New Zealand is subject to wide range of natural disasters including: "earthquakes, volcanic eruptions, tsunamis, storms, floods and landslides".1 In particular, the country lies in a geologically unstable zone with major fault lines running for much of the length of the country. Most recently an earthquake on 22 February 2011 caused widespread destruction of Christchurch with 182 deaths and 6659 people injured in the initial 24 hours.2Flooding, due to intense or prolonged rain, is by far the most frequent natural disaster to impact on New Zealand.1 Flooding disasters, as well as severe storms, may also become more common with climate change.3 New Zealand's population growth may also contribute to the impact of flooding disasters, if house building continues on flood plains and low-lying coastal areas. Pandemics and economic disasters can also potentially cause disruptions to basic societal functions, including food supply.Due to the risk of these disasters, New Zealand civil defence authorities encourage preparation measures—including emergency food storage. Food storage of "non-perishable food (canned or dried food)" for a minimum of 3 days is encouraged and a civil defence website provides tips on the type of dried and canned foods which can be stored.4 More specific lists of foods, but with no explicit consideration of cost or nutritional value, are detailed on local government websites (e.g., Porirua City Council5).Yet there is evidence that such types of disaster preparations are not fully made by the New Zealand population (e.g., while 92% of respondents in a flood-prone area reported having canned food, only 27% had bottled water).6 Indeed, food insecurity is a significant problem for low-income populations in New Zealand,7-9 and so it is likely that such households often have no emergency food supplies. A recent study found that in the preceding 12 months, 50% of the families in a longitudinal study reported that they had been "forced to buy cheaper food in order to afford other necessities" and 13% of the families "reported having used food grants or food banks".10Due to the high prevalence of obesity and over-weight, most New Zealanders actual carry many days of stored energy in their bodies in fat deposits. Nevertheless, for optimal physical functioning in a disaster setting, on-going access to food containing carbohydrates, protein and fat is highly desirable. This is particularly the case for those contributing to disaster rescue and relief work and those subject to increased energy requirements (e.g., via physical activity and exposure to cold). Similarly, food can provide psychological comfort, prevent additional anxiety associated with hunger, and facilitate going to sleep at night. Preparing and eating food with others may also contribute to a sense of normality and communal experience in a disaster setting.A particular method for identifying low-cost foods that meet nutrient requirements is through linear programming. For example, this technique has been used to consider optimisation of diets in a number of studies (e.g., in France,11 for a cancer prevention diet,12 a diet without processed foods,13 and for designing the "Thrifty Food Plan"14—albeit using a non-linear programming for the latter). Given this background, the aim of this study was to perform optimisation analyses for the New Zealand context to inform emergency food stockpiling policies that organisations can promote (e.g., civil defence) and that citizens can consider.Methods Initial food selection—Given the thousands of different food products for sale in New Zealand, we had to take a simplified approach for selecting food products to include in the modelling. We therefore used dried, processed or canned foods from: The foods used in compiling the country's Food Price Index (FPI).15 A list in previous work that identified low-cost sources of protein in New Zealand.16 Those unprocessed foods that were found in the "bulk bins" of a supermarket and the low-cost canned foods (convenience sample in the capital city, Wellington). The lists of selected foods from a previous nutrition optimisation study in France.11 Scenarios—The first scenario (EP-B) considered achieving daily dietary energy intake for men at the lowest cost, and included foods which required cooking (Table 1). The second (EP-NC) added the requirement that foods did not need to be cooked (while also allowing some foods to be able to be sprouted or soaked before eating). The next scenario (EP-H) included foods that were optimised for low-cost, but also to meet all nutrient requirements for men (albeit with a higher iron requirement to increase relevance for women). The last scenario (EP-NS) considered the situation of zero spoilage (e.g., well-organised storage by institutions). Table 1. Specific scenarios used for the optimisation modelling for determining emergency foods for storage Aim of specific scenario Additional details on the