differences in mean diastolic blood sumed more caffeine than African Americans. We rec- pressure between the highest caffeine-intake group com- ognize that this was a single evaluation of a small group pared with the other 2 groups were 3.4 mm Hg (95% CI, of adolescents able to participate in a structured dietary −1.8 to 8.5) for those consuming 0 to 50 mg/d and 4.7 protocol. Replication of these findings is needed with a mm Hg (95% CI, −0.5 to 9.9) for subjects consuming more larger sample size of adolescents who are following their than 50 to 100 mg/d. For white participants, the differ- normal dietary patterns. ences in mean diastolic blood pressure between the high- The effects of caffeine consumption on blood pres- est caffeine-intake group compared with the other 2 sure in adults have been widely reported.16 Individuals groups were 1.5 mm Hg (95% CI, −2.7 to 5.7) for those who habitually consumed caffeine experienced eleva- consuming 0 to 50 mg/d and 2.7 mm Hg (95% CI, –1.3 tions in blood pressure throughout the day in response to 6.8) for subjects consuming 50 to 100 mg/d. to a single caffeine dose.17 There is limited information about the effects of caffeine on the blood pressure of ado- COMMENT lescents. Among healthy, nonobese young women (aged 17-22 years), Strickland et al18 examined the effects of 2 In this observational study, we identified a group of ado- levels of caffeine (3 mg and 250 mg) on cardiovascular lescents who chose and subsequently consumed large reactivity in a crossover study of a sample stratified ac- amounts of caffeinated beverages during a 3-day sodium- cording to race (African American and white subjects) controlled diet. The African Americans in the highest caf- and parental history of hypertension. Systolic blood pres- feine-intake category had higher systolic blood pressure sure during stress was 5 mm Hg higher for subjects re- readings than all other adolescents in the study, includ- ceiving the higher dose of caffeine, but there were no ef- ing white participants in the highest caffeine-intake cat- fects for race or parental history of hypertension. egory. There was a modest main effect of caffeine for dia- These results did not demonstrate a dose-response stolic blood pressure that was present when the highest relationship for systolic or diastolic blood pressure. The caffeine-intake group was compared with the middle in- 2 groups of African American adolescents who con- take group. Unlike systolic blood pressure, this effect oc- sumed smaller amounts of caffeine had the same sys- curred across both races. These results are consistent with tolic blood pressure readings, and there was no effect our earlier findings in a different sample of only African across the 3 caffeine-intake categories for white sub- American youth that found a greater change in blood pres- jects. The African American adolescents who selected large sure (diastolic) in response to a competitive stress chal- quantities of soft drinks may have been more sensitive lenge (video game) for subjects who consumed large quan- to the effects of caffeine than the white subjects who con- tities of caffeinated beverages, as assessed by the same sumed even larger amounts of caffeine. Wide individual dietary protocol.7 variations in caffeine metabolism may influence adapta- The dietary protocol, which was designed to control tion to its long-term use and its effects on blood pres- sodium intake, provided a unique opportunity to observe sure.16 African Americans, who have a higher risk of hy- the food choices of these participants. The adolescents were pertension, may be more susceptible to the pressor effects able to choose from a wide array of beverages, so their of caffeine than other populations. An alternative expla- choices likely reflect their preferences for certain foods and nation is that although some African American and white beverages instead of others.13 In this study, we used sub- subjects consumed large amounts of soft drinks, their jects’ menu selections as a surrogate measure for a bever- other lifestyle behaviors such as exercise, smoking, and age preference survey. We were then able to adjust intake other food choices differed. These behavioral differ-

