- Email: [email protected]
Using Latent Class Growth Modeling to Examine Longitudinal Patterns of Body Mass Index Change from Adolescence to Adulthood
RESEARCH
Original Research: Brief
Using Latent Class Growth Modeling to Examine Longitudinal Patterns of Body Mass Index Change from Adolescence to Adulthood Jennifer N. Becnel, PhD; Amanda L. Williams, PhD ARTICLE INFORMATION Article history: Submitted 17 October 2018 Accepted 26 April 2019
Key words: Obesity Body mass index Life span Latent class growth modeling Sex 2212-2672/Copyright ª 2019 by the Academy of Nutrition and Dietetics. https://doi.org/10.1016/j.jand.2019.04.025
ABSTRACT Background Few studies use longitudinal designs to assess patterns of body mass index (BMI) change from adolescence to adulthood or incorporate severe obesity as a unique subgroup. Objective To examine patterns of BMI trajectories from adolescence to adulthood and identify demographic characteristics associated with each BMI trajectory pattern. Design Height, weight, and demographic characteristics were drawn from Waves I to V of the nationally representative school-based sample of the National Longitudinal Study of Adolescent to Adult Health (Add Health) conducted from 1994 to 2018 (data collection is ongoing). Participants/setting Participants included 3,315 (55.5% female) subjects responding to in-home interviews across all five Waves of Add Health. Main outcome measures BMI at each wave modeled over time. Statistical analyses Latent class growth modeling and logistic regression analysis using population sample weights. Results Five classes of weight patterns best fit the sample. Twenty-nine percent of the sample had an always healthy BMI (class 1) and 34.9% changed from healthy weight to overweight (class 2). Moving from healthy weight to obese comprised 21.8% of the sample (class 3). BMI patterns increasing from overweight to obese (class 4) and from obese to severely obese (class 5) comprised 7.6% and 7.1% of the sample, respectively. Weight change was similar for males and females with some racial or ethnic minority participants more likely to be severely obese in adulthood. Conclusions Results emphasize the importance of tracking weight longitudinally and point to a nationally representative trend of increasing BMI during the transition to adulthood. There was no substantive decreasing trend identified in the sample. Findings highlight the need for effective early and ongoing intervention and prevention strategies and can aid in identification of vulnerable youth who are at the highest risk for moving to problematic weight categories. J Acad Nutr Diet. 2019;-:---.
O
BESITY CONTINUES TO BE AN IMPORTANT PUBLIC health problem. Estimated health care costs associated with treating obesity and its comorbidities are expected to reach $860 to $956 billion by 2030.1 Despite some reports that obesity rates were leveling off,2,3 other evidence suggests that obesity prevalence rates continue to rise, particularly for subsets of the population.4 Moreover, prevalence of the most severe forms of obesity (body mass index [BMI] 120% over the 95th percentile for age and sex or BMI35 for pediatric populations and BMI40 for adult populations) are increasing at faster rates than overweight and obesity (30BMI39.9).5 Current trends also forecast the likelihood of excess weight persisting through the life span along with a mounting associated comorbidity risk.6 Although there is heterogeneity in weight fluctuation, with subjects gaining or losing weight within the parameters of Centers for Disease Control (CDC)-defined5 actuarial ª 2019 by the Academy of Nutrition and Dietetics.
categories, it is critical to understand long-term patterns of weight change aligned with these benchmark categories to form and shape timely prevention strategies.7 The objective of this study was to use latent class growth modeling (LCGM) to explore potential BMI trajectory patterns in a group of adolescents followed over 21 years and examine how BMI trajectories differ by sex and racial or ethnic background. Longitudinal designs provide a nice complement to cross-sectional designs and aid in our understanding how BMI changes within and across subjects over time.8-11 LCGM is a valuable analytic tool for summarizing data across multiple time points and identifies patterns within distinct groups of people. This study extends previous research by focusing on underlying patterns of categorical BMI change and follows adolescents as they enter midadulthood. To our knowledge, this is the first study to use the newly released Wave V data from the National Longitudinal Study of JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
1
RESEARCH Adolescent to Adult Health (Add Health) to examine weight trajectories.
MATERIALS AND METHODS Add Health Data in this study came from Add Health. Add Health is an ongoing nationally representative school-based sample of adolescents followed into adulthood. The study used a multistage, stratified, clustered sampling design. A sample of 80 high schools and 52 middle schools from the United States were selected to ensure the sample was representative in terms of region of the country, urbanicity, school size, school type, and ethnicity. Of the 20,745 adolescents surveyed in Wave I in 1994-1995 (ages 11 to 20 years), 14,738 participants were reinterviewed at Wave II in 1996 (ages 12 to 21 years). At Wave III in 2001-2002 and Wave IV in 2008-09, 15,197 and 15,701 participants were interviewed again (ages 18 to 27 years and 24 to 33 years, respectively). Wave V is in ongoing data collection (2016-2018) with a preliminary sample of 3,872 adults (ages 32 to 42 years). Detailed information on survey procedures and sampling frames applied in Add Health are available on the website (http://www.cpc.unc.edu/ projects/addhealth/design).12 This study was deemed exempt by the University of Arkansas Institutional Review Board under federal regulation 45 46.101 CFR.
Measures For BMI, weight and height were measured in Waves I to V during in-home interviews from trained and certified field interviewers. Weight was taken using a high capacity digital scale and height was taken using a steel measure tape and carpenters square. BMI ([weight] kg/[height] m2]) categories were created from height and weight using age- and sexspecific CDC growth charts.13 Participants were categorized as healthy weight (18.5BMI24.9), overweight (25BMI29.9), obese (30BMI39.9), and severely obese (BMI 120% over the 95th percentile or BMI40). BMI categories were used because these cutoffs are often useful in determining the appropriate treatment options by health care professionals.14 After taking into account those with complete data from Waves I to V, underweight participants were excluded due to the small sample size (n¼34). Although BMI z scores are considered the ideal measure of BMI at a single time point for children, raw BMI scores have been recommended when examining longitudinal changes in BMI in children and adolescents.15,16 Age was constructed based on interview date (month and year) and participants’ self-reported birthday at Wave I. Sex was self-reported by participants at Wave I as male or female. Race or ethnicity was self-reported at Wave I and categorized as white, black, Hispanic, Asian, and Native American. If subjects selected more than one race or ethnicity, they were prompted to select one that best described their racial or ethnic background.
Statistical Analyses Statistical analyses were carried out using IBM SPSS v.24 and Mplus v.8.1.5.17,18 Missing data were handled using full information maximum likelihood estimation. Longitudinal sample weights were used in the model estimation. BMI data 2
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
RESEARCH SNAPSHOT Research Question: How have weight patterns changed over the last decade as adolescents transition to adulthood? Key Findings: Weight, as reflected by body mass index, generally increases from adolescence to adulthood for both young men and women. However, youth of racial or ethnic minority are more likely to have problematic weight gain over time, including reaching adulthood with a severely obese status. were somewhat positively skewed per the KolmogorovSmirnov test with Lilliefors correction; however, this makes sense given the nature of BMI and weight cutoffs, as well as the large sample size that yields overly conservative normality tests.19,20 Jung and Wickrama’s21 five-step approach to class specification was used as follows: step 1: specify a single-class LGCM; step 2: specify unconditional LCGM without covariates or outcomes; step 3: determine the number of classes comparing k-1 models (combined with developmental and conceptual knowledge); step 4: review estimates to address any convergence issues, and step 5: specify LCGM with covariates. This process was used to identify weight trajectory classes among participants, based on BMI-defined weight status at Waves I to V. Model fit was evaluated based on AIC/BIC, LMR-LRT, and BLRT, and entropy values in Mplus, combined with understanding weight gain trends from adolescence to adulthood and to provide sufficient class sizes for meaningful comparisons (ie, “combination of fit indices, research question, parsimony, theoretical justification, and interpretability”).21(p305) BMI values for classes were carefully evaluated to determine the number of classes with the best statistical fit and that made the most conceptual sense. The five-class model was a better fit than the four-class model (eg, entropy value increased from 0.76 to 0.78 and LMR-LRT significantly improved to P<.05).21 Although the six-class model had slightly improved fit in terms of LMR-LRT, entropy values remained the same and conceptually the additional class BMI values did not move them to a different weight category, per CDC guidelines. The latent classes were defined and exported to SPSS for logistic regression analysis to examine the likelihood of being in each class by sex and race or ethnicity.