constraints EP-B) Baseline scenario, trying to achieve the lowest cost for dried and canned foods To minimise the daily cost while obtaining the target dietary energy (kJ) detailed in Table 3 and no other nutrient targets. To ensure some modest level of variety (and to protect from some types of foods perishing without being replaced in time), a maximum limit of 100 g per specific food was set except for the amount of vegetable oil (set at 4 tablespoons, i.e., 60 g). EP-NC) As per Scenario EP-B but only for no-cooking required foods As above for Scenario EP-B, but only for foods that can be eaten without cooking (e.g., including all canned items in Table 2 such as canned meat/fish, canned pulses, canned fruit, canned vegetables, canned spaghetti, vegetable oil, olive oil, powdered milk etc). But we included grains that could be soaked and then eaten e.g., wholemeal oats. Also included were foods that can be soaked in water and sprouted within a few days e.g., lentils, and dried peas. In this scenario the maximum amount of vegetable oil was set lower than for EP-B (at 2 tablespoons, i.e., 30 g). The lower limit for all foods that were selected in the optimisation process was 10 g. EP-H) As per Scenario EP-B but to meet all the daily nutritional requirements for men (maximising health) As above for Scenario EP-B, but where all the nutritional requirements for men are achieved (see Table 3). This can be considered a luxury in disaster circumstances but would give reassurance to those groups who are particularly interested in maintaining good nutrition e.g., for children, pregnant women, lactating women, and adults with chronic health problems. Also in this Scenario, we increased the maximum limit to 200 g per any particular food to ensure that there was a feasible solution in the optimisation process. EP-NS) As per Scenario EP-B but with no spoilage (e.g., very well-organised storage) As above for Scenario EP-B, but where there is zero spoilage assumed e.g., very well-organised storage by citizens or by an institution or disaster relief organisation. Data inputs (price and nutrients)—For food items from the FPI we used the relevant price data (monthly data averaged over multiple stores nationally for the 12 months of 2011).15 But for other food items, we used online supermarket data (Countdown, January 2012), or the lowest in-store (e.g., bulk bin) prices from New World or Countdown supermarkets (both in Karori, Wellington). We ignored prices on "specials" and only considered non-bulk products (i.e., ≤1.5 kg). Nutrient values for the foods were obtained from the "New Zealand food composition database" (New nutrient database in 2012: http://www.foodcomposition.co.nz/foodfiles). Nutrient intakes were adjusted to account for food spoilage (see below). Spoilage estimates—We found no data on the rates of spoilage of stored food in New Zealand (and international data on household food wastage was not considered applicable17). So we made informed guesses as follows for the condition of stored emergency food at the one-year point: "High spoilage" at 20% for stockpiled foods that are at particular risk of being damaged by "pantry moths" (e.g.,Plodia interpunctella), and weevils e.g., the wheat weevil (Sitophilus granaries) and the rice weevil (Sitophilus oryzae) that are established in New Zealand. Pantry moths may also be an increasing problem in New Zealand.18 "Moderate" at 10% for food that can still be damaged by rodents which can eat through plastic wrapping. "Low" at 5% for cans which can rust or glass containers which can break. We applied these estimates as per the details in Table 2, but also in one Scenario (EP-NS) we assumed no loss from spoilage or other storage related losses. Table 2. Foods entered into the model, price data inputs and spoilage factors (foods ordered by increasing price within each food category) Food category / items Scenario EP-B (All) Scenario EP-NC (no cooking required) Price (NZD per 100 g) Spoilage assumption: point estimate (%) Spoilage uncertainty interval (%)# Grains and cereals Wholemeal flour Yesw 0.14 20 12-28 White flour (wheat, standard) Yesw 0.14 20 12-28 Pasta (dried) Yesw 0.21 20 12-28 Rice (white) Yesw 0.25 20 12-28 Oats (whole grain, raw) Yesw Yesw 0.32 20 12-28 Semolina Yesw 0.38 20 12-28 Spaghetti (canned) Yesw ** Yesw ** 0.40 5 3-7 Pop corn Yes 0.79 20 12-28 Couscous Yesw 0.99 20 12-28 Breakfast biscuits (e.g.,"Weetbix") Yesw Yesw 0.55 20 12-28 Fruit & vegetables Canned vegetables ("generic brand" tomatoes diced, lowest cost canned vegetables) Yes* Yes* 0.24 5 3-7 Canned fruit ("generic brand" fruit salad in syrup, lowest cost canned fruit) Yes* Yes* 0.26 5 3-7 Canned fruit ("generic brand" apricot halves) Yes* Yes* 0.27 5 3-7 Fruit juice (apple) Yes Yes 0.31 5 3-7 Peaches (canned) Yes Yes 0.42 5 3-7 Sultanas Yes Yes 0.57 10 6-14 Raisins Yes Yes 0.79 10 6-14 Meat, fish and dairy Milk (powdered, skim) Yesw Yesw 1.00 10 6-14 Sardines (canned) Yes*

