The tide of dietary risks for noncommunicable diseases in Pacific Islands: an analysis of population NCD surveys

STEPs surveys apply standardized and internationally recognized methods to collect data on a range of NCD risk factors including dietary behaviors (including sodium, sugar, fruit and vegetable intake), risk factors (hypertension, hypercholesterolemia, overweight and obesity) and health outcomes (diabetes) [44]. Since 2002, STEPs have been conducted in countries across the Pacific every 5 to 10 years. We conducted a secondary analysis using summary data from STEPS reports. The survey targets a representative sample of adults aged between 18 and 69 years and gathers data via questionnaires, physical measurements and biochemical measurements. It has been designed so that each county measures a core set of risk factors using standardized methods so that comparisons can be made over time within a country and between countries. Countries have the option of adding modules on additional risk factors or questions that capture more information on the core set of risk factors [43]. Full detail on STEPS survey methodology is described elsewhere [44]. Published STEPS reports were accessed online from WHO and/or governments websites. At the time of this analysis, no Pacific country had published more than two STEPs reports. Because we sourced publicly available data ethics approval was not sought.

We extracted data on modifiable dietary risk factors (fruit and vegetable intake) and specific dietary conditions (overweight and obesity, hypertension, hypercholesterolemia) that were collected in a similar manner across two time points. Most recent surveys in PICTs have added behavioral questions on intakes of sugar or sodium. Because of a growing awareness of the NCD risk associated with sugar and salt in PICTs [31, 45, 46] and focus in food policy [47, 48] we also report sugar sweetened beverage (SSB) consumption and sodium intake where they were measured in the second round (these were largely absent from the first round). Data was extracted into an excel form by two different authors. Because we were interested in risk profiles by sex, data were disaggregated by sex and age strata, usually capturing samples between 25 and 64 years of age in 5 years, 10 years or twenty-year groups. We elected not to extract data on hyperglycemia given issues with blood glucose measurement in some STEPs surveys [49]. Table 1 provides definitions for the extracted risk factors and conditions.

We employed a direct standardization technique to calculate age standardized rates for each countries in preference to using crude age specific rates could be misleading because of the differences in underlying composition of the populations. The WHO standard population grouped in 5-year intervals [50] was used to calculate age-standardized rates for each indicator using dstdize command in Stata v17.0 [51]. A 95% confidence interval was calculated using the methods described by Breslow and Day [52]. For Tokelau, the confidence interval was not calculated as the whole target population was included in the survey. Data was only from the STEPs surveys in bands of 20 years or greater than 20 years (45–64 years / 45–69 years) and the Cook Islands and Wallis and Futuna used a non-standard age group band of 18–44 years in the second-round surveys. Hence, unstandardized rates have been presented for these countries along with confidence intervals that have been computed using exact binomial method.

We present data for individual countries and pooled prevalence between survey periods to give an indication of overall changes in risk factor prevalence for these 8 countries. The age-standardised rates were pooled using metaprop command to calculate the pooled prevalence using a fixed effect model [53]. The pooled weighted estimate was calculated using the inverse variance method after Freeman-Tukey Double Arcsine Transformation to stabilize the variances [53]. Exact binomial confidence interval was calculated for each pooled estimate. Test of proportion was conducted to examine the statistical difference between two rounds of surveys.

Figure 1 reports age-standardized prevalence of adults consuming less than 5 serves of fruits and vegetables per day. Prevalence decreased significantly in Tonga from 92.2% (95%CI: 90.4, 94.0) to 73.4% (95%CI: 71.6, 75.1) over 8 years, and in the Solomon Islands from 93.8% (95%CI: 92.6, 94.9) to 87.4% (95%CI: 85.9, 88.9) over 9 years. In both countries statistically significant reductions were observed for both women and men (see Supplementary File 1). In Nauru and Wallis and Futuna, prevalence decreased statistically significantly for men only, from 98.4% (95%CI: 97.6, 99.4) to 94.84% (95%CI: 92.5, 97.2) and 96.3% (95%CI: 92.3, 100.3) to 88.3 (95%CI: 83.9, 91.8) respectively. In Tokelau on the other hand, prevalence increased from 90.8 to 96.5% over the 9 years between 2006 and 2015.

The pooled analysis revealed a significant decrease in the proportion of adults consuming less than 5 serves of fruit and vegetables per day, from 94% (95%CI: 93.9, 94.5) to 88% (95%CI: 87.5, 88.2), significant for both men and women.

Four of the countries measured sugary drink consumption in Survey 2. Adults in Kiribati, Nauru and Tokelau (across both sexes) reported consuming over 3.5 sugary drinks each per day. In contrast, Solomon Islands adults reported consuming an average of 0.4 sugary drinks per day. SSB consumption did not vary significantly between men and women (Table 4).