RESULTS The present study consisted of 3,315 participants who provided data for key study variables across all five waves (55.5% female; Wave I mean age¼15.41, standard deviation [SD]¼ 1.82). Of the participants, 74.7% reported being non-Hispanic white, 13.3% non-Hispanic black, 7.8% Hispanic, 2.9% Asian, and 1.3% Native American. Average participant BMI increased from 22.5 (SD¼4.5) at Wave I to 29.9 (SD¼7.6) at Wave V. First, correlations among all study variables were examined. Then, fit indices, obesity trends, and power considerations were used to specify and define the following five classes of weight status trajectories based on whether they were always the same weight status (across the five waves of data), or had an increasing or decreasing weight status: class 1: always healthy weight (AHW; n¼951; 28.6%); class 2: --
2019 Volume
-
Number
-
RESEARCH healthy weight to overweight (H-OW; n¼1,172; 34.9%); class 3: healthy weight to obese (H-OB; n¼686; 21.8%); class 4: overweight to obese (OW-OB; n¼260; 7.6%); class 5: obese to severely obese (OB-SOB; n¼246; 7.1%; the Figure). The Table reports odds ratios (ORs) and CIs for likelihood of class membership based on sex and race or ethnicity. Male and female participants had similarly increasing weight status trajectories from adolescence to adulthood. However, logistic regression results indicate that women are significantly more likely to be in the AHW or H-OB classes compared with men (OR¼1.41, 95% CI [1.40-1.41], P<0.001 and OR¼1.39, 95% CI [1.38-1.39], P<0.001, respectively). Women were less likely than men to be classified as H-OW (OR¼0.66, 95% CI [0.66-0.67], P<0.001), OW-OB (OR¼0.71, 95% CI [0.71-0.72], P<0.001), or OB-SOB (OR¼0.92, 95% CI [0.92-0.93], P<0.001). Similar patterns by sex were found
among white and Asian participants, although the ORs were more pronounced between Asian men and women than for white men and women. Sex patterns in class odds were also similar among Hispanic participants, with the exception of Hispanic women being 59% more likely to be classified as OBSOB than Hispanic males (95% CI [1.55-1.63], P<0.001). Black female participants were more likely to be classified as H-OB, OW-OB, and OB-SOB than black males, but less likely to fall into AHW or H-OW classes. Although women in the overall sample were more likely to have a period of healthy weight in their weight status trajectories, this was not true for black women in the sample, and Hispanic women were at greater risk for the least healthy weight class than their male counterparts. Asian participants were more likely to be classified as AHW than white participants (OR¼1.75, 95% CI [1.73-1.77],
Figure. Trajectory classes of body mass index (BMI) from Wave 1 (1995) through Wave 5 (2018) for the full sample (n¼3,315). --
2019 Volume
-
Number
-
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
3
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
Class 1 (n[951) AHWb
Variables
Class 2 (n[1,772) HWc/OWd
Class 3 (n[686) HW/OBe
Class 4 (n[260) OW/OB
Class 5 (n[246) OB/ severely OB
OR (95% CI)!
Sex differencesf Female
1.41*** (1.40, 1.41)
0.66*** (0.66, 0.67)
1.39*** (1.38, 1.39)
0.71*** (0.71, 0.72)
0.92*** (0.92, 0.93)
White
1.46*** (1.45, 1.47)
0.74*** (0.74, 0.75)
1.29*** (1.28, 1.30)
0.61*** (0.60, 0.61)
0.76*** (0.76, 0.77)
Black
0.78*** (0.77, 0.79)
0.47*** (0.47, 0.48)
1.88*** (1.86, 1.90)
1.84*** (1.81, 1.88)
1.45*** (1.43, 1.48)
Hispanic
1.25*** (1.24, 1.28)
0.49*** (0.48, 0.50)
1.78*** (1.75, 1.80)
0.65*** (0.63, 0.66)
1.59*** (1.55, 1.63)
Asian
2.45*** (2.40, 2.50)
0.27*** (0.26, 0.27)
2.13*** (2.07, 2.19)
0.27*** (0.25, 0.29)
1.06 (0.99, 1.14)
0.48*** (0.48, 0.49)
0.88*** (0.87, 0.89)
1.86*** (1,84, 1.87)
0.93*** (0.92, 0.94)
1.92*** (1.90, 1.94)
Female (n¼379)
0.38*** (0.38, 0.38)
0.72*** (0.71, 0.73)
2.12*** (2.11, 2.14)
1.42*** (1.40, 1.44)
2.48*** (2.45, 2.50)
Male (n¼193)
0.71*** (0.71, 0.72)
1.13*** (1.12, 1.14)
1.45*** (1.44, 1.47)
0.47*** (0.46, 0.48)
1.30*** (1.28, 1.32)
0.57*** (0.56, 0.57)
0.89*** (0.89, 0.90)
1.69*** (1.67, 1.70)
1.05*** (1.03, 1.06)
1.59*** (1.57, 1.61)
0.53*** (0.53, 0.54)
0.73*** (0.73, 0.74)
1.91*** (1.89, 1.93)
1.09*** (1.07, 1.11)
2.15*** (2.12, 2.19)
Racial or ethnic differences Black
Hispanic Female (n¼214) Male (n¼143) Asian
g
0.62*** (0.61, 0.63)
1.12*** (1.10, 1.13)
1.38*** (1.37, 1.40)
1.02* (1.01, 1.04)
1.03*** (1.02, 1.06)
1.75*** (1.73, 1.77)
0.98*** (0.97, 0.99)
0.84*** (0.83, 0.85)
0.28*** (0.27, 0.29)
0.31*** (0.30, 0.32)
Female (n¼117)
2.24*** (2.21, 2.28)
0.55*** (0.54, 0.56)
1.03** (1.01, 1.05)
0.16*** (0.15, 0.17)
0.34*** (0.32, 0.36)
Male (n¼90)
1.34*** (1.32, 1.36)
1.55*** (1.52, 1.57)
0.62*** (0.61, 0.64)
0.36*** (0.35, 0.37)
0.28*** (0.26, 0.29)
Native Americanh
0.27*** (0.26, 0.27)
0.76*** (0.74, 0.77)
2.49*** (2.45, 2.53)
0.70*** (0.67, 0.72)
2.96*** (2.90, 3.02)
a
OR¼odds ratio. AHW¼always healthy weight. c HW¼healthy weight. d OW¼overweight. e OB¼obese. f Reference group is male. g Reference group is white. h Insufficient number of cases to compute sex differences for Native American participants (n¼23 females and 18 males; 36 participants did not report their race or ethnicity). *P<0.05. **P<0.01. ***P<0.001. b
RESEARCH
4
Table. Weighted ORsa (95% CIs) indicating likelihood of weight status trajectory class membership based on sex and race or ethnicity using data from National Longitudinal Study of Adolescent to Adult Health
--
2019 Volume -
Number -
RESEARCH P<0.001), and black, Hispanic, and Native American participants were less like to have an AHW trajectory (OR¼0.48, 0.57, and 0.27, respectively, P<0.001). All racial and ethnic minorities were less likely than white participants to fall into the H-OW class. Although Asian participants were less likely than white participants to be classified as H-OB (OR¼0.84, 95% CI [0.83-0.85], P<0.001), black, Hispanic, and Native American minority groups were more likely to increase from H-OB status (OR¼1.86, 1.69, 2.49, respectively, P<0.001). Only Hispanic participants were more likely than white to be classified as OW-OB (OR¼1.05, 95% CI [1.03-1.06], P<0.001); all other minority categories were less likely than white to increase from OW-OB status. Black participants were 92% more likely to increase from obese to severely obese, Hispanic participants were 59% more likely, and Native Americans were 196% more likely (P<0.001). Racial or ethnic differences among females in the sample indicate that Asian women were more likely than white women to have a consistently healthy weight status across the transition from adolescence to adulthood. In fact, Asian women were less likely to fall into any of the increasing weight categories (compared with white women), with the exception of a slightly higher chance of being classified as HOB (OR¼1.03, 95% CI [1.01-1.05], P<0.01). Patterns of probabilities for being in the AHW class were similar between men, women, and the full sample (with Asian participants having the greatest likelihood). However, although ORs in the overall model and among women indicated that racial and ethnic minorities were less likely than white participants to be classified in the H-OW trajectory, black (OR¼1.13), Hispanic (OR¼1.12), and Asian (OR¼1.55) men were more likely to be in this trajectory class than white men (P<0.001).
