To assess the affordability of an item, it is necessary to understand the cost of the item and consider that in the context of an individual’s (or household’s) economic capacity to purchase that particular item. This study utilised secondary data analysis to calculate the cost of purchasing recommended amounts of NSFV in Port Vila, Vanuatu, and considered that in relation to a household’s total expenditure, a measure of economic capacity .
The relatively low RSEs for this study indicate that the data are of good quality and provide reliable estimates of true expenditure patterns. Approximately 80% of households in Port Vila did not spend enough on NSFV to meet the recommendations, even when the cheapest varieties were considered. This is higher than the estimate determined from the WHO STEPS survey  which reported that 62% of adults in Vanuatu consumed insufficient quantities of NSFV. However, the WHO STEPS survey sampled both rural and urban areas across Vanuatu, whereas the present study focused only on Port Vila, Vanuatu’s largest urban centre where population growth and urbanisation introduce unique challenges to food security and NSFV access and affordability [19,20]. The present study also considered the NSFV requirements of all household occupants; thereby capturing children for whom poor diets can have lifelong detrimental effects on growth and development . Despite the different methodologies and results, both studies report largely inadequate NSFV intakes in Vanuatu when compared against the minimum recommendations for these foods.
To the best of our knowledge, NSFV cost and affordability has not been assessed in a comparable manner elsewhere in Pacific Island countries. The findings of this paper are consistent with international evidence which reports that insufficient NSFV intakes are common in low and middle-income countries, particularly amongst the poorest [16,34-36]. In order to purchase the recommended amount of NSFV for all household members, the poorest households in Port Vila would need to spend almost 20% of their total budget and more than 40% of their food budget on NSFV. This would leave little capacity to purchase other essential foods, as well as necessities such as fuel, electricity, school fees, transport and medical costs.
Overall, global food prices are becoming more expensive and increasingly volatile . The poor are most affected by this situation, especially those in developing countries who may spend up to 75% of their income on food . Countries reliant on food imports, which includes the majority of Pacific Islands, are particularly vulnerable to international food price increases and volatility .
High food costs can be a barrier for low-income households in making healthy food purchases [26,38,39] and can drive low-income farming households to sell their produce for a profit and purchase cheaper and less nutritious foods instead . Low-income households tend to consume nutrient poor and energy dense diets, which are inconsistent with dietary guidelines and associated with increased risk of obesity, NCDs and malnutrition [26,38,39]. Households with lower expenditure on fruits and vegetables are also at greater risk of under-5 child mortality .
A reduction in population dietary energy density, which may be achieved through increased consumption of energy dilute fruits and vegetables, is a recognised public health strategy for obesity prevention and improving macronutrient intakes . This may increase diet costs as an inverse relationship between a food’s energy density and its energy cost, i.e. cost per kilojoule, has been observed in various contexts including the United States and France , urban South Africa  and remote Australian Aboriginal communities . However, the use of this metric has been questioned [44,45] and foods that are consistent with dietary guidelines have been reported to cost less than foods high in added sugars, added fats or sodium when compared on a cost per edible weight or cost per portion basis .
The results of this study indicate that improving the affordability and consumption of energy-dilute and nutrient-dense NSFV will be important in the prevention of NCDs and micronutrient deficiencies in Vanuatu. Research in Brazil has estimated that reducing fruit and vegetable prices by 20% would lead to an increase of approximately 16% of the proportion of these foods in the national diet .
One opportunity to improve NSFV affordability is through reducing wastage and increasing the efficiency of production and supply systems . Approximately one-third of food produced for human consumption globally is wasted  and this can significantly reduce developing countries’ capacity to meet NSFV intake recommendations . Due to their perishable nature and short shelf-life, fruits and vegetables are particularly vulnerable to wastage throughout the harvesting and distribution process [49,50]. Opportunities are available to support developing countries to improve their fruit and vegetable production processes , and these are worthy of consideration.
The findings from this study can be used to support targeted food and agricultural policy development in Vanuatu. Guidelines for the promotion of local produce in the Pacific Islands are available  and are based on a fruit and vegetable promotion program that was developed in the Federated States of Micronesia. In this program, multi-sectorial community-based promotion of local food production and consumption was found to be associated with an increased frequency of household consumption of fruits and local vegetables and an increased variety of local foods generally . The present study has identified the cost per edible serve of locally produced NSFV, which will assist to promote the cheapest ways to meet NSFV recommendations in Vanuatu.
