Table 2.

Qualitative review of research addressing determinants or correlates of obesity in Native Hawaiian and Other Pacific Islander (NHOPI) youth

Citation	Study population	Study Design	Predictors/correlates	Outcome variable measure	Results
Prenatal
CDC, 2011	Washington 28,671 Infant-mother pairs 2,442 NHOPI & 26,229 Asian pairs	Cross-sectional	Maternal characteristics: Pre-pregnancy height & weight Initiation of prenatal care Maternal age Smoking during pregnancy	Infant characteristics: Birth weight (BW) Length of gestation	Infants born to NHOPI mothers significantly more likely than Asian counterparts to be born preterm (P<.001), at high BW (P<.001), or to have received late/no prenatal care (P<.001). NHOPI mothers significantly more likely than Asian mothers to be OWOB (P<.001) before pregnancy, to smoke during pregnancy (P<.001), or to be adolescents (P<.001).
Utz, et al, 2012	Utah 229,598 Adolescent (16 or 17y/o) - mother pairs 2,743 NHOPI, 987 Asian, & 225,868 Non-Hispanic white pairs	Retrospective	Maternal exposure to prenatal care (initiation & utilization) Potential covariates: Adolescent variables (birth weight, exact age at BMI measurement, birth year, gender) Maternal variables (educational attainment, pre-pregnancy BMI, gestational weight gain, smoking / drinking behaviors)	Overweight or obesity (OWOB) in adolescence (BMI)	Adolescents of others who received early (first trimester initiation) or adequate (defined by Kotelchuck Index) prenatal care were significantly less likely to be OWOB (data not shown). NHOPI mothers less likely to receive adequate prenatal care, largely driven by late initiation (P<.05). NHOPI demonstrated the largest protective effect of early prenatal care on the risk of adolescent obesity (P<.001). The effect remained after controlling for maternal education (P<.05), but became non-significant (P=.1) after controlling for pre-pregnancy BMI.
Infancy (birth – 2 y/o)
Hawley, et al, 2014	American Samoa 795 Infants Ages 0–15 mo.	Longitudinal	Feeding mode at 4 mo. (+/− 2 mo.): Breastfed Formula fed Mixed-fed (breast milk, formula, or solid foods)	OWOB at 15 mo. (BMI z-score) Rapid growth (RG) (Conditional gain >.67 z-scores)	Formula-fed infants gained weight & length faster than breastfed infants (P<.05). Formula-fed boys were significantly more likely to be obese at 15 mo. than breastfed boys (P<.01). There were no significant differences in girls at 15 mo., but OWOB was greater in the mixed-fed group. There was a significant difference in RG by feeding mode among boys (27% RG in formula-fed, 17% breastfed, & 6.4% mixed fed) (P<.01), but not girls.
Novotny, et al, 2007	Common - wealth of the Northern Marianas Islands (CNMI) 420 children participating in WIC Ages 6 mo.–10 y/o 54 native Chamorro, 8 native Carolinian, 69 Filipino, & majority mixed ethnicity or of other ethnicities	Cross-sectional	Primary caregiver's report of past breast feeding	OWOB in childhood (BMI)	Any breastfeeding was negatively associated with BMI (after adjusting for age, sex, BW, & mother's years of education) (P=.043). The association of BMI w/ exclusive breastfeeding & duration of breastfeeding were not significant.
Okihiro, et al, 2012	Hawai‘i 389 children Ages 4–5 y/o 66% Native Hawaiian (NH), 21.6% Samoan, & 12.3% Filipino	Retrospective	Growth acceleration during first 2 years of life (consecutive time intervals: 2 days–5 mo., 6–11 mo., & 12–23 mo.) Severe RG (increase in weight-for-length z-score of ?1.0 SD over an age interval) Moderate RG (increase in weight-for-length z-score of ?.67 SD, but <1.0 SD over an age interval)	OWOB at pre-kindergarten (PreK) (BMI)	Severe RG from 12–23 mo. was strongly associated w/ PreK obesity (OR 4.36, 95% CI 1.85–10.27), w/ 48% of these children obese at PreK, compared to 16.7% of children w/ moderate RG & 19.3% w/out RG.
Childhood (3 – 10 y/o)
Pobutsky, et al, 2013	Hawai‘i 12,823 children Ages 4–5 y/o	Cross-sectional	School Complexes	OWOB (BMI)	28.6% of children were overweight or obese (14.4% &14.2%, respectively). Proportions of OWOB were persistently higher (32.5%+) in certain school complexes on O‘ahu (Farrington, Kahuku, Waialua, & Waipahu), as well as some rural & neighbor island school complexes (Lana‘i & Lahainaluna)
Stark, et al, 2011	Hawai‘i 554 children Ages 2–10 y/o 42.6% NH/ part-NH, 6.9% Pacific Islander (PI), 18.8% Asian/part- Asian, 2.7% Hispanic, 18.4% Filipino, 7.6% White, & 5.8% Hispanics, Black & Other	Cross-sectional	Socio-economic status (SES) (Medicaid vs non-Medicaid) Place of residence	OWOB (BMI)	Boys had a higher incidence of OWOB (54%) than girls (46%). No association between SES & OWOB. PI had highest incidence of OWOB (40%) followed by NH/part-NH (19%) & Filipinos (19%). PI 3.6 times more likely to be OWOB. There was a significant relationship between OWOB & place of residence (P=.008). Children from West O‘ahu, Honolulu, & Central O‘ahu/North Shore areas were 2-3 times more likely to be OWOB compared to those from the Windward side.