Mechanisms for measuring salt varies significant across the included surveys. Five countries asked about ‘always or often’ adding salt before eating or when eating (Cooks, Kiribati, Tokelau, Solomon Islands, Nauru) (Table 5). Nauru and Cook Islands reported the per cent of participants ‘always or often’ eating processed food high in salt, and applied a likert scale querying participants on the importance of lowering dietary salt. Because of this variation we only extracted data on the percent of adults in Survey 2 ‘always or often’ adding salt to meals before eating. The proportion of adults ‘always or often’ adding salt to meals before eating ranged from 31.6% in Tokelau (higher for women than men) to 65.4% (60.5–70.3) in Nauru. Based on urinary analysis, adults in Tokelau, consumed an average of 10.1 g/day of salt, and consumption was higher for men (12.0 g/day) than women (8.6 g/day). In Wallis and Futuna salt consumption was 10.2 g/day, also higher for men (11.7 g/day) than women (8.8 g/day).

Figure 2 reports age-standardized prevalence of adults living with overweight and obesity. There was a statistically significant increase in prevalence from 59.1% (95%CI: 57.5, 60.5) to 67.96% (95%CI: 66.1, 69.8) in Fiji largely attributable to an increase for women from 75.2% (95%CI: 74.1, 76.3) to 85.3% (95%CI: 84.4, 86.3). Prevalence also increased in Tokelau from 93.3% to 95.2%, particularly for women (94.5% to 95.4%). Women lived with a higher prevalence of overweight and obesity than men in all countries except Nauru. No significant changes in prevalence were observed for the Cook Islands, Kiribati, the Solomon Islands or Tonga. The pooled analysis revealed a significant increase from 76.9% (95%CI: 76.1, 77.7) to 82.1% (95%CI: 81.3, 82.9) in the proportion of adults living with overweight or obesity.

Prevalence of hypertension increased in 6 countries (Fig. 3). In Kiribati prevalence increased from 18.4% (95%CI: 16.4, 20.4) to 42.13% (95%CI: 38.9, 45.4), in the Solomon Islands from 9.6% (95%CI: 8.1, 11.1) to 26.83% (95%CI: 23.5, 27.9), in Nauru from 29.5% (95%CI: 27.3, 31.8) to 37.6% (95%CI: 33.9, 41.2)], in Tokelau from 35.6% to 42.4%), in Tonga from 23.9% (95%CI: 21.1, 26.7) to 29.8% (95%CI: 28.1, 31.6) and in Fiji from 25.7% (95%CI: 24.6, 26.8) to 30.81% (95%CI: 29.2, 32.5) (Fig. 3). Increases were significant for women in all countries and for men except in Nauru and Tonga. Against this pattern, hypertension prevalence decreased from 58.6 (95%CI: 55.5, 61.8) to 47.2 (95%CI: 42.3, 52.2) in the Cook Islands driven by a large decrease for men.

The pooled analysis showed an overall increase in the prevalence of hypertension from 25.4% (95%CI: 24.7, 26.2) to 33.41% (95%CI: 32.5, 34.4) across the 8 countries.

Six countries had comparable measures for hypercholesterolaemia (Fig. 4). Prevalence increased from 25.1% (95%CI: 21.1, 29.1) to 35.8% (33.2, 38.4) in the Solomon Islands and from 42.2% to 65.96% in Tokelau. Prevalence decreased from 80.0% (95%CI: 77.3, 82.8) to 58.2% (95%CI: 63.2, 52.9) in the Cook Islands, and from 27.7% (95%CI: 24.4, 30.9) to 17.8% (95%CI: 20.4, 15.2) in Kiribati. Significant reductions were observed for men and women in both countries.

Our results align with literature describing a steady increase in overweight and obesity in Pacific Island countries [23, 24, 54]. However, in our study this increase was driven by just two countries, Fiji and Tokelau, and in particular by an increase in prevalence for women in Fiji. Fiji was one of the first Pacific countries to complete a STEPS survey, and the timing of the survey (eariery in the processed food transition) may have contributed to lower baseline prevalence compared to other countries. Our observation of increasing overweight and obesity in women compared to men in Fiji is consistent with other studies in LMICs [55‐57]. Gender weight disparities may be a result of sociocultural factors, or because men are more often engaged in highly physical occupations compared with women, and involvement in sports is still less common in women [57]. In 6 of the 8 countries, there was no significant increase in overweight and obesity prevalence and mean BMIs were also relatively stable. This contrasts with other countries, including the US [58], where rates of obesity (BMI > 30) have accelerated faster than rates of overweight (BMI > 25) in recent years. High baseline levels of overweight and obesity in Pacific countries may have contributed to this stabilisation, noting that some Pacific populations have less fat mass at a given BMI than Caucasian populations [59]. It is also possible that preventive measures are starting to make a difference in some countries.

