Intervenciones en jurisdicciones gubernamentales para la reducción del sodio dietético en la población
Resumen
Antecedentes
El consumo excesivo de sodio dietético es un factor de riesgo de hipertensión, accidente cerebrovascular y enfermedades cardiovasculares. En la actualidad, el consumo de sodio dietético es demasiado alto en casi todo el mundo. La ingesta excesiva de sodio se asocia con la hipertensión, una patología frecuente y costosa, que representa una carga de morbilidad significativa. Un gran número de jurisdicciones en todo el mundo han puesto en marcha iniciativas para la reducción del sodio dietético en la población. Ninguna revisión sistemática ha examinado el impacto de estas iniciativas.
Objetivos
• Evaluar la repercusión de las intervenciones para la reducción del sodio dietético en la población a partir de jurisdicciones gubernamentales.
• Evaluar la repercusión diferencial de esas iniciativas mediante indicadores sociales y económicos.
Métodos de búsqueda
Se hicieron búsquedas en las siguientes bases de datos electrónicas desde su fecha de inicio hasta el 5 de enero de 2015: el Registro Cochrane Central de Ensayos Controlados (Cochrane Central Register of Controlled Trials, CENTRAL); el Registro Especializado del Grupo Cochrane de Salud Pública (Cochrane Public Health Group Specialised Register); MEDLINE; MEDLINE In Process & Other Non‐Indexed Citations; EMBASE; Effective Public Health Practice Project Database; Web of Science; bases de datos del Registro de Ensayos de Promoción de Intervenciones Sanitarias (Trials Register of Promoting Health Interventions, TRoPHI); y Latin American Caribbean Health Sciences Literature (LILACS). También se realizaron búsquedas en la literatura gris, en otras fuentes nacionales y en las referencias de los estudios incluidos.
Esta revisión se realizó en paralelo con una revisión integral sobre los esfuerzos para la reducción del sodio en todo el mundo (Trieu 2015), por medio de la cual se obtuvo información adicional directa de los contactos en los países.
No se impusieron restricciones en base al idioma o el estado de la publicación.
Criterios de selección
Se incluyeron iniciativas en la población (es decir, intervenciones orientadas a poblaciones enteras, en este caso, jurisdicciones gubernamentales, a nivel mundial) para la reducción del sodio dietético, con al menos una toma de datos antes de la intervención y al menos una después de la misma, de una jurisdicción equivalente. Se incluyeron poblaciones de todas las edades y los siguientes tipos de diseños de estudio: aleatorizado en grupos, controlados pre y post, series de tiempo interrumpido y sin control pre y post. Se estableció contacto con los autores de los estudios en diferentes puntos de la revisión para solicitar información faltante.
Obtención y análisis de los datos
Dos autores de la revisión extrajeron los datos y dos autores de la revisión evaluaron el riesgo de sesgo para cada iniciativa incluida.
Se analizó la repercusión de las iniciativas mediante cálculos del consumo de sodio a partir de encuestas sobre la dieta o muestras de orina. Todos los cálculos se convirtieron a una medida común: la ingesta de sal en gramos por día. Se analizó la repercusión mediante el cálculo del cambio medio en la ingesta de sal (gramos por día) desde antes de la intervención hasta después de ésta.
Resultados principales
Se revisó un total de 881 documentos de texto completo. De éstos, se identificaron 15 iniciativas nacionales, con más de 260 000 personas, que cumplieron con los criterios de inclusión. Ninguna de las iniciativas se aplicó en países de ingresos bajos a medios o bajos. Todas las iniciativas excepto una usaron un diseño de estudio no controlado pre y post.
Debido a los niveles altos de heterogeneidad de los estudios (I2 > 90%), se centró la atención en las iniciativas individuales en lugar de los resultados agrupados.
Diez iniciativas aportaron datos suficientes para el análisis cuantitativo de la repercusión (64 798 participantes). Según el método Grades of Recommendation, Assessment, Development and Evaluation (GRADE), se calificó la calidad de la evidencia como muy baja por el riesgo de sesgo de los estudios incluidos, así como la variación en la dirección y el tamaño del efecto entre los estudios. Cinco estudios mostraron una disminución media en la ingesta de sal diaria promedio por persona desde antes hasta después de la intervención, con una variación de 1,15 g/día menos (Finlandia) a 0,35 g/día menos (Irlanda). Dos iniciativas mostraron un aumento medio de la ingesta de sal desde antes de la intervención a después de la misma: Canadá (1,66) y Suiza (0,80 gramos/día más por persona); sin embargo, en ambos países el punto temporal de datos preintervención fue de varios años antes de que comenzara la intervención. Las iniciativas restantes no mostraron un cambio medio estadísticamente significativo.
Siete de las 10 iniciativas eran actividades de naturaleza estructural de intervenciones incorporadas y de componentes múltiples (p.ej. reformulación de productos alimenticios, política de adquisición de alimentos en ámbitos específicos). De estas siete iniciativas, cuatro mostraron una disminución media estadísticamente significativa en la ingesta de sal desde antes de la intervención hasta después de la intervención, con una variación de Finlandia a Irlanda (ver anteriormente), y una mostró un aumento medio estadísticamente significativo de la ingesta de sal desde antes de la intervención hasta después de la intervención (Suiza, véase arriba).
Nueve iniciativas permitieron el análisis cuantitativo de la repercusión diferencial por sexo (hombres y mujeres por separado). Para las mujeres, tres iniciativas (China, Finlandia, Francia) mostraron una disminución media estadísticamente significativa; cuatro (Austria, los Países Bajos, Suiza, Reino Unido) no mostraron ningún cambio significativo; y dos (Canadá, Estados Unidos) mostraron un aumento medio estadísticamente significativo de la ingesta de sal desde antes de la intervención hasta después de la intervención. Para los hombres, cinco iniciativas (Austria, China, Finlandia, Francia, Reino Unido) mostraron una disminución media estadísticamente significativa; tres (los Países Bajos, Suiza, Estados Unidos) no mostraron ningún cambio significativo; y una (Canadá) mostró un aumento medio estadísticamente significativo de la ingesta de sal desde antes de la intervención hasta después de la intervención.
