Monitoring metabolic health in the adolescent population of Rarotonga: a multi-year cross-sectional study from 2016 to 2023

Abstract

Background

The Cook Islands, a small Pacific nation, carries a high burden of obesity (75%) and non-communicable disease (NCD) among adults. Primary prevention before adulthood, including school-based strategies, is a priority. Local evidence is crucial but challenging in resource-limited, low-population settings. This study presents data on the prevalence of risk factors associated with metabolic syndrome (MetS) among adolescents in Rarotonga, the main island of the Cook Islands.

Methods

A multi-year cross-sectional survey of NCD risk factors in Year 9 students (median age: 13.8 years) between 2016–2019 and 2022–2023 (n = 783) represented 65% of the cohort (n = 1214). Data analysis used descriptive statistics with 5% significance testing.

Results

The prevalence of risk factors include: 22.1% overweight, 37.6% obese; 39.1% Waist-to-Height ratio (WtHR) > 0.5 (study years: χ² p < 0.001); 9.4% high Blood Pressure (BP) (gender: χ² p < 0.001, study years: χ² p = 0.005); 8.1% elevated Total blood Cholesterol (TC) (study years: χ² p = 0.003); 37.1% elevated fasting Blood Sugar Levels (fBSL) (study years: χ² p = 0.007). Where all measures were available (n = 261), 27.6% of participants were living with obesity and ≥ 2 risk factors. Body-Mass-Index-for-Age (BMI-for-age) percentile and WtHR risk category proportions were positively correlated (χ² p < 0.001) with 78.9% of those with obesity falling into the increased risk WtHR category. Significant associations (all p < 0.001) were identified between elevated BP and BMI-for-age categories, WtHR, and TC levels.

Conclusions

The multi-year cross-sectional approach enabled valid data within a low-resource setting that confirmed extremely high levels of overweight and obesity in the Year 9 student population. We argue that including waist circumference measures alongside BMI-for-age was a first step to enabling greater depth of understanding of the risk levels in this population. Some risk factors associated with MetS have been assessed, but investment in improving glucose and lipid monitoring is strongly recommended. This study highlights the importance of monitoring paediatric MetS in the Cook Islands. It suggests that it should be used longitudinally to inform the impact of interventions from early life through to adolescence.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12889-026-28637-x.

Introduction

The Cook Islands is comprised of 15 small islands in the southern Pacific Ocean, with a resident population of 15,000 [1]. As a Small Island Developing State (SIDS), the Cook Islands manages challenges associated with geographic isolation, vulnerability to natural disasters, limited economic and human capacity, and difficulty providing health services [2–4]. Like most Pacific Island Countries and Territories (PICTs), the Cook Islands is experiencing a crisis of obesity and related non-communicable diseases (NCDs), primarily cardiovascular diseases (CVDs), diabetes, cancer, and chronic respiratory diseases [5–7]. In parallel with significant changes in the food environment, the prevalence of overweight, obesity, and NCDs in the Cook Islands has steadily increased over the past century [7–9]. Evidence for adults ages 18–69 years indicates 92.6% with Body-Mass-Index (BMI) ≥ 25 and 75% with BMI ≥ 30; 99% live with at least one, and 62% ≥ three NCD risk-factors; 34% live with type 2 diabetes [6]. NCDs constitute the leading causes of disability and premature mortality, thereby imposing significant social and economic burdens for the Cook Islands [6, 7, 10].

Metabolic Syndrome (MetS) is a cluster of conditions associated with NCDs, particularly cardiovascular diseases, type 2 diabetes mellitus, and chronic kidney disease [11]. MetS is characterised by several risk factors, including abdominal obesity, insulin resistance, hyperglycaemia, dyslipidaemia, and hypertension [11]. The correlation between the aggregation of these factors (typically three or more) and the risk of NCDs is primarily attributed to their role in promoting oxidative stress accompanied by or linked to vascular inflammation [12].

