Community level interventions to improve food security in developed countries
Abstract
Objectives
This is a protocol for a Cochrane Review (intervention). The objectives are as follows:
To systematically review the effects of community level interventions in improving food security in developed countries, both across whole communities and for disadvantaged or at‐risk individuals or groups within a community. We also aim to identify features of community food security interventions that enable or impede the effective implementation of these interventions.
Background
Description of the condition
Adequate nutrition is an important determinant of health (Mann 2007). Adequate nutrition is itself determined by food security. This entails people having, at a minimum, the ready availability of nutritionally adequate and safe foods and an assured ability to acquire acceptable foods in socially acceptable ways (Anderson 1990). Conversely, food insecurity “refers to the social and economic problem of lack of food due to resource or other constraints, not voluntary fasting or dieting, or because of illness, or for other reasons" (Wunderlich 2006). Food insecurity is a significant issue in both developed and developing countries. In all settings, food security can result from limitations in the availability of food, access to food supplies, or both (WHO 2008). In developing countries, however, food security is often further compromised by the poor utilitisation of food due to a lack of clean water, sanitation or health care; or unstable access to adequate food at all times due to sudden shocks such as economic, climatic or political crises, or cyclical events such as seasonal food insecurity (FAO 2006 ). As the causes and characteristics of food insecurity can differ in these two contexts, this review will focus only on interventions to improve food security in developed countries.
In developed countries, food insecurity is associated with a lower consumption of fruit, vegetables, and meat (Tarasuk 2001a) leading to a decreased nutrient and micronutrient intake (Cook 2002; Tarasuk 1999). It has also been proposed that food insecurity in these countries may lead to increased consumption of inexpensive, energy dense foods (Drewnowski 2004). This inadequate intake of nutritious foods or overconsumption of nutrient poor foods can lead to a number of physical health problems, for example lower resistance to infection (Ball 1997; Wahlqvist 2002), or high blood pressure and other diet‐related risk factors for non‐communicable chronic diseases (WHO 2003). More specifically, evidence suggests that food insecurity in developed countries can be related to a higher prevalence of obesity in women (Dinour 2007) and that it is associated with poorer self reported health status (Stuff 2004; Tarasuk 2001a) and diabetes (Seligman 2007). Food insecurity has also been shown to be related to mental distress, including symptoms of anxiety and depression, loss of dignity, lack of control, distress, alienation and compromised social health (Hamelin 2002). It has been suggested that food insecurity has significant social and economic consequences for society in developed countries, including increased social exclusion and greater reliance on the healthcare system (Hamelin 1999). Because good nutrition is essential for growth and cognitive development, food insecurity is particularly problematic for children. In one study, food insecure children were nearly twice as likely to have fair or poor health and were a third more likely to have been hospitalised since birth (Cook 2004). Children who come from food insecure families are more likely to display psychosocial dysfunction, poor behaviour, and lower rates of learning and academic achievement (Jyoti 2005).
Description of the intervention
Measurement of food insecurity in developed countries has tended to concentrate on household or individual food insecurity, with a focus on the relationship between food and financial resources (Wunderlich 2006). This review will address interventions to improve community food security, where community food security can be considered as the collective experience of the households' food security. A community is defined as a group of people living in a specified geographical area (Bellows 2002).This definition also covers a group who share a social identity in terms of class, race, national origin or environmental vulnerability, or a community within a wider geographic community (Smith 1993).
A community is food secure when “all community residents obtain a safe, culturally acceptable, nutritionally adequate diet through a sustainable food system that maximizes self‐reliance and social justice” (Hamm 2003). Community food security is dependent on all components in the food system. In this context the food system includes "the foundations for food production, the social aspects of consumption, and relevant government and other policies, as well as the actual growing, processing, and distributing of substances that results in foods that people consume" (Gillespie 2000).
To assist in the identification of community food security interventions and appropriate primary outcome measures, we will use a logic model to illustrate the determinants of community food security which is itself conceptualised as food supply and access to food (see Figure 1).
