Introduction 

 The increase in global temperature has affected the global climatic system and subsequently impacted the natural environment and its inhabitants. The 5th Assessment Report of the intergovernmental panel on climate change (IPCC) reported that climate change is unequivocal since the 1950's and the world continues to warm through the decades \citep{RN1260}. Climate change is the change in average weather over an extended period such as decades or millenniums.  The United Nations Framework Convention on Climate Change (UNFCCC) defines climate change as "the direct and indirect impacts of human activities altering the global composition of the atmosphere in addition to the natural climate variability over specified time " \citep{Pachauri2014}. The Intergovernmental Panel on Climate Change (IPCC) defines climate change as climate determined by statistical analysis, which is the changes in means and properties that prolonged over time, usually in decades or more \cite{Pachauri2014}. Scientists can trace back past climates using the natural resources in the environment. Paleoclimate is the study of climate before the use of monitoring equipment. Tree rings are precise measurements as it contains dated information about the previous environment \cite{RN1281}. Corals also provide information on the past climate variability of the tropical and subtropical oceans\citep{RN4870}. Climate change is said to be caused by internal and external factors. External factors cause the variations in solar energy reaching the earth caused sunspots on the sun. The little ice age of the 17th century, when the world was unusually cold correlated with a few sunspots \cite{RN4708}. Discovered in the early 20th century Milankovich cycle is the varying of the earth and the sun geometry over time, and it has three effects. The eccentricity which occurs every 100,000 years cause’s solar irradiance to vary between seasons, a tilt which occurs every 41,000 years and precession influences heat distribution between latitudes and play essential roles in the triggering of the ice ages when radiation is reaching the higher latitudes. The tilting of the poles towards and away from the sun varies and has a direct effect on the temperature and the seasons on the planet\cite{RN1301,RN194,RN4698}.  The IPCC glossary defines anthropogenic climate change is when humans interfere with the natural gases in the atmosphere which in turn influences the change in average weather and climatic conditions \citep{Pachauri2014}. The former United States President, Mr Barack Obama claimed that "climate change is the most important threat to humanity in this century"  during the Climate Summit in New York in 2014\citep{RN195}
 Climate change is a threat to human health. Climate change challenges the basic health and human rights necessities which is proper shelter, clean air and water and adequate supply of nutritious food  \cite{RN4671}. The density of the population, economic status, existing environmental conditions, and accessibility to quality health services will determine the level of vulnerability \cite{RN1283}. Climate change increases the likelihood of crop failures leading to food shortages and increasing the risk of malnutrition in countries that are dependent on agriculture to supply food to their citizens. Climate change will affect food availability through direct effects, the variation in rainfall causes either floods or droughts, the variation in temperature modifies the length of the crop growing seasons \cite{RN159}.
The variation in rainfall caused by climate change will have an effect on crop productions in the Pacific. The rising temperatures, climate variabilities such as El Nino and La Nina conditions, sea level rise and more intense cyclones also affect the agriculture sector. Many of the rural dwellers in the Pacific still rely on locally grown food as their main staple diet. Most of these crops are dependent on rains in the summer (November to April), so the crops are dependent on rainfall. Climate projections for the region are variations in total rainfall which would be catastrophic to the crops \cite{RN4834}.  The South Pacific Regional Environment Program (SPREP) warned Fiji that if it ignores the possible effects of climate change, which include increased vector-borne disease and increased malnutrition due to food shortages during extreme weather events, it will lose US $5 -19 million by 2050 \cite{RN489}.The United Nations Children Education Fund (UNICEF) defines malnutrition as " when the body does not get the proper amount of energy (calories), proteins, carbohydrates, fats, vitamins, minerals and other nutrients required to keep the organs and tissues healthy and functioning well''\cite{RN1334}. The World Health Organisation classifies malnutrition into three categories, undernutrition which includes wasting (low weight for height), stunting (low height for age), and underweight which is a low weight for age. Micronutrient malnutrition which is either lack of or excess of specific nutrients in the body and overnutrition which is excess nutrients in the body causing overweight and obesity are the other types of malnutrition\cite{RN4687}.  Malnutrition is a complex illness caused by several factors, and to be able to prevent or minimise it, adequate and nutritious food must be considered \cite{RN4774}. The importance of the study is to determine that the effects of climate change which is the increase in temperature, variation in rainfall and the frequency of extreme weather due to that variation are one of the contributing factors to the prevalence of malnutrition in Fiji.  The Fijian Ministry of Health website\cite{RN4836} reported that malnutrition is the number one childhood killer in the country, this is an alarming statement coming from the health ministry. Fiji has the abundance of fresh fruits, vegetables, root crops, marine and aquatic food and yet reports of malnutrition and fatalities caused by malnutrition are still happening. The United Nations Food and Agriculture Organisation issued a statement in the media that malnutrition was prevalent in Fiji was caused by shortages of vegetables, four months after category five Cyclone Winston hit Fiji in February of 2016\cite{RN4869}.  Research on the relationship between climate change and malnutrition were conducted in Africa and Asia, and there has never been any similar kind of study conducted in Fiji.  This is one of the first studies of its kind in investigating the association between climate-related variables and the prevalence of malnutrition in Fiji.