Sustainable development addressing the NCD crisis in PICTs is driven by whole-of-government system-based approaches [13, 14]. Region-wide collaboration is guided by the Pacific NCD Roadmap [15, 16], with the Monitoring Alliance for NCD Action (MANA) tracking indicators, including risk factors for MetS, used to inform evidence-based actions [7]. Recommendations for actions promoting sustainable development for health within the roadmap prioritise primary prevention alongside secondary prevention and treatment, acknowledging the complex interactions between genetics, behaviours, and environment that impact NCD risk [16]. Associations between nutritional and non-nutritional environmental exposures in early life and vulnerability towards later-life development of NCDs underscore the importance of primary prevention [17–19]. Thus, strategies from pre-conception through adolescence and early adulthood contribute towards approaches to NCD risk reduction, impacting current and future generations [20–24]. While definitions of paediatric MetS are contested [25–27], there is growing recognition of increasing levels of risk factors associated with MetS in children and adolescents, which, when clustered, increases the risk for the development of CVDs and type 2 diabetes in adulthood [28]. Collectively, this evidence underscores the urgent need to assess the prevalence of risk factors associated with MetS in childhood and adolescence and to intervene to reduce risk factor development in this demographic [29].

Over the past decade, the Cook Islands have strengthened the contribution of school-based approaches to primary NCD risk reduction via the Pacific Science for Health Literacy Project (PSHLP), recently renamed Oraʻanga Tūmanava [30]. This education-health-science-community partnership involving the Ministries of Education and Health and the University of Auckland draws on evidence from the field of Developmental Origins of Health and Disease [17–19] to inform primary prevention opportunities in adolescence and parenthood [21, 22, 24]. School-based programmes address education and health goals linked to scientific and health literacies, critically engaged citizenship and NCD risk reduction [31]. Teenagers explore evidence related to the NCD crisis via programmes that support core curriculum goals in science, health and social studies. Evidence that feeds into the MANA dashboard from the WHO STEPwise approach to NCD risk factor surveillance (STEPS) survey [32] and the WHO Global School Based Student Health Survey (GSHS) [33, 34] is used to create age-appropriate learning resources exploring local data. STEPS measures key behavioural and biological factors to monitor population-level NCD risk in adults aged 18–69; GSHS measures behavioural risk and protective factors in young people ages 13–17. The lack of biological data in the adolescent population was noted as a gap when PSHLP was initiated in 2013. By 2015, teenagers engaging in PSHLP learning modules were requesting greater access to personal metabolic health data such as blood sugar and cholesterol [35]. Acquiring local, context-specific evidence is vital for crafting and monitoring targeted interventions. However, in PICTs, this task is complicated by small populations and resource constraints, which create challenges related to capability, capacity, and community trust and acceptance [36].

Addressing the gap identified by professionals and teenagers, the Cook Islands Year 9 (Y9) Metabolic Health Monitoring programme was established to provide an integrated health-education approach to informing health promotion related to NCD prevention for teenagers in the Cook Islands. An earlier publication presented aspects of preliminary data (2016–2019) [37]. Our recent article reporting on the evolution of the programme from pilot to sustainable practice focused primarily on the co-design process and qualitative data related to this but also included a high-level overview of health data (2016—2023), enabling clarity of context [35]. We now present previously unreported in-depth analysis of data from the programme (2016–2023), investigating the prevalence and characteristics of risk factors associated with MetS among adolescents in Rarotonga.

Methods

This multi-year cross-sectional study used data from the Y9 Metabolic Health Programme conducted in Rarotonga from 2016–2019 and 2022–2023 [35]. Routine health monitoring was suspended during the COVID-19 pandemic resulting in no data collection during 2020 and 2021. To address the lack of biological evidence associated with MetS related factors in adolescents, the existing biennial School Physical Health Examination (SPHE) programme was extended for Y9 classes. The SPHE, which included height and weight but no other MetS-relevant factors, was expanded in Y9 classes to include waist circumference, blood pressure, blood sugar and total blood cholesterol and conducted annually on a trial basis. This enabled assessment of individual MetS related factors and investigation of clustering of ≥ 2 risk factors.