DISCUSSION This study used LCGM to examine patterns of BMI change from adolescence to adulthood using preliminary Wave V data from Add Health. Five BMI trajectory classes were identified with meaningful class differences based on sex and race or ethnicity. In general, results revealed that participants either stayed a healthy weight or increased in BMI status from adolescence to adulthood. In addition, maintaining a healthy weight into adulthood only represented 29% of the sample, leaving a large portion (71%) of the subjects moving into a heavier weight status over time. There was no statistical class indicating a decreasing BMI trajectory. However, this does not mean that no participants in the sample lost weight (eg, 22 subjects moved from an excess weight category at Wave I to a healthy weight at Wave V). The low incidence of weight loss and persistence of excess weight is consistent with previous work9,22 and points to the reality that very few adolescents or young adults are able to lose weight as they age, making prevention efforts critical. Based on class sizes and BMI means, most subjects go from being healthy weight to overweight from adolescence to adulthood. It is possible that moving from healthy weight to overweight is a reflection of normative weight gain as one ages (ie, average is 1 to 2 lb per year).22,23 More substantial weight gains are likely due to secular changes in obesogenic environments, including changes in diet, and physically inactive lifestyles.24 Of particular concern are those who are --
2019 Volume
-
Number
-
jumping from healthy weight to obese or obese to severely obese. Although weight gain in adulthood is common,23,25 it is associated with adverse health outcomes, and substantial weight gains are associated with elevated health risk.23 In addition, severe obesity has limited effective treatment options,26 making the move to severe obesity worrisome. People with severe obesity are a uniquely vulnerable group given their rates of life-threatening comorbidities in which traditional diet, exercise, or behavioral modification programs have low efficacy.26 Presently, bariatric surgery is the most effective long-term treatment for severe obesity, yet has the risk of major complications, making prevention critical.27 Given that health risks are amplified with higher levels of weight gain and higher BMI status,23 these results highlight the importance of understanding patterns of BMI change to identify groups at risk for the most substantial weight gain for targeted intervention. Results from cross-sectional data analyzing trends over time reveal that obesity and severe obesity prevalence rates continue to rise, meaning that each year more youth are reporting heavier weight status, especially for the most severe forms of obesity.3,5,8,28 Specifically, rates of childhood obesity have risen from 13.9% in 1999-2000 to 18.5% in 20152016,8 and rates of severe obesity increasing from 4.7% in 1999-2000 to 8% in 2011-2012,5 with the most substantial gains for adolescents 12 to 19 years.28 The present longitudinal study complements this cross-sectional work, and, taken together, the results illustrate troubling trajectories for cohorts of youth each year. Each year, more youth are developing a trajectory of unhealthy weight status into adulthood, and those adolescents who have obesity are likely to become adults with severe obesity as weight tracks strongly from adolescents to adulthood.11,29 Even those who are overweight will likely move into adulthood with an obese status, as indicated in this study. Current results also highlight the importance of examining the intersection of time, sex, and race or ethnicity. For example, cross-sectional studies note the low prevalence of obesity in those of Asian descent3; however, the present longitudinal data indicate that although Asian participants were more likely to have a healthy BMI status into adulthood relative to white participants, Asian women were more likely to either always be healthy weight or to move from healthy weight to obese, compared with Asian men and white women. Thus, two different trajectory profiles exist for this group that has been traditionally characterized as having very low obesity prevalence rates. In addition, cross-sectional studies highlight stark increases in severe obesity for white females and black males.5 However, present findings show two different trajectory profiles for white females who are more likely to either always be healthy weight or move from healthy weight to obese, compared with white males. An increasing BMI trend was found for black women who were more likely to become obese or severely obese, relative to black men. Therefore, population trends in obesity prevalence are incomplete if not accounting for the interplay of time within the same age, sex, and racial or ethnic groups from year to year (repeated crosssections) as well as patterns within and across subjects followed over decades (longitudinal). Taken together, both perspectives can aid in the meaningful identification of time points for intervention as well as who is most in need of early or sustained intervention. JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
5
RESEARCH In addition to continuing to explore the intersection of time, sex, and race or ethnicity, future studies should also examine the various biological, social, and cultural factors that contribute to differing trajectory profiles that were not addressed in the present study. For example, in Asian participants, an AHW trajectory might be a function of a healthier diet or less sedentary behavior and the increasing BMI pattern may be reflective of generational acculturation to a less healthy diet and lifestyle.30 Moreover, Asian individuals experience higher rates of metabolic syndrome at lower BMIs,31,32 suggesting that healthy weight might need to be defined at a lower range for this population. In addition, black and Hispanic participants were at greater risk of more severe BMI change, relative to white participants. This may reflect cultural values where black and Hispanic individuals have an increased tolerance for adiposity, value a larger body frame, and find larger bodies aesthetically pleasing.33 These biological and sociocultural factors are important considerations for future intervention and prevention studies.
Strengths and Limitations First, LCGM is a useful tool for evaluating patterns in BMI data. LCGM has been used in studies of mental health and dietary patterns34,35; however, it has rarely been used to characterize patterns of weight and can provide valuable information on weight change across the life span. Second, instead of using arbitrary cutoffs or leaving BMI as a continuous variable, we use measured height and weight to form CDC-defined BMI categories (ie, healthy weight, overweight, obese, severely obese) to create trajectories to model real weight change. This categorization is also useful because these cutoffs are used by health practitioners to determine access to treatment options.14 Third, the present study utilized preliminary Wave V data, extending the ages available for analysis beyond previous studies using Add Health.9,10 However, Wave V data are preliminary and have yet to be fully released. As a result, the sample sizes of some groups (ie, Native American) were too small for meaningful analysis. In addition, the objective of this study was to examine patterns of BMI change by sex and race or ethnicity, and the analysis does not capture sociocultural or environmental factors that likely affect BMI change during the transition from adolescence to adulthood, which warrant further study.
CONCLUSIONS This study utilized data from a longitudinal, nationally representative sample to examine BMI trajectory patterns from adolescence to adulthood. Five distinct trajectory patterns were identified, in which subjects either maintained or increased in weight over time. The majority (71%) of subjects increased in weight, leaving only 29% of the sample maintaining a healthy weight trajectory through adulthood. Given the comorbidities of excessive weight gain and severe obesity, the most concerning trajectories were moving from a healthy weight to obese and obese to severely obese. In this study, black and Hispanic females were most likely to fit in these patterns.
References 1.
6
Wang Y, Beydoun MA, Liang L, Caballero B, Kumanyika SK. Will all Americans become overweight or obese? Estimating the progression and cost of the US obesity epidemic. Obesity. 2008;16:2323-2330.
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
2.
Flegal KM, Carroll MD, Kit BK, Ogden CL. Prevalence of obesity and trends in the distribution of body mass index among US adults, 1999-2010. JAMA. 2012;307:491-497.
3.