La información fue insuficiente para indicar si hubo un cambio diferencial en la ingesta media de sal desde antes de la intervención hasta después de la intervención por otros ejes de la equidad incluida en el marco PROGRESS (p.ej. educación, lugar de residencia).
No se identificó ningún efecto adverso con estas iniciativas.
El número de iniciativas fue insuficiente para permitir otros análisis de subgrupos, incluida la estratificación por tipo de intervención, el estado económico del país y la duración (o año de comienzo) de la iniciativa.
Muchos estudios tenían fortalezas metodológicas, incluidas las grandes muestras, representativas a nivel nacional, de la población y una medición rigurosa de la ingesta de sodio dietético. Sin embargo, todos los estudios se calificaron como de alto riesgo de sesgo, lo que demuestra la naturaleza observacional de la investigación y el uso de un diseño de estudio no controlado. La calidad de la evidencia para el resultado principal fue baja. Se pudo realizar un análisis de sensibilidad sólo para la repercusión.
Conclusiones de los autores
Las intervenciones en jurisdicciones gubernamentales para la reducción del sodio dietético en la población pueden dar lugar a reducciones de la ingesta de sal en la población desde antes de la intervención hasta después de la intervención, en particular si son componentes múltiples (más de una actividad de intervención) e incorporan actividades de naturaleza estructural (p.ej. reformulación de productos alimenticios), y especialmente en hombres. La heterogeneidad entre los estudios fue significativa, lo que representa diferentes contextos (población y ámbitos) y las características de la iniciativa. La implementación de las iniciativas futuras debe incluir medios más efectivos de evaluación para ayudar a una mejor comprensión de la variación en los efectos.
Resumen en términos sencillos
Iniciativas de los gobiernos nacionales para reducir la ingesta de sal en la población
El problema
En casi todos los países del mundo, la mayoría de las personas consumen demasiada sal. Constituye un problema porque el consumo excesivo de sal puede causar hipertensión, lo que puede llevar a problemas de salud como cardiopatías y accidente cerebrovascular. Para reducir la cantidad de sal consumida, los gobiernos de muchos países han desarrollado iniciativas nacionales de reducción de la sal.
Pregunta de la revisión
El objetivo fue examinar si las iniciativas nacionales de reducción de la sal han sido efectivas en la reducción de la cantidad de sal consumida en esas poblaciones.
Características de los estudios
Se realizaron búsquedas en artículos de investigación e informes oficiales y se contactó en forma directa con individuos que trabajaran en la reducción del consumo de sal en los países respectivos. La evidencia está actualizada al 5 de enero de 2015, cuando se realizó la última búsqueda en las bases de datos electrónicas. Las iniciativas de 15 países cumplieron con los criterios de inclusión. Diez de estos países aportaron datos suficientes para el análisis cuantitativo, a partir de estudios que incorporaron a 64 798 participantes. Las iniciativas variaron entre una actividad (p.ej. en Japón, que en el momento de la redacción tenía una campaña de información pública) y muchas actividades (p.ej. en el Reino Unido, que proporcionó cinco actividades incluida la información nutricional en el empaquetado, restricciones de comercialización a niños y reformulación de productos alimentarios). De los 15 países que cumplieron con los criterios de inclusión, siete presentaron información acerca de la fuente de financiamiento y de éstos, seis informaron financiamiento no proveniente de la industria. Los otros ocho países no informaron una fuente de financiamiento para uno o más puntos de datos.
Resultados clave y calidad de la evidencia
Cinco de los 10 países incluidos en el análisis cuantitativo (China, Finlandia, Francia, Irlanda e Inglaterra) mostraron una disminución en la ingesta de sal después de la intervención. Dos de los 10 países (Canadá, Suiza) presentaron un aumento de la ingesta de sal después de la intervención, sin embargo, en ambos países los únicos datos disponibles fueron de varios años antes de que comenzara la intervención. No se puede presentar un resultado general sobre si estas iniciativas son útiles debido a las significativas diferencias entre ellas.
Cuando se centró la atención en el subconjunto de siete países cuyas iniciativas de reducción de la sal incluyeron componentes múltiples y no se dedicaron exclusivamente a la educación de la población, se halló que más de la mitad (cuatro de siete) mostró una disminución en la ingesta de sal desde antes de la intervención hasta después de la intervención.
Cuando se examinaron las nueve iniciativas que analizaron a mujeres y hombres por separado, se halló que entre los hombres, más de la mitad (cinco de nueve) mostró una disminución en la ingesta de sal después de la intervención. Entre las mujeres, el patrón de los resultados fue menos claro: tres de nueve intervenciones mostraron una disminución; dos, un aumento; y cuatro no mostraron ningún cambio en la ingesta de sal.
Los diseños de estudio de bajo sesgo, como los ensayos controlados aleatorizados, habitualmente no son apropiados para evaluar iniciativas complejas como éstas; por lo tanto, se calificaron todos los estudios incluidos en esta revisión como de baja calidad metodológica. Las grandes muestras poblacionales representativas a nivel nacional y la medición cuidadosa de la ingesta de sodio dietético fueron las fortalezas de varios estudios. Sin embargo, no está clara la fiabilidad de los resultados de los estudios debido a las limitaciones en el diseño de estudio.
En términos generales, los resultados muestran que las iniciativas de los gobiernos nacionales pueden lograr reducciones con alcance en la población de la ingesta de sal, especialmente entre los hombres y en particular si emplean más de una estrategia e incluyen actividades estructurales como la reformulación de productos alimenticios (es decir, empresas alimentarias que incluyen menos sal en los productos alimenticios). La amplia variación de los resultados en los estudios hallados implica un desafío para la interpretación de la evidencia actual. Este hecho explica la necesidad de más estudios de investigación que ayuden a una mejor comprensión de la situación.