Food insecurity has been experienced in developed countries for centuries. As a result, charitable approaches to emergency food provision have arisen, providing services such as food banks and soup kitchens. However, such models fail to address the underlying causes of food insecurity and their capacity is far too limited to provide adequate food (Raine 2003; Tarasuk 2001). In recent times, there has been an anti‐hunger and food security movement (Bellows 2002) which is predicated on the human right to food (Chilton 2009). This has led to a range of interventions aimed at increasing community supply and access to adequate amounts of nutritious food by improving food availability, affordability, and distribution within communities and through expanding economic opportunities for community residents and businesses (Bellows 2002). At the same time, the sustainable agriculture movement has been developing, which aims to preserve and promote local farming, create direct markets for local produce, and promote the use of regenerative methods for local food production (Gottlieb 1996). The types of interventions that target the underlying causes of community food insecurity (McCullum 2005; Rychetnik 2003) are the focus of the current review.
How the intervention might work
In this review we will include interventions which address the contributing factors to food supply and access as given in our logic model (see Figure 1). From a supply perspective these may include local food production and changes in planning and zoning to facilitate urban agriculture, preserving arable peri‐urban land for agriculture, and improving the location of stores selling healthy food. In addressing food access, strategies may include food pricing, food delivery services, and providing vouchers for farmers markets (among others) to address economic access, distance and transportation to shops. Some community level food security interventions may be conducted within specific settings such as stores, restaurants, schools, or workplaces. There may also be interventions such as community level policy coalitions that influence or implement a number of strategies within a specific community (Neff 2009 ). The outcomes of community food security interventions can be measured by a range of variables that are indicative of food supply, access or food use (for example, the amount of food produced by a community garden or the food expenditure or intake levels of a sample of community members). A number of other factors impact on effectiveness, including the fit between the intervention and the context, the reach of the intervention, and the intensity with which it is delivered (for example, amount of resources devoted to it).
Why it is important to do this review
Many initiatives to improve food security exist in high and middle income countries such as the United States, UK, and Australia. While some broad initiatives have been externally evaluated (for example, the food assistance programs in the US (Barrett 2002) and a prospective and controlled evaluation of large scale food retailing (Cummins 2005)), many other programs are not rigorously evaluated. Currently, there are no known systematic reviews of the effectiveness of community food security interventions (Tarasuk 2001). There have been a small number of non‐systematic reviews of community food security interventions conducted and these have identified the need for more rigorous evaluation and the importance of highlighting the process issues in program implementation (McGlone 1999). In the face of ongoing financial crises and climate change that will impact on the sustainability of the food supply, we are likely to face growing food insecurity. It is now more important than ever to understand how to best intervene at a community level to improve food security.
Objectives
To systematically review the effects of community level interventions in improving food security in developed countries, both across whole communities and for disadvantaged or at‐risk individuals or groups within a community. We also aim to identify features of community food security interventions that enable or impede the effective implementation of these interventions.
Methods
Criteria for considering studies for this review
Types of studies
Randomised controlled trials (RCTs), cluster randomised controlled trials (cRCTs), controlled before and after studies (CBAs), interrupted time series (ITS), historically controlled studies (HCS), and prospective controlled cohort studies will be included. CBAs may come in the form of cohort (same persons used before and after) or cross‐sectional (sometimes called independent panel) studies with sampling redrawn from the whole community either randomly or ad hoc. In this review, ITS studies must involve at least two observations before the intervention and two interventions after the intervention. ITS studies may use data routinely collected for monitoring and surveillance in 'natural experiments' (Petticrew 2005) or data collected for a specific interventions. Results for each type of study will be tabulated and analysed separately. Process evaluations, or process measures attached to included studies, will also be sought to provide insight into reach, adoption, and implementation.