New Zealand is subject to wide range of natural disasters including: "earthquakes, volcanic eruptions, tsunamis, storms, floods and landslides".1 In particular, the country lies in a geologically unstable zone with major fault lines running for much of the length of the country. Most recently an earthquake on 22 February 2011 caused widespread destruction of Christchurch with 182 deaths and 6659 people injured in the initial 24 hours.2Flooding, due to intense or prolonged rain, is by far the most frequent natural disaster to impact on New Zealand.1 Flooding disasters, as well as severe storms, may also become more common with climate change.3 New Zealand's population growth may also contribute to the impact of flooding disasters, if house building continues on flood plains and low-lying coastal areas. Pandemics and economic disasters can also potentially cause disruptions to basic societal functions, including food supply.Due to the risk of these disasters, New Zealand civil defence authorities encourage preparation measures—including emergency food storage. Food storage of "non-perishable food (canned or dried food)" for a minimum of 3 days is encouraged and a civil defence website provides tips on the type of dried and canned foods which can be stored.4 More specific lists of foods, but with no explicit consideration of cost or nutritional value, are detailed on local government websites (e.g., Porirua City Council5).Yet there is evidence that such types of disaster preparations are not fully made by the New Zealand population (e.g., while 92% of respondents in a flood-prone area reported having canned food, only 27% had bottled water).6 Indeed, food insecurity is a significant problem for low-income populations in New Zealand,7-9 and so it is likely that such households often have no emergency food supplies. A recent study found that in the preceding 12 months, 50% of the families in a longitudinal study reported that they had been "forced to buy cheaper food in order to afford other necessities" and 13% of the families "reported having used food grants or food banks".10Due to the high prevalence of obesity and over-weight, most New Zealanders actual carry many days of stored energy in their bodies in fat deposits. Nevertheless, for optimal physical functioning in a disaster setting, on-going access to food containing carbohydrates, protein and fat is highly desirable. This is particularly the case for those contributing to disaster rescue and relief work and those subject to increased energy requirements (e.g., via physical activity and exposure to cold). Similarly, food can provide psychological comfort, prevent additional anxiety associated with hunger, and facilitate going to sleep at night. Preparing and eating food with others may also contribute to a sense of normality and communal experience in a disaster setting.A particular method for identifying low-cost foods that meet nutrient requirements is through linear programming. For example, this technique has been used to consider optimisation of diets in a number of studies (e.g., in France,11 for a cancer prevention diet,12 a diet without processed foods,13 and for designing the "Thrifty Food Plan"14—albeit using a non-linear programming for the latter). Given this background, the aim of this study was to perform optimisation analyses for the New Zealand context to inform emergency food stockpiling policies that organisations can promote (e.g., civil defence) and that citizens can consider.Methods Initial food selection—Given the thousands of different food products for sale in New Zealand, we had to take a simplified approach for selecting food products to include in the modelling. We therefore used dried, processed or canned foods from: The foods used in compiling the country's Food Price Index (FPI).15 A list in previous work that identified low-cost sources of protein in New Zealand.16 Those unprocessed foods that were found in the "bulk bins" of a supermarket and the low-cost canned foods (convenience sample in the capital city, Wellington). The lists of selected foods from a previous nutrition optimisation study in France.11 Scenarios—The first scenario (EP-B) considered achieving daily dietary energy intake for men at the lowest cost, and included foods which required cooking (Table 1). The second (EP-NC) added the requirement that foods did not need to be cooked (while also allowing some foods to be able to be sprouted or soaked before eating). The next scenario (EP-H) included foods that were optimised for low-cost, but also to meet all nutrient requirements for men (albeit with a higher iron requirement to increase relevance for women). The last scenario (EP-NS) considered the situation of zero spoilage (e.g., well-organised storage by institutions). Table 1. Specific scenarios used for the optimisation modelling for determining emergency foods for storage Aim of specific scenario Additional details on the