Article Text

Download PDFPDF

Cohort profile: Pacific Islands Families (PIF) growth study, Auckland, New Zealand
  1. E Rush1,
  2. M Oliver2,3,
  3. L D Plank4,
  4. S Taylor2,3,
  5. L Iusitini3,
  6. S Jalili-Moghaddam1,
  7. F Savila3,
  8. J Paterson3,
  9. E Tautolo3
  1. 1Faculty of Health and Environmental Sciences, Child Health Research Centre, Auckland University of Technology, Auckland, New Zealand
  2. 2Department of Epidemiology and Biostatistics, Faculty of Health and Environmental Science, Auckland University of Technology, Auckland, New Zealand
  3. 3Centre for Pacific Health and Development Research, School of Public Health and Psychosocial Studies, Auckland University of Technology, Auckland, New Zealand
  4. 4Department of Surgery, Faculty of Medical and Health Sciences, University of Auckland, Auckland, New Zealand
  1. Correspondence to Professor E Rush; elaine.rush{at}


Purpose This article profiles a birth cohort of Pacific children participating in an observational prospective study and describes the study protocol used at ages 14–15 years to investigate how food and activity patterns, metabolic risk and family and built environment are related to rates of physical growth of Pacific children.

Participants From 2000 to 2015, the Pacific Islands Families Study has followed, from birth, the growth and development of over 1000 Pacific children born in Auckland, New Zealand. In 2014, 931 (66%) of the original cohort had field measures of body composition, blood pressure and glycated haemoglobin. A nested subsample (n=204) was drawn by randomly selecting 10 males and 10 females from each decile of body weight. These participants had measurement of body composition by dual-energy X-ray absorptiometry, food frequency, 6 min walk test and accelerometer-determined physical activity and sedentary behaviours, and blood biomarkers for metabolic disease such as diabetes. Built environment variables were generated from individual addresses.

Findings to date Compared to the Centres for Disease Control and Prevention (CDC) reference population with mean SD scores (SDS) of 0, this cohort of 931 14-year-olds was taller, weighed more and had a higher body mass index (BMI) (mean SDS height >0.6, weight >1.6 and BMI >1.4). 7 of 10 youth were overweight or obese. The nested-sampling frame achieved an even distribution by body weight.

Future plans Cross-sectional relationships between body size, fatness and growth rate, food patterns, activity patterns, pubertal development, risks for diabetes and hypertension and the family and wider environment will be examined. In addition, analyses will investigate relationships with data collected earlier in the life course and measures of the cohort in the future. Understanding past and present influences on child growth and health will inform timely interventions to optimise future health and reduce inequalities for Pacific people.

  • Birth cohort
  • food habits
  • physical activity
  • growth
  • body size

This is an Open Access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See:

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Strengths and limitations of this study

  • This is the only adolescent cohort of Pacific Islands children followed from birth.

  • This multifactorial study measures at age 14 years a diverse range of factors (food and sleep habits, physical activity and sedentary behaviour, and environmental factors) in relation to growth and physical development.

  • Blood was sampled for the first time in the nested substudy for biomarkers of health and disease.

  • This study has experienced high attrition of participants due primarily to high emigration from New Zealand.


Pacific people are the fourth largest ethnic group in New Zealand (NZ), and the third largest group in Auckland, NZ's biggest city.1 The rate of population growth for Pacific people in NZ is approximately three times that of the general population. Estimates suggest a rise from 7.4% who identify with one or more Pacific ethnic groups in 20131 to 10% in 2026. This population is youthful and highly urbanised—almost half are aged <20 years, and over two-thirds reside in Auckland. Within the umbrella grouping of being ‘Pacific’, substantial diversity exists among Pacific communities in terms of language, heritage, ethnic affiliations and cultural practices. Pacific people, and in particular Pacific children and young people, have higher health risks and less healthy behaviours than other NZ people.2

With ageing, the prevalence of obesity and overweight increases,3 and from the 1980s the rate of increase within a generation has also increased. For NZ Pacific people in 2015, the prevalence of obesity was disproportionately high at 66% for adults and 30% for children, compared with 30% and 11% in the general population, respectively.2

The food and physical activity environment is recognised as having a strong effect on both physical activity and nutrition behaviours; however, until now almost all research in this field has been limited by cross-sectional design and a focus on adults or children only. Food insecurity, where food sometimes or often runs out due to lack of money, is associated with increased prevalence of obesity.4 In 2008/2009, the NZ Adult Nutrition Survey5 reported that food insecurity increased from 22% of respondents in 1997 to 40%, with 50% of Pacific women reporting food insecurity.