Types of participants
The participants in this review will all be community members exposed to community food security interventions. These interventions are defined here as either community wide: co‐ordinated widespread programmes which support improvements to food supply or access in a particular geographic area (for example, a specific town); or community level: for groupings of people who share common interests or needs within a geographic area (for example, urban indigenous populations). This review will include interventions conducted in developed countries. We define developed countries as those classified by the World Bank as high‐income economies (www.worldbank.org). This review will particularly focus on the effects of these interventions on those population groups most at risk of food insecurity. These include people who: live in an area identified as socio‐economically disadvantaged (for example, low income area, ghetto, social housing projects); are part of a socio‐economically disadvantaged group (for example, working class or described as at risk or disadvantaged); or marginalised due to their socio‐economic circumstances (for example, receiving social assistance, below low‐income cut‐offs, unemployed).
Types of interventions
We recognize that achieving community food security may require multiple strategies. Strategies to improve food supply or access are largely environmental (Story 2008) though health and nutrition education can be used to improve knowledge and skills that can be considered as determinants of access to food (see Figure 1). For this review, a community wide intervention is defined as one which has at least one environmental strategy aimed at improving food supply or food access. Interventions involving multiple strategies can include educational strategies but must have an environmental strategy as one comparative arm of the study. Only studies with a minimum of six months follow up from the start of the intervention to measurement of outcomes will be included. The six‐month period is considered as the minimum time frame in which environmental interventions, in particular, can achieve outcomes.
Included interventions
1. Interventions to improve food supply, such as:
-
agricultural policy to increase local production of nutritious foods;
-
urban planning and land zoning changes to increase the production or retail of nutritious foods;
-
gardening initiatives operating at community level, across multiple schools or households;
-
economic development programmes to support establishing grocery stores in areas that lack easy access to such stores;
-
changes in food retailing or marketing practice for nutritious food.
2. Interventions to improve access to food, such as:
-
social policies to improve socio‐economic opportunities or economic resources for food procurement;
-
food delivery or distribution services;
-
changes in transport services for consumers or community members to food retailers;
-
reducing the price of nutritious food.
Excluded interventions
-
Interventions that provide short term food relief (without a focus on ongoing food security), such as food banks or soup kitchens
-
Interventions to address food insecurity in developed countries in emergency situations such as natural disasters
-
Nutritional labelling (covered in another review)
While a safe food supply is part of the technical definition of food security, we do not address food safety or hygiene in this review.
Types of outcome measures
Primary outcomes
Studies will be included if they have quantifiable data on measures of the primary outcomes listed below or if studies undertook measurement of these primary outcomes (as stated in a published protocol or conference paper about the study) but only the secondary or process measures are reported in the publication. Following on from our logic model (Figure 1) the primary outcomes are as follows.
1. Primary outcomes at the community level
-
Food production and availability ‐ local production of nutritious food
-
Variety and availability in food outlets ‐ range and amount of nutritious food available for purchase
-
Price ‐ price of nutritious food
-
Location of food outlets ‐ physical access to nutritious food (distance or travel time to access nutritious food)
2. Primary outcomes at the household or individual level
-
Financial resources ‐ prevalence of household food insecurity (measured using a valid and reliable tool)
-
Resources and capacity to acquire and consume nutritious food ‐ expenditure on, purchase or consumption of nutritious food, prevalence of people consuming the recommended amounts of nutritious food
Nutritious food is defined as coming from a nutrient rich and physiologically required food group including milk and dairy foods, fruit and vegetables, meat and meat alternatives, and grains and cereal foods and is available or consumed in amounts providing 70% of the nutritional requirements and 95% of the estimated energy requirements of groups of individuals depending on age and gender as specified in the Australian Guide to Healthy Eating (Smith 1998).
Secondary outcomes
Secondary outcomes will be examined if reported in an eligible study. Studies reporting only secondary outcomes without reporting any objective measure of primary outcomes will not be eligible for the review.
