Brief Epidemiologic Report

Obesity (2006) 14, 1107–1112; doi: 10.1038/oby.2006.126

Trends in Overweight from 1980 through 2001 among Preschool-Aged Children Enrolled in a Health Maintenance Organization*

Juhee Kim*,†, Karen E. Peterson*,†, Kelley S. Scanlon‡, Garrett M. Fitzmaurice§,, Aviva Mustparallel, Emily Oken**, Sheryl L. Rifas-Shiman**, Janet W. Rich-Edwards**,†† and Matthew W. Gillman†,**

  1. *Department of Society and Human Development and Health, Harvard School of Public Health, Boston, Massachusetts;
  2. †Department of Nutrition, Harvard School of Public Health, Boston, Massachusetts;
  3. §Department of Biostatistics, Harvard School of Public Health, Boston, Massachusetts;
  4. ‡Division of Nutrition and Physical Activity, Centers for Disease Control and Prevention, Atlanta, Georgia;
  5. Division of General Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston, Massachusetts;
  6. parallelDepartment of Public Health and Family Medicine, Tufts University School of Medicine, Boston, Massachusetts;
  7. **Department of Ambulatory Care and Prevention, Harvard Pilgrim Health Care and Harvard Medical School, Boston, Massachusetts
  8. ††Department of Epidemiology, Harvard School of Public Health, Channing Laboratory, Brigham and Women's Hospital, Boston, Massachusetts.

Correspondence: Juhee Kim 677 Huntington Avenue, Building III-R616, Department of Society and Human Development and Health, Harvard School of Public Health, Boston, MA 02115. E-mail: juheekim@hsph.harvard.edu

*The costs of publication of this article were defrayed, in part, by the payment of page charges. This article must, therefore, be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Received 8 June 2005; Revised  0000; Accepted 27 April 2006.

Top

Abstract

Objective: To examine overweight trends over a 22-year period among preschool-aged children from primarily middle-income families enrolled in a health maintenance organization.

Research Methods and Procedures: From well-child care visits to a Massachusetts health maintenance organization, we randomly selected one visit per child per calendar year, yielding a study sample of 120,680 children seen at 366,109 visits from 1980 through 2001. Using multivariate logistic regression models accounting for repeated observations of individual children across years, we estimated trends in prevalence of overweight (weight-for-length/height greater than or equal to 95th percentile) and at-risk-for-overweight (85th to 95th percentile).

Results: Over the 22-year study period, the observed prevalence of overweight increased from 6.3% to 10.0% and at-risk-for-overweight increased from 11.1% to 14.4% . These increases were evident among all groups of children including infants < 6 months of age. Overall, the adjusted odds ratios were 1.21 per decade (95% confidence interval, 1.17 to 1.25) for overweight and 1.06 per decade (95% confidence interval, 1.03 to 1.08) for at-risk-for-overweight.

Discussion: Rates of overweight are increasing in very young children, even infants, from primarily middle-class families.

Keywords:

HMO, overweight, surveillance, preschool-aged children, infants

Top

Introduction

During the past 30 years in the United States, the prevalence of overweight has substantially increased among youth (1, 2, 3). The first few years of life constitute a key period for development of overweight in childhood and its long-term health consequences. In a U.S. cohort study, rapid weight gain from ages 0 to 4 months was associated with overweight at age 7 years and in early adulthood (4). In other studies, early accelerated weight gain among children born small was associated with higher BMI, especially central obesity in early childhood (5) and with higher adult systolic blood pressure (6). Two nutritional surveillance systems, the National Health and Nutrition Examination Survey (NHANES)1 and the Pediatric Nutrition Surveillance System (PedNSS), have documented an increase in the prevalence of overweight among children younger than 5 years (1, 2, 7). Neither study, however, reported infancy as a separate age group, and in the NHANES, the number of children younger than 24 months was small. Further, the PedNSS is restricted to a low-income population, and its most recent trend data published in a peer-reviewed journal are from almost a decade ago (2).

The purpose of this study was to estimate the changes in prevalence of overweight and at-risk-for-overweight in preschool children enrolled in a Massachusetts health maintenance organization (HMO) during a 22-year period from 1980 through 2001.