Methods

The design is that of an ecological study.  Secondary data collection was used to obtain the dataset for malnutrition indicators, climate variables and the other covariates in the study. Climate variables were collected for the period between 1st January 2009 and 31st December 2016.  These are the monthly averaged maximum temperature,  the monthly averaged minimum temperature, the monthly averaged total rainfall and the monthly averaged humidity. In this study, the predictor variables were the indicators of climate change ("minimum, maximum, and average monthly temperature'', "minimum, maximum, and average monthly rainfall'', "and average humidity''). The response variable was the prevalence of malnutrition (underweight children, growth faltering, severe malnutrition, and anaemia). Ethics approval was obtained from the Ministry of Health Ethics committee and the UC Human Ethics Committee.

Climate Data

Climate data were extracted from the Fiji Meteorological Service (FMS) for the years January 2006 to December 2016. The FMS database had the average monthly maximum temperature,average monthly minimum temperature, the average monthly humidity and the average monthly total rainfall. The weather monitoring stations are located in the four divisions of Fiji and provide a representative reading of the climate variables for the country. The FMS has given the monthly readings as the average of the monthly readings. The monthly recordings of maximum, minimum temperature, total rainfall and relative humidity for each of the stations were grouped together on a spreadsheet The columns had the climate variables and the months and the rows had theweather stations. The average of the monthly climate readings from the nine weather stations was the reading for Fiji.  

Malnutrition Data

Malnutrition was classified by the Fijian Ministry of Health as underweight, growth faltering, severe malnutrition and anaemia.  The malnutrition data for the years between 2006 and 2008 was excluded from the study because it had different classifications compared with malnutrition data for the years 2009 through 2016. The new classifications for malnutrition are mildly underweight, moderately underweight and severe underweightbased on the consolidated monthly return format.   The Fijian Ministry of Health has been using the National Center for Health Statistics (“NCHS”) definitions until 2008, where malnutrition was classified as mild, moderate and severe underweight as low weight.  The malnutrition counts include underweight, growth faltering, severe malnutrition and anaemia. The researcher created a separate dataset because the raw data for growth faltering, severe malnutrition and anaemia was only available from 2009 to 2016. The dataset for 2006 to 2008 has only raw counts of mild, moderate and severe underweight as their classifications.   

Quantitative

  The researcher conducted time series analysis to describe the trend of the climate variables using the dataset created where only the readings of each of the climate variables were on the spreadsheet. A time series analysis is a sequence of data points recorded at different time points. It is commonly used when there are 50 or more data points in the series, and when considering seasonality, there should be at least 4 or 5 cycles in the data. Time series data have a natural ordering, such as the days, months, quarters, years and so forth which makes it distinct from cross-sectional studies. Time series data can be either stationary or non-stationary. A stationary time series are those whose properties such as mean, variance and autocorrelation are all constant over time and a non-stationary time series is a series where its properties change over time. Data was cleaned with the Open Refine software and exported to the R Studio software for analysis. Step one of analysis is descriptive analysis, hence the construction of a table for climate variables (average maximum and minimum temperature, average rainfall and average humidity). The table had in the header the number of months, median, mean, standard deviations and the percentiles which were calculated using the R Studio software. The R studio software was used to construct bar plots for the average maximum and the average maximum monthly temperatures, the average monthly rainfall and the average monthly humidity to determine how the climate variables behaved over the ten year period. Each of the bar plots represented the months the data was collected, and it was placed on the x-axis and the y-axis showed the readings for the climate variables. The trend of the climate variables was established using the dataset that was created where only the readings of each of the climate variables were recorded on the spreadsheet. Once it was transferred to the R Studio software it was then converted into time series data using the ts function from the tseries library. The frequency was set at 12 since the data were collected on a monthly basis and the first input was identified as 2006 as this was the time of the first recording. The association between climate variables (that is, minimum temperature, maximum temperature, average humidity, and average rainfall) and malnutrition (underweight, severe malnutrition, and anaemia) were modelled using a generalised additive model (GAM) using autoregressive integrated over moving averages (ARIMA) as Poisson process.