Authors' conclusions
Summary of findings
| Population‐level interventions in national government jurisdictions to reduce sodium consumption | ||||||
| Patient or population: people of all ages Settings: government jurisdictions Intervention: population‐level intervention to reduce dietary sodium consumption Comparison: mean change in salt intake (grams/d) from pre‐intervention to post‐intervention | ||||||
| Outcomes | Illustrative comparative risks | Relative effect | Number of participants | Quality of the evidence | Comments | |
| Pre‐intervention estimate | Post‐intervention estimate | |||||
| Salt intake in grams per day ‐ all eligible studies that provided sufficient data for quantitative analysis of overall impact (n = 10 countries) | Mean salt intake at pre‐intervention ranged from 6.1 to 12.7 grams/d | Mean salt intake at post‐intervention ranged from 5.5 to 11.9 grams/d | Not shown owing to high study heterogeneity | 64,798 participants | ⊕⊝⊝⊝ | As the result of high study heterogeneity (I2 > 90% in all analyses), we focus on individual studies rather than pooled effects. We downgraded the quality of evidence from low to very low due to all studies having high risk of bias (confounding domain) and the substantial inconsistency of effect sizes and the direction of the effect, which varied across the studies. |
| Salt intake in grams per day ‐ subset of initiatives that are multi‐component and incorporate activities of a structural nature (n = 7 countries) | Mean salt intake at pre‐intervention ranged from 6.3 to 11.8 grams/d | Mean salt intake at post‐intervention ranged from 5.5 to 10.6 grams/d | Not shown owing to high study heterogeneity | 34,227 participants (9 studies)* | ⊕⊝⊝⊝ | |
| Salt intake in grams per day ‐ women only (n = 9 countries) | Mean salt intake at pre‐intervention ranged from 4.7 to 12.2 grams/d | Mean salt intake at post‐intervention ranged from 4.1 to 11.4 grams/d | Not shown owing to high study heterogeneity | 27,184 participants (11 studies)* | ⊕⊝⊝⊝ | |
| Salt intake in grams per day ‐ men only (n = 9 countries) | Mean salt intake at pre‐intervention ranged from 6.1 to 13.2 grams/d | Mean salt intake at post‐intervention ranged from 5.2 to 12.5 grams/d | Not shown owing to high study heterogeneity | 22,977 participants (11 studies)* | ⊕⊝⊝⊝ | |
| GRADE Working Group grades of evidence | ||||||
| *Includes evaluations of the United Kingdom for England only, Scotland only and Great Britain. Including only Great Britain does not change overall findings We rated the overall quality of evidence using the GRADE (Grades of Recommendation, Assessment, Development and Evaluation Working Group) framework (GRADE n.d.), which is based on 5 considerations: risk of bias, imprecision, inconsistency, indirectness and publication bias. All of our studies started at 'low' quality because of their observational, uncontrolled nature. | ||||||
Background
We have made minor changes to this section. See Differences between protocol and review.
Please note that, throughout the review, we use both terms ‐ "sodium" and "salt" ‐ to respect the term used in the cited document. For our analysis (see below), we converted all estimates to a common metric: salt intake in grams per day.
Description of the issue
Excess dietary sodium consumption is a risk factor for high blood pressure, stroke, cardiovascular disease and other adverse health outcomes (He 2009; Mohan 2009; Strazzullo 2009). Particularly strong evidence suggests a direct relationship between sodium intake and high blood pressure (He 2009). Hypertension is common (Vasan 2002) and costly (Gaziano 2009) and accounts for a significant burden of disease and death. The World Health Organization (WHO) considers high blood pressure to be the leading preventable risk factor for death in the world (Ezzati 2002; Lopez 2006). It has been estimated that mean salt consumption in 181 of 187 countries exceeded the daily intake of salt recommended by WHO in 2010, and 51 of these countries reported mean intakes greater than double the recommended amount (WHO 2007; Powles 2013).
Dietary sodium reduction may be addressed through a population‐level intervention approach. Population‐level interventions target whole populations (e.g. jurisdictions), including individuals with higher risk profiles and those with more moderate risk profiles (Rose 1992; McLaren 2010). This contrasts with a high‐risk strategy whereby efforts are focused on individuals at highest level of risk. The population‐level approach may provide significant leverage for impact because the societal impact of moving the entire distribution of risk to the left (i.e. in the direction of lower risk) may be very large (Figure 1) (Rose 1992). Dietary sodium reduction lends itself to a population‐level intervention approach for several reasons. First, the association between sodium intake and blood pressure is linear with no obvious threshold (SWG 2010) and thus no clear cutoff from which a discrete high‐risk group of individuals can be identified for targeted intervention. Second, modelling studies have shown that the population impact of a widespread reduction in sodium intake is potentially very large (e.g. Joffres 2007; Smith‐Spangler 2010) and could yield significant cost savings (Joffres 2007; Dall 2009) such that cost‐effectiveness exceeds that of clinical interventions (e.g. Bibbins‐Domingo 2010). A study of 23 low‐income and middle‐income countries, which account for 80% of global deaths from chronic disease, estimated that 8.5 million deaths could be averted over 10 years through a 15% reduction in salt intake across the population (Asaria 2007). Third, knowledge of the main sources of sodium in the diet of a given population provides guidance about leverage points for population intervention in different countries. Specifically, in affluent countries, most (75% to 80%) sodium consumed comes from processed foods (Mattes 1991; Andersen 2009); therefore the food industry represents an important leverage point for intervention. In lower‐income and middle‐income countries, most sodium is consumed through addition of salt during the cooking or eating process to sauces and seasonings that are very high in salt (WHO 2007); thus a sodium reduction intervention in these countries might take the form of reduced salt content in condiments, increased use of salt substitutes and development of information campaigns aimed at individual behaviour change.
Depiction of the hypothetical impact of a population‐level intervention, achieved by shifting the frequency distribution of the risk factor in a direction of lower risk (adapted from Rose 1992).