Pacific health and agricultural agencies have proactively promoted fruit and vegetable consumption [60, 61] in recent years, and offered agricultural support programs for farmers [60, 62] which may have contributed to the decrease over time in the proportion of adults consuming < 5 servings of fruit and vegetables each day. Despite this decrease, 88% of Pacific adults still report inadequate consumption. That this is consistent with the global dietary transition away from plant-based diets makes it no less concerning, and it points to the need to strengthen the efforts mentioned above. Inadequate fruit and vegetable consumption is an important but often neglected risk factor for NCDs [63], and a challenge across most regions of the world [64], particularly in LMICs [65]. A study of fruit and vegetable consumption in 28 LMICs between 2005 and 16 found that only 18% (16.6–19.4%) of adults over 15 years consumed WHO recommended amounts [65]. Consumption increased with GDP and secondary education but decreased with food pricing instability. Fiji and Tonga both relaxed import duties on fruit and vegetables, although evidence from Tonga suggest that this may have only benefited traders [66]. These findings point to the need to strengthen food systems approaches that promote production of resilient, biodiverse crops, and address post-harvest losses and market access [67].

An emerging concern for Pacific countries is high SSB consumption [68]. Adults in Nauru, Tokelau, and Kiribati consumed more than 3.5 serves of sugary beverages per day. Similarly high average daily serves have been observed in Tuvalu (3 serves/day) based on their STEPS survey. A recent study of trade data from 12 Pacific Island countries documented a 65% increase to sugary drink imports between 2000 and 2015 [69]. In this study, the Solomon Islands stood out from other countries with adults reporting consuming 0.3 average daily serves of SSBs. This may be attributable to the Solomon Islands being at an earlier stage of the global dietary transition than other Pacific countries, remoteness from markets, or the high volume of sweetened tea/coffee beverage powders consumed [70], which may not have been adequately captured by STEPS. Many Pacific Islands countries having adopted taxes on SSBs [31], but these may need to be increased in order to make meaningful shifts to consumption, and the sale of SSBs in and around educational institutions could be tightened [71, 72]. The Solomon Islands in particular may benefit from introducing an SSB tax to keep consumption levels low [70].

The average prevalence of hypertension increased from a quarter to a third of all adults in these Pacific countries, with prevalence levels similar to Australia (34%) [73], possibly due to the high un-met needs in controlling blood pressure in Pacific countries [74, 75]. While further surveys are needed to confirm a trend, persistently high (and potentially increasing) rates of hypertension signal a future pipeline of vascular diseases with a potentially overwhelming impact on Pacific health systems and economies [76]. Dietary sodium, saturated fats and trans fats are major dietary contributors to hypertension [75] and saturated and trans fat are major dietary contributors to hypercholesterolemia [77, 78]. In this study, overall prevalence of hypercholesterolemia was over 40%, and in the two countries where salt intake was measured, it was over 10 g/day, more than double that recommended by WHO. These indicators support the need to disrupt current dietary patterns in the Pacific, specifically excessive consumption of fatty meat, hydrogenated vegetable oil [26, 29, 79], and foods high in sodium [28, 80].

Our analysis highlights the ongoing challenge that Pacific countries face in responding to dietary causes of NCDs. Unhealthy dietary patterns are fueled by increased trade liberalization [81‐83], the penetration of food marketing [84, 85], and by food environments that promote affordable and convenient processed foods that are high in energy, salt, sugar and fat [27]. Additionally, policymakers have faced strong opposition from food and beverage companies trying to diminish policies [86], and pressures to minimize impacts of food policy on trade participation [37]. The multisectoral nature of food policy has made it difficult for Pacific leaders to implement and then enforce all recommended policy measures [87, 88], leading many to favour ‘softer’ approaches (i.e. guidelines and promotional materials) over regulatory approaches. Further, Pacific Island countries have struggled to find capacity to carry out regular dietary surveys and demonstrate the potentially positive impact of food environment policies on consumption [40]. Pacific Island MANA as a component of Framework of action for revitalization of healthy islands in the Pacific has been an important step to promoting political accountability to NCD prevention [38, 39], but countries will need to adopt a stronger cross sectoral approaches towards regulating, monitoring and enforcing food environment policies [33, 39, 89].

The purpose of STEPS is to provide a standardized method for collecting, analysing and disseminating data on key NCD risk. In the Pacific, STEPs surveys are used to inform high-level economic discussions [90], for regional monitoring and accountability strategies [39], to contribute to global monitoring, and to underpin evidence-based policymaking at the national level [70, 91]. By gathering STEPs data from multiple countries and over two time points we identified several opportunities to strengthen NCD risk factor monitoring in Pacific countries. Firstly, standardizing age grouping between survey rounds and countries would aid interpretation of published survey reports. For instance, Nauru, Kiribati and Solomon Islands reported results in the groupings of 25–34, 35–44, 45–54 and 55–64 years in the first round while the groupings were 18–29, 30–44 and 45–69 years in the second round. Secondly, standardizing risk factor thresholds or cut points between countries and survey rounds. In Fiji for instance, 2002 fruit and vegetable consumption was reported as the percent of people reporting < 1 serve of fruit and vegetable per day, whereas in 2011, it was the percent consuming < 5 serves. Thirdly, standardized time intervals between the surveys. Fourthly reducing the lag between data collection and publication of study reports so timely action can be taken. Finally salt intake, in particular, needs to be reported consistently, perhaps in place of less useful measures such as self-reported oil intake [92].