This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Research Protocols, is properly cited. The complete bibliographic information, a link to the original publication on http://www.researchprotocols.org, as well as this copyright and license information must be included.
Respiratory disease is the third most common cause of death in New Zealand, with Pacific people living in New Zealand bearing the greatest burden of this type of disease. Although some epidemiological outcomes are known, we lack the specifics required to formulate targeted and effective public health interventions. The Pacific Islands Families (PIF) birth cohort study is a study that provides a unique source of data to assess lung function and current respiratory health among participants entering early adulthood and to examine associations with early life events during critical periods of growth.
This paper aims to provide an overview of the design, methods, and scope of the
From 2000-2019, the PIF study has followed, from birth, the growth, and the development of 1398 Pacific children born in Auckland, New Zealand. Participants were nested within the overall PIF study (at ages 18-19 years) from June 2018, and assessments were undertaken until mid-November 2019. The assessments included respiratory and general medical histories, a general physical examination, assessment of lung function (forced expiratory volume and forced vital capacity), self-completed questionnaires (St George’s Respiratory Questionnaire, European Quality of Life 5 Dimensions-3 Level, Epworth Sleepiness Scale for Children and Adolescents, and Leicester Cough Questionnaire), blood tests (eosinophils, Immunoglobulin E, Immunoglobulin G, Immunoglobulin A, Immunoglobulin M, and C-reactive protein), and chest x-rays. Noninferential analyses will be carried out on dimensionally reduced risk and protective factors and confounders.
Data collection began in June 2018 and ended in November 2019, with a total of 466 participants recruited for submission of the paper. Collection and collation of chest x-ray data is still underway, and data analysis and expected results will be published by November 2020.
This is the first longitudinal observational study to address the burden of respiratory disease among Pacific youth by determining factors in early life that impose long-term detriments in lung function and are associated with the presence of respiratory illness. Identifying risk factors and the magnitude of their effects will help in adopting preventative measures, establishing whether any avoidable risks can be modified by later resilient behaviors, and provide baseline measurements for the development of respiratory disease in later adult life. The study results can be translated into practice guidelines and inform health strategies with immediate national and international impact.
DERR1-10.2196/18916
Respiratory disease is the third most common cause of death in New Zealand [
Pacific people (Samoan, Tongan, Cook Islands Māori, Niuean, and Tokelauan) living in New Zealand are the fourth largest population group, and the third largest living in Auckland, New Zealand’s biggest city [
Lung development starts in utero with substantial structural development and continues through early childhood, with the alveoli likely increasing in number, size, and complexity through adolescence [
Current scientific theory indicates that many adult respiratory diseases arise from early events during the period of rapid growth from infancy through childhood [
Intrauterine growth restriction is a risk factor for reduced lung function during infancy [
Once airway restriction has occurred, the composite picture from other overlapping studies in healthy and asthmatic populations suggests that lung function centiles tend to track with time [
In seminal research, Barker et al [
In the Dunedin Study birth cohort, 26.9% of the participants had continuing symptoms of asthma, with half persistent and half recurring in early adulthood [
Fewer studies have examined the impact of risk or protective factors such as vitamin deficiencies, physical activity levels, and breastfeeding on respiratory function and later respiratory disease. Despite early pathological studies that indicated that the alveolar structure was complete by 2 years of age, recent magnetic resonance imaging suggests that the alveoli continue to develop into young adulthood [
Vitamin deficiencies (vitamins A, D, and E) seem to have a greater effect on alveolar development than on airway development [
There are some known gender differences in respiratory disease patterns. Hospitalization rates for total respiratory admissions are higher in boys than girls when less than 15 years of age but become more common in women than in men as adults [
The Pacific Islands Families (PIF) study birth cohort is an ongoing observational study of the health and development of a birth cohort of children of Pacific ethnicity and their parents. The selected findings from the study included a high immunization uptake (89%) among the cohort over the first 2 years of life [
This study will address 3 objectives: (1) estimate the effect of early life (eg, birthweight, antenatal smoke exposure, postnatal smoke exposure) and childhood risk factors (eg, allergies, dwelling conditions from the first 2 years of life, child smoking at 14 years) on peak lung function attainment and respiratory outcomes in Pacific youth aged 18 to 19 years; (2) determine modifiable childhood risk and protective factors; including breastfeeding, immunization, and nutrition during the first 2 years of life; exercise at ages 4, 11, and 14 years; peak flow at ages 6 and 9 years; respiratory infections, respiratory condition–related hospital admissions, and reported breathing problems in the first 2 years of life; and asthma in childhood) on lung function attainment and respiratory outcomes in Pacific youth aged 18 to 19 years; and (3) estimate the population attributable fraction and population avoidable fraction of modifiable early life risk factors and childhood resilience factors on these outcomes. We hypothesize that at age 18 to 19 years, (1) early life risks result in poorer lung function and respiratory outcomes in early adulthood in Pacific youth and (2) protective or resilience factors throughout childhood moderate the impact of these early life risks on these poorer lung function and respiratory outcomes in early adulthood.
