Systematic Review of Prevalence of Young Child Overweight And
Total Page:16
File Type:pdf, Size:1020Kb
SYSTEMATIC REVIEW Systematic Review of Prevalence of Young Child Overweight and Obesity in the United States–Affiliated Pacific Region Compared With the 48 Contiguous States: The Children’s Healthy Living Program We estimated overweight Rachel Novotny, PhD, Marie Kainoa Fialkowski, PhD, Fenfang Li, PhD, Yvette Paulino, PhD, Donald Vargo, PhD, and obesity (OWOB) prev- Rally Jim, MO, Patricia Coleman, BS, Andrea Bersamin, PhD, Claudio R. Nigg, PhD, Rachael T. Leon Guerrero, alence of children in US- PhD, Jonathan Deenik, PhD, Jang Ho Kim, PhD, and Lynne R. Wilkens, DrPH Affiliated Pacific jurisdic- tions (USAP) of the Children’s THERE ARE FEW DATA ON Hawaii was 33% (13% over- Study Selection Healthy Living Program com- overweight and obesity (OWOB) weight and 20% obese) and the Peer-reviewed literature. For our pared with the contiguous fi United States. of children in the US-Af liated risk for OWOB varied by eth- primary data sources, we searched fi We searched peer-reviewed Paci c Islands, Hawaii, and nicity,from2-foldinAsiansto electronic databases (PubMed, literature and government Alaska, collectively referred to as 17-fold in Samoans, compared US National Library of Medicine; 8,9 reports (January 2001–April the US-Affiliated Pacific region with Whites. Data from the EBSCO Publishing; and Web of 2014) for OWOB prevalence (USAP) in this article (Figure A, Commonwealth of the Northern Science) for articles published be- of children aged 2 to 8 years available as a supplement to the Mariana Islands (CNMI) showed tween January 2001 and April in the USAP and found 24 online version of this article at similar OWOB prevalence.10 2014 with the following search sources. We used 3 articles http://www.ajph.org). The USAP Aggregating prevalence esti- terms: child, obesity, overweight, from National Health and has not been included in the Na- mates for the region and by juris- Pacific, Alaska, Samoa, Micronesia, Nutrition Examination Sur- tional Health and Nutrition Exam- diction will allow programs to Hawaii, Marshall Islands, Mariana, veys for comparison. Mixed ination Survey (NHANES) or other target their activities and poli- Palau, Guam. We found 323 models regressed OWOB prevalence on an age poly- national surveillance systems with cies. The purpose of this article articles; 223 were unique and 1,2 nomial to compare trends measured anthropometric data. is to (1) estimate prevalence of we reviewed these for other (n = 246 data points). Native ethnic populations (Native OWOB of children aged 2 to 8 inclusion criteria. In the USAP, OWOB prev- Hawaiians, Pacific Islanders, Alaska years living in the USAP and (2) Publicly available government alence estimates increased Natives) of the USAP have not determine how that prevalence agency data. For our secondary with age, from 21% at age 2 been reported on in national sur- compares with children aged 2 to data sources, we located other re- years to 39% at age 8 years, veillance,3 yet Native Hawaiians 8 years living in the 48 contigu- ports on OWOB from the USAP increasing markedly at age 5 and other Pacific Islanders consti- ous states. by Internet search engine (Google) years; the proportion obese tute 1.2 million people (0.4% of with the same search terms. We increased from 10% at age 2 the total US population) and have METHODS further limited search hits in ex- years to 23% at age 8 years. increased 40% in the past decade,4 cess of 1 million to government The highest prevalence was in American Samoa and and Native Alaskans constitute Investigators from the Chil- agencies that focused on the 2- to 5 ’ Guam. (Am J Public Health. another quarter million people. dren s Healthy Living for the 8-year-old age group (e.g., Head Published online ahead of The USAP has political ties to the Remote Underserved Minority Start, Department of Education, print November 13, 2014: United States (Table A, available as Populations of the Pacific Program Department of Health and Human e1–e14. doi:10.2105/AJPH. a supplement to the online version searched peer-reviewed literature Services, Special Supplemental 2014.302283) of this article at http://www.ajph. and publicly available agency Feeding Program for Women, org).4 data for OWOB prevalence rates Infants, and Children [WIC]). In The high prevalence of obe- in the USAP relative to the Centers addition, we contacted child obe- sity and noncommunicable for Disease Control and Preven- sity experts in the Pacific region diseases in USAP adult popula- tion (CDC) body mass index (BMI; for the relevant government tions6 and consequent state of definedasweightinkilograms agency reports. We found 14 emergency declared7 underpins divided by the square of height reports and reviewed these for the urgency of obesity preven- in