Secondary analysis of the Vanuatu HIES and CPI datasets provided the opportunity to explore the issue of NSFV cost and affordability in a cost-effective and efficient manner. Secondary analysis of household expenditure survey data has been estimated to be 75 times cheaper than conducting a 24-hour food recall survey . The periodical nature of these data collections also offers the opportunity to monitor trends in NSFV cost and affordability over time and evaluate the impact of relevant health and agricultural policies. Options for improving the use of household expenditure surveys for nutrition analysis should be considered in future survey enumerations [53,54].
This study used household NSFV expenditure data as a proxy measure for NSFV consumption. This type of data cannot account for food wastage at the household level, which may have resulted in an over-estimation of the proportion of households meeting the NSFV recommendations. However, food wastage is in developing countries is generally lower than that of developed countries, particularly at the household level  which may lessen the impact of this error.
Household food expenditure data also assumes an even distribution of foods amongst household members. This may not be the case, as gender differences exist in the amount of NSFV consumed by adults in Vanuatu, with men more likely than women to meet daily intake recommendations (42% compared to 35%) . Consumption may also differ by age of household member . Despite these limitations, estimates from household expenditure surveys compare well against 24-hour recalls for estimates of food consumption at the household level .
The food cost information in this study was obtained from the Vanuatu CPI. While the NSFV prices in the CPI collected were extensive, they are not exhaustive, as market food prices were only collected from the central market in Port Vila and smaller roadside markets were excluded. If produce at roadside markets was cheaper than the central markets, we may have underestimated the proportion of households purchasing recommended amounts of NSFV.
The CPI food prices were used to calculate the cost per edible serve of each NSFV item. As the Pacific Islands Food Composition Tables  does not contain information on the edible portions, US Food Composition Tables  were used to adjust food prices for the inedible portions. It is possible that people in Vanuatu prepare NSFV with more or less wastage than what is reported in the US Food Composition Tables, which may have also introduced potential error into the results of this study.
Another study limitation of this study is the use of household expenditure, rather than income, as a measure of household wealth. A valid and reliable measure of household income is difficult to obtain in developing countries due to widespread under-reporting and the often ad-hoc nature of wages in the informal work sector, as well as part-time or seasonal work opportunities [27,32]. The 2010 HIES household income data were also affected by respondent error . Thus analysts prefer to use expenditure data in developing countries as an indicator of household wealth as it tends to smooth-out income fluctuations and is easier to measure . In addition, errors may have been introduced in the expenditure data if the estimated monetary value participants assigned to the subsistence produce they consumed, and the amount they consumed, was under-reported [23,57]. However, this issue may be more relevant to rural areas where subsistence agriculture is more common .
The HIES was conducted between October and December 2010, and the corresponding CPI prices were selected to match the HIES enumeration period. Therefore this analysis may not represent annual NSFV cost and affordability in Port Vila, as food costs and households’ finances may be influenced by seasonal fluctuations in factors such as crop yields, employment opportunities, and household expenditure.
This study assessed the cost and affordability of meeting the WHO/FAO expert panel recommendations for fruits and vegetable intake , and therefore starchy crops such as sweet potato, yam, taro, potato, plantain banana and breadfruit were excluded from the analysis. This is justified as regional nutrition recommendations  group starchy crops with carbohydrate-based foods; estimates of the burden of disease attributable to low fruit and vegetable intake exclude starchy crops [58,59]; and other studies [16,34] that assess fruit and vegetable intakes, including the WHO STEPS survey  have excluded starchy crops. Including starchy crops in this analysis would require information on households’ dietary energy requirements, which was not available through the information collected in the HIES survey. We acknowledge that the exclusion of starchy crops in this analysis limits the ability to compare the results with studies that have classified fruit and vegetable items differently. Inconsistency in the definition of fruits and vegetables is a recognised issue affecting comparability of research in this field [34,59,61].
To be consistent with regional recommendations which promote consumption of local foods , imported NSFV items were excluded in this study. This is unlikely to significantly affect the results, as the CPI data indicated that these items were expensive relative to local produce, and the Vanuatu HIES survey  reported that these items were generally not commonly purchased foods, accounting for only 3.1% and 0.4% of respective fruit and vegetable expenditures in urban areas.
Lastly, as there are no separate recommendations for children’s NSFV consumption, the population-wide recommendation of 400 g of NSFV per day  was applied for all household members. Thirty two percent of the urban population in Vanuatu is younger than 15 years . If children’s recommended intakes are lower than adults, this study may have over-estimated the proportion of households with insufficient NSFV expenditure, and the relationship between NSFV affordability and household size. Establishing and applying children’s NSFV requirements would assist to improve the accuracy of the results of this study.