The PIF birth cohort study is a multidisciplinary study [
In recognition of attrition and residential mobility of the initial PIF cohort (n=1398, 681 females and 717 males, to n=954, 463 females and 468 males) by age 14 years, we anticipated achieving a sample size of 750 youths from the original cohort. Since June 2018, an assessment of Pacific youth aged 18 to 19 years was initiated. The only exclusion criteria were (1) exclusion of cohort members whose current sickness would prohibit them from producing maximal effort during lung function testing and (2) exclusion of cohort members resident outside of Auckland, as assessments could only be undertaken in Auckland.
Ethical approval for this study was obtained from the Central Health and Disability Ethics Committee on May 24, 2018 (reference 18/CEN/24). Written informed consent was obtained from the youth to participate in an assessment involving a series of physical and clinical assessments and self-administered web-based questionnaires. Arrangements were made for 2 research assistants of Pacific ethnicity to transport participants to and from a clinic set up for the respiratory assessments at the University of Auckland (Tāmaki campus) and Ascot hospital where chest x-rays and blood tests were performed. The youth were thanked with a gift voucher for their participation. The data sets used during this study are available from the corresponding author upon reasonable request.
The primary outcome of this study was the FEV1 Z-score standardized for height, gender, and age (American Thoracic Society/European Respiratory Society criteria using Global Lung Initiative reference values) as a continuous variable. The FEV1 Z-score will be dichotomized to an indicator of a Z-score <−1.64 for estimating the population attributable risk.
Other clinical assessments involve the documentation of participants’ respiratory and general medical histories, medications, a clinical examination (respiratory rate, cardiovascular, and ear, nose, and throat examinations), and a Bacillus Calmette–Guérin vaccine scar presentation, clubbing severity (categorized into mild, moderate, or severe), tonsil score (according to the Brodsky and Friedman Scales [
The number of pulmonary exacerbations in the previous 12 months was documented. Symptom severity (cough, sputum color, and dyspnea) was recorded on validated 5-point scales [
In addition, oxygen saturation (SPO2; Medtronic, Nellcor PM10N) and spirometry pre- and postsalbutamol (EasyOne Air Spirometer, NDD Medical Technologies) were assessed. Spirometry was performed according to American Thoracic Society standards [
Nonfasting blood tests were completed at Ascot Hospital by Labtests, a pathology laboratory service accredited by the International Accreditation New Zealand. Trained phlebotomists drew 10 mL of blood for testing the levels of Immunoglobulin E, Immunoglobulin G, Immunoglobulin A, and Immunoglobulin M, eosinophils, and C-reactive protein. A further 6 mL of serum was stored at Middlemore Hospital tissue bank for future analysis of biomarkers; a separate consent for this analysis was obtained.
Chest x-rays (posteroanterior and lateral) were performed by Ascot Radiology. The radiation dose is 0.02 millisieverts with background radiation in comparison being 3 to 4 millisieverts, which is equivalent to 3 days of usual background radiation exposure. No chest x-rays were taken if a participant was pregnant. Chest x-rays were scored by 2 scorers using the Brasfield scoring system. The Brasfield system [
Body size and composition measurements included height (Seca 213), weight (Tanita BC545), waist circumference with a nonstretchable tape, standing hand-to-foot bioimpedance analysis (ImpediMed Single Frequency 50 kHz Bioimpedance Analyzer, Tanita BC545), and blood pressure using an automated sphygmomanometer (Omron Auto Blood Pressure monitor IA2, Omron Healthcare) with appropriate cuff sizes.
Anthropometric and blood pressure measurements were repeated until 2 measurements were recorded within a predetermined tolerance (weight ±0.5 kg, height and waist ±0.5 cm, and systolic and diastolic blood pressure ±10 mm Hg). BMI was calculated as weight in kg/height in meters squared, and prevalence of obesity, overweight, and thinness were derived, standardized for age and gender using the Cole cutoffs [
Participants self-administered 4 questionnaires on a tablet computer with the research nurse present to assist as necessary: St. George’s Respiratory Questionnaire (SGRQ), European Quality of Life 5 Dimensions-3 Level, Epworth Sleepiness Scale for Children and Adolescents (ESS-CHAD), and the Leicester Cough Questionnaire (LCQ).
The SGRQ [
The European Quality of Life-5 Dimensions is a generic measure of self-reported health status [
ESS-CHAD [
Any abnormal results were discussed on a case-by-case basis between the research nurses conducting the assessments and the coinvestigators of the study with backgrounds in respiratory medicine. In cases of incidental findings or results of concern, a referral letter was addressed to their general practitioner, along with notifying general practitioners of their patients’ involvement in the study and that the chest x-ray and blood tests of their patients will be available from the standard clinical information portal (Concerto).
A research electronic data capture (REDCap) database (hosted on the Auckland University of Technology server and fully compliant with International Organization for Standardization standards and international data management) was set up to capture all data. REDCap is a web-based system that can be used for direct data input or secondary input from paper-based clinical record forms and questionnaires.
The key objective of the analysis was to obtain causal effect estimates of risk factors and modifiable protective factors on respiratory outcomes, conditional on preexisting risk factors and confounders. The analyses will be carried out in 3 stages: (1) dimensional reduction of the covariates, (2) causal modeling of the risk and protective factors, and (3) estimation of the population attributable and avoidable risks associated with the risk and protective factors.