Ogden CL, Carroll MD, Kit BK, Flegal KM. Prevalence of childhood and adult obesity in the United States, 2011-2012. JAMA. 2014;311:806814.
4.
Skinner AC, Ravanbakht SN, Skelton JA, Perrin EM, Armstrong SC. Prevalence of obesity and severe obesity in US children, 1999-2016. Pediatrics. 2018;141:e20173459.
5.
Skinner AC, Skelton JA. Prevalence and trends in obesity and severe obesity among children in the United States, 1999-2012. JAMA Pediatr. 2014;168:561-566.
6.
Merten MJ. Weight status continuity and change from adolescence to young adulthood: Examining disease and health risk conditions. Obesity. 2010;18:1423-1428.
7.
Brownell K, Wadden T. The heterogeneity of obesity: fitting treatments to individuals. Behav Ther. 2016;47:950-965.
8.
Hales CM, Fryar CD, Carroll MD, Freedman DS, Ogden CL. Trends in obesity and severe obesity prevalence in US youth and adults by sex and age, 2007-2008 to 2015-2016. JAMA. 2018;319:1723-1725.
9.
Gordon-Larsen P, Adair LS, Nelson MC, Popkin BM. Five-year obesity incidence in the transition period between adolescence and adulthood: The National Longitudinal Study of Adolescent Health. Am J Clin Nutr. 2004;80:569-575.
10.
Gordon-Larsen P, The NS, Adair LS. Longitudinal trends in obesity in the United States from adolescence to the third decade of life. Obesity. 2010;18:1801-1804.
11.
Balantekin KN, Hohman EE, Adams EL, et al. More rapid increase in BMI from age 5-15 is associated with elevated weight status at age 24 among non-Hispanic white females. Eat Behav. 2018;31: 12-17. 7.
12.
Harris KM. The National Longitudinal Study of Adolescent to Adult Health (AddHealth), Waves I & II, 1994-1996; Wave III, 2001-2002; Wave IV 2007-2009. Chapel Hill, NC: Carolina Population Center, University of North Carolina at Chapel Hill; 2009.
13.
Centers for Disease Control and Prevention, National Center for Health Statistics. Growth charts. http://www.cdc.gov/growthcharts/. Published 2017. Accessed June 26, 2019.
14.
North American Association for the Study of Obesity, National Heart Lung and Blook Institute. The Practical Guide: Identification, Evaluation and Treatment for Overweight and Obesity in Adults. Bethesda, MD: US Department of Health and Human Services, Public Health Service, National Institutes of Health, National Heart, Lung, and Blood Institute; 2000.
15.
Berkey CS, Colditz GA. Adiposity in adolescents: change in actual BMI works better than change in BMI z-score for longitudinal studies. Ann Epidemiol. 2007;17:44-50.
16.
Cole TJ, Faith MS, Pietrobelli A, Heo M. What is the best measure of adiposity change in growing children: BMI, BMI %, BMI z-score or BMI centile? Eur J Clin Nutr. 2005;59:419-425.
17.
IBM SPSS Statistics for Windows. Version 24 [computer program]. Armonk, NY: IBM Corp; 2016.
18.
Muthén LK, Muthén BO. Mplus User’s Guide. 6th ed. Los Angeles, CA: Muthén & Muthén; 1998-2016.
19.
Ghasemi A, Zahediasl S. Normality tests for statistical analysis: A guide for non-statisticians. Int J Endocrinol Metab. 2012;10:486-498.
20.
van de Schoot R, Sijbrandij M, Winter SD, Depaoli S, Vermunt JK. The GRoLTS-checklist: Guidelines for reporting on latent trajectory studies. Struct Equ Modeling. 2017;24:451-467.
21.
Jung T, Wickrama KAS. An introduction to latent class growth analysis and growth mixture modeling. Soc Personal Psychol Compass. 2008;2:302-317.
22.
Malhotra R, Ostbye T, Riley CM, Finkelstein EA. Young adult weight trajectories through midlife by body mass category. Obesity. 2013;21:1923-1934.
23.
Zheng Y, Manson JE, Yuan C, et al. Associations of weight gain from early to middle adulthood with major health outcomes later in life. JAMA. 2017;318:255-269.
24.
Lake A, Townshend T. Obesogenic environments: exploring the built and food environments. J R Soc Promot Health. 2006;126:262-267.
25.
Hutfless S, Gudzune KA, Maruthur N, et al. Strategies to prevent weight gain in adults: A systematic review. Am J Prev Med. 2013;45:e41-e51.
--
2019 Volume
-
Number
-
RESEARCH 26.
McTigue KM, Harris R, Hemphill B, et al. Screening and interventions for obesity in adults: Summary of the evidence for the U.S. Preventive Services Task Force. Ann Intern Med. 2003;139: 933-949.
31.
Hsu WC, Araneta MRG, Kanaya AM, Chiang JL, Fujimoto W. BMI cut points to identify at-risk Asian Americans for type 2 diabetes screening. Diabetes Care. 2015;38:150-158.
32.
27.
Chang S, Stoll CRT, Song J, Varela JE, Eagon CJ, Colditz GA. The effectiveness and risks of bariatric surgery: An updated systematic review and meta-analysis, 2003-2012. JAMA Surg. 2014;149:275-287.
Palaniappan LP, Wong EC, Shin JJ, Fortmann SP, Lauderdale DS. Asian Americans have greater prevalence of metabolic syndrome despite lower body mass index. Int J Obes. 2011;35:393-400.
33.
28.
Ogden CL, Carroll MD, Lawman HG, et al. Trends in obesity prevalence among children and adolescents in the United States, 19881994 through 2013-2014. JAMA. 2016;315:2292-2299.
Padgett J, Biro FM. Different shapes in different cultures: body dissatisfaction, overweight, and obesity in African-American and Caucasian females. J Pediatr Adolesc Gynecol. 2003;16:349-354.
34.
29.
Simmods M, Llewellyn A, Owen CG, Woolacott N. Predicting adult obesity from childhood obesity: A systematic review and metaanalysis. Obes Rev. 2015;17:95-107.
Hunter AM, Muthen BO, Cook IA, Leuchter AF. Antidepressant response trajectories and quantitative electroencephalography (QEEG) biomarkers in major depressive disorder. J Psychiatr Res. 2010;44:90-98.
30.
Diep CS, Foster MJ, McKyer ELJ, Goodson P, Guidry JJ, Liew J. What are Asian-American youth consuming? A systematic literature review. J Immigr Minor Health. 2015;17:591-604.
35.
Sotres-Alvarez D, Herring AH, Siega-Riz AM. Latent transition models to study women’s changing of dietary patterns from pregnancy to 1 year postpartum. Am J Epidemiol. 2013;177:852-861.
AUTHOR INFORMATION J. N. Becnel and A. L. Williams are assistant professors, University of Arkansas, Fayetteville. Address correspondence to: Jennifer N. Becnel, PhD, HOEC 210, University of Arkansas, Fayetteville, AR 72701. E-mail: [email protected]
STATEMENT OF POTENTIAL CONFLICT OF INTEREST No potential conflict of interest was reported by the authors.
FUNDING/SUPPORT There is no funding to disclose.
ACKNOWLEDGEMENTS This research uses data from Add Health, a program project directed by Kathleen Mullan Harris and designed by J. Richard Udry, Peter S. Bearman, and Kathleen Mullan Harris at the University of North Carolina at Chapel Hill, and funded by grant P01-HD31921 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development, with cooperative funding from 23 other federal agencies and foundations. Special acknowledgment is due to Ronald R. Rindfuss and Barbara Entwisle for assistance in the original design. Information on how to obtain the Add Health data files is available on the Add Health website (http://www.cpc.unc.edu/addhealth). No direct support was received from grant P01-HD31921 for this analysis.
AUTHOR CONTRIBUTIONS J. N. Becnel conceptualized the manuscript and was the primary writer. A. L. Williams conceptualized the manuscript, conducted the data analyses, and was the secondary writer.