When population‐level sodium reduction interventions are undertaken, it is important that they do not worsen socio‐economic inequities in health (Whitehead 2007; Frohlich 2008), including those involving sodium intake (Ji 2014; McLaren 2014). Concern about the presence of and increase in socio‐economic inequities in health has been voiced by various national and international health organisations, including the World Health Organization (WHO) (CSDH 2008), the US Department of Health and Human Services (USDHHS) and the Public Health Agency of Canada (PHAC 2006). Despite enduring existence of and concern about socio‐economic inequities in health worldwide (CSDH 2008), we currently know very little about whether, or the extent to which, population‐level interventions are equitable in their impact (Whitehead 2007; Bambra 2010). By using dietary sodium reduction as an example, this review provides insight into the broader question of whether and how population‐level interventions can achieve impact that is both overall and equitable.
Description of the intervention
Review authors focused on population‐level initiatives (i.e. interventions that target whole populations, in this case, government jurisdictions, worldwide) for dietary sodium reduction.
From the dietary sodium reduction literature, we identified six types of population‐level interventions for dietary sodium reduction that may occur within government jurisdictions. These intervention types may be conceptualised as existing on a continuum anchored by more agentic interventions (i.e. act on behaviours) at one end, and more structural interventions (i.e. act on circumstances in which behaviours occur) at the other end (McLaren 2010).
-
Food product reformulation: large‐scale efforts to lower the sodium content of food products at the time of production. By 'large‐scale', we mean efforts that characterise whole jurisdictions (e.g. provinces, countries), transcending specific settings within jurisdictions (see 'food procurement policy', below).
-
Pricing interventions: large‐scale strategies (e.g. taxation, subsidisation) designed to manipulate the price of food products in a way that encourages the purchase of healthier foods and discourages the purchase of less healthy foods. By 'large‐scale', we mean efforts that span whole jurisdictions (e.g. provinces, countries), transcending specific settings within jurisdictions. This may include strategies to manipulate the price of table salt or of high salt‐containing products (e.g. condiments) in lower‐income and middle‐income countries.
-
Food procurement policy in specific settings: nutrition policy (e.g. limits on sodium content in foods) implemented within a system of contained food service settings or environments, particularly publicly funded environments such as schools, colleges/universities, childcare settings, workplaces, recreational facilities, prisons, hospitals and long‐term care facilities.
-
Restrictions on marketing to children: efforts by government to restrict the extent or nature of promotional/marketing activities conducted by companies. As an example, the province of Quebec, Canada, has banned television advertising to children since 1980 under sections 248 and 249 of the Consumer Protection Act (CPA 1978). By 'marketing', we mean use of various media, including but not limited to television.
-
On‐package nutrition information: nutrition information provided directly on food packaging. We have identified three subtypes of on‐package nutrition information: (1) information on calories, nutrients and percentage daily value of nutrients (e.g. Canada's Nutrition Facts Table); (2) on‐package symbols provided to assist consumers with healthy food selection (e.g. the 'traffic light' system of the UK, which tells consumers at a glance whether a food is high (red), medium (amber) or low (green) in salt, sugar, fat and saturated fat, respectively); and (3) on‐package high‐salt warning labels (e.g. mandatory warning labels on foods in Finland whose sodium content exceeds defined limits).
-
Information campaigns: public information/education campaigns that focus specifically on salt/sodium or more broadly on diet (including sodium). Campaigns may be delivered via diverse media, including television, radio, posters/billboards, newspapers, other print materials, social media (e.g. Facebook, Twitter), RSS (really simple syndication) feeds, email subscriptions, online advertising and cell phones (health 'apps', text message updates) but must pertain to a whole government jurisdiction to be eligible for our review.
How the intervention might work
This review is anchored in scholarly literature on social determinants of health, population health and health promotion (e.g. Ottawa Charter 1986; Whitehead 2007; CSDH 2008; Raphael 2009). From this perspective, a general starting point is a major, although sometimes unclear, distinction between interventions that are individually oriented and those that are structurally oriented. Individually oriented interventions include efforts that aim to strengthen individuals by providing information, education or skills. The underlying theory (implicit or explicit) is that the health issue (in this case, excess sodium consumption and associated health problems) reflects personal deficits such as lack of information, knowledge or skills, and the interventions are designed to offset, or make up for, these deficiencies (Whitehead 2007). Structurally oriented interventions aim to improve the environment, settings or conditions in which individuals live, work, go to school and recreate. The underlying theory (implicit or explicit) is that the health issue reflects structural deficits such as limited availability of appropriate foods or limited access or opportunity to procure appropriate foods as the result of restricted social and economic resources such as income, social status, place of residence and cultural/ethnic group (Whitehead 2007; Raphael 2009). Structurally oriented interventions are designed to offset, or make up for, these deficiencies.
Building on the foundational work of Rose (Rose 1992), we (McLaren 2010; Sumar 2011) distinguished between population‐level interventions that are more agentic (target behaviour change amongst individuals) and those that are more structural (target conditions in which behaviours occur). The intervention types listed above convey the breadth of approaches available for dietary sodium reduction, ranging from individually oriented or agentic, to structurally oriented, as illustrated in our logic model (Figure 2) and accompanying glossary (Appendix 1).
Logic model.
We have identified four broad, overlapping categories of mechanisms by which interventions may affect food selection and purchase, sodium intake and ultimately health.
-
Availability and accessibility: number or proportion of lower‐sodium food products available on the market or within settings of interest.
-
Cost: purchase price of lower‐sodium products, relative to higher‐sodium products, especially relative to higher‐sodium versions of the same product.
-
Appeal and familiarity: symbolic desirability of or attachment to a food product as experienced by consumers, established through effective marketing or some other means (e.g. personal tastes and preferences).
-
Knowledge and awareness: consumer cognizance of dietary sodium specifically or of diet generally, including associations with health.
A single, one‐to‐one relationship between intervention type and mechanism may not be evident. For example, large‐scale pricing strategies may facilitate selection of lower‐sodium food via 'cost' while impeding selection via 'appeal and familiarity' if the lower‐sodium products do not appeal to the individual.