1. Secondary outcomes ‐ community level (built, social, economic, natural environments)
-
Land used for agriculture
-
Density of supermarkets or retail outlets selling nutritious foods
2. Secondary outcomes ‐ adverse outcomes
-
Stigmatization
-
Dependency
-
Obesity or excessive weight loss
Search methods for identification of studies
Electronic searches
The discussion of the concept of food security and interventions to improve food security in developed countries only emerged in the peer‐reviewed literature as a broad phenomenon in the 1980s (Tarasuk 2001a). Therefore, we will search electronic databases from 1980 to the present for relevant studies in any language. The following databases will be searched: the Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library), Cochrane Public Health Register, MEDLINE, EMBASE, CINAHL, Global health, Bibliomap, TRoPHI, ERIC, PsycINFO, Sociological Abstracts, Web of science (SCI‐EXPANDED, SSCI), AGRICOLA, Econlit, Food Science and Technology Abstracts, Greenfile.
Some of the selected databases index a combination of published and unpublished studies (for example, doctoral dissertations and conference abstracts), therefore unpublished studies will be partially captured through the electronic search process. See 'Grey literature' below for details of more targeted searches for unpublished studies.
Both the range of potential interventions and the terminology to describe the outcomes of interest are diverse. Therefore, we have developed a sensitive search strategy with terms describing interventions or outcomes used in conjunction with terms describing study design. The search strategy for MEDLINE can be found in Appendix 1. This will be modified for other databases as appropriate.
Searching other resources
Grey literature
The following strategies will be used to identify unpublished studies.
-
Searches of the grey literature databases OpenSIGLE and HMiC.
-
Searches of key websites including the World Health Organization, Food and Agriculture Organization, Community Food Security Coalition, United States Department of Agriculture, Toronto Food Policy Council, Food Standard Agency (UK), Food Research and Action Centre (USA), Food Secure Canada (Bits & Bytes).
Handsearching
We will handsearch from 1980 relevant English language journals not indexed in the electronic databases above (for example, Journal of Hunger and Poverty). Handsearching will also be conducted for the previous five years' issues of the journals that yield the greatest number of relevant studies from the database searches.
Communication
-
A panel of experts on community food security has been convened to provide direction and advice to the review process. These experts will be consulted about any relevant unpublished or ongoing studies that they may be aware of, and may be asked to review the final list of included studies to check for potential gaps.
-
We will post messages on listserves of various national and provincial food security networks.
Data collection and analysis
Selection of studies
Due to the breadth of the concepts around food security and the diversity of sources searched, the search process is expected to return in excess of 10,000 citations. An initial screening of titles and abstracts will therefore be undertaken to exclude studies that are obviously irrelevant (for example, animal and laboratory studies). The abstracts and titles of the remaining citations will be independently assessed by two review authors (CB and BK) for eligibility according to the inclusion criteria. Full copies of all those deemed eligible by one review author, or articles for which eligibility is unclear, will be retrieved for closer examination (CB and BK). All studies which initially appear to meet inclusion criteria but on inspection of the full text paper do not meet the inclusion criteria will be detailed in the table 'Characteristics of excluded studies' together with reasons for their exclusion. Disagreements regarding eligibility of studies will be resolved via consensus and consultation with an arbitrator (SC), if needed. EndNote software will be used to manage the records retrieved.
Data extraction and management
Data will be independently extracted by two review authors (CB, BK). Our data extraction forms will be based on the data collection forms from the Effective Practice and Organisation of Care (EPOC) Review Group and the Cochrane Public Health Review Group, albeit modified for the purposes of this review. If there any discrepancies between the two author in the extraction of data, a third review author (SC) will be used to resolve these discrepancies.
We will extract data on:
-
study design,
-
sampling method of outcomes,
-
description of the intervention (including process factors),
-
context,
-
details about participants (including number in each group),
-
length of intervention and follow up,
-
definition of socio‐economic status of participant groups,
-
primary outcomes at a community and household level,
-
secondary outcomes at a community and household level,
-
cost‐effectiveness,
-
statistical analysis.