Top

Research Methods and Procedures

Setting and Subjects

Our study population consisted of children younger than 72 months who were seen for well-child visits at any of the 14 health centers of Harvard Vanguard Medical Associates (HVMA) in eastern Massachusetts from 1980 through 2001. Since its inception, this practice has cared for a predominantly employed and insured population. Children with Medicaid insurance were accepted from 1987 onwards. Throughout the study period, the HMO used a completely electronic medical record system that contained demographic and growth data. Originally a staff model HMO, HVMA evolved into a group practice in 1998. From 1998, however, race/ethnicity was less often recorded, and Medicaid insurance information was no longer available.

We extracted data from all well-child visits, including weight and height from the clinic visit record and demographic information from the enrollment record. From a total of 767,538 visits by 126,024 children, we excluded 7.5% of the visits because age, gender, height, or weight data were not available. An additional 1.1% of visits were excluded because weight-for-age z score or height-for-age z (HAZ) score was below - 6 or above 6 (8) or because weight-for-height z (WHZ) score was below - 4 or above 5 (2). Of the complete 702,456 visits, we randomly selected 366,109 to provide one visit per child per calendar year, thus avoiding double counting of children with multiple records per year. For the entire study period, the mean number of visits per child was 3.0. The study protocol was reviewed and approved by Institutional Review Boards of the Harvard School of Public Health and Harvard Pilgrim Health Care.

Measures

Medical assistants measured length or height and weight according to the written protocol of the HVMA health centers. Weight was measured to the nearest 0.25 pounds on a pediatric scale. In general, length in children younger than 24 months was measured recumbent. For children older than 36 months, height was generally measured standing. Because of uncertainty about the body position of the 24- to 36-month-old children, we considered that age group separately. Medical assistants used a paper-and-pencil technique for children younger than 24 months rather than the recommended recumbent measuring board (9).

In a measurement validation study conducted at one of the participating HMO health centers, we found that this paper-and-pencil method systematically overestimated children's length compared with the standard method (10). This overestimate of length yielded lower estimates of overweight (4.7% ) and higher estimates of children whose weight-for-length was less than the 5th percentile (13.0% ) compared with national reports (10). After applying a regression correction factor to adjust for systematic underestimation: [ (0.953 times length measured by paper-and-pencil method) + 1.8 cm] , the prevalence of overweight was 9.3% , and 5.7% were below the 5th percentile. Correspondingly, we corrected recumbent length for children younger than 24 months in our study.

Analysis

We computed growth outcomes using the SAS program provided by the Centers for Disease Control and Prevention (CDC), which used the CDC 2000 growth reference (11). No established definition of overweight exists for children younger than 24 months of age. Because our study goal was to monitor overweight status over time among children younger than 24 months, we chose the same definition used in an analysis of trends in overweight among young children participating in the PedNSS (12). That is, we defined at-risk-for-overweight as weight-for-length (or height, for children at least 24 month of age) greater than or equal to85th but <95th percentile and overweight as weight-for-length greater than or equal to95th percentile, specific for gender and age in months. In addition, for children 24 months and older, we used the same percentile cut-off points defined by BMI-for-age, calculated by dividing weight in kilograms by height in meters squared. In a secondary analysis, we examined overweight prevalence using the definition of the International Obesity Task Force (13). For ease of discussion, we use the term overweight for all age groups in this study. Race/ethnicity was recorded by clinicians as white, black, Hispanic, Native American/Alaskan Native, Asian, or other. Because of small numbers in some minority groups, we categorized race/ethnicity as white, black, Hispanic, or other.

We checked the assumption of constant variance in HAZ and WHZ as a data quality assessment (2). After excluding outliers, all standard deviation (SD) units of HAZ and WHZ were between 0.9 and 1.2 in each age group, year, and clinic site, which are acceptable limits.

Because of missing data, we restricted analyses on race/ethnicity to the years 1980 to 1981 through 1996 to 1997 for the unadjusted prevalence estimates. Because Medicaid insurance was not accepted before 1987 and some data were missing after 1997, we restricted analyses to the years 1987 to 1988 through 1996 to 1997 to estimate the unadjusted prevalence.

We estimated the adjusted odds of being overweight and at-risk-for-overweight for overall time trends from multivariate logistic regression models using generalized estimating equations to account for the correlation among repeated observations of the same children across years (14). Because the correlation among weight-for-height percentiles on the same child decreased over time, we assumed an autoregressive correlation structure. Binary status of overweight or at-risk-for-overweight was the outcome of interest, and time (visit year) was the main predictor for trend with the covariates of gender, age group, and clinic site (main model). Overall years, information on race/ethnicity, and Medicaid status were not available for 52% and 12% of the sample, respectively. Because of the extent of missing data on race/ethnicity and Medicaid status, the multivariate logistic regressions did not control for these variables. Instead, we did separate analyses of the subsample that had complete data on race/ethnicity to estimate race-specific trends. To estimate Medicaid-specific trends, we used multiple imputations (10 replications) for missing Medicaid status assuming prevalence of Medicaid status was constant over time (6.4% on Medicaid insurance).