Why it is important to do this review
We focus on population‐level dietary sodium reduction interventions that are implemented by governments. This focus is important because an increasing number of jurisdictions worldwide are developing and/or implementing population‐level dietary sodium reduction interventions (Webster 2011; Trieu 2015). Although some countries have a long history of population‐level sodium reduction efforts (e.g. Finland, where efforts have been ongoing since the 1970s), most initiatives have been implemented more recently. The number of countries with active programmes has increased enormously since the early 2000s, when the World Health Organization first publicised its global target of < 5 grams salt/d (corresponding to < approximately 1967 mg/d of sodium, which sometimes is rounded to < 2000 mg/d of sodium) (WHO 2003; Penney 2011; Trieu 2015).1 It is important and timely to systematically assess (1) the overall impact and (2) the differential impact of these initiatives.
Other Cochrane reviews on dietary sodium reduction have focused on patient populations and intensive clinical interventions (Hooper 2004), or have combined diverse target populations and interventions in the interest of focusing specifically on the main effects of sodium reduction per se (Taylor 2011). Further, none have examined the equity of intervention impact. Our focus on population‐level dietary sodium reduction interventions in government jurisdictions, including how equitable they are in terms of their impact, is unique and has important implications for policy‐makers worldwide in terms of identifying potentially impactful and cost‐effective mechanisms for sodium reduction that lie outside the health sector.
Objectives
-
To assess the impact of population‐level interventions for dietary sodium reduction in government jurisdictions worldwide.
-
To assess the differential impact of those initiatives by social and economic indicators.
Methods
Criteria for considering studies for this review
Types of studies
This section is substantively the same as the protocol (McLaren 2013).
Because of the unit of intervention (populations, not individuals) and the scale and scope of the interventions of interest (government jurisdictions), inclusion criteria go beyond randomised controlled trials to include other study designs.
The jurisdictional nature of intervention delivery is such that the interventions of interest often resemble a natural experiment whereby assessment of impact is based on jurisdictional (e.g. national, provincial) data gathered before and after the intervention, as we found in a similar review (Sumar 2011), in which almost all (18 of 19) included studies used an interrupted time series design or an uncontrolled pre‐post design.
The following study characteristics were included.
-
Cluster‐randomised studies (i.e. studies in which populations (jurisdictions) are randomised to one intervention condition or another).
-
Controlled pre‐post studies (i.e. studies with pre‐intervention and post‐intervention data from the intervention population and a comparison population).
-
Interrupted time series studies (i.e. studies reporting more than one data point before, and more than one data point after, the intervention, in a single population).
-
Uncontrolled pre‐post studies (i.e. studies reporting at least one pre‐intervention data point and at least one post‐intervention data point from the same or a comparable jurisdiction (e.g. both pre‐intervention and post‐intervention data points were national estimates).
Please note that we use the term "study" to refer to the interventions or initiatives considered in this review.
Types of participants
This section is the same as the protocol (McLaren 2013).
We included populations of males and females, of any age, living in any geographic region worldwide.
To be included, the study had to focus on a population, which for this review was defined as a government jurisdiction (e.g. country, state/region/province, municipality).
We excluded studies that focused on (1) a population subgroup defined by health or socio‐demographic indicators or (2) an area or setting that was not a government jurisdiction (e.g. school, community).
Types of interventions
We have made minor changes to this section. See Differences between protocol and review.
We included population‐level interventions in government jurisdictions (e.g. national, state/regional/provincial, municipal) for dietary sodium reduction in which activities were under way (vs in the planning stages) and for which a start date could be identified (for the purpose of confirming pre‐intervention and post‐intervention data points). Interventions included at least one of the following activities.
-
Public information/education campaign.
-
On‐package nutrition information.
-
Restriction on marketing to children.
-
Food procurement policy in specific settings.
-
Pricing intervention.
-
Food product reformulation.
We excluded the following.
-
Interventions targeting those at high risk (e.g. individuals with pre‐existing hypertension).
-
Interventions, regardless of target population, delivered in a one‐on‐one or small group format.
-
Pharmaceutical interventions.
Types of outcome measures
We have made important changes to this section. See Differences between protocol and review.
Primary outcomes
-
Any measure of dietary sodium consumption, including the following.
-
-
Dietary survey (e.g. 24‐hour recall, food frequency questionnaire, consumption as estimated from reported food purchasing).
-
Urine sample (e.g. 24‐hour urine, spot urine).
-
All estimates were converted to a common metric ‐ salt intake in grams per day ‐ through the following conversions.
-
1 gram of salt = 393.4 milligrams of sodium.
-
Salt intake (grams/d) = urinary sodium concentration (mmol/d) * molecular weight of sodium chloride (0.058 grams/mmol).
One gram of salt, or 393.4 milligrams of sodium, corresponds to approximately 1/6 of a teaspoon of table salt, which represents approximately 1/5 of the World Health Organization global target of 5 grams of salt per day (WHO 2003).
Secondary outcomes
None (see Differences between protocol and review).
Search methods for identification of studies
Electronic searches
This section is the same as in the protocol (McLaren 2013).
We searched the following research databases from their start date to 5 January 2015.
-
Cochrane Central Register of Controlled Trials (CENTRAL).
-
Cochrane Public Health Group Specialised Register (via communication with the Cochrane Public Health Group Trials Search Co‐ordinator).
-
MEDLINE (Ovid platform).
-
MEDLINE In Process & Other Non‐Indexed Citations (Ovid platform).
-
EMBASE (Ovid platform).
-
Effective Public Health Practice Project Database.
-
Web of Science.
-
Trials Register of Promoting Health Interventions (TRoPHI) databases.
-
Latin American Caribbean Health Sciences Literature (LILACS) database published by BIREME (http://new.paho.org/bireme/, a Pan American Center of the Pan American Health Organization, Regional Office of WHO).