--
2019 Volume
-
Number
-
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
7
Original Research: Brief
Using Latent Class Growth Modeling to Examine Longitudinal Patterns of Body Mass Index Change from Adolescence to Adulthood Jennifer N. Becnel, PhD; Amanda L. Williams, PhD ARTICLE INFORMATION Article history: Submitted 17 October 2018 Accepted 26 April 2019
Key words: Obesity Body mass index Life span Latent class growth modeling Sex 2212-2672/Copyright ª 2019 by the Academy of Nutrition and Dietetics. https://doi.org/10.1016/j.jand.2019.04.025
ABSTRACT Background Few studies use longitudinal designs to assess patterns of body mass index (BMI) change from adolescence to adulthood or incorporate severe obesity as a unique subgroup. Objective To examine patterns of BMI trajectories from adolescence to adulthood and identify demographic characteristics associated with each BMI trajectory pattern. Design Height, weight, and demographic characteristics were drawn from Waves I to V of the nationally representative school-based sample of the National Longitudinal Study of Adolescent to Adult Health (Add Health) conducted from 1994 to 2018 (data collection is ongoing). Participants/setting Participants included 3,315 (55.5% female) subjects responding to in-home interviews across all five Waves of Add Health. Main outcome measures BMI at each wave modeled over time. Statistical analyses Latent class growth modeling and logistic regression analysis using population sample weights. Results Five classes of weight patterns best fit the sample. Twenty-nine percent of the sample had an always healthy BMI (class 1) and 34.9% changed from healthy weight to overweight (class 2). Moving from healthy weight to obese comprised 21.8% of the sample (class 3). BMI patterns increasing from overweight to obese (class 4) and from obese to severely obese (class 5) comprised 7.6% and 7.1% of the sample, respectively. Weight change was similar for males and females with some racial or ethnic minority participants more likely to be severely obese in adulthood. Conclusions Results emphasize the importance of tracking weight longitudinally and point to a nationally representative trend of increasing BMI during the transition to adulthood. There was no substantive decreasing trend identified in the sample. Findings highlight the need for effective early and ongoing intervention and prevention strategies and can aid in identification of vulnerable youth who are at the highest risk for moving to problematic weight categories. J Acad Nutr Diet. 2019;-:---.
O
BESITY CONTINUES TO BE AN IMPORTANT PUBLIC health problem. Estimated health care costs associated with treating obesity and its comorbidities are expected to reach $860 to $956 billion by 2030.1 Despite some reports that obesity rates were leveling off,2,3 other evidence suggests that obesity prevalence rates continue to rise, particularly for subsets of the population.4 Moreover, prevalence of the most severe forms of obesity (body mass index [BMI] 120% over the 95th percentile for age and sex or BMI35 for pediatric populations and BMI40 for adult populations) are increasing at faster rates than overweight and obesity (30BMI39.9).5 Current trends also forecast the likelihood of excess weight persisting through the life span along with a mounting associated comorbidity risk.6 Although there is heterogeneity in weight fluctuation, with subjects gaining or losing weight within the parameters of Centers for Disease Control (CDC)-defined5 actuarial ª 2019 by the Academy of Nutrition and Dietetics.
categories, it is critical to understand long-term patterns of weight change aligned with these benchmark categories to form and shape timely prevention strategies.7 The objective of this study was to use latent class growth modeling (LCGM) to explore potential BMI trajectory patterns in a group of adolescents followed over 21 years and examine how BMI trajectories differ by sex and racial or ethnic background. Longitudinal designs provide a nice complement to cross-sectional designs and aid in our understanding how BMI changes within and across subjects over time.8-11 LCGM is a valuable analytic tool for summarizing data across multiple time points and identifies patterns within distinct groups of people. This study extends previous research by focusing on underlying patterns of categorical BMI change and follows adolescents as they enter midadulthood. To our knowledge, this is the first study to use the newly released Wave V data from the National Longitudinal Study of JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
1
RESEARCH Adolescent to Adult Health (Add Health) to examine weight trajectories.
MATERIALS AND METHODS Add Health Data in this study came from Add Health. Add Health is an ongoing nationally representative school-based sample of adolescents followed into adulthood. The study used a multistage, stratified, clustered sampling design. A sample of 80 high schools and 52 middle schools from the United States were selected to ensure the sample was representative in terms of region of the country, urbanicity, school size, school type, and ethnicity. Of the 20,745 adolescents surveyed in Wave I in 1994-1995 (ages 11 to 20 years), 14,738 participants were reinterviewed at Wave II in 1996 (ages 12 to 21 years). At Wave III in 2001-2002 and Wave IV in 2008-09, 15,197 and 15,701 participants were interviewed again (ages 18 to 27 years and 24 to 33 years, respectively). Wave V is in ongoing data collection (2016-2018) with a preliminary sample of 3,872 adults (ages 32 to 42 years). Detailed information on survey procedures and sampling frames applied in Add Health are available on the website (http://www.cpc.unc.edu/ projects/addhealth/design).12 This study was deemed exempt by the University of Arkansas Institutional Review Board under federal regulation 45 46.101 CFR.
Measures For BMI, weight and height were measured in Waves I to V during in-home interviews from trained and certified field interviewers. Weight was taken using a high capacity digital scale and height was taken using a steel measure tape and carpenters square. BMI ([weight] kg/[height] m2]) categories were created from height and weight using age- and sexspecific CDC growth charts.13 Participants were categorized as healthy weight (18.5BMI24.9), overweight (25BMI29.9), obese (30BMI39.9), and severely obese (BMI 120% over the 95th percentile or BMI40). BMI categories were used because these cutoffs are often useful in determining the appropriate treatment options by health care professionals.14 After taking into account those with complete data from Waves I to V, underweight participants were excluded due to the small sample size (n¼34). Although BMI z scores are considered the ideal measure of BMI at a single time point for children, raw BMI scores have been recommended when examining longitudinal changes in BMI in children and adolescents.15,16 Age was constructed based on interview date (month and year) and participants’ self-reported birthday at Wave I. Sex was self-reported by participants at Wave I as male or female. Race or ethnicity was self-reported at Wave I and categorized as white, black, Hispanic, Asian, and Native American. If subjects selected more than one race or ethnicity, they were prompted to select one that best described their racial or ethnic background.
Statistical Analyses Statistical analyses were carried out using IBM SPSS v.24 and Mplus v.8.1.5.17,18 Missing data were handled using full information maximum likelihood estimation. Longitudinal sample weights were used in the model estimation. BMI data 2
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
RESEARCH SNAPSHOT Research Question: How have weight patterns changed over the last decade as adolescents transition to adulthood? Key Findings: Weight, as reflected by body mass index, generally increases from adolescence to adulthood for both young men and women. However, youth of racial or ethnic minority are more likely to have problematic weight gain over time, including reaching adulthood with a severely obese status. were somewhat positively skewed per the KolmogorovSmirnov test with Lilliefors correction; however, this makes sense given the nature of BMI and weight cutoffs, as well as the large sample size that yields overly conservative normality tests.19,20 Jung and Wickrama’s21 five-step approach to class specification was used as follows: step 1: specify a single-class LGCM; step 2: specify unconditional LCGM without covariates or outcomes; step 3: determine the number of classes comparing k-1 models (combined with developmental and conceptual knowledge); step 4: review estimates to address any convergence issues, and step 5: specify LCGM with covariates. This process was used to identify weight trajectory classes among participants, based on BMI-defined weight status at Waves I to V. Model fit was evaluated based on AIC/BIC, LMR-LRT, and BLRT, and entropy values in Mplus, combined with understanding weight gain trends from adolescence to adulthood and to provide sufficient class sizes for meaningful comparisons (ie, “combination of fit indices, research question, parsimony, theoretical justification, and interpretability”).21(p305) BMI values for classes were carefully evaluated to determine the number of classes with the best statistical fit and that made the most conceptual sense. The five-class model was a better fit than the four-class model (eg, entropy value increased from 0.76 to 0.78 and LMR-LRT significantly improved to P<.05).21 Although the six-class model had slightly improved fit in terms of LMR-LRT, entropy values remained the same and conceptually the additional class BMI values did not move them to a different weight category, per CDC guidelines. The latent classes were defined and exported to SPSS for logistic regression analysis to examine the likelihood of being in each class by sex and race or ethnicity.