We examined differences in overweight trends in subgroups by including interaction terms between visit year and the covariates. Although we modeled time in years, for interpretability we express time trends for a period of 10 years (per decade). We report adjusted odds ratios (ORs) and 95% confidence intervals (CIs) for overweight and at-risk-for-overweight. Because the demographic profile of study population changed over time, we also calculated predicted prevalence of overweight and at-risk-for-overweight by applying the constant proportions of gender, age group, and clinic site over time to the estimated beta coefficients from the main model. All analyses used the SAS statistical analysis system (version 8.01; SAS Institute, Cary, NC).

Top

Results

Over the 22-year study period, the prevalence of both overweight and at-risk-for-overweight increased nearly linearly. The unadjusted prevalence of overweight increased from 6.3% in 1980 to 1981 to 10.0% in 2000 to 2001, and at-risk-for-overweight increased from 11.1% to 14.4% (Table 1). We observed large relative increases in the unadjusted prevalence of overweight among children 0 to 5.9 months and children at least 2 years old and a large increase in at-risk-for-overweight among children 0 to 5.9 months (Table 1). Relative increases in the prevalence of overweight were greater among girls than boys, Hispanic children than white or black children, and children insured by Medicaid than those not on Medicaid. Not all of these disparities were apparent in increases in at-risk-for-overweight (Table 1). Overall, the adjusted OR of being overweight per additional decade was 1.21 (95% CI, 1.17 to 1.25); for at-risk-for-overweight, it was 1.06 (95% CI, 1.03 to 1.08) (Table 1).


We observed a corresponding increase in mean body size over the 22 years: mean (SD) for weight-for-age z score increased from - 0.05 (1.02) to 0.27 (1.08), HAZ score from 0.05 (1.02) to 0.23 (1.00), and WHZ score from 0.01 (1.11) to 0.30 (1.10). The proportion of children with weight-for-height less than 5th percentile decreased from 7.2% in 1980 to 1981 to 4.3% in 2000 to 2001. Relative to the reference group, infants 0 to 5.9 months, there was a steeper increase in odds of overweight among children 24 to 35.9 months (Table 1; Figure 1).

Figure 1.
Figure 1 - Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, please contact help@nature.com or the author

Age-specific predicted prevalence of overweight from 1980 through 2001 among 120,680 children 0 to 71.9 months seen at 366,109 well-child care visits at a Massachusetts HMO by age group. The predicted prevalences were estimated based on a multivariate logistic regression with covariates gender, age group, clinic site, and visit year; interaction terms between visit year and gender and age group accounted for repeated observations of individual children across years.

Full figure and legend (95K)

In a separate analysis, we used the subsample with complete data to estimate the adjusted trends in overweight by racial/ethnic group. Increases in the prevalence of overweight were greater for Hispanic and black children than for white children. Compared with the odds for white children, the adjusted odds of overweight trend per decade were 1.41 (95% CI, 1.15 to 1.71) for Hispanic and 1.20 (95% CI, 1.07 to 1.35) for black children. We also confirmed that the estimates for prevalence and for trends in overweight that were adjusted for race/ethnicity, in addition to all other covariates, were similar to our adjusted estimates that did not include race/ethnicity in the model.

To compare our results with those of other published studies, we examined trends in overweight among children 2 to 5.9 years. Using the definition of BMI greater than or equal to 95th percentile, the observed prevalence of overweight increased from 6.7% in 1980–1981 to 12.0% in 2000–2001. Overall, the adjusted OR of being overweight for each additional decade was 1.31 (95% CI, 1.26 to 1.37). When we instead applied the IOTF definition, the observed prevalence of overweight increased from 2.2% in 1980–1981 to 5.5% in 2000–2001; the adjusted OR of being overweight for each additional decade was 1.45 (95% CI, 1.37 to 1.54).

Top

Discussion

Our data demonstrate that overweight is increasing in very young children, including infants. Accumulating evidence supports the importance of overweight status and weight gain during early infancy as a risk factor for persistent overweight during preschool years (15) and later adulthood overweight (16). Our observation of a trend of increasing weight among young infants may portend continued increase in childhood and adult obesity.