(REPRINTED) ARCH PEDIATR ADOLESC MED/ VOL 158, MAY 2004 WWW.ARCHPEDIATRICS.COM 476

©2004 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 10/01/2021 ences along with soft drink consumption may have con- dynamics in African- and European-American youth. Hypertension. 2003;41: tributed to the higher systolic blood pressure in the Af- 1196-1201. 4. James JE. Is habitual caffeine use a preventable cardiovascular risk factor? Lan- rican American adolescents. cet. 1997;349:279-281. Our findings demonstrate an association between 5. US Department of Agriculture, Agricultural Research Service. Food and Nutrient systolic blood pressure and caffeine intake in African Intakes by Children, 1994-1996, 1998, Table Set 17. Washington, DC: ARS Food American adolescents. These subjects did not have hyper- Surveys Research Group. Available at: http://www.barc.usda.gov/bhnrc/foodsurvey /pdf/scs_all.pdf. Accessed July 21, 2003. tension, so it is not known if caffeine is related to the de- 6. Smiciklas-Wright H, Mitchell DC, Mickle SJ, Cook AJ, Goldman JD. Foods com- velopment of hypertension or to blood pressure levels in monly eaten in the United States: quantities consumed per eating occasion and adolescents who have hypertension. For adolescents who in a day, 1994-1996: US Department of Agriculture NFS Report No. 96-5. Janu- consume large quantities of caffeinated soft drinks, 2 ques- ary 2002. Available at: http://www.barc.usda.gov/bhnrc/foodsurvey/pdf/Portion tions should be addressed. First, does caffeine have a di- .pdf. Accessed July 21, 2003. 7. Savoca MR, Evans CD, Wilson ME, Ludwig DA, Harshfield GA. Caffeine is rect effect on blood pressure control for adolescents who related to stress-induced changes in diastolic blood pressure among African consume amounts of caffeine similar to levels consumed American adolescents [International Society for Behavioral Nutrition and by adults? Second, does high soft drink consumption in- Physical Activity Web site]. Available at: http://www.isbnpa.org/Meeting_2003 dicate an array of dietary and lifestyle practices that to- /Saturday-%20oral%20abstracts.pdf. Accessed July 21, 2003. 19 8. Harshfield GA, Wilson ME, Hanevold C, et al. Impaired stress-induced pressure gether increase the risk of developing hypertension? Fur- natriuresis increases cardiovascular load in African American youths. Am J Hy- ther research is needed to separate the direct effect of caffeine pertens. 2002;15(suppl 10, pt 1):903-906. on blood pressure from soft drink consumption as 1 of sev- 9. Update on the 1987 Task Force Report on High Blood Pressure in Children and eral lifestyle behaviors associated with the risk of hyper- Adolescents: a working group report from the National High Blood Pressure Edu- tension among young African Americans. cation Program. Pediatrics. 1996;98(suppl 4, pt 1):649-658. 10. Kavey RE, Daniels SR, Lauer RM, Atkins DL, Hayman LL, Taubert K. American Heart Association guidelines for primary prevention of atherosclerotic cardio- Accepted for publication December 30, 2003. vascular disease beginning in childhood. J Pediatr. 2003;142:368-372. This study was supported by grants HL-59954 and HL- 11. Wilson ME, Baxter S. The successful implementation of a protcol to normalize so- 64225 from the National Heart, Lung, and Blood Institute, dium intake in African-American youths [abstract]. Am J Hypertens. 2001;14:241. 12. Food and nutrient database 33, version 4.06_34. St Paul, Minn: University of Min- National Institutes of Health, Bethesda, Md. nesota; 2002. Updated July 2003. Corresponding author and reprints: Margaret R. 13. Drewnowski A, Hann C. Food preferences and reported frequencies of food con- Savoca, PhD, Georgia Prevention Institute, Department of sumption as predictors of current diet in young women. Am J Clin Nutr. 1999; Pediatrics, Medical College of Georgia, 1499 Walton Way, 70:28-36. 14. Harnack L, Stang J, Story M. Soft drink consumption among US children and HS 1640, Augusta, GA 30912-3715 (e-mail: msavoca adolescents: nutritional consequences. J Am Diet Assoc. 1999;99:436-441. @mail.mcg.edu). 15. Arbeit ML, Nicklas TA, Frank GC, Webber LS, Miner MH, Berenson GS. Caffeine intakes of children from a biracial population: the Bogalusa Heart Study. JAm REFERENCES Diet Assoc. 1988;88:466-471. 16. Nurminen ML, Niittynen L, Korpela R, Vapaatalo H. Coffee, caffeine and blood pressure: a critical review. Eur J Clin Nutr. 1999;53:831-839. 1. Sorof JM. Prevalence and consequence of systolic hypertension in children. Am 17. Lane JD, Pieper CF, Phillips-Bute BG, Bryant JE, Kuhn CM. Caffeine affects car- J Hypertens. 2002;15(suppl 2, pt 2):57S-60S. diovascular and neuroendocrine activation at work and home. Psychosom Med. 2. Dekkers JC, Snieder H, Van Den Oord EJ, Treiber FA. Moderators of blood pres- 2002;64:595-603. sure development from childhood to adulthood: a 10-year longitudinal study. 18. Strickland TL, Myers HF, Lahey BB. Cardiovascular reactivity with caffeine and stress J Pediatr. 2002;141:770-779. in black and white normotensive females. Psychosom Med. 1989;51:381-389. 3. Snieder H, Harshfield GA, Treiber FA. Heritability of blood pressure and hemo- 19. Fishbein L. Causes of obesity. Lancet. 2001;357:1977-1978.

(REPRINTED) ARCH PEDIATR ADOLESC MED/ VOL 158, MAY 2004 WWW.ARCHPEDIATRICS.COM 477