constraints EP-B) Baseline scenario, trying to achieve the lowest cost for dried and canned foods To minimise the daily cost while obtaining the target dietary energy (kJ) detailed in Table 3 and no other nutrient targets. To ensure some modest level of variety (and to protect from some types of foods perishing without being replaced in time), a maximum limit of 100 g per specific food was set except for the amount of vegetable oil (set at 4 tablespoons, i.e., 60 g). EP-NC) As per Scenario EP-B but only for no-cooking required foods As above for Scenario EP-B, but only for foods that can be eaten without cooking (e.g., including all canned items in Table 2 such as canned meat/fish, canned pulses, canned fruit, canned vegetables, canned spaghetti, vegetable oil, olive oil, powdered milk etc). But we included grains that could be soaked and then eaten e.g., wholemeal oats. Also included were foods that can be soaked in water and sprouted within a few days e.g., lentils, and dried peas. In this scenario the maximum amount of vegetable oil was set lower than for EP-B (at 2 tablespoons, i.e., 30 g). The lower limit for all foods that were selected in the optimisation process was 10 g. EP-H) As per Scenario EP-B but to meet all the daily nutritional requirements for men (maximising health) As above for Scenario EP-B, but where all the nutritional requirements for men are achieved (see Table 3). This can be considered a luxury in disaster circumstances but would give reassurance to those groups who are particularly interested in maintaining good nutrition e.g., for children, pregnant women, lactating women, and adults with chronic health problems. Also in this Scenario, we increased the maximum limit to 200 g per any particular food to ensure that there was a feasible solution in the optimisation process. EP-NS) As per Scenario EP-B but with no spoilage (e.g., very well-organised storage) As above for Scenario EP-B, but where there is zero spoilage assumed e.g., very well-organised storage by citizens or by an institution or disaster relief organisation. Data inputs (price and nutrients)—For food items from the FPI we used the relevant price data (monthly data averaged over multiple stores nationally for the 12 months of 2011).15 But for other food items, we used online supermarket data (Countdown, January 2012), or the lowest in-store (e.g., bulk bin) prices from New World or Countdown supermarkets (both in Karori, Wellington). We ignored prices on "specials" and only considered non-bulk products (i.e., ≤1.5 kg). Nutrient values for the foods were obtained from the "New Zealand food composition database" (New nutrient database in 2012: http://www.foodcomposition.co.nz/foodfiles). Nutrient intakes were adjusted to account for food spoilage (see below). Spoilage estimates—We found no data on the rates of spoilage of stored food in New Zealand (and international data on household food wastage was not considered applicable17). So we made informed guesses as follows for the condition of stored emergency food at the one-year point: "High spoilage" at 20% for stockpiled foods that are at particular risk of being damaged by "pantry moths" (e.g.,Plodia interpunctella), and weevils e.g., the wheat weevil (Sitophilus granaries) and the rice weevil (Sitophilus oryzae) that are established in New Zealand. Pantry moths may also be an increasing problem in New Zealand.18 "Moderate" at 10% for food that can still be damaged by rodents which can eat through plastic wrapping. "Low" at 5% for cans which can rust or glass containers which can break. We applied these estimates as per the details in Table 2, but also in one Scenario (EP-NS) we assumed no loss from spoilage or other storage related losses. Table 2. Foods entered into the model, price data inputs and spoilage factors (foods ordered by increasing price within each food category) Food category / items Scenario EP-B (All) Scenario EP-NC (no cooking required) Price (NZD per 100 g) Spoilage assumption: point estimate (%) Spoilage uncertainty interval (%)# Grains and cereals Wholemeal flour Yesw 0.14 20 12-28 White flour (wheat, standard) Yesw 0.14 20 12-28 Pasta (dried) Yesw 0.21 20 12-28 Rice (white) Yesw 0.25 20 12-28 Oats (whole grain, raw) Yesw Yesw 0.32 20 12-28 Semolina Yesw 0.38 20 12-28 Spaghetti (canned) Yesw ** Yesw ** 0.40 5 3-7 Pop corn Yes 0.79 20 12-28 Couscous Yesw 0.99 20 12-28 Breakfast biscuits (e.g.,"Weetbix") Yesw Yesw 0.55 20 12-28 Fruit & vegetables Canned vegetables ("generic brand" tomatoes diced, lowest cost canned vegetables) Yes* Yes* 0.24 5 3-7 Canned fruit ("generic brand" fruit salad in syrup, lowest cost canned fruit) Yes* Yes* 0.26 5 3-7 Canned fruit ("generic brand" apricot halves) Yes* Yes* 0.27 5 3-7 Fruit juice (apple) Yes Yes 0.31 5 3-7 Peaches (canned) Yes Yes 0.42 5 3-7 Sultanas Yes Yes 0.57 10 6-14 Raisins Yes Yes 0.79 10 6-14 Meat, fish and dairy Milk (powdered, skim) Yesw Yesw 1.00 10 6-14 Sardines (canned) Yes*