We developed a detailed search strategy by combining established search terms and free text terms for each database. Searches combined dietary sodium search terms with terms designed to capture studies on the breadth of population‐level intervention types as described above (food product reformulation, pricing interventions, food procurement policy, marketing restrictions, on‐package nutrition information, information campaigns). We validated the search against key articles known to be important for our review. We developed the search strategy for MEDLINE and adapted it for the other databases to take account of differences in search terms and syntax rules. The search strategies used for the research databases are shown in Appendix 2.
We applied no date or language limits.
Searching other resources
We have made changes to this section. See Differences between protocol and review.
Grey literature
We searched the following grey literature websites and resources, using the search strategy indicated.
| Grey literature website or resource | Search term/strategy applied |
| OpenGrey | Sodium or salt (discipline ‘Medicine’) |
| World Health Organization | Dietary sodium or salt and reduc* or decrease and strategies or intervention |
| Public Health Agency of Canada | Dietary sodium or salt and reduc* or decrease |
| Centers for Disease Control and Prevention | Salt (http://www.cdc.gov/salt/publications.htm) → Publications |
| Pan American Health Organization (Caribbean Food and Nutrition Institute) | Health topics → Salt reduction (http://www.paho.org/hq/index.php?option=com_content&view=article&id=2015&Itemid=4024) |
| World Action on Salt and Health (WASH) | World Action and search for information about relevant literature from each country |
| Institute of Medicine | Keyword search: sodium or salt or sodium chloride |
Country questionnaires
On the basis of the searches described above, and led by review authors Webster and Trieu (Trieu 2015), we established a list of national sodium reduction initiatives and sent it to international experts and representatives of the World Health Organization to identify whether we had missed any countries with initiatives.
Led by Webster and Trieu (Trieu 2015), we prepared a questionnaire (Appendix 3) and sent it to 87 country programme leaders identified through the expert review. We followed up queries with country programme leaders or the relevant WHO regional expert or through a targeted search. Those findings are published in our companion review (Trieu 2015), and we considered all 75 initiatives reported in that paper for inclusion in this review.
Other modes of obtaining information
We handsearched the reference lists of included studies and performed related reference searches on PubMed and citation searches on Web of Science to ensure that we had identified all available published material for each intervention. As we refined the list of initiatives (countries) to be included in this review, we undertook purposive Internet searching for materials pertaining to those initiatives (e.g. government statistical agency reports or websites describing survey methods).
In the event of missing information or uncertainty, we attempted to contact study authors or country contacts via email.
Data collection and analysis
Selection of studies
This section is substantively the same as in the protocol (McLaren 2013).
Two review authors (NS and LM, or AB and LM) independently screened the titles and abstracts (when available) of all reports identified through the online searches. We retrieved in full all reports that appeared to meet inclusion criteria, or for which titles and abstracts provided insufficient detail. Two review authors (NS and LM, or AB and LM) independently assessed the full reports to determine whether they met inclusion criteria, resolving disagreements by discussion.
Data extraction and management
This section is substantively the same as in the protocol (McLaren 2013).
Two review authors (AB and LM) collaboratively extracted data using data extraction forms designed for this review, which had been pilot tested and revised before finalisation. We resolved disagreements on data extraction via discussion and contacted authors of primary studies and/or country contacts to request clarification of unclear data or to obtain missing information. When we received responses, we included this information in the review.
For each initiative, we extracted the following information.
-
Study design.
-
Participants (age, sex, region of residence).
-
Sampling strategy.
-
Sample size.
-
Response rate.
-
Details of the intervention, including time frame and main activities.
-
Outcomes (measures and units of dietary sodium intake).
-
Axes of inequality (PROGRESS indicators included, if any. PROGRESS, a framework for incorporating focus on equity in systematic reviews, includes the following axes: place of residence; race/ethnicity/culture/language; occupation; gender/sex; religion; education; socioeconomic status; social capital).
-
Funding source(s).
-
Conflicts of interest.
-
Sources of data points and references.
Assessment of risk of bias in included studies
We have made changes to this section. See Differences between protocol and review.
We assessed risk of bias using an adapted version of the Cochrane risk of bias tool (Higgins 2011). Our adapted risk of bias tool is similar to, and was informed by, the one used in a recently published Cochrane systematic review (Iheozor‐Ejiofor 2015). We assessed the following seven bias domains: sampling, confounding, reliability/validity of outcome measures, blinding of outcome assessment, representativeness of sample of underlying population, risk of selective outcome reporting and other sources of bias.
Two review authors (AB and LM, with the assistance of a research assistant (non‐author)) independently assessed risk of bias (high, low, unclear) for each bias domain for each initiative. Risk of bias assessment was based on information obtained from multiple, diverse materials (e.g. scientific journal article, government report or website, country questionnaire) for each data point. Assessment for each bias domain was based on the worst (highest risk of bias) rating, across data points. For example, for the sampling domain, if an initiative was assessed as having high risk of bias for one data point and low risk of bias for the other data point, the overall assessment for the sampling domain for that initiative assigned high risk of bias.
To assign a summary risk of bias assessment for an initiative (i.e. across all domains), the procedure is to assign low risk of bias overall when risk of bias is low for all domains, unclear risk of bias overall when risk of bias is unclear for one or more domains, and high risk of bias overall when risk of bias for one or more domains is high. However, because all of our interventions had at least one domain rated as having high risk of bias (and thus was scored as having high risk of bias overall), we sought a way to capture variation in methodological quality across our included studies. Accordingly, we computed the proportion of bias domains (number out of 7) scored as high or uncertain risk for each intervention.