Adolescence represents a critical time of growth and puberty is marked from the first appearance of secondary sexual characteristics to full sexual maturation.6 Adolescence, a complex and interesting time, links childhood to adulthood. Patterns of growth and maturation during adolescence may be explained by previous life course events or help explain future risk for cardiometabolic health.7

The Pacific Islands Families (PIF) study provides a unique source of data for research on growth, development and psychosocial functioning at critical developmental stages within the family environment. In 2000 (at the time of the birth of their child), 40% of mothers in the longitudinal PIF study reported food insecurity.8 At 4 years, a positive association was observed between body size and consumption of protein and dairy foods,9 and at 6 years obesity (27%) and overweight (31%) prevalence were positively associated with birth weight.10 Another examination with this cohort showed that at 6 years (the transition to school), television watching and maternal waist circumference were associated with child body fatness,11 and physical activity (measured by an accelerometer) of the mother was positively associated with activity of her child.12

At age 14–15 years, we began a study to investigate the relationships of food and activity patterns, pubertal development and risks for diabetes and hypertension with body size; biomarkers for metabolic syndrome, liver function and inflammation; body fatness and growth rates; and individual, social and built environment factors associated with growth. This article describes the study protocol and presents preliminary descriptive results on body size and composition.

Cohort description

From 2000 to 2015, the longitudinal PIF birth cohort study has followed the growth and development of 1398 Pacific children recruited following birth at Middlemore Hospital, South Auckland. This sample size was defined to enable findings to be generated that were specific to the predominant Pacific groups residing in NZ (Samoan, Tongan, Cook Islands Māori). Core measures include items on general health, child behaviour, relationships, acculturation and growth and development. Where specific priority social or health issues have arisen for Pacific peoples throughout the study, additional substudies have been conducted as appropriate. The last data collection wave occurred in 2011.

Between May 2014 and July 2015, informed consent was obtained from youth and their primary caregivers to participate in an assessment involving a series of physical measurements and a self-administered online questionnaire completed by the youth on a tablet computer with the assessor present to assist as necessary. A team of three assessors conducted the assessments in groups in secondary schools. Youth not able to be assessed at school were assessed at their home or other location convenient to them. Youth were thanked with a gift voucher for their participation. A total of 931 youth from the cohort were followed up. In cases where children exhibited metabolic risk (systolic blood pressure >130 mm Hg and/or glycated haemoglobin (HbA1c) >40 mmol/mol), parents were notified by telephone and by post and advised to take their child to their general practitioner for further tests. Thirty-two children were identified with HbA1c >40 mmol/mol and only four had previously been diagnosed, one with type 1 diabetes. Where parental consent was obtained, measurements of concern were provided to the child's general practitioner.

At approximately the half-way point through data collection with the full cohort, a nested substudy started with a random selection from those youth who had already been measured in the full cohort assessment (up until that point), stratified by sex and body weight decile at 11 years. Youth selected in this way were visited by a Pacific research assistant and invited to participate in the substudy. Informed consent (separate to that for the full cohort assessment) was obtained from the parent and the youth who wished to participate and a food frequency and sleep habits questionnaire was administered at this time. Arrangements were made for the research assistant to transport the child to and from the Body Composition Laboratory of the University of Auckland, Department of Surgery, based at Auckland City Hospital. Children arrived at the laboratory in an overnight fasted state and at the completion of the assessment they were provided breakfast and thanked for their participation with a gift voucher. They were fitted with an ActiGraph accelerometer on an elasticated belt to be worn at the waist during waking hours over the next seven consecutive days, to be removed only during bathing, swimming or sleeping and playing contact sports (eg, rugby) and then transported back home or to school. On the eighth day, the accelerometer was collected from the youth's home by the research assistant. The target sample size was 10 participants in each stratum. Between October 2014 and February 2016, a total of 204 youth were assessed in the nested substudy.

For both the full cohort and the nested substudy assessments, parental consent was obtained to use the youth's current and historical home and school addresses in geographic information system analyses, for the purposes of assessing the child's physical and built environment.