Process measures
Measures relating to the process of implementing an intervention will also be extracted. Process measures will be defined by Section 6 of the Cochrane Public Health Group systematic review guidelines and augmented from a list by Audrey 2006. These include:
-
context,
-
resources devoted to intervention,
-
intensity,
-
cost,
-
reach,
-
level of community and stakeholder engagement in planning and implementation of the intervention,
-
characteristics of agencies, people, delivering intervention, including how they are perceived by community members,
-
appropriateness (cultural, education level, fit to need) of intervention,
-
incentives to participate in program, and
-
was intervention informed by a needs assessment?
We shall also employ the 12‐item checklist developed by the Evidence for Policy and Practice Information and Co‐ordinating Centre (EPPI Centre) to assist with our interpretation of process evaluations (www.eppi.ioe.ac.uk). Authors will be contacted for missing information or clarification, as needed. Where possible, we will record effects by socio‐economic position and by other socio‐demographic variables, including place of residence, gender, race or ethnicity, age, and food security status. The data extraction form will be pilot tested on two included studies to minimise error. The outcome data will be assembled with information obtained from concurrent process evaluations to enhance interpretation of the findings.
Assessment of risk of bias in included studies
Two review authors (BK, CB) will independently assess the risk of bias for each study using the following criteria.
a) RCTs, cRCTs, controlled before and after studies (CBAs)
We shall use the Cochrane risk of bias tool and questions from the Quality Assessment Tool for Quantitative Studies (www.ephpp.ca) to assess the validity of the design of included studies. In assessing risk of bias in the RCTs and CBAs, we will consider selection bias, allocation bias, confounding, blinding, data collection methods, withdrawals and dropouts, statistical analysis, and intervention integrity (adherence, exposure, participation). No overall score will be given.
b) Interrupted time series (ITS)
Our appraisal criteria for ITS studies will be adapted from the risk of bias checklist developed by the Cochrane EPOC Group. In assessing risk of bias in the ITS designs, we will consider protection against secular changes, protection against detection bias, reliability of the outcome measures, co‐intervention, and completeness of the data set.
We will not restrict analysis to studies by degree of risk of bias. However, we will identify the risk of bias in our analysis and interpretation of findings. Any discrepancies between review authors will be resolved in a discussion with a third author.
Unit of analysis issues
Where trials have used clustered randomisation, we anticipate that study investigators would have presented their results after appropriately controlling for clustering effects (robust standard errors or hierarchical linear models). If it is unclear whether a cluster randomised controlled trial has used appropriate controls for clustering, the study investigators will be contacted for further information.
Where appropriate controls were not used, individual participant data will be requested and an estimate of the intraclass correlation coefficient will be calculated. The data will be re‐analysed using multi‐level models that control for clustering. Following this, effect sizes and standard errors will be meta‐analysed in RevMan using the generic inverse method (Higgins 2008).
If appropriate controls were not used and individual participant data are not available, statistical support will be provided from members of our team who have expertise in systematic review methodology and analysis. If there is insufficient information to control for clustering, outcome data will be entered into RevMan using individuals as the units of analysis and then sensitivity analysis will be used to assess the potential biasing effects of inadequately controlled cluster trials (Klar 2001).
Where re‐analyses occur, we will clearly mark the new results as 're‐analysed'. We will also note where re‐analysis is not possible.
Dealing with missing data
Authors will be contacted directly by email if outcome data are unclear or have not been fully reported, or appear in a published protocol but these are not reported as outcomes. In particular, if studies report variables of interest at baseline but these are not reported as outcomes, authors will be contacted. All missing outcome data for included studies will be captured on the data extraction form and reported in the risk of bias table. The study will be excluded from the review if insufficient information on the primary outcome of concern can not be obtained (due to inability to contact authors, lost or unavailable data). Reasons for exclusion will be given in the table 'Characteristics of excluded studies'.
Assessment of heterogeneity
Identifying heterogeneity
We will use the following methods to assess heterogeneity.
-
Common sense (e.g., are the interventions, participants, or outcomes so different that they cannot be combined). This will be based on a synthesis of the process elements.
-
Chi2 test for heterogeneity (P < 0.10) and I2 statistic.