RESULTS The present study consisted of 3,315 participants who provided data for key study variables across all five waves (55.5% female; Wave I mean age¼15.41, standard deviation [SD]¼ 1.82). Of the participants, 74.7% reported being non-Hispanic white, 13.3% non-Hispanic black, 7.8% Hispanic, 2.9% Asian, and 1.3% Native American. Average participant BMI increased from 22.5 (SD¼4.5) at Wave I to 29.9 (SD¼7.6) at Wave V. First, correlations among all study variables were examined. Then, fit indices, obesity trends, and power considerations were used to specify and define the following five classes of weight status trajectories based on whether they were always the same weight status (across the five waves of data), or had an increasing or decreasing weight status: class 1: always healthy weight (AHW; n¼951; 28.6%); class 2: --
2019 Volume
-
Number
-
RESEARCH healthy weight to overweight (H-OW; n¼1,172; 34.9%); class 3: healthy weight to obese (H-OB; n¼686; 21.8%); class 4: overweight to obese (OW-OB; n¼260; 7.6%); class 5: obese to severely obese (OB-SOB; n¼246; 7.1%; the Figure). The Table reports odds ratios (ORs) and CIs for likelihood of class membership based on sex and race or ethnicity. Male and female participants had similarly increasing weight status trajectories from adolescence to adulthood. However, logistic regression results indicate that women are significantly more likely to be in the AHW or H-OB classes compared with men (OR¼1.41, 95% CI [1.40-1.41], P<0.001 and OR¼1.39, 95% CI [1.38-1.39], P<0.001, respectively). Women were less likely than men to be classified as H-OW (OR¼0.66, 95% CI [0.66-0.67], P<0.001), OW-OB (OR¼0.71, 95% CI [0.71-0.72], P<0.001), or OB-SOB (OR¼0.92, 95% CI [0.92-0.93], P<0.001). Similar patterns by sex were found
among white and Asian participants, although the ORs were more pronounced between Asian men and women than for white men and women. Sex patterns in class odds were also similar among Hispanic participants, with the exception of Hispanic women being 59% more likely to be classified as OBSOB than Hispanic males (95% CI [1.55-1.63], P<0.001). Black female participants were more likely to be classified as H-OB, OW-OB, and OB-SOB than black males, but less likely to fall into AHW or H-OW classes. Although women in the overall sample were more likely to have a period of healthy weight in their weight status trajectories, this was not true for black women in the sample, and Hispanic women were at greater risk for the least healthy weight class than their male counterparts. Asian participants were more likely to be classified as AHW than white participants (OR¼1.75, 95% CI [1.73-1.77],
Figure. Trajectory classes of body mass index (BMI) from Wave 1 (1995) through Wave 5 (2018) for the full sample (n¼3,315). --
2019 Volume
-
Number
-
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
3
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
Class 1 (n[951) AHWb
Variables
Class 2 (n[1,772) HWc/OWd
Class 3 (n[686) HW/OBe
Class 4 (n[260) OW/OB
Class 5 (n[246) OB/ severely OB
OR (95% CI)!
Sex differencesf Female
1.41*** (1.40, 1.41)
0.66*** (0.66, 0.67)
1.39*** (1.38, 1.39)
0.71*** (0.71, 0.72)
0.92*** (0.92, 0.93)
White
1.46*** (1.45, 1.47)
0.74*** (0.74, 0.75)
1.29*** (1.28, 1.30)
0.61*** (0.60, 0.61)
0.76*** (0.76, 0.77)
Black
0.78*** (0.77, 0.79)
0.47*** (0.47, 0.48)
1.88*** (1.86, 1.90)
1.84*** (1.81, 1.88)
1.45*** (1.43, 1.48)
Hispanic
1.25*** (1.24, 1.28)
0.49*** (0.48, 0.50)
1.78*** (1.75, 1.80)
0.65*** (0.63, 0.66)
1.59*** (1.55, 1.63)
Asian
2.45*** (2.40, 2.50)
0.27*** (0.26, 0.27)
2.13*** (2.07, 2.19)
0.27*** (0.25, 0.29)
1.06 (0.99, 1.14)
0.48*** (0.48, 0.49)
0.88*** (0.87, 0.89)
1.86*** (1,84, 1.87)
0.93*** (0.92, 0.94)
1.92*** (1.90, 1.94)
Female (n¼379)
0.38*** (0.38, 0.38)
0.72*** (0.71, 0.73)
2.12*** (2.11, 2.14)
1.42*** (1.40, 1.44)
2.48*** (2.45, 2.50)
Male (n¼193)
0.71*** (0.71, 0.72)
1.13*** (1.12, 1.14)
1.45*** (1.44, 1.47)
0.47*** (0.46, 0.48)
1.30*** (1.28, 1.32)
0.57*** (0.56, 0.57)
0.89*** (0.89, 0.90)
1.69*** (1.67, 1.70)
1.05*** (1.03, 1.06)
1.59*** (1.57, 1.61)
0.53*** (0.53, 0.54)
0.73*** (0.73, 0.74)
1.91*** (1.89, 1.93)
1.09*** (1.07, 1.11)
2.15*** (2.12, 2.19)
Racial or ethnic differences Black
Hispanic Female (n¼214) Male (n¼143) Asian
g
0.62*** (0.61, 0.63)
1.12*** (1.10, 1.13)
1.38*** (1.37, 1.40)
1.02* (1.01, 1.04)
1.03*** (1.02, 1.06)
1.75*** (1.73, 1.77)
0.98*** (0.97, 0.99)
0.84*** (0.83, 0.85)
0.28*** (0.27, 0.29)
0.31*** (0.30, 0.32)
Female (n¼117)
2.24*** (2.21, 2.28)
0.55*** (0.54, 0.56)
1.03** (1.01, 1.05)
0.16*** (0.15, 0.17)
0.34*** (0.32, 0.36)
Male (n¼90)
1.34*** (1.32, 1.36)
1.55*** (1.52, 1.57)
0.62*** (0.61, 0.64)
0.36*** (0.35, 0.37)
0.28*** (0.26, 0.29)
Native Americanh
0.27*** (0.26, 0.27)
0.76*** (0.74, 0.77)
2.49*** (2.45, 2.53)
0.70*** (0.67, 0.72)
2.96*** (2.90, 3.02)
a
OR¼odds ratio. AHW¼always healthy weight. c HW¼healthy weight. d OW¼overweight. e OB¼obese. f Reference group is male. g Reference group is white. h Insufficient number of cases to compute sex differences for Native American participants (n¼23 females and 18 males; 36 participants did not report their race or ethnicity). *P<0.05. **P<0.01. ***P<0.001. b
RESEARCH
4
Table. Weighted ORsa (95% CIs) indicating likelihood of weight status trajectory class membership based on sex and race or ethnicity using data from National Longitudinal Study of Adolescent to Adult Health
--
2019 Volume -
Number -
RESEARCH P<0.001), and black, Hispanic, and Native American participants were less like to have an AHW trajectory (OR¼0.48, 0.57, and 0.27, respectively, P<0.001). All racial and ethnic minorities were less likely than white participants to fall into the H-OW class. Although Asian participants were less likely than white participants to be classified as H-OB (OR¼0.84, 95% CI [0.83-0.85], P<0.001), black, Hispanic, and Native American minority groups were more likely to increase from H-OB status (OR¼1.86, 1.69, 2.49, respectively, P<0.001). Only Hispanic participants were more likely than white to be classified as OW-OB (OR¼1.05, 95% CI [1.03-1.06], P<0.001); all other minority categories were less likely than white to increase from OW-OB status. Black participants were 92% more likely to increase from obese to severely obese, Hispanic participants were 59% more likely, and Native Americans were 196% more likely (P<0.001). Racial or ethnic differences among females in the sample indicate that Asian women were more likely than white women to have a consistently healthy weight status across the transition from adolescence to adulthood. In fact, Asian women were less likely to fall into any of the increasing weight categories (compared with white women), with the exception of a slightly higher chance of being classified as HOB (OR¼1.03, 95% CI [1.01-1.05], P<0.01). Patterns of probabilities for being in the AHW class were similar between men, women, and the full sample (with Asian participants having the greatest likelihood). However, although ORs in the overall model and among women indicated that racial and ethnic minorities were less likely than white participants to be classified in the H-OW trajectory, black (OR¼1.13), Hispanic (OR¼1.12), and Asian (OR¼1.55) men were more likely to be in this trajectory class than white men (P<0.001).