To address concerns relating to student wellbeing and personal data access, the programme was integrated into existing Y9 PSHLP cross-curricula learning modules exploring biological, environmental and social factors associated with the NCD epidemic [38, 39]. A resource was developed for students to record health data in a personal booklet with explanations of measures [40]. This integration of health data collection with curriculum-linked learning in the Y9 Metabolic Health Programme is underpinned by cultural values supporting a trusting environment acceptable to students, families, and professionals (Fig. 1) [35].

Fig. 1.

PSHLP/Oraʻanga Tūmanava Y9 Metabolic Health Programme. Interconnectivity of curriculum-linked learning with health data collection enabled by culturally embedded values [35]

The challenge of small population size was addressed by engaging a multi-year cross-sectional study design [41]. Capacity limitations for Te Marae Ora, Cook Islands Ministry of Health (TMO) relating to data collection, processing, and analysis capacity were addressed through professional development. Training and standard operating procedures were developed to upskill and mentor additional public health staff for data collection and cleaning. University of Auckland collaborators initially led data analysis, which gradually transitioned to the TMO Public Health team through professional development, supporting resources and mentoring [42].

Participants and consents

This study involved Y9 students from four schools in Rarotonga, the most populated island of the Cook Islands. These schools have a combined average annual Y9 cohort of 211 and serve 96% of Y9 students in Rarotonga. The associated learning and teaching modules were business-as-usual in participating schools and thus are offered to all students. Consent for additional data collection within routine health checks was obtained from TMO, the Ministry of Education, and school principals. Routine data collection was offered to all students in the target year group (Y9). Parent/caregiver consent is always confirmed before students participate in routine health data collection in Cook Islands schools. Evidence from health checks was reviewed by clinicians, offering potential for follow-up where needed.

Ethics approval for University of Auckland staff to work with TMO staff on de-identified data from the extended routine school health checks was obtained from Auckland Health Research Ethics Committee Ref. 24,668. The Cook Islands National Research Committee did not require formal approval for this study.

Data collection

Data collection was carried out by TMO as an add-on within regular school-based health monitoring. Annual training was provided for data collection teams using Standard Operating Procedures developed for the programme [43] and following protocols agreed between the Ministries of Education and Health. Training included assessment of inter-observer consistency. Consistent supervision of data collection was used to support measurement reliability. All measurement devices were calibrated in compliance with standard protocols for the Cook Islands Ministry of Health. Stations operated by health professionals were set up in schools for measurement during class time, assessing the risk factors detailed in Table 1. Students could opt out of any measure. Where this occurred, a null result was recorded. Where blood sugar testing was conducted, staff used a conversation to check whether participants had adhered to the 8-h fasting regime. Results were recorded in spreadsheets and each student’s Health Profile booklet [40]. For measurements outside the normal range, parents/caregivers were contacted within a week to discuss results and arrange necessary referrals.

Table 1.

Summary of health measures and assessment metrics [35]

Measurement	Hardware/Consumables	Process Requirements	Assessment metric	Categories
Weight	SECA 813 digital scale	Barefoot, light clothing; to nearest 0.1 kg	Body-Mass-Index-for-Age (BMI-for-age, kg/m2) [44]	Underweight: < 5th percentile Healthy Weight: 5th to < 85th Overweight: 85th to < 95th Obesity: ≥ 95th percentile
Height	Portable, wall mounted stadiometer	Barefoot; to nearest 0.1 cm
Waist Circumference	MyoTape flexible measuring tape	Midpoint between top of hip bone and bottom of ribs following exhalation; to nearest 0.1cm	Central Obesity Risk via Waist-to-Height Ratio (WtHR) calculation [45] or Waist circumference percentiles [46]	WtHR Low risk: ≤ 0.5 Increased risk: > 0.5	WC Increased risk: ≥ 90th percentile
Blood Pressure	Blood pressure cuff and Omron HEM-7203 automatic monitor	3–5 min seated resting; three measures, discard 1st measure;	Age-based Paediatric Blood Pressure (BP, mmHg) [47]	Normal: SBP < 120 and DBP < 80 Elevated: SBP 120–129 and DBP < 80 HTN 1: SBP 130–139 and/or DBP 80–89 HTN 2: SBP ≥ 140 and/or DBP ≥ 90
Blood Sugar	Glucose strips, disposable lance, and SD Codefree glucose meter	Finger prick discarding 1st drop; record in mmol/L; minimum 8h fasting status recorded	Blood Sugar Levels (BSL, mmol/L) [48]	Fasting BSL (fBSL) Normal: < 5.6 mmol/L Raised: ≥ 5.6 mmol/L	Random BSL (rBSL) Normal: <11.1 mmol/L Raised: ≥11.1 mol/L
Total Blood Cholesterol	Cholesterol strips, disposable lance, Accutrend cholesterol meter	Finger prick discarding 1st drop; record in mmol/L; fasting status recorded	Total Cholesterol (TC, mmol/L) [49]	Normal: < 5.0 mmol/L Raised: ≥ 5.0 mmol/L