meters) reference, as is re- other inclusion criteria. tion, starting with children. The ported by NHANES and has The other inclusion criteria in- mean OWOB prevalence for been used in past reports for cluded (1) English language, the children aged 5 to 8 years in the USAP.9,11,12 main language used for business in Published online ahead of print November 13, 2014 | American Journal of Public Health Novotny et al. | Peer Reviewed | Systematic Review | e1 SYSTEMATIC REVIEW the region; (2) children aged 2 to 8 the best estimate available for any overall estimates were not overly bootstrap analysis performed years were included in the report; particular single age; for instance, influenced by jurisdictions with 500 iterations in which a random and (3) OWOB prevalence (%) if the prevalence was 10% for more publications. The weights selection of data sources with re- in the USAP defined with CDC children aged 2 to 4 years, the best were adjusted so that the total placement was made within juris- body mass index criteria13---15 estimate of the probability of obe- sample size n, defined as the sum diction maintaining the number (‡ 85th percentile and < 95th sity for a 2-year-old child is 10%. of the weights, equaled the num- of data sources per jurisdiction percentile for age and sex was Therefore, a record was created ber of children included in the at each iteration. We performed labeled “overweight”; ‡ 95th per- for each single age in the age model to maintain the correct further subgroup analyses (based centile was labeled “obese”16). group with the age group---specific type I and II errors. Thus, for on jurisdiction, year, source of prevalence and an equal propor- estimation of the overall USAP data, and type of sampling) as Data Extraction and Synthesis tion of the sample size (e.g., a prevalence, each jurisdiction was sensitivity analyses. To test disag- One experienced reviewer prevalence estimate for the age assigned a sample size of pJ n, gregation of published estimates 17 (M. K. F.) independently identi- group aged 2 to 4 years would where pJ is the number of children of age groups into single ages, we fied eligible data sets and recorded lead to 3 records). One investiga- younger than 10 years in the did analysis of variance modelling study year, authors, publication tor (L. R. W.) entered data into 2010 census of jurisdiction (J) of prevalence by age group, using year, location, racial/ethnic group(s), a spreadsheet and a separate in- divided by the total number of the same weighting scheme as ages, sample size, OWOB vestigator (F. L.) reviewed the children younger than 10 years described previously and assign- prevalence, and notes on OWOB data. We performed an inverse across jurisdictions included in ing each data source to age group criteria (list of eligible data sets variance---weighted, fixed-effect the model.42,43 This poststratifi- 2 to 5 years or 6 to 8 years; we available on request). A second meta-regression39 to produce cation weighting44 allows for the assigned estimates to one of these reviewer (F. L.) confirmed the curves for OWOB prevalence overall USAP estimate to reflect categories. data. by single ages. A mixed model the distribution of children across We identified 11 primary and regressed the OWOB prevalence jurisdictions as in a simple random RESULTS 14 secondary data sources from on a polynomial of age (years) sample. 2001 to 2014 from Alaska, accounting for the variance of We used one set of models to Two hundred forty-six single- American Samoa, CNMI, Feder- the prevalence estimates,40 with predict prevalence by single ages year data points resulted from ated States of Micronesia (FSM; polynomials up to the fifth power. for each jurisdiction within the 27 data sources: 3 from the Yap, Kosrae, Pohnpei, and Chuuk), As the power functions were corre- USAP and to test for differences contiguous states (27 data points Guam, and Hawaii (Table 1). Be- lated, we used orthogonal polyno- between jurisdictions using for single ages),11,37,38 3 from cause 2 primary data sources18,19 mials41 to determine the signifi- a global F test of all age power Alaska,25---27 5 from American reported on the same data set, we cance of each independent power components. We used another Samoa,28---31 2 from CNMI,10,18 dropped 1, yielding 10. We found component (linear, squared, cubic, model to predict prevalence by 2 from Guam,33,34 10 from no data sources for the Marshall etc.) and the maximum degree single ages for the USAP region Hawaii,8,9,20---23,35---37,45and 2 Islands or Palau. We used data needed to fit the curve. We also overall and to compare the overall from the FSM.24,32 The 24 USAP from NHANES from 2009 to performed random models and USAP and contiguous US curves sources contributed 219 data 2010, 2007 to 2008, and 2003 the results were similar to the across ages with a global F test.