DISCUSSION This study used LCGM to examine patterns of BMI change from adolescence to adulthood using preliminary Wave V data from Add Health. Five BMI trajectory classes were identified with meaningful class differences based on sex and race or ethnicity. In general, results revealed that participants either stayed a healthy weight or increased in BMI status from adolescence to adulthood. In addition, maintaining a healthy weight into adulthood only represented 29% of the sample, leaving a large portion (71%) of the subjects moving into a heavier weight status over time. There was no statistical class indicating a decreasing BMI trajectory. However, this does not mean that no participants in the sample lost weight (eg, 22 subjects moved from an excess weight category at Wave I to a healthy weight at Wave V). The low incidence of weight loss and persistence of excess weight is consistent with previous work9,22 and points to the reality that very few adolescents or young adults are able to lose weight as they age, making prevention efforts critical. Based on class sizes and BMI means, most subjects go from being healthy weight to overweight from adolescence to adulthood. It is possible that moving from healthy weight to overweight is a reflection of normative weight gain as one ages (ie, average is 1 to 2 lb per year).22,23 More substantial weight gains are likely due to secular changes in obesogenic environments, including changes in diet, and physically inactive lifestyles.24 Of particular concern are those who are --
2019 Volume
-
Number
-
jumping from healthy weight to obese or obese to severely obese. Although weight gain in adulthood is common,23,25 it is associated with adverse health outcomes, and substantial weight gains are associated with elevated health risk.23 In addition, severe obesity has limited effective treatment options,26 making the move to severe obesity worrisome. People with severe obesity are a uniquely vulnerable group given their rates of life-threatening comorbidities in which traditional diet, exercise, or behavioral modification programs have low efficacy.26 Presently, bariatric surgery is the most effective long-term treatment for severe obesity, yet has the risk of major complications, making prevention critical.27 Given that health risks are amplified with higher levels of weight gain and higher BMI status,23 these results highlight the importance of understanding patterns of BMI change to identify groups at risk for the most substantial weight gain for targeted intervention. Results from cross-sectional data analyzing trends over time reveal that obesity and severe obesity prevalence rates continue to rise, meaning that each year more youth are reporting heavier weight status, especially for the most severe forms of obesity.3,5,8,28 Specifically, rates of childhood obesity have risen from 13.9% in 1999-2000 to 18.5% in 20152016,8 and rates of severe obesity increasing from 4.7% in 1999-2000 to 8% in 2011-2012,5 with the most substantial gains for adolescents 12 to 19 years.28 The present longitudinal study complements this cross-sectional work, and, taken together, the results illustrate troubling trajectories for cohorts of youth each year. Each year, more youth are developing a trajectory of unhealthy weight status into adulthood, and those adolescents who have obesity are likely to become adults with severe obesity as weight tracks strongly from adolescents to adulthood.11,29 Even those who are overweight will likely move into adulthood with an obese status, as indicated in this study. Current results also highlight the importance of examining the intersection of time, sex, and race or ethnicity. For example, cross-sectional studies note the low prevalence of obesity in those of Asian descent3; however, the present longitudinal data indicate that although Asian participants were more likely to have a healthy BMI status into adulthood relative to white participants, Asian women were more likely to either always be healthy weight or to move from healthy weight to obese, compared with Asian men and white women. Thus, two different trajectory profiles exist for this group that has been traditionally characterized as having very low obesity prevalence rates. In addition, cross-sectional studies highlight stark increases in severe obesity for white females and black males.5 However, present findings show two different trajectory profiles for white females who are more likely to either always be healthy weight or move from healthy weight to obese, compared with white males. An increasing BMI trend was found for black women who were more likely to become obese or severely obese, relative to black men. Therefore, population trends in obesity prevalence are incomplete if not accounting for the interplay of time within the same age, sex, and racial or ethnic groups from year to year (repeated crosssections) as well as patterns within and across subjects followed over decades (longitudinal). Taken together, both perspectives can aid in the meaningful identification of time points for intervention as well as who is most in need of early or sustained intervention. JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
5
RESEARCH In addition to continuing to explore the intersection of time, sex, and race or ethnicity, future studies should also examine the various biological, social, and cultural factors that contribute to differing trajectory profiles that were not addressed in the present study. For example, in Asian participants, an AHW trajectory might be a function of a healthier diet or less sedentary behavior and the increasing BMI pattern may be reflective of generational acculturation to a less healthy diet and lifestyle.30 Moreover, Asian individuals experience higher rates of metabolic syndrome at lower BMIs,31,32 suggesting that healthy weight might need to be defined at a lower range for this population. In addition, black and Hispanic participants were at greater risk of more severe BMI change, relative to white participants. This may reflect cultural values where black and Hispanic individuals have an increased tolerance for adiposity, value a larger body frame, and find larger bodies aesthetically pleasing.33 These biological and sociocultural factors are important considerations for future intervention and prevention studies.
Strengths and Limitations First, LCGM is a useful tool for evaluating patterns in BMI data. LCGM has been used in studies of mental health and dietary patterns34,35; however, it has rarely been used to characterize patterns of weight and can provide valuable information on weight change across the life span. Second, instead of using arbitrary cutoffs or leaving BMI as a continuous variable, we use measured height and weight to form CDC-defined BMI categories (ie, healthy weight, overweight, obese, severely obese) to create trajectories to model real weight change. This categorization is also useful because these cutoffs are used by health practitioners to determine access to treatment options.14 Third, the present study utilized preliminary Wave V data, extending the ages available for analysis beyond previous studies using Add Health.9,10 However, Wave V data are preliminary and have yet to be fully released. As a result, the sample sizes of some groups (ie, Native American) were too small for meaningful analysis. In addition, the objective of this study was to examine patterns of BMI change by sex and race or ethnicity, and the analysis does not capture sociocultural or environmental factors that likely affect BMI change during the transition from adolescence to adulthood, which warrant further study.
CONCLUSIONS This study utilized data from a longitudinal, nationally representative sample to examine BMI trajectory patterns from adolescence to adulthood. Five distinct trajectory patterns were identified, in which subjects either maintained or increased in weight over time. The majority (71%) of subjects increased in weight, leaving only 29% of the sample maintaining a healthy weight trajectory through adulthood. Given the comorbidities of excessive weight gain and severe obesity, the most concerning trajectories were moving from a healthy weight to obese and obese to severely obese. In this study, black and Hispanic females were most likely to fit in these patterns.
References 1.
6
Wang Y, Beydoun MA, Liang L, Caballero B, Kumanyika SK. Will all Americans become overweight or obese? Estimating the progression and cost of the US obesity epidemic. Obesity. 2008;16:2323-2330.
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
2.
Flegal KM, Carroll MD, Kit BK, Ogden CL. Prevalence of obesity and trends in the distribution of body mass index among US adults, 1999-2010. JAMA. 2012;307:491-497.
3.
Ogden CL, Carroll MD, Kit BK, Flegal KM. Prevalence of childhood and adult obesity in the United States, 2011-2012. JAMA. 2014;311:806814.