A technical challenge to overcome in this study is the large dimensionality of the covariates involved, potentially leading to overfitting. To alleviate this problem, we will apply dimensional reduction techniques to the covariates, taking care to maintain interpretability where necessary, especially with regard to population attributable and avoidable risks. The dimensional reduction will proceed using variations in sliced inverse regression (SIR) [
With
where α, β, and γ are the causal parameter vectors of interest;
For categorical outcomes, a logit link function is applied to the right-hand side of the above equation (see also the Inferential Setting section). The rest of the data analytical approach follows without modification.
We note that the estimation of α addresses hypothesis 1, whereas the estimation of β and the interaction term parameter γ addresses hypothesis 2. The causal model considered above treats early life respiratory conditions on an equal footing with confounders. We will also attempt the analysis by simultaneously fitting the following components, creating a mediation analysis where early life risk factors may affect later respiratory outcomes:
Direct path:
Mediated path:
In our experience, mediated analyses such as equation 2 can easily become intractable numerically. We are likely to resort to the mediation model when considering specific early childhood conditions with reasonably high prevalence, such as asthma, as opposed to a full set of early childhood conditions.
To allay the bias potentially associated with attrition in the cohort, we applied inverse probability weighting (IPW), a recognized technique to compensate for selection bias [
The population attribute and avoidable risks (PAR) associated with each risk and protective factor will be estimated using a Monte Carlo approach [
As missingness in early childhood data is minimal, we will use a singly imputed data set obtained from a full conditional specification of the covariate distribution using a discriminant function for categorical values, for our analyses. There will be no attempt to impute missing outcome values, if any, from the clinical assessments.
All tests will be carried out at the 5% significance level, against two-sided hypotheses. Estimates will be reported as point estimates and 95% CIs. Standard descriptive quantities and simple regression results will be presented for all outcomes and main risk and protective factors. The main analysis will be handled through generalized additive models under a normal family with an identity link. Assumptions of residual normality will be checked visually and using standard distributional tests; departures from normality will be dealt preferentially with the selection of an alternative family and link, and, in case of variance behavior inconsistent with a known family, with a generalized additive model for location, scale, and shape [
This study was funded in October 2017 and received ethical approval in May 2018. Data collection began in June 2018 and ended in November 2019, with a total of 466 participants recruited for submission of the paper. Collection and collation of chest x-ray data is still underway, and data analysis and expected results will be published by November 2020.
To the best of our knowledge, this is the first observational study to address the high burden of respiratory disease in Pacific youth aged between 18 and 19 years by estimating its causal relationship with factors in early life (risk and protective), which impose long-term detriments in lung function and are associated with the presence of respiratory illness as this population moves into young adulthood and nears the lifetime peak lung function. It is also at the time when the difference in hospitalization rates for any respiratory illnesses moves from being higher in males throughout childhood to being higher in females in adulthood.
Although utilizing the PIF cohort entails some loss in representativeness with regard to the Pacific population at large, it brings advantages by avoiding the costs and risks of establishing a new study de novo. Attrition must be acknowledged in utilizing this PIF cohort, largely because of (1) transient residential mobility by the age of 18 years and (2) in part because of logistical difficulties. However, as an analytical epidemiological study, the sample needs only be representative with regard to the effect of interest to be an internally valid study. Even if, in the proposed study, attrition interacts with the relationship between covariates and outcomes, selection bias will be allayed at the analysis stage by IPW and plausible causal relationships, revealed [
This study will measure current lung function and assess the presence or absence of respiratory disease in the PIF study birth cohort, a group at increased risk of respiratory disease, which mainly resides in a region with a high prevalence of respiratory disease. Findings may be relevant for Māori, with approximately 8% of the original PIF cohort having Māori heritage, and Māori experiencing a similarly unacceptable high rate of respiratory illnesses in New Zealand. Moreover, this information can be directly used to formulate public health strategies to reduce future disease in this high-risk group for life, which will be relevant to the population as a whole.
chronic obstructive pulmonary disorder
Epworth Sleepiness Scale for Children and Adolescents
forced expiratory volume in 1 second
forced vital capacity
inverse probability weighting
Leicester Cough Questionnaire
population attribute and avoidable risks
Pacific Islands Families
research electronic data capture
St. George’s Respiratory Questionnaire
sliced inverse regression
The authors would like to acknowledge the participants and their families for their constant support and participation. The authors also sincerely thank the research team for their tireless efforts with data collection and project completion. This study was funded by the Health Research Council of New Zealand (grant number 17/155). This funding body played no role in the design of the study and collection, analysis, and interpretation of data, or in writing this manuscript.
ET, CB, CW, AV, AT, and LI participated in the study design. ET, SJ, LI, and CB drafted the manuscript. All authors revised the paper and agreed with the final version. All authors agree both to be personally accountable for the author’s own contributions and to ensure that questions related to the accuracy or integrity of any part of the work, even those in which the author was not personally involved, are appropriately investigated, resolved, and the resolution documented in the literature.
None declared.