School-based BMI screening and parent notification: A statewide natural experiment

doi:10.1001/archpediatrics.2011.127

Arch Pediatr Adolesc Med. Author manuscript; available in PMC 2012 November 1.

Published in final edited form as:

Arch Pediatr Adolesc Med. 2011 November; 165(11): 987–992.

Published online 2011 July 4. doi: 10.1001/archpediatrics.2011.127

PMCID: PMC3206196

NIHMSID: NIHMS311087

School-based BMI screening and parent notification: A statewide natural experiment

Kristine A. Madsen, MD, MPH

Kristine A. Madsen, University of California, San Francisco, Dept of Pediatrics, 3333 California Street, Box 0503, San Francisco, CA 94118, Office: 415-514-2445, Fax: 415-476-6106;

Kristine A. Madsen: madsenk/at/peds.ucsf.edu

The publisher's final edited version of this article is available at Arch Pediatr Adolesc Med

See other articles in PMC that cite the published article.

Abstract

Objective

School districts nationwide are adopting school-based body-mass index (BMI) screening to address the pediatric obesity epidemic. The effect of school-based BMI screening and parental notification on pediatric obesity, however, remains unknown. We sought to assess the impact of BMI screening with parental notification on weight status for California public school students.

Design

A natural experiment wherein nearly all California school districts conducted annual BMI screening in 5^th, 7^th, and 9^th grade, but parental notification of BMI screening results was optional.

Setting

Data from mandatory fitness testing in California public schools between 2001 and 2008.

Participants

Participants were 6,967,120 5^th, 7^th, and 9^th grade youth (73% of enrolled).

Main outcome measure

BMI z-score was the main outcome in adjusted mixed effects linear regression models, assessing whether notifying parents of their child’s BMI in a given year predicted BMI z-score 2 years hence.

Results

Rates of parental notification of BMI screening results increased from 35% in 2001 to 52% in 2008. BMI notification in 5^th and/or 7^th grade had no impact on subsequent BMI z-scores (95% CI -0.03, 0.01) compared to no notification. No differences in the impact of notification were seen by race/ethnicity. Results did not vary with sensitivity analyses.

Conclusions

These findings suggest that while BMI screening itself could have benefits, parental notification in its current form may not reduce pediatric obesity. Until effective methods of notification are identified, schools should consider directing resources to policies and programs proven to improve student health.

INTRODUCTION

The broad scope of the pediatric obesity epidemic^{1,2, 3} calls for public health approaches.⁴ School-based strategies have the potential to reach the vast majority of youth and may address increasing racial and ethnic disparities in obesity.⁵ The Institute of Medicine (IOM) recommends school-based body-mass index (BMI) screening with parental notification of results as a public health means of reducing pediatric obesity.⁶ Theoretically, notifying parents of BMI screening results can inform parents that their child is at risk, thereby allowing parents to take effective action to improve their child’s weight status. However, the Center for Disease Control and Prevention, the American Heart Association, and US Preventive Services Task Force do not include school-based BMI screening in their recommendations, citing a lack of evidence to support its effect.^7–9

Despite the lack of evidence, as of 2006, 41% of school districts nationwide mandated height and weight assessments, and 72% of those districts required reporting results to parents.¹⁰ The state of California was an early adopter of BMI screening. For the last decade, almost all California public schools have collected BMI data annually on 5^th, 7^th, and 9^th grade students. However, notifying parents of BMI screening results remains optional. Mandated screening with optional notification creates a natural experiment through which to examine the effect of the now widespread use of parental notification. We used California’s natural experiment to determine if notifying parents of school-based BMI screening results reduces obesity at the population level and if notification has the potential to reduce health disparities.

METHODS

Study Population

We studied 5^th, 7^th, and 9^th grade youth who underwent mandatory fitness testing in California public schools between 2001 and 2008. The Committee on Human Research at the University of California, San Francisco certified this study as exempt.