The full cohort assessment included body size and composition (height (Seca 213 Hamburg, Germany), weight (Tanita BC545, Tokyo), waist and hip circumference with a non-stretchable tape and standing hand-to-foot bioimpedance analysis (ImpediMed Single Frequency 50 kHz Bioimpedance Analyser, Imp-DF50; Impedimed, Brisbane, Queensland, Australia)), grip strength (Mentone, Melbourne), blood pressure using automated sphygmomanometers (Omron Auto Blood Pressure monitor T8;Omron Healthcare, Kyoto, Japan) with appropriate cuff sizes and a screening test for HbA1c (Afinion AS100 Analyser, Oslo, Norway), with a non-fasting finger prick blood sample of 1.5 μL. Validated questionnaires assessed physical activity and sedentary behaviours,13 food patterns14 ,15 and pubertal status.16

Body mass index (BMI) was calculated as weight in kg/height in metres squared and prevalence of obesity, overweight and thinness were derived, standardised for age and gender using the Cole cut-offs.17 The resistance measurements were multiplied by 1.028 to account for the standing position.18 Fat-free mass and percentage body fat were derived using the prediction equation validated previously with Pacific children and the same bioimpedance device.19

The substudy assessment included food frequency and sleeping times by online questionnaire and collection of a 15 mL fasting venous blood sample for biomarkers for metabolic syndrome (blood pressure, dyslipidaemia, glucose), liver function and inflammation. A further 5 mL of plasma was stored for future analysis of other non-communicable disease markers (where separate consent for this was obtained). Body composition analysis was by bioimpedance analysis as for the full study and by dual-energy X-ray absorptiometry (DXA, model iDXA, GE-Lunar, Madison, Wisconsin, USA) for total body and abdominal fatness and bone mineral content. Height, weight, waist circumference and blood pressure and resting pulse were measured. Anthropometric and blood pressure measurements for both the full cohort and substudy were repeated until within a predetermined tolerance (weight ±0.5 kg, height and waist ±0.5 cm and bioimpedance ±5 Ω, systolic and diastolic blood pressure ±10 mm Hg). After screening with the Physical Activity Readiness Questionnaire, physical function and fitness were measured with the standard 6-min walk test20 and a grip strength test.21 ,22 Physical activity was objectively assessed using GT3X+ ActiGraph accelerometers (Actigraph, Pensacola, Florida, USA). The reliability and validity of these accelerometers are well established.23–25

Units were initialised using ActiLife 6 (Actigraph, Pensacola, Florida, USA), using a raw data sample frequency of 30 Hz. Data were extracted into ActiLife as raw .gt3x files, and aggregated into .agd and .dat files using 15 s epochs and all download options checked (ie, steps, lux, inclinometer, low frequency extension enabled). Files were then converted to .csv within Meterplus (Santech, San Diego, California, USA). Downloaded data were screened on a weekly basis to identify any obvious accelerometer malfunctions or outliers.26 Accelerometer count thresholds of Evenson et al27 were employed to classify time spent sedentary, and in light, moderate and vigorous intensity physical activity (sedentary 0–25 counts 15 s/, light 26–573 counts 15 s/, moderate 574–1002 counts 15 s/, vigorous ≥1003 counts 15 s/).27 Non-wear time was classified as 60 min or more of consecutive zero counts.28 Criteria for valid days were at least 7 hours of data; participants with three or more valid days will be included in analyses (Paper under consideration).

Built environment

Objective built environment measures will be calculated in ArcGIS V.10.2 (ESRI, Redlands, California, USA), as detailed below. Variables were calculated around individual residential addresses and school locations using three street network buffers (400, 800 and 1200 m). This range of buffer sizes was employed to facilitate adaptability in the light of future emerging research and flexibility for differing research questions.29 ,30


Drawing on the work of Frank et al,31 three individual walkability components were derived around individual residential addresses: (1) Net residential density, the ratio of residential dwellings to the residential land area, (2) Street network connectivity, the ratio of number of intersections with three or more intersecting streets per square kilometre to land area and (3) Land use mix, an entropy index based on the presence or absence of five types of land use.

Accessibility to destinations: in the neighbourhood (eg, shops, doctor) was assessed using the Neighbourhood Destination Accessibility Index around individual addresses.32 ,33

Availability of public open space: was determined by calculating the ratio of green space, including parks and playgrounds, that can be freely accessed by the public to the defined area for each individual address.34

Accessibility to green and blue spaces: was derived by calculating the shortest street network distance from each individual's home address to the nearest green space and to the nearest blue space (eg, waterways, rivers, coastline).