-
Visual examination of graphs for outliers and between‐study differences.
If substantial heterogeneity is detected (I2 > 50%) we will present data in a narrative style rather than proceed to meta‐analysis.
Assessment of reporting biases
Where we suspect reporting bias, as detected in the risk of bias assessment, we will contact authors of studies for missing data.
If the number of studies is less than 10 we will follow the methods described in the Cochrane Handbook. If there are more than 10 studies, the possible trial effect will be plotted against standard error and shown as funnel plots (Higgins 2008). Asymmetry could be caused by a relationship between effect and sample size or by publication bias (Egger 1997). If asymmetry is detected, the results will be interpreted in combination with visual examination of the plots. This will allow exploration of small study effects and outliers. If small study effects are identified, sensitivity analysis will be conducted to explore the impact on the meta‐analysis.
Data synthesis
a) RCTs, cRCT, quasi‐experimental or CBA studies
Where baseline data are available from RCTs and CBAs, pre‐intervention and post‐intervention means will be reported for both the study and control groups and the absolute change from baseline will be calculated (change in study group values minus change in control group values), along with standard deviations and 95% confidence intervals (CI). If the standard deviations (sd) for change are not given, we will calculate them using the formula: SQRT | ((sd2baseline+ sd2endofstudy) ‐2*?*(sdbaseline *sdendofstudy)) | where? represents the correlation between baseline and end of study (estimated at 0.4). When baseline data are not available, results will be expressed as the relative percentage change (difference between post‐intervention value in the study and control group expressed as a percentage of the post‐intervention value in the control group).
b) Interrupted time series (ITS)
We will calculate relative and absolute mean difference in the before and after values. When possible, we will use time series regression to calculate mean change in level and mean change in slope.
c) Binary outcomes
For binary outcomes, we will present the relative risk (RR) of the outcome compared to the control group. We will also calculate the risk difference, which is the absolute difference in the proportions in each treatment group.
d) Meta‐analyses
Meta‐analysis will be carried out when there is a minimum of two studies of any type of intervention to be combined. If we do not find enough studies for a meta‐analysis or for the assessment of heterogeneity (I2 > 50%), the review will be reported as a descriptive narrative only.
Meta‐analyses will be performed separately for each outcome, by single versus multiple strategies, and study design (RCTS (including cRCTs and controlled clinical trials (CCTs), CBAs, and ITS).
We shall use a random‐effects model for all analyses (Higgins 2008). When outcomes are measured on the same scale, we will use the mean difference; when they are on different scales we shall report the standardized mean difference. Data synthesis will be performed using RevMan 2008.
We will provide an estimate and 95% confidence interval (CI) and generate a forest plot for each meta‐analysis and will discuss the extent of evidence against homogeneity.
Subgroup analysis and investigation of heterogeneity
When possible, comparisons will be reported by socio‐economic group as well as by other relevant socio‐demographic variables, including baseline nutritional status, gender, race or ethnicity, and place of residence or food security status. Where results by socio‐economic variables are not available in the primary articles and reports, we will request this data from the authors and re‐calculate effect sizes and P values. We will then conduct quantitative meta‐analysis, if possible. We will conduct separate analyses for each outcome according to: 1) different study designs (that is, ITS, RCT, and CBA); 2) different countries; 3) community wide versus high risk populations; and 4) single versus multiple strategies. The results will be interpreted using clinical significance as well as statistical significance.
Sensitivity analysis
Sensitivity analysis will be used to evaluate whether the pooled effect sizes are robust across components of methodological quality. For methodologic quality, we will conduct sensitivity analyses for each major component of the quality checklists (for example, blinded assessment, randomisation, reliable primary outcome). We will also conduct sensitivity analyses to determine whether differences exist in results if we exclude those studies with estimated standard deviations and to assess the implications of using the imputed correlation coefficient in estimating standard deviations for change. We will carry out a sensitivity analysis of studies with a low risk of bias as reported within the Cochrane 'Risk of bias' table.