Fitness Assessment Data

Pursuant to California Education code,¹¹ California public schools assessed the fitness status of 5^th, 7^th, and 9^th graders each spring in 1999 and from 2001–2008 using the FITNESSGRAMR.¹² The FITNESSGRAM, a battery of six fitness assessments including body composition, is widely used nationally; of 32 states that either require or recommend BMI or fitness assessment in schools, a majority (56%) recommend use of the FITNESSGRAM.¹³ The California Department of Education (CDE) allows schools to use BMI, skin-fold measures or bioelectrical impedance to estimate body composition; over 95% of California schools use BMI (personal communication from CDE). School districts are required to send student-level data to the CDE and may contract with outside vendors to obtain student-level reports of FITNESSGRAM performance to provide to parents.¹⁴

De-identified student-level FITNESSGRAM data were available from the CDE for the years 2001–2008. Individual identifiers were not available and data from an individual student could not be linked across years. Data records include student grade, sex, age in years, height to the nearest inch, weight to the nearest pound, and race/ethnicity (African American, American Indian/Alaskan Native, Asian, Filipino, Hispanic/Latino, Pacific Islander, or White - Not of Hispanic Origin). To protect student confidentiality, the CDE included Filipino and Pacific Islander students in the Asian category. The CDE redacted data for students who, based on sex, grade, and race, were among a group of 10 or fewer students in their school district. BMI z-scores were calculated using the CDC’s SAS program.¹⁵

District-Level Data

We obtained data directly from the CDE website¹⁶ for total enrollment; percent of students eligible for free or reduced-price meals; urban level (based on an 8-point scale from the Census Bureau, ranging from a large city to a rural area); and an index of retention (the percentage of students enrolled in the district continuously from October 2007 until standardized testing in spring 2008).

Interviews with Districts

We used structured telephone interviews to determine if, when and how districts notified parents of student fitness assessment results. Researchers called the main phone number for all 438 K-8 and 376 K-12 school districts and asked to speak to the person with the greatest knowledge of physical fitness test procedures. The identified individual was asked if current district policy was to notify parents of FITNESSGRAM results and, if so, in what year notification began, and by what method parents were notified. Interviewees who stated that their district did not currently notify parents of results were asked if that policy had ever changed. All interviewees were asked if they thought any schools within the district deviated from the district-level policy regarding parent notification. Interviewees in 17 districts reported not knowing the answer to one or more questions and either referred researchers to another individual or sought information from another district employee in order to provide complete information.

Statistical Analyses

We used linear regression, adjusted for year and clustering by district code, to identify predictors of missing data (Table 1) and the Wilcoxon Rank Sum test to identify predictors of notification status in 2001 and 2008 (Table 2). To determine if parent notification when a child is in 5^th grade predicts BMI z-score in 7^th grade, and if notification in 7^th grade predicts BMI z-score in 9^th grade, we used a mixed effects linear regression model with a random effect for district (to account for repeated measures within districts over time and clustering of students within districts) and a random intercept. The primary outcome was student-level BMI z-score and the primary predictor was district-level prior notification, a time-varying binary indicator of a district’s notification status two years prior. Prior notification was coded as 0 for all 5^th graders; 7^th and 9^th grade students were assigned prior notification values based on their current district’s prior notification policy. The full model included the following student-level factors: grade (5^th grade as reference), sex, race/ethnicity (non-Hispanic white as reference), and cohort year (to account for temporal trends in obesity - categorical with students in 5^th grade in 2001 as reference group). The model was adjusted for district-level factors: proportion of students eligible for free or reduced-price meals, and urban level (categorical with large city as reference). To determine if parent notification might reduce health disparities, we included an interaction term in the full model for prior notification*race/ethnicity. Statistical analyses were done in SAS version 9.2 (SAS Institute Inc., Inc., Cary, NC, USA).

TABLE 1

Enrollment and BMI record data

Table 2

Differences in district-level factors by notification status in 2001 and 2008

Sensitivity analyses

To isolate the impact of timing of notification, we created a categorical term to compare no prior notification to notification in 5^th only, notification in 7^th only, and notification in both 5^th and 7^th. Additionally, we reran the full model restricting data to: 1) 380 districts with retention ≥95% (included 64% of enrolled students); 2) 700 districts with a retention index >87% (included 98.3% of enrolled students); 3) districts with enrollment ≤75,000 (excluded 4 large districts); 4) the years 2004–2008 (to account for potential lack of institutional memory for earliest years of FITNESSGRAM administration). Finally, we reran the full model excluding 106 districts reporting that individual schools might deviate from the districts’ policy, or reporting that the district did not have a policy requiring notification and left the decision to individual schools.