We rated the overall quality of evidence using the GRADE (Grades of Recommendation, Assessment, Development and Evaluation Working Group) framework (GRADE n.d.), which is based on five considerations: risk of bias, imprecision, inconsistency, indirectness and publication bias. We assigned four possible GRADE ratings: high, moderate, low and very low. Observational studies start at a GRADE rating of low, which may be increased or decreased. The grade may be decreased by one or (if very serious) two levels in the following circumstances: serious or very serious limitations to study quality; important inconsistency; some or major uncertainty about directness; imprecise or sparse data; or high probability of reporting bias. The grade may be increased in the following circumstances: strong evidence of association based on consistent evidence from two or more observational studies, with no plausible confounders (+1); very strong evidence of association based on direct evidence with no major threats to validity (+2); evidence of a dose‐response gradient (+1); or all plausible confounders with reduced effect (+1) (GRADE n.d.).
Measures of treatment effect
We have made minor changes to this section. See Differences between protocol and review.
Two review authors (AB and LM) collaboratively conducted data entry.
For analysis, we included estimates of daily average sodium intake obtained by any method (e.g. dietary survey, urine sample). We converted reported estimates into daily salt intake in grams per day with standard deviation (if not already provided in that format), when possible. We based analyses on estimates from comparable measurement tools (i.e. dietary survey at both time points, or urine sample at both time points).
We calculated overall impact based on mean change in salt intake (grams/d) from pre‐intervention to post‐intervention. We examined differential impact by sex on the basis of mean change in salt intake (grams/d) from pre‐intervention to post‐intervention, for men and women separately.
We intended to examine differential impact by other axes of social inequality based on PROGRESS indicators (place of residence, race/ethnicity, occupation, gender, religion, education, social capital, socio‐economic position); however, the data did not permit this.
When multiple data points were available for an initiative, we based analysis of impact on the pre‐intervention data point closest in time to the start year of the intervention, and the post‐intervention data point farthest away in time from the start year of the initiative. As a sensitivity check, we re‐ran analysis of impact using alternative post‐intervention data points, along with alternative pre‐intervention data points, when available.
Unit of analysis issues
Unit of analysis issues were not relevant to this review because of the nature of eligible studies. However, review authors may need to consider unit of analysis issues in future iterations of the review; for example, if eligible cluster‐randomised controlled trials are identified.
Dealing with missing data
This section is the same as in the protocol (McLaren 2013).
When overall estimates (for the full population) were not provided, we calculated them (when possible) using other information (e.g. using data reported for men and women separately to compute an overall estimate). When this was not possible, we contacted study authors or country contacts in an attempt to obtain the missing information. When we could not obtain missing data, despite these efforts, we documented this on our data extraction and risk of bias form.
Assessment of heterogeneity
This section is substantively the same as in the protocol (McLaren 2013).
We conducted statistical tests for heterogeneity (I2) to determine whether a meta‐analysis was feasible. We considered clinical heterogeneity (i.e. related to the populations, interventions, comparators and outcomes framework (PICO)) and heterogeneity related to study design.
Assessment of reporting biases
This section is substantively the same as in the protocol (McLaren 2013)
We undertook extensive efforts to ensure a comprehensive search, specifically, a grey literature search that cast a wide net, consultation with experts in the field and direct contact with individuals engaged in national sodium reduction efforts worldwide (see Searching other resources). We anticipated that the comprehensiveness of our search process and the significant proportion of grey literature resources obtained (reports, websites, personal contacts) would reduce the likelihood of publication bias.
Nonetheless, we also created a funnel plot to examine possible reporting biases in keeping with the recommendation to do so if more than 10 interventions are included (Higgins 2011).
Data synthesis
We have made changes to these three sections: Data synthesis, Subgroup analysis and investigation of heterogeneity, Sensitivity analysis (see Differences between protocol and review).
We based the primary analysis on all included studies (for which required information was not missing), regardless of risk of bias.
We used Review Manager 5.3 to calculate mean change in salt intake (grams/d) from pre‐intervention to post‐intervention. See Data and analyses.
Subgroup analysis and investigation of heterogeneity
We intended to examine differences in overall impact by type(s) of intervention activities (see Description of the intervention). However, we could not do this because most initiatives involved more than one intervention activity. However, the pool of initiatives did vary in the extent to which they included activities of a more structural nature, as we have conceptualised it (e.g. food product reformulation, food procurement policy in specific settings), with some countries providing some structural activities and others providing none. Therefore, we examined separately the subset of interventions that included some activities of a structural nature.
We intended to examine differences by economic status of country (e.g. high vs low‐middle) and by duration of the initiative (see Differences between protocol and review). However, the small number of studies identified overall and limited variation amongst them on these dimensions (see Included studies below) precluded these subgroup analyses.
Sensitivity analysis
For initiatives with multiple available data points, we based the analysis of impact on the pre‐intervention data point closest in time to the start year of the intervention, and the post‐intervention data point farthest away in time from the start year of the initiative. As a sensitivity check, we re‐ran analyses of impact using alternative post‐intervention data points, along with alternative pre‐intervention data points, when available.
The data did not permit sensitivity analysis based on other dimensions (e.g. risk of bias).
Presentation of main results
We presented results and overall quality of the evidence (GRADE n.d.) in summary of findings Table for the main comparison.
Research ethics board review
We secured research ethics board approval for this study from the Conjoint Health Research Ethics Board at the University of Calgary (Ethics ID E‐24264). Regarding country questionnaires, we informed programme leaders of the purpose of the study through an introductory email sent with the questionnaire; consent for use of the information as part of the study was conveyed through return of the questionnaire. The University of Sydney Human Research Ethics Committee granted approval for the questionnaire portion of this work (#14923).
Results
Description of studies
Results of the search
We identified 15,706 unique records (19,768 total records) through the database search (conducted in December 2013 and updated in January 2015), of which 14,995 were eliminated as irrelevant upon screening of titles/abstracts (leaving 711 retained). The grey literature search yielded 170 documents.
We assessed 881 full‐text documents/sources (711 from the published literature + 170 from the grey literature) for eligibility. We eliminated 828 of these. Although we eliminated many documents for multiple reasons, the approximate breakdown by main reason was as follows: not empirical (e.g. commentaries, letters), 48%; did not fit our definition of a jurisdiction (e.g. non‐governmental regions such as communities), 23%; simulation studies (e.g. predicted impact of interventions), 6%; assessed salt content of foods rather than individual salt intake, 5%; otherwise not relevant, 18%.