The similarity between our overall trend data and those of other published studies lends confidence to our estimates of trend by subgroup. The predicted prevalence of overweight (BMI greater than or equal to 95th percentile) among the HMO children 2 to 5.9 years was 10.5% in 2001, which is similar to the 10.3% reported for children participating in NHANES 1999 to 2002 (3) and a little lower than the 13.4% reported for low-income children participating in the PedNSS in 2001 (17). The relative increase in prevalence of overweight among HMO children 2 to 5.9 years from 1992 to 2001 was 24.9% , similar to the 26.4% increase among PedNSS children from 1992 to 2001, but somewhat lower than the 44.4% increase reported by NHANES from 1988 to 1994 to 1999 to 2002 (3). The rate of increase in overweight was greater among black and Hispanic HMO children than among white children, a finding consistent with the NHANES and PedNSS reports.

Our HMO-based surveillance system had a large sample of children younger than 24 months, providing estimates for that age group that might not be possible with the relatively small numbers of infants in NHANES (1, 3). Given the current lack of weight information for this age group, this study provides a potential data source for monitoring or tracking weight status for very young children. An additional strength of this study is that it provides an example of how to use medical records routinely collected in medical practice. Additional uses of an HMO-based surveillance system include needs assessment for planning purposes and evaluation of programs and policies addressing overweight that have been implemented in a clinical setting.

We recognize that weight-for-height estimates of children younger than 24 months are derived from growth charts created with a fewer number of children than what was available for construction of growth charts for older children (18). For this reason, we recommend caution in comparing prevalences and rates of increase across age groups. Nevertheless, we used the same outcome definitions and reference population over time, so that trend estimates within age groups should be stable.

With the use of medical record data, we cannot ensure the quality of anthropometric measurements. Our validation study found systematic overestimation of length measurement in children younger than 2 years, for which we attempted to correct statistically in this study (10). Because the mean and SD of HAZ were relatively stable across age groups, across clinics, and over the years, it is unlikely that a systematic bias in height measurement affected the observed trends in overweight. Further, lack of standardization in anthropometric measurements would likely attenuate the true association between overweight status and time due to increased random error.

Missing information on race/ethnicity and on Medicaid status after 1998 is another limitation of the HMO surveillance data. Thus, in estimating secular trends in overweight, we were not able to fully control for the demographic changes over time in our HMO population. We estimated the age-specific predicted prevalences adjusted for changes in gender, age, and clinic site, but not for race/ethnicity and Medicaid status. We ran subgroup analyses on children with complete race/ethnicity information and children without Medicaid insurance. The adjusted OR (95% CI) for overweight among children who had complete race/ethnicity information was 1.24 per decade (1.09 to 1.42) and for children without Medicaid insurance was 1.19 (1.09 to 1.30), which were similar to our reported estimates of 1.21 (1.17 to 1.25). Our sensitivity analyses confirmed that the overall estimates in trends did not seem to be biased by missing data. The sensitivity analyses also suggested the importance of those variables for the subgroup analyses; the prevalence and the magnitude of increase in both overweight and at-risk-for-overweight were higher among minority race/ethnicity groups and among children with Medicaid insurance.

The current obesity epidemic warrants an active role by the medical community both in improving treatment methods and in providing and evaluating prevention efforts for childhood obesity. An active monitoring system of weight status for patients represents an important step toward the evaluation of obesity prevention efforts in the medical community. Our study indicates that even our youngest children are not exempt from the well-documented trends of overweight seen among older children and adults in the United States.

Top

Notes

1 Nonstandard abbreviations: NHANES, National Health and Nutrition Examination Survey; PedNSS, Pediatric Nutrition Surveillance System; HMO, health maintenance organization; HVMA, Harvard Vanguard Medical Associates; HAZ, height-for-age z score; WHZ, weight-for-height z score; CDC, Centers for Disease Control and Prevention; SD, standard deviation; OR, odds ratio; CI, confidence interval.