4.
Skinner AC, Ravanbakht SN, Skelton JA, Perrin EM, Armstrong SC. Prevalence of obesity and severe obesity in US children, 1999-2016. Pediatrics. 2018;141:e20173459.
5.
Skinner AC, Skelton JA. Prevalence and trends in obesity and severe obesity among children in the United States, 1999-2012. JAMA Pediatr. 2014;168:561-566.
6.
Merten MJ. Weight status continuity and change from adolescence to young adulthood: Examining disease and health risk conditions. Obesity. 2010;18:1423-1428.
7.
Brownell K, Wadden T. The heterogeneity of obesity: fitting treatments to individuals. Behav Ther. 2016;47:950-965.
8.
Hales CM, Fryar CD, Carroll MD, Freedman DS, Ogden CL. Trends in obesity and severe obesity prevalence in US youth and adults by sex and age, 2007-2008 to 2015-2016. JAMA. 2018;319:1723-1725.
9.
Gordon-Larsen P, Adair LS, Nelson MC, Popkin BM. Five-year obesity incidence in the transition period between adolescence and adulthood: The National Longitudinal Study of Adolescent Health. Am J Clin Nutr. 2004;80:569-575.
10.
Gordon-Larsen P, The NS, Adair LS. Longitudinal trends in obesity in the United States from adolescence to the third decade of life. Obesity. 2010;18:1801-1804.
11.
Balantekin KN, Hohman EE, Adams EL, et al. More rapid increase in BMI from age 5-15 is associated with elevated weight status at age 24 among non-Hispanic white females. Eat Behav. 2018;31: 12-17. 7.
12.
Harris KM. The National Longitudinal Study of Adolescent to Adult Health (AddHealth), Waves I & II, 1994-1996; Wave III, 2001-2002; Wave IV 2007-2009. Chapel Hill, NC: Carolina Population Center, University of North Carolina at Chapel Hill; 2009.
13.
Centers for Disease Control and Prevention, National Center for Health Statistics. Growth charts. http://www.cdc.gov/growthcharts/. Published 2017. Accessed June 26, 2019.
14.
North American Association for the Study of Obesity, National Heart Lung and Blook Institute. The Practical Guide: Identification, Evaluation and Treatment for Overweight and Obesity in Adults. Bethesda, MD: US Department of Health and Human Services, Public Health Service, National Institutes of Health, National Heart, Lung, and Blood Institute; 2000.
15.
Berkey CS, Colditz GA. Adiposity in adolescents: change in actual BMI works better than change in BMI z-score for longitudinal studies. Ann Epidemiol. 2007;17:44-50.
16.
Cole TJ, Faith MS, Pietrobelli A, Heo M. What is the best measure of adiposity change in growing children: BMI, BMI %, BMI z-score or BMI centile? Eur J Clin Nutr. 2005;59:419-425.
17.
IBM SPSS Statistics for Windows. Version 24 [computer program]. Armonk, NY: IBM Corp; 2016.
18.
Muthén LK, Muthén BO. Mplus User’s Guide. 6th ed. Los Angeles, CA: Muthén & Muthén; 1998-2016.
19.
Ghasemi A, Zahediasl S. Normality tests for statistical analysis: A guide for non-statisticians. Int J Endocrinol Metab. 2012;10:486-498.
20.
van de Schoot R, Sijbrandij M, Winter SD, Depaoli S, Vermunt JK. The GRoLTS-checklist: Guidelines for reporting on latent trajectory studies. Struct Equ Modeling. 2017;24:451-467.
21.
Jung T, Wickrama KAS. An introduction to latent class growth analysis and growth mixture modeling. Soc Personal Psychol Compass. 2008;2:302-317.
22.
Malhotra R, Ostbye T, Riley CM, Finkelstein EA. Young adult weight trajectories through midlife by body mass category. Obesity. 2013;21:1923-1934.
23.
Zheng Y, Manson JE, Yuan C, et al. Associations of weight gain from early to middle adulthood with major health outcomes later in life. JAMA. 2017;318:255-269.
24.
Lake A, Townshend T. Obesogenic environments: exploring the built and food environments. J R Soc Promot Health. 2006;126:262-267.
25.
Hutfless S, Gudzune KA, Maruthur N, et al. Strategies to prevent weight gain in adults: A systematic review. Am J Prev Med. 2013;45:e41-e51.
--
2019 Volume
-
Number
-
RESEARCH 26.
McTigue KM, Harris R, Hemphill B, et al. Screening and interventions for obesity in adults: Summary of the evidence for the U.S. Preventive Services Task Force. Ann Intern Med. 2003;139: 933-949.
31.
Hsu WC, Araneta MRG, Kanaya AM, Chiang JL, Fujimoto W. BMI cut points to identify at-risk Asian Americans for type 2 diabetes screening. Diabetes Care. 2015;38:150-158.
32.
27.
Chang S, Stoll CRT, Song J, Varela JE, Eagon CJ, Colditz GA. The effectiveness and risks of bariatric surgery: An updated systematic review and meta-analysis, 2003-2012. JAMA Surg. 2014;149:275-287.
Palaniappan LP, Wong EC, Shin JJ, Fortmann SP, Lauderdale DS. Asian Americans have greater prevalence of metabolic syndrome despite lower body mass index. Int J Obes. 2011;35:393-400.
33.
28.
Ogden CL, Carroll MD, Lawman HG, et al. Trends in obesity prevalence among children and adolescents in the United States, 19881994 through 2013-2014. JAMA. 2016;315:2292-2299.
Padgett J, Biro FM. Different shapes in different cultures: body dissatisfaction, overweight, and obesity in African-American and Caucasian females. J Pediatr Adolesc Gynecol. 2003;16:349-354.
34.
29.
Simmods M, Llewellyn A, Owen CG, Woolacott N. Predicting adult obesity from childhood obesity: A systematic review and metaanalysis. Obes Rev. 2015;17:95-107.
Hunter AM, Muthen BO, Cook IA, Leuchter AF. Antidepressant response trajectories and quantitative electroencephalography (QEEG) biomarkers in major depressive disorder. J Psychiatr Res. 2010;44:90-98.
30.
Diep CS, Foster MJ, McKyer ELJ, Goodson P, Guidry JJ, Liew J. What are Asian-American youth consuming? A systematic literature review. J Immigr Minor Health. 2015;17:591-604.
35.
Sotres-Alvarez D, Herring AH, Siega-Riz AM. Latent transition models to study women’s changing of dietary patterns from pregnancy to 1 year postpartum. Am J Epidemiol. 2013;177:852-861.
AUTHOR INFORMATION J. N. Becnel and A. L. Williams are assistant professors, University of Arkansas, Fayetteville. Address correspondence to: Jennifer N. Becnel, PhD, HOEC 210, University of Arkansas, Fayetteville, AR 72701. E-mail: [email protected]
STATEMENT OF POTENTIAL CONFLICT OF INTEREST No potential conflict of interest was reported by the authors.
FUNDING/SUPPORT There is no funding to disclose.
ACKNOWLEDGEMENTS This research uses data from Add Health, a program project directed by Kathleen Mullan Harris and designed by J. Richard Udry, Peter S. Bearman, and Kathleen Mullan Harris at the University of North Carolina at Chapel Hill, and funded by grant P01-HD31921 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development, with cooperative funding from 23 other federal agencies and foundations. Special acknowledgment is due to Ronald R. Rindfuss and Barbara Entwisle for assistance in the original design. Information on how to obtain the Add Health data files is available on the Add Health website (http://www.cpc.unc.edu/addhealth). No direct support was received from grant P01-HD31921 for this analysis.
AUTHOR CONTRIBUTIONS J. N. Becnel conceptualized the manuscript and was the primary writer. A. L. Williams conceptualized the manuscript, conducted the data analyses, and was the secondary writer.
--
2019 Volume
-
Number
-
JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS
7