SBP Systolic Blood Pressure, DBP Diastolic Blood Pressure, HTN 1 Stage 1 Hypertension/High-Normal, HTN 2 Stage 2 Hypertension/High

Data analysis

Data analysis used descriptive statistics and significance testing between sex and collection years (p < 0.05). This included the Mann–Whitney U test (for sex differences) and Kruskal–Wallis H test (for study year differences) for BMI-for-age and chi-square tests for BMI-for-age, Waist-to-Height Ratio (WtHR), total cholesterol (TC), blood sugar (fasting Blood Sugar Levels: fBSL, random Blood Sugar Levels, rBSL), and Age-Based Paediatric Blood Pressure (BP). Statistical analyses were conducted using R (V4.3.1) or SPSS (V31.0.1). Missing data were eliminated from analysis of individual and collective measures.

Results

Demographic data

Between 2016–19 and 2022–23, data was collected from 783 Y9 students [35]. Participant distribution aligned with the schools' populations, exhibiting a balanced gender ratio and a median age of 13.8 years (Table 2). The collection of ethnicity data is not usual in the Cook Islands. Approximately 85% of Rarotonga’s population identify as Cook Islands Māori [1].

Table 2.

Participation by study year and school [35]

Year		Participants by School (n) (NA is recorded where a school did not participate in a particular year)							Participation Rate
		A	B	C	D	Total	Median Age (years)	% Female	% Y9 cohort in participating schools	% Total Y9 cohort in Rarotonga
2016	Yr 9	50	13	NA	7	70	13.8	54.3	42.4	38.5
2017	Yr 9	117	22	NA	NA	139	13.8	51.1	83.2	73.9
2018	Yr 9	84	17	10	9	120	14.1	52.5	67.0	64.9
2019	Yr 9	93	32	NA	11	136	13.7	51.5	65.7	63.6
2022	Yr 9	128	24	7	11	170	13.8	45.3	81.7	75.2
2023	Yr 9	100	26	4	18	148	13.6	58.8	72.5	67.6
Total		572	134	21	56	783	13.8	51.9	67.6	64.5

BMI-for-age and WtHR

Figure 2 presents the percentage distribution of BMI-for-age for the combined Y9 cohorts (n = 778) and across study years. Similarly, Fig. 3 presents the percentage distribution of WtHR for the combined Y9 cohorts (n = 624) and across study years. Risk factors observed include 0.7% underweight, 22.1% overweight, 37.6% obese, and 39.6% with increased WtHR. There was no significant difference in the distribution of BMI-for-age categories across years compared to what would be expected if independent (χ² p = 0.19, Kruskal–Wallis H test p = 0.61). However, there was a significant difference in the distribution of WtHR categories across years (χ² p < 0.001). Differences in observed data patterns in 2018 and 2023 compared to other years were key contributors to this significant p-value. No statistically significant difference between sexes was observed for BMI-for-age (p = 0.12) or WtHR (p = 0.16).

Fig. 2.

Percentage distribution of BMI-for-age for Y9 cohort, cumulative and by study year. a shows the cumulative proportion in each BMI-for-age category across all study years, while (b) presents individual study year proportions. There was no significant difference in the proportion of BMI-for-age categories between study years (χ2 p = 0.19, Kruskal–Wallis H test p = 0.61)

Fig. 3.