RESULTS

Valid BMI data were available for 6,967,120 students, representing 72.7% of all 5^th, 7^th, and 9^th graders enrolled over the years 2001–2008 (Table 1). The number of districts with no FITNESSGRAM data records available declined from 70 in 2001 to 29 in 2008 (representing 4.5% and 0.2% of enrolled students, respectively). There were small but significant district-level correlations between the amount of missing data and the proportion of students eligible for free-and-reduced price meals and the mean student BMI z-score (Table 1).

Researchers conducted telephone interviews with 429 (98%) of 438 K-8 districts and 355 (94%) of 376 K-12 school districts. Districts that did not respond to interview requests were not different in enrollment, proportion of students eligible for free meals, or mean BMI z-score in 2008 from districts interviewed. The number of districts notifying parents of BMI screening results increased from 261 (35%) in 2001 to 386 (51%) in 2008 (proportion notifying in 2008 was 53% when 26 districts that did not know notification status prior to 2008 were included). Districts that notified were larger, had fewer missing FITNESSGRAM data, lower mean student BMI z-score (in 2001 but not in 2008), slightly more African-American and Asian students, and were more urban than districts that did not notify (Table 2). Associations between missing data and free meals or BMI z-score (Table 1) did not vary by notification status (p>0.186 for all interaction terms). To notify parents of FITNESSGRAM results, most districts sent a letter either via U.S. mail (70% of those notifying) or with students to bring home to their parents (19%). Of all districts that ever notified, only 12 stopped doing so.

In the mixed effects model, prior parental notification had no impact on subsequent BMI z-scores (effect size −0.01 BMI z-score units; 95% CI -0.03, 0.01) after adjusting for grade, sex, race/ethnicity, urban level, and free or reduced-price meal eligibility (Table 3). In a similarly adjusted model with type of notification as a categorical predictor, neither mailing a letter nor sending a letter home with the child significantly affected BMI z-score, nor were the effects different from each other in a pair-wise comparison. No differences in the impact of notification were seen by race/ethnicity (p value for interaction term = 0.64). To determine if the proportion of obese students in a district modified the effect of BMI reporting, we included an interaction term between prior notification and district-level quartiles for the proportion of obese students (mean proportion of students with a BMI ≥95^th percentile across all data years). While there was significant interaction between prior notification and quartile of proportion of obese students (p<0.001), the effect was not significant within any quartile, nor was there a step-wise effect with increasing proportion of obese students (Table 4). No other sensitivity analyses yielded results that differed from the primary model.

Table 3

Regression coefficients from mixed effects model predicting BMI z-score

Table 4

Impact of prior notification on BMI z-score, by quartile of proportion of obese youth (mean over years 2001–2008)

COMMENT

This study, which took advantage of California’s statewide natural experiment, did not find that notifying parents of school-based BMI screening results for 5^th and 7^th grade students had an effect on pediatric obesity. While notifying parents of their child’s weight status might be part of a multi-faceted approach to reducing obesity, these results suggest that current notification methods in 5^th and 7^th grade are not sufficiently effective to warrant the practice on a large scale. California, which is home to almost 1 in 8 youth living in the United States, has the largest Latino population in the country.¹⁷ This study, therefore, provided a unique opportunity to evaluate the impact of school-based BMI screening and notification in one of the ethnic groups most susceptible to obesity and type 2 diabetes.¹⁸

Theoretically, BMI screening and reporting can notify parents that their child is overweight or obese and prompt them to act on this knowledge. There is consistent evidence demonstrating that many parents of overweight and obese children aren’t cognizant of their child’s weight status.¹⁹ However, perceptions of weight status and the risks associated with obesity are complex and may not be changed by experts’ reports of risk.²⁰ Many adults who are obese or otherwise at risk for cardiovascular disease do not perceive themselves to be at increased risk, despite experts’ opinions to the contrary.^{21, 22} The long-term risks of childhood obesity are particularly difficult to convey as parents frequently believe their child will “grow into their weight.”²³