Top

References

  1. Ogden, C. L., Flegal, K. M., Carroll, M. D., Johnson, CL. (2002) Prevalence and trends in overweight among US children and adolescents, 1999–2000. JAMA. 288: 1728–1732. | Article | PubMed | ISI |
  2. Mei, Z., Scanlon, K. S., Grummer-Strawn, L. M., Freedman, D. S., Yip, R., Trowbridge, FL. (1998) Increasing prevalence of overweight among US low-income preschool children: the Centers for Disease Control and Prevention pediatric nutrition surveillance, 1983 to 1995. Pediatrics. 101: E12 | Article | PubMed | ChemPort |
  3. Hedley, A. A., Ogden, C. L., Johnson, C. L., Carroll, M. D., Curtin, L. R., Flegal, KM. (2004) Prevalence of overweight and obesity among US children, adolescents, and adults, 1999–2002. JAMA. 291: 2847–2850. | Article | PubMed | ISI | ChemPort |
  4. Stettler, N., Kumanyika, S. K., Katz, S. H., Zemel, B. S., Stallings, VA. (2003) Rapid weight gain during infancy and obesity in young adulthood in a cohort of African Americans. Am J Clin Nutr. 77: 1374–1378. | PubMed | ChemPort |
  5. Ong, K. K., Ahmed, M. L., Emmett, P. M., Preece, M. A., Dunger, DB. (2000) Association between postnatal catch-up growth and obesity in childhood: prospective cohort study. BMJ. 320: 967–971. | Article | PubMed | ChemPort |
  6. Law, C. M., Shiell, A. W., Newsome, C. A., et al (2002) Fetal, infant, and childhood growth and adult blood pressure: a longitudinal study from birth to 22 years of age. Circulation. 105: 1088–1092. | Article | PubMed | ISI | ChemPort |
  7. Ogden, C. L., Troiano, R. P., Briefel, R. R., Kuczmarski, R. J., Flegal, K. M., Johnson, CL. (1997) Prevalence of overweight among preschool children in the United States, 1971 through 1994. Pediatrics. 99: E1 | Article | PubMed | ChemPort |
  8. Centers for Disease Control and Prevention (Accessed January 15, 2003) BMI for Children and Teens. http://www.cdc.gov/nccdphp/dnpa/bmi/bmi-for-age.htm.
  9. World Health Organization (1995) Physical Status: The Use and Interpretation of Anthropometry. World Health Organization: Geneva, Switzerland.
  10. Rifas-Shiman, S. L., Rich-Edwards, J. W., Scanlon, K. S., Kleinman, K. P., Gillman, MW. (2005) Misdiagnosis of overweight and underweight children younger than 2 years of age due to length measurement bias. MedGenMed. 7: 56
  11. Centers for Disease Control and Prevention (Accessed January 10, 2004) A SAS Program for the CDC Growth Charts. http://www.cdc.gov/nccdphp/dnpa/growthcharts/sas.htm
  12. Mei, Z., Grummer-Strawn, L. M., Pietrobelli, A., Goulding, A., Goran, M. I., Dietz, WH. (2002) Validity of body mass index compared with other body-composition screening indexes for the assessment of body fatness in children and adolescents. Am J Clin Nutr. 75: 978–985. | PubMed | ISI | ChemPort |
  13. Cole, T. J., Bellizzi, M. C., Flegal, K. M., Dietz, WH. (2000) Establishing a standard definition for child overweight and obesity worldwide: international survey. BMJ. 320: 1240–1243. | Article | PubMed | ChemPort |
  14. Diggle, P., Heagerty, P., Liang, K., Zeger, S. (1994) Analysis of Longitudinal Data. Oxford University Press: New York, NY.
  15. Mei, Z., Grummer-Strawn, L. M., Scanlon, KS. (2003) Does overweight in infancy persist through the preschool years? An analysis of CDC pediatric nutrition surveillance system data. Soz Praventivmed. 48: 161–167.
  16. Stettler, N., Stallings, V. A., Troxel, A. B., et al (2005) Weight gain in the first week of life and overweight in adulthood: a cohort study of European American subjects fed infant formula. Circulation. 111: 1897–1903. | Article | PubMed | ISI |
  17. Centers for Disease Control and Prevention (2003) Pediatric Nutrition Surveillance 2001 Report. US Department of Human and Human Services: Washington, DC.
  18. Kuczmarski, R. J., Ogden, C. L., Grummer-Strawn, L. M., et al (2000) CDC growth charts: United States. Adv Data. 1–27. | PubMed |
Top

Acknowledgments

This work was supported, in part, by the NIH (Grant HL 68041), by the CDC (Task Order 0957-007), by the Seiden-Denny Scholarship Fund for Maternal and Child Health, Harvard School of Public Health, Berkowitz Fellowship in Public Health Nutrition, and by the Harvard Pilgrim Health Care Foundation and Harvard Medical School.

Extra navigation

.
ADVERTISEMENT