Route to school

(1) Distance to school was calculated using the shortest street network distance from home to school and (2) Steepness of the route to school was assessed, and the percentage of school route segments with a slope exceeding 5% calculated.

Pedestrian network connectivity: was calculated around both the home and the school using the ratio of the pedestrian network area to the maximum possible area within the defined boundary.35

Traffic speeds exposure: around both the home and school was assessed using road hierarchy as a proxy. The total lengths of all higher speed roads (≥ 60 km/hour) and all lower speed roads (<60 km/hour) were calculated within the defined boundary. Thereafter the ratio of high to low speed roads was calculated.

Data analysis

Both the physical measurements data and the questionnaire data for the full cohort and substudy were entered into SurveyMonkey (with validation rules applied) and exported as SPSS (V.22, IBM) files. The SPSS files were restructured and cleaned. Descriptive statistics were calculated from the cleaned data sets, including means and SDs for numeric measurements and percentages for categorical variables.

Findings to date

For the full cohort study conducted between May 2014 and July 2015, retention was 66% of the original birth cohort. Of the 931 children seen, 788 were assessed at school, 133 were assessed at home and 10 who were not able to be physically seen completed the questionnaire part of the assessment via an online survey (figure 1). Of the 140 children assessed at home or online, 87 were assessed in Australia where they lived. Relocation to areas other than Auckland had occurred for 12% of children (108/931). For the nested substudy, 204 children living in South Auckland were assessed at the laboratory with an average measurement duration of 53 min.

Figure 1

Consort style diagram of recruitment and retention by gender and location at 14 years.

Measurements undertaken to understand physical growth build on relevant measures from previous years (table 1). The participants in the nested substudy (table 2) were not (statistically) significantly different to the full cohort in terms of their anthropometry, blood pressure, body composition and HbA1c. Compared with the CDC reference growth curves37 with mean SD scores (SDS) of 0, this cohort of 14-year-olds was taller, weighed more and had a higher BMI (mean SDS height >0.6, weight >1.6 and BMI >1.4) (table 2). Seven of 10 youth were overweight or obese. The prevalence of obesity and overweight had increased from 2011 to 2014, less in boys (65.5% to 66.4%) than in girls (66.7% to 81.0%) according to International Obesity Taskforce grades.

Table 1

From 2000 to 2015, age and number of participants measured for variables associated with physical growth and risk for future disease

Table 2

Physical characteristics of youth at 14 years in full cohort and subsample

Strengths and limitations

Culturally specific longitudinal research is essential to identify antecedent factors for growth trajectories from birth to adolescence.38 Despite these research needs, Pacific youth are generally under-represented in physical activity and health research. Building on 14 years of family engagement, this is a ground-breaking epidemiologically robust cohort focusing on determining the pathways leading to successful adaptation as the youth and their families negotiate critical developmental transitions. Unique findings from this research can be used to drive effective policies and programmes that serve the needs of Pacific families, promote well-being and address social disparities. The generalisability of potential outcomes of PIF research to migrant populations in other countries is an important potential contribution to international research in health and related areas. The major limitation of this study is the high attrition rate due in part to the transience/migration of the cohort and the logistical difficulties and follow-up costs for those in other geographical areas.

Understanding past and present influences on child growth and health will inform timely interventions and policies to optimise future health and reduce inequalities for Pacific peoples.


The authors wish to acknowledge the participants and their families for their ongoing support and participation. They also sincerely thank Desmond Ryan and Wafa Elasheg for their tireless efforts with data collection and completion of the project.



  • Contributors ER, MO, LDP, ST and JP participated in the study design. ST, LI and ER performed the editing and analysis of data. ER, MO and LI drafted the manuscript. All authors revised the article and agree with the final version and findings.

  • Funding This study was funded by a grant from the Health Research Council of New Zealand (14-156).

  • Competing interests None declared.

  • Ethics approval Ethical approval for the full cohort study was obtained from the Southern Health and Disability Ethics Committee on 4 December 2013 (ref. 13/STH/159) and for the nested substudy from the Central Health and Disability Ethics Committee on 28 July 2014 (ref. 14/CEN/108).

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement Researchers can apply to obtain raw data from the director of the Pacific Islands Families Study, Dr El-Shadan Tautolo,