Percentage distribution of WtHR for Y9 cohort, cumulative and by study year. a shows the cumulative proportion in WtHR categories across all study years (excluding 2019, no measurements recorded), while (b) presents individual study year proportions. The average WtHR was 0.48, just below the at-risk cut-off (increased risk if WtHR > 0.5). There was a significant difference in the proportion of WtHR categories between study years (χ2 p < 0.001), attributed to fewer participants in the low-risk WtHR category than expected in 2018 and more participants in the low-risk WtHR category than expected in 2023, as indicated by the arrows shown in the graph

The correlation between BMI-for-age and the proportion of individuals categorised as high-risk based on WtHR demonstrated a positive trend (Fig. 4). Most participants in the Not Overweight/Obese BMI-for-age group (93.6%) and a substantial portion of those in the Overweight category (68.4%) were classified as Low-Risk according to their WtHR. In contrast, the group living with obesity predominantly comprised individuals falling into the Increased Risk WtHR category, accounting for 78.9% of the observations.

Fig. 4.

BMI-for-age percentiles against WtHR, by risk category, Y9 cohort 2016–2023. Coloured segments differentiate between categories; for example, the darkest blue segment in the top-right-most quadrant highlights those in the increased-risk WtHR category and Obese category. There was a significant difference in the distribution of BMI-for-age categories across WtH R categories (χ2 p < 0.001), attributed to more participants in the healthy weight/low-risk than expected if independent and more participants in the obese/high-risk than expected

BP, TC, and fBSL

Figure 5 presents the cumulative and study year data for BP, TC, and fBSL. Risk factors observed include 9.4% with high BP, 8.1% with elevated TC, and 37.1% with elevated fBSL. All three variables show a significant difference across years, with variations in observations in 2018 compared to other years being a key contributor to significant p-values. There was only a significant difference between sexes for BP (χ² p < 0.001). Specifically, female normal BP levels were higher than expected, while male normal BP levels were lower than expected.

Fig. 5.

Y9 cohort 2016–2023, cumulative and by year, percentage distribution for BP, TC, and fBSL. a shows the cumulative proportion in BP categories across all study years, while (b) presents individual study year proportions, with a significant difference in the proportion of BP categories between study years (χ2 p = 0.005), attributed to fewer participants in the normal category than expected in 2017; fewer elevated in 2023; more high-normal in 2022 and 2023. c shows the cumulative proportion in TC categories across all study years, (d) presents individual study year proportions, with a significant difference in the proportion of TC categories between study years (χ2 p = 0.003), attributed to fewer participants in the raised category than expected in 2016, 2018; more raised in 2023. e shows the cumulative proportion in fBSL categories across all study years, (f) presents individual study year proportions, with a significant difference in the proportion of fBSL categories between study years (χ2 p = 0.007), attributed to fewer participants in the raised category than expected in 2018; more raised in 2022; as indicated by the arrows shown in the graph

Other variables and relationships

Additional variable analyses reveal that elevated BP is significantly associated with BMI-for-age categories, WtHR, and TC levels (Fig. 6). Apart from these findings, no other significant correlations between variables were observed.

Fig. 6.

Percentage distribution—cumulative Y9 cohort 2016–2023, BP compared to other variables. a BP compared to BMI-for-age, distribution across categories significantly different than if independent (χ2 p < 0.001), attributable to more participants in the healthy-normal group than expected; more obese-high normal; more obese-high; fewer healthy-high normal; fewer obese-normal. b BP compared to WtHR, distribution across categories significantly different (χ2 p < 0.001), attributable to more low-normal; more increased-high normal; fewer increased-normal. c BP compared to TC, distribution across categories significantly different (χ2 p < 0.001), attributable to: more raised-high; as indicated by the arrows shown in the graph

Metabolic risk factors

Five risk factors relating to metabolic health were further considered: BMI-for-age in the obese category, Waist Circumference (WC) at or above 90th percentile (replacing WtHR for comparability to other studies), at or above elevated BP, raised fBSL, and raised TC levels (as a proxy for high-density lipoproteins (HDL)/triglycerides (TG)) [25, 29, 46].