Two studies have demonstrated that BMI reporting can improve the accuracy of parents’ perceptions of their child’s weight status.^{24, 25} West et al demonstrated this effect among both African American and white parents.²⁵ Chomitz et al further found that parents of children in Kindergarten through 8^th grade reported being motivated to attempt lifestyle changes as a result of BMI reporting,²⁴ though it should be noted that the parents in Chomitz’s study were of relatively high socioeconomic status. Also noteworthy is the fact that in Chomitz’s study, 1 to 6 weeks after a BMI report was sent, only 63% of parents recalled having received the report, suggesting that BMI reporting may be a weak intervention. In a large-scale effort among a diverse population in West Virginia, Harris et al found that BMI reporting did not change parents’ perceptions of their child’s weight.²⁶ Further work to enhance the impact of BMI reporting should explore parents’ perceptions of the causes of obesity, its associated risks, and what can and should be done at the individual, family, and community levels. This “mental models” approach has been successfully used to improve risk communications in other arenas.²⁷ A better understanding of parents’ mental models might suggest communication methods to provide critical missing information and dispel misconceptions around pediatric obesity that affect parents’ willingness or ability to make changes. It will be particularly important to explore mental models among distinct race/ethnic and socioeconomic subgroups, given these factors’ impact on weight perception.²⁸

Even if BMI reporting can alter perceptions in diverse groups, school-based BMI reporting fails one of the most salient aspects of a useful screening test: having an effective therapy if the disease (or condition) is detected.²⁹ Lifestyle interventions to treat pediatric obesity are largely ineffective³⁰ and recommending individual behavior change is unlikely to meet with success, if the experience of multidisciplinary pediatric obesity clinics is any guide.^31,32 Thus, expecting a single BMI report to parents to have a meaningful effect on a child’s weight status, in the absence of environmental changes, may be wishful thinking.

Arkansas did see a halt in the progression of obesity after implementing BMI screening and notification as part of Act 1220 of 2003.³³ However, Act 1220 simultaneously called for changes in cafeteria food offerings, increased physical activity requirements, and healthier vending machine options, making attribution to any one intervention difficult. A recent study employing a similar multi-faceted approach to alter the school environment demonstrated a significant impact on obesity, without implementing BMI screening and notification.³⁴ There is evidence that focused interventions can have a positive impact on pediatric obesity. For example, policies banning the sale of sugar-sweetened beverages and snacks high in fat or sugar during the school day appear to be related to declines in obesity seen after 2005 in California.³⁵ Increased quality and quantity of physical education has been associated with decreased obesity and improved fitness.^36–38 Until a cost-effective method of BMI notification can be found, notification resources would be better invested in changing youths’ environment, particularly in low-income communities.

The present study could not assess the impact of BMI screening itself, and it is possible that screening alone may heighten community awareness, which, in turn, could lead to changes in school or community policies over a period of years. These changes might have an impact on obesity that our model, which looked to see if notification predicted weight status 2 years hence for a cohort of children, could not detect. It will be important to study the impact of screening itself on obesity.

Limitations

Several limitations should be considered in interpreting these findings. Misclassification of the predictor variable, which would decrease our ability to see an effect of BMI reporting, could occur if institutional memory is poor or if individual schools deviate from the district policy regarding notification. Additionally, if students changed districts between 5^th and 9^th grade and notification status differed between the old and new districts, their data would be misclassified. Sensitivity analyses to address misclassification (limiting data to the most recent years, excluding districts indicating that individual schools might deviate from district policy, and excluding districts with high mobility) yielded similar findings. Parents who did not receive results as intended would be misclassified, and sensitivity analyses would not address this.

We could not link data across years for individual students and while it would be unusual, associations at the student level could be different from those at the district level, due to confounding at either the student or district level. Data were differentially missing for districts with heavier students and districts with a higher proportion of students eligible for free meals; however, notification status did not modify this association so it is unlikely that this bias in missing data would affect our findings beyond potentially limiting generalizability.

The quality of school-based BMI screening data is unknown. There is no surveillance of FITNESSGRAM test administration, and the integrity of data collection methods likely vary (which will decrease precision of estimates), and may vary by school (which might bias results). Nonetheless, random error is not likely an issue given our very small 95% CI for the effect. Bias in measurements is possible, but unless there is also bias in the change over time, it should not affect our results for the effect of notification.

The widespread use of BMI screening and reporting is heartening as it reflects schools’ willingness to dedicate resources to address the obesity epidemic. However, current methods of reporting school-based BMI screening results to parents do not appear to have an impact on pediatric obesity. While BMI screening itself may have value, further work to evaluate different approaches to providing parents with BMI screening information should be pursued before BMI reporting is implemented on a large scale. In addition, research could explore how this type of information might be used more broadly with other stakeholders and in policy. In the meantime, schools will likely see greater benefits if resources are used to increase opportunities for physical activity and improve nutrition.