Novotny, et al, 2013	Hawai‘i 4,608 children Ages 5–8 y/o 13.6% White, 9.4% Asian, 9.9% Filipino, 7.1% NH, 15.2% NH-Asian mixed, 1.9% Samoan, 33.1% other mixed (Incl. African American, American Indian/Aleutian/Eskimos, or other PI), & 9.8% Other	Cross-sectional	Neighborhood education level A subsample (n=2,169) had Vital Records data on: Maternal education level Maternal age	OWOB (BMI z-score)	All children, except Asians, significantly more likely to be OWOB compared to Whites (P<.05). Excess risk varied by ethnic group (Samoan & NH had the highest; OR = 9.4, OR = 2.5, respectively). There was a significant association between ethnic group & neighborhood education level (P<.001), which held after adjusting for age & sex (data not shown). Older maternal age groups (P<.04) & higher maternal education levels (P=.001) were associated w/ lower BMI among children.
Brown, et al, 2011	Hawai‘i 125 kindergarten (K) & third grade students K: mean age 5.6 y/o Third grade: Mean age 8.7 y/o 48.8% NH, 57.8% non-NH	Cross-sectional	Cohort (K or third grade) Sex Hawaiian ancestry Parental educational attainment Household	OWOB (BMI z-score & other adiposity measures eg, waist circumference, abdominal circumference, etc.)	BMI z-scores were significantly higher in third grade male NH children (P<.01). There was no significant ethnic difference in adiposity measures in kindergarteners. Among third grade girls, father's educational attainment was significantly & inversely related to adiposity measures (P<.1). Hawaiian ancestry & income was not significantly related to adiposity measures.
Bruss, et al, 2003 & Bruss, et al, 2005	CNMI 32 primary caregivers of children (ages 6–10 y/o)	Observational	Sociocultural & familial factors	Child feeding practices Perceptions of weight normalcy	Themes: Caregivers, esp. mothers, demonstrate inner dissonance when child-feeding practices conflict w/ cultural values related to food, identify challenges posed by the community as a barrier to healthful eating habits for their children. Cultural differences among ethnic groups regarding children's weight status. Intergenerational conflict related to child feeding between mothers & grandmothers. Both mothers & fathers report intra-family conflict related to child feeding. Parents report avoiding emotional conflicts related to child feeding.
DeRenne, et al, 2008	Hawai‘i 68 K- sixth grade students, enrolled in the A+ afterschool program at two schools About 75% NH	Quasi-experiment	Primary objective: assess feasibility of incorporating physical activity (PA) into an afterschool program Secondary objective: compare effectiveness of two intervention programs: - School A: model curriculum led by trained after school leaders - School B: structured activity program designed & taught by PE teacher	Anthropometric measures (stature, weight, skin fold thickness to determine BMI & estimate body fat) Health-related physical fitness, knowledge & attitudes on PA	After 12 weeks, children from both groups had a mean decrease of 1.2mm in the sum of skinfolds (P<.05) & a significant increase in mean distance covered in the 3-min walk-run test (P<.001). Students in School B had better scores on all variables & significantly lowered BMI (P<.05), did more sit-ups (P<.001), & covered longer distances on the walk-run test (P<.05) than School A.
Adolescence (12 – 18 y/o)
Teranishi, et al, 2011	Hawai‘i 874 children & adolescents Ages 10–17 y/o Over 33% multiracial, 25%White only, 20% Asian only, 20% NHOPI	Cross-sectional	Child's health status (reported excellent-to-poor by parents) Potential covariates: Demographics (Parental education, federal poverty level, insurance type, primary household language, parent nativity)	OWOB (BMI)	Children reported to be in poorer overall health were 2.92 times more likely to be OWOB than those in better health (after accounting for age, race, gender, parental education). Compared to Asian children, NHOPI & multiracial children were 3.04 & 2.31 times as likely to be OWOB. Boys were 1.94 times more likely than girls to be OWOB. Children whose parents' highest level of education was <12 years were 4.40 times more likely to be OWOB than children w/ at least one parent w/ >12 years of education.
LeonGuerrero, et al, 2004	Guam 643 middle & 590 high school students 54% & 53% Chamorro, 32% & 31% Filipino, 6% & 3% Micronesian Islander, 5% Asian, 5% Other ethnicity	Cross-sectional	Demographic characteristics Drug use behaviors	OWOB (BMI)	Adolescent males more likely to be OWOB than adolescent females (P<.01). Filipino adolescents had significantly lower BMI than all other ethnic groups (P<.01). There was a significant difference in percent of OWOB in Chamorro adolescent girls (31.01%) vs Filipino adolescent girls (11.42%) (P<.0001). OWOB adolescents significantly more likely to try marijuana (P<.01). & cocaine (P<.05) than “healthy weight” counterparts. OWOB adolescent girls significantly more likely to smoke cigarettes (30%) than “healthy weight” counterparts (P<.05).

^*NHOPI=Native Hawaiian or Pacific Islander; NH = Native Hawaiian; OWOB=overweight or obese; BMI = body mass index; BW = birth weight; mo. = months; y/o = years old; wks = weeks; w/ = with; & = and