Data for both BMI-for-age and WC were available for 620 participants, representing 79.2% of the sample. Both risk factors were evident in 29.5% of this group. No significant difference was found between females and males (χ² p = 0.52).

Data for all five measures was available for 261 participants, representing a third of the sample. To evaluate sample representativeness, baseline characteristics were compared between this sub-group and those for whom less than five measures were available. One-way ANOVA and Chi-square tests revealed no statistically significant differences in BMI percentiles (p = 0.50), fBSL (p = 0.40), CHOL (p = 0.75), sex (p = 0.06), waist circumference (p = 0.35), or blood pressure (p = 0.20). A significant difference was observed for data collection year (p < 0.001), which was expected due to temporal sampling shifts; otherwise, the cohorts remained highly comparable across key baseline parameters. In the subsample where five measures were available, 27.6% were living with obesity and two or more other risk factors. Recognising the limited cholesterol measures, with TC used as a substitute for HDL and TG, 14.2% of participants were identified as living with three or more risk factors (WC, BP, fBSL, TC) (Table 3).

Table 3.

MetS factor cumulative percentage

Number of MetS Risk Factors (From WC, BP, fBSL, TC)	0	1	2	3	4
3 or more	100.0%	75.5%	41.0%	14.2%	1.5%
Obesity and 2 or more	42.1%	38.3%	27.6%	11.1%	1.5%

In general, BMI-for-age and WC risk combined was often associated with a participant living with obesity and two or more risk factors (92.3% had the same result for both risk factor measures, combining true positive and true negative results). Of those who were identified as living with obesity and at least two other risk factors, 85.1% were captured by evaluating only obesity and at-risk WC. If BMI-for-age and WC risk were to be used as a method for predicting the proportion of participants with obesity and two or more risk factors, it was found that only 4.2% were identified as having both BMI-for-age and WC risk but not as living with obesity and two or more risk factors (false negative). This means they were at risk according to BMI-for-age and WC but not according to their BP, TC, or BSL. By comparison, 3.4% of participants were not identified as at-risk using BMI-for-age and WC factors but were identified using the other method (false positive).

Discussion

Overall, the findings of this study indicate high levels of risk factors associated with MetS in Y9 students in Rarotonga.

Among Y9 students (median age 13.8y), the prevalence of BMI-for-age ≥ 85th and ≥ 95 percentiles were 59.7% and 37.6%. This aligns with data from the 2015 WHO GSHS for Cook Islands students aged 13–15 years (n = 363) (58% ≥ 85th percentile; 31% ≥ 95th percentile), 75% of which were measured, and 25% self-reported [50]. The findings are consistent with other PICTs [10, 51–53] and adolescents of Pacific ancestry in New Zealand [54]. Global age-standardized data show that childhood (5–19 years) obesity for the Cook Islands and Niue (also a very small PICT) are the highest in the world and had the greatest levels of increase between 1990 and 2022 at 20 percentage points for girls and 25 percentage points for boys [10].

BMI serves as a robust indicator of adiposity for adolescents [55]. The addition of waist circumference in this study enabled the assessment of central obesity via WtHR, an accessible assessment metric for low resource settings that is independent of sex and age [56] and addresses recognised limitations associated with BMI [57]. Adolescents whose WtHR indicated increased risk were most likely to also be living with obesity (79%). However, 21% in the obese category and 68% in the overweight category were assessed via WtHR as low risk, indicating lower cardiometabolic risk [58, 59].

The finding of 26.5% of participants with blood pressure readings in the high normal (HTN1) or high (HTN2) range is similar to findings of a NZ-based study of slightly younger children [60]. However, measurement of blood pressure is operationally challenging in school-based screening programmes [61] and these data should be treated with some caution. While our protocol addresses the known issues of cuff size, seated rest period and repeat readings, issues of anxiety and the provision of a controlled environment are challenging in this setting [47]. Further, consideration of the issues associated with accuracy of measurement in children with obesity [62] and the impact of hormonal activity on BP during adolescence [63] should be made when interpreting these data.