Acknowledgments

Support: RWJF Active Living Research Grant 65715, NICHD 1K23HD054470-01A1, and American Heart Association 0865005F. No funders were involved in any aspect of the analyses contained herein or in the preparation of this manuscript.

References

1. Olshansky SJ, Passaro DJ, Hershow RC, et al. A potential decline in life expectancy in the United States in the 21st century. N Engl J Med. 2005;352(11):1138–1145. [PubMed]

2. Bibbins-Domingo K, Coxson P, Pletcher MJ, Lightwood J, Goldman L. Adolescent overweight and future adult coronary heart disease. N Engl J Med. 2007;357(23):2371–2379. [PubMed]

3. Freedman DS, Khan LK, Serdula MK, Dietz WH, Srinivasan SR, Berenson GS. The relation of childhood BMI to adult adiposity: the Bogalusa Heart Study. Pediatrics. 2005;115(1):22–27. [PubMed]

4. Ludwig DS. Childhood obesity--the shape of things to come. N Engl J Med. 2007;357(23):2325–2327. [PubMed]

5. Madsen KA, Weedn AE, Crawford PB. Disparities in peaks, plateaus, and declines in prevalence of high BMI among adolescents. Pediatrics. 2010;126(3):434–442. [PMC free article] [PubMed]

6. Preventing Childhood Obesity: Health in the Balance. Washington, D.C: Institute of Medicine, Committee on Prevention of Obesity in Children and Youth. National Academies Press; 2005.

7. U. S Preventive Services Task Force. Screening and interventions for overweight in children and adolescents: recommendation statement. Pediatrics. 2005;116(1):205–209. [PubMed]

8. Policy Position Statement on Body Mass Index (BMI) Surveillance and Assessment in Schools. American Heart Association; 2008. [accessed 22 Oct 2010.]. http://www.americanheart.org/presenter.jhtml?identifier=3060761.

9. Ikeda JP, Crawford PB, Woodward-Lopez G. BMI screening in schools: helpful or harmful. Health Educ Res. 2006;21(6):761–769. [PubMed]

10. Nihiser AJ, Lee SM, Wechsler H, et al. Body mass index measurement in schools. J Sch Health. 2007;77(10):651–671. [PubMed]

11. The Pupil Nutrition, Health, and Achievement Act of 2001, Chapter #1 California Statutes of 2001, Chapter #913 (Senate Bill 19 Escutia) (2001).

12. Welk GJ, Morrow JRJ, Falls HBE. Fitnessgram Reference Guide. Dallas, TX: The Cooper Institute; 2002. [accessed 22 Oct 2010.]. www.cooperinstitute.org/products/grams/documents/FITNESSGRAM_ReferenceGuide.pdf.

13. Linchey J, Madsen KA. Snapshot of state policies on fitness assessment: 2010. Paper presented at: Pediatric Academic Societies Annual Meeting; 2011; Denver, CO.

14. Appendix G. 2009–10 Physical Fitness Test Frequently Asked Questions. California Department of Education; 2009. [accessed 22 Oct 2010.]. http://www.cde.ca.gov/ta/tg/pf/documents/app09g.pdf.

15. A SAS program for the CDC Growth Charts. Atlanta, GA: Division of Nutrition, Physical Activity and Obesity, National Center for Chronic Disease Prevention and Health Promotion; 2007. [Accessed on Nov 2, 2009.]. http://www.cdc.gov/nccdphp/dnpa/growthcharts/resources/sas.htm.

16. [Accessed June, 2010.];DataQuest. http://data1.cde.ca.gov/dataquest/

17. Guzman B. The Hispanic Population. [accessed 22 Oct 2010.];Census 2000 Brief: US Census Bureau. 2001 http://www.census.gov/prod/2001pubs/c2kbr01-3.pdf.

18. Diabetes: Disabling, Deadly, and on the Rise. Centers for Disease Control and Prevention; 2005. [accessed on Oct 5, 2009.]. www.cdc.gov/nccdphp/aag/aag_ddt.htm.