At that point, we considered the 75 countries that had been identified in our companion review (Trieu 2015) for possible inclusion in this review. We identified that 45 of the 75 countries had two or more data points, and thus were potentially eligible for inclusion in our review. We excluded 18 of these 45 for one or more of three reasons: (1) The country lacked a pre‐intervention data point (i.e. lacked a data point that preceded the start year of the initiative); (2) existing data points were based on non‐comparable jurisdictions (e.g. one national data point and one provincial or regional data point); (3) the start date of the initiative remained unclear, despite efforts to clarify the date. We classified another 12 countries as "ongoing" because, at the time of writing, they lacked a post‐intervention data point (see Characteristics of ongoing studies). For countries designated as excluded or ongoing owing to missing or non‐comparable data points, we made efforts to locate useable data points (e.g. contacted country contacts, searched national health ministry or statistics agency websites, in addition to performing our comprehensive literature search) before finalizing the designation as excluded or ongoing.
We ultimately included 15 initiatives (countries). All initiatives were national in scope. Key documentation for those 15 countries consisted of 25 published articles, 15 grey literature documents and 13 country questionnaires and associated correspondence.
It is important to note that in many instances, we did not evaluate the impact of an initiative by relying on a single research article, but rather we pieced information together from several discrete materials, including published articles, grey literature documents and country questionnaires and associated correspondence. Accordingly, we deemed it necessary to treat the country or initiative as the unit of analysis. For each initiative, we identified one main publication to be used in in‐text citations throughout the Results and Discussion sections of the review. The main publication is the one that, in our view, provides the most information about that initiative. However, the main publication usually is not comprehensive, and for a more thorough understanding of each initiative, we recommend that all corresponding documents should be consulted.
We have presented results of the search in Figure 3.
Flow diagram.
Included studies
Description of included initiatives
We have described the 15 included countries (15 national initiatives) in the Characteristics of included studies tables. The initiative in the United Kingdom counts as one initiative, but because it has been evaluated through data for England only (Millett 2012), for Scotland only (Scottish Centre for Social Research 2011) and for Great Britain as a whole (Wyness 2012), we have prepared three tables to describe it. The total sample size for the 15 initiatives exceeds 260,000 participants. Because many of the initiatives include multiple samples and subsamples (e.g. full dietary survey vs urine sample; full sample vs working‐age subsample), computing the precise total sample size for the 15 initiatives is not a straightforward process. Our analyses (see later) are based on a subset of initiatives, and on a subset of data points within those initiatives. We have included in the Characteristics of included studies table all data points that we identified, including those that were not eligible for analysis.
For Finland (Laatikainen 2006), our comparisons are based on estimates for the Kuopio region. The Finland national sodium reduction initiative began in the early 1970s in North Karelia, and after a five‐year pilot phase was extended to other regions (Puska 2008‐ see Finland). Because our earliest data point for that country was 1979, Kuopio was the only region that satisfied our requirement of at least one comparable data point pre‐initiative and post‐initiative.
The included studies are diverse in terms of settings and interventions, as we have described in the following sections.
Twelve of the 15 countries are classified by the World Bank as 'high‐income', and three (China Du 2014, Thailand Supornsilaphachai 2013, and Turkey Erdem 2010) as 'upper‐middle‐income'. As reported in our companion review (Trieu 2015), national sodium reduction efforts have been identified in countries representing all of the income categories of the World Bank. However, we have included in this review none of the initiatives provided in 'lower‐middle‐income' or 'low‐income' categories because we determined that they were in planning stages (i.e. no substantive activity at the time of writing), or because we found limitations such as no pre‐intervention data.
Of the six World Health Organization regions, four are represented in our review: Europe (Austria ‐ Austria country questionnaire 2014‐2015, Denmark ‐ Denmark country questionnaire 2014‐2015, Finland ‐ Laatikainen 2006, France ‐ Dubuisson 2010, Ireland ‐ Perry 2010, Netherlands ‐ Hendriksen 2013, Switzerland ‐ Chappuis 2011, Turkey ‐ Erdem 2010, United Kingdom ‐ Millett 2012Wyness 2012Scottish Centre for Social Research 2011); Western Pacific (China ‐ Du 2014, Japan ‐ Udagawa 2008, New Zealand ‐ New Zealand Ministry for Primary Industries); the Americas (Canada ‐ McLaren 2014, United States of America ‐ Pfeiffer 2014); and South‐East Asia (Thailand ‐ Supornsilaphachai 2013). Although our companion review (Trieu 2015) identified national sodium reduction efforts in all six regions, we have included no initiatives from the African or Eastern Mediterranean regions in this review because we determined that they were in planning stages (i.e. no substantive activity at the time of writing), or because we found limitations such as no pre‐intervention data.
We identified the intervention activities for each initiative and classified them as aligning with the intervention types listed under Description of the intervention. It is important to note that we made these classifications on the basis of substantive activities that were under way during the time frame (data points) considered in this review. In some cases (e.g. Canada ‐ McLaren 2014, Japan ‐ Udagawa 2008), other substantive activities have been initiated more recently (e.g. working with industry towards food reformulation) and were not captured within available data points. Updates of this review will need to consider the evolution of some initiatives over time, as additional post‐intervention data points become available.
With this in mind, we have classified initiatives as follows. Most initiatives (12/15) included multiple (more than one) intervention activities. The United Kingdom (Scottish Centre for Social Research 2011; Millett 2012; Wyness 2012) initiative included five intervention activities, and initiatives in four countries (Denmark ‐ Denmark country questionnaire 2014‐2015, Finland ‐ Laatikainen 2006, Netherlands ‐ Hendriksen 2013, Thailand ‐ Supornsilaphachai 2013) included four activities. Three countries (Canada ‐ McLaren 2014, China ‐ Du 2014, Japan ‐ Udagawa 2008) were characterised as