Our study fell short in terms of two measures required for MetS risk factor assessment in children and adolescents [27, 29, 64]. Resource availability enabled measurement of total cholesterol but not the required HDL. Furthermore, triglycerides were not measured, and low adherence limited the proportion of students with valid fasting blood sugar measures. While we have assessed some risk factors for MetS, we recognise that this is not a complete assessment; therefore, we cannot reliably compare our data with studies examining adolescent MetS. Acknowledging this, we note that Grant et al. (2008), in a study of Pacific Island adolescents aged 9 −14 years in New Zealand, identified 7.5% with ≥ 3 MetS risk factors, while Sellers et al. (2008) found that in a cohort of Aboriginal adolescents aged 9 −14 years in Darwin, 14.0% were living with ≥ 3 MetS risk factors [65, 66]. Reisinger et al. (2021) [25] present a review of the prevalence of paediatric MetS conducted from 2014–2019. In studies with age ranges similar to ours (12–16 years), prevalence levels based on ≥ 3 risk factors ranged from 1.9% to 9.5%. Acknowledging limitations, the evidence we have presented of obesity plus at least two other risk factors at 14.2% suggests that alongside the high prevalence of overweight and obesity, MetS is likely to be at concerning levels in the Y9 adolescent population in Rarotonga. We recommend that changes be made to the protocols in the Y9 Metabolic Health Programme in the Cook Islands to ensure that a full assessment of MetS is possible in the future.

Our data points to the potential benefits of jointly assessing WtHR/WC and BMI-for-age, as supported elsewhere in the literature [67–69]. BMI-for-age and WtHR/WC were shown to be reliable measures associated with obesity and two or more risk factors in our study. Incorporating waist circumference measurements into the Cook Islands biennial SPHE program would be advantageous in enhancing population-level evidence of emerging risk factors associated with MetS in children and adolescents. The potential of this approach to allow the combined use of BMI and WtHR to monitor risk across the 5- to 18-year-old demographic, including those in remote outer islands, has led to a protocol change, with the inclusion of waist measurements being gradually introduced into the Cook Islands SPHE from 2024.

Larger samples are preferred for anonymity, accurate analysis (e.g. chi-squared test assumptions) [70], and reducing variability [71]. However, the sample sizes in this research were not small relative to the population size [72]. The sample size calculation showed that the sample size exceeded expectations [73]. The cumulative Y9 population across the study years [2016–2019 2022–2023] was estimated to be 1214 students. For a population of N = 1214, the recommended sample size (with finite population correction) was n = 287 (α = 95%, ± 5%, and estimated population proportion = 57.57%, averaged from three sources [74–76]. In fact, n = 783 for the metabolic health data in this study is above the recommended sample size of 726 people, calculated using an increased confidence level of 99% and a smaller margin of error of 3%.

Limitations

In addition to the limitations already discussed, we note the following issues to provide context for the interpretation of findings.

The 2018 data presented unexplained variability. This inconsistency may be attributed to measurement errors, as despite the inclusion of annual training and consistent oversight, the data collection was conducted by multiple individuals. This is particularly pertinent given that repeated measurements can lead to minor discrepancies [77]. Such variability underscores the necessity of ongoing professional development to ensure adherence to established standard operating procedures. This limitation also highlights the importance of employing a multi-year cross-sectional design to detect trends and maintain consistency over time.

Low adherence to fasting limited the integrity of the data relating to blood sugars. To overcome this issue HbA1c assessment was proposed and trialled in 2017. While operationally stable in the clinic, testing in a classroom produced unstable results, possibly due to extreme heat and humidity and no mechanical ventilation. We also note the potential for pubertal stage to influence factors related to MetS and acknowledge that this was not assessed [78, 79]. Furthermore, while full lipid testing is ideal, we note the uncertainty regarding recommendations for this age group due to the instability of lipid levels during puberty [80–82]. The introduction of HbA1c and lipid fractions in place of fBSL and total cholesterol testing is recommended. It would also be valuable to explore the cultural and social acceptability of the potential for assessment of pubertal stage, noting the challenges this presents [83].