19. Ritchie LD, Welk G, Styne D, Gerstein DE, Crawford PB. Family environment and pediatric overweight: what is a parent to do? J Am Diet Assoc. 2005;105(5 Suppl 1):S70–79. [PubMed]

20. Slovic P. Trust, emotion, sex, politics, and science: surveying the risk-assessment battlefield. Risk Anal. 1999;19(4):689–701. [PubMed]

21. van der Weijden T, van Steenkiste B, Stoffers HE, Timmermans DR, Grol R. Primary prevention of cardiovascular diseases in general practice: mismatch between cardiovascular risk and patients’ risk perceptions. Med Decis Making. 2007;27(6):754–761. [PubMed]

22. Gregory CO, Blanck HM, Gillespie C, Maynard LM, Serdula MK. Perceived health risk of excess body weight among overweight and obese men and women: differences by sex. Prev Med. 2008;47(1):46–52. [PubMed]

23. Jain A, Sherman SN, Chamberlin LA, Carter Y, Powers SW, Whitaker RC. Why don’t low-income mothers worry about their preschoolers being overweight? Pediatrics. 2001;107(5):1138–1146. [PubMed]

24. Chomitz VR, Collins J, Kim J, Kramer E, McGowan R. Promoting healthy weight among elementary school children via a health report card approach. Arch Pediatr Adolesc Med. 2003;157(8):765–772. [PubMed]

25. West DS, Raczynski JM, Phillips MM, Bursac Z, Heath Gauss C, Montgomery BE. Parental recognition of overweight in school-age children. Obesity (Silver Spring) 2008;16(3):630–636. [PubMed]

26. Harris CV, Neal WA. Assessing BMI in West Virginia schools: parent perspectives and the influence of context. Pediatrics. 2009;124 (Suppl 1):S63–72. [PMC free article] [PubMed]

27. Atman CJ, Bostrom A, Fischhoff B, Morgan MG. Designing Risk Communications: Completing and Correcting Mental Models of Hazardous Processes, Part I. Risk Anal. 1994;14(5):779–788. [PubMed]

28. Chang VW, Christakis NA. Self-perception of weight appropriateness in the United States. Am J Prev Med. 2003;24(4):332–339. [PubMed]

29. Newman TB, Kohn MA. Evidence-Based Diagnosis. New York, NY: Cambridge University Press; 2009. Screening Tests.

30. Oude Luttikhuis H, Baur L, Jansen H, et al. Interventions for treating obesity in children. Cochrane Database Syst Rev. 2009;(1):CD001872. [PubMed]

31. Madsen KA, Garber AK, Mietus-Snyder ML, et al. A clinic-based lifestyle intervention for pediatric obesity: efficacy and behavioral and biochemical predictors of response. JPediatr Endocrinol Metab. 2009 in press. [PMC free article] [PubMed]

32. Pinhas-Hamiel O, Lerner-Geva L, Copperman N, Jacobson MS. Insulin resistance and parental obesity as predictors to response to therapeutic life style change in obese children and adolescents 10–18 years old. J Adolesc Health. 2008;43(5):437–443. [PubMed]

33. Thompson JW, Card-Higginson P. Arkansas’ experience: statewide surveillance and parental information on the child obesity epidemic. Pediatrics. 2009;124 (Suppl 1):S73–82. [PubMed]

34. Foster GD, Linder B, Baranowski T, et al. A school-based intervention for diabetes risk reduction. N Engl J Med. 2010;363(5):443–453. [PMC free article] [PubMed]

35. Sanchez-Vaznaugh EV, Sanchez BN, Baek J, Crawford PB. ‘Competitive’ food and beverage policies: are they influencing childhood overweight trends? Health Affair. 2010;29(3):436–446. [PubMed]

36. Datar A, Sturm R. Physical education in elementary school and body mass index: evidence from the early childhood longitudinal study. Am J Public Health. 2004;94(9):1501–1506. [PubMed]

37. Madsen KA, Gosliner W, Woodward-Lopez G, Crawford PB. Physical activity opportunities associated with fitness and weight status among adolescents in low-income communities. Arch Pediatr Adolesc Med. 2009;163(11):1014–1021. [PMC free article] [PubMed]

38. Sollerhed AC, Ejlertsson G. Physical benefits of expanded physical education in primary school: findings from a 3-year intervention study in Sweden. Scand J Med Sci Sports. 2008;18(1):102–107. [PubMed]