Conclusions

The findings of this study confirm a notably high prevalence of overweight and obesity among Y9 students in Rarotonga and, for the first time, provide evidence of elevated levels of key risk factors associated with MetS in this demographic. The combined assessment of BMI-for-age and WtHR highlights the benefit of incorporating waist circumference measurements into school-based health monitoring in low-resource settings, such as the Cook Islands. Some risk factors associated with MetS have been assessed, but investment in improving glucose and lipid monitoring is strongly recommended. The use of a multi-year cross-sectional approach addressed the challenges posed by the small population size, thereby facilitating the presentation of evidence that underscores the urgent need for sustained investment in interventions both before and during adolescence to mitigate risk levels in this population. This study emphasises the importance of the reported work that has been monitoring risk factors associated with paediatric MetS in the Cook Islands and suggests that this should be strengthened to enable full paediatric MetS assessment and employed longitudinally to assess the impact of health promotion interventions.

Abbreviations

BP

Blood Pressure

BMI

Body Mass Index

BMI-for-age

Body-Mass-Index-for-Age

CVDs

Cardiovascular Diseases

DBP

Diastolic Blood Pressure

fBSL

Fasting blood sugar levels

HbA1C

Haemoglobin A1C

HDL

High-Density Lipoproteins

HTN 1

Stage 1 Hypertension / High-Normal

HTN 2

Stage 2 Hypertension / High

GSHS

Global School Based Student Health Survey

NCD

Non-Communicable Diseases

MANA

Monitoring Alliance for NCD Action

MetS

Metabolic Syndrome

PSHLP

Pacific Science for Health Literacy Project

PICTs

Pacific Island Countries and Territories

rBSL

Random Blood Sugar Levels

SBP

Systolic Blood Pressure

SIDS

Small Island Developing State

SPHE

School Physical Health Examination

STEPS

STEPwise approach to NCD risk factor surveillance

TC

Total Blood Cholesterol

TG

Triglycerides

TMO

Te Marae Ora, Cook Islands Ministry of Health

WC

Waist Circumference

WHO

World Health Organization

WtHR

Waist-to-Height Ratio

Y9

Year 9

Authors’ contributions

JLB, Karen Tairea (deceased), and MHV conceptualised and, with TJ and CBW, designed the study. Data acquisition was led by TJ with CBW, KN, DTC, MB, SAT, MHV and JLB. Data analysis and interpretation were conducted by JLB, KN and AR. JLB and AR drafted the manuscript. All authors reviewed the final manuscript and approved the submitted version.

Funding

This work was supported by the New Zealand Ministry of Foreign Affairs and Trade Partnerships for International Development Fund PF-1–280 and PF-11–613; The Health Research Council of New Zealand, Ref 20/284; The Ministry of Education, Cook Islands; Te Marae Ora, Ministry of Health Cook Islands.

Data availability

Data are owned by Te Marae Ora, Ministry of Health, Cook Islands. Summary data are provided within the manuscript or supplementary information files.

Declarations

Ethics approval and consent to participate

This study was conducted using deidentified health system data from the Te Marae Ora Cook Islands Ministry of Health database of school health check records. These data are routine health records collected with informed parental consent from children under the age of 16 following protocols specified by Te Marae Ora, Cook Islands Ministry of Health. The Cook Islands National Research Committee deemed that as the data were deidentified, the study did not require formal approval. No identifiable personal health information was used in this research. Approval for University of Auckland staff to work with Te Marae Ora staff on deidentified data from the Cook Islands database of school health check records was obtained from Auckland Health Research Ethics Committee Ref. 24,668. All procedures followed were in accordance with the ethical standards of Te Marae Ora, Cook Islands Ministry of Health, the Cook Islands National Research Committee, the Auckland Health Research Ethics Committee and with the Helsinki Declaration of 1975, as revised in 2024.

Consent for publication

Not Applicable.

Competing interests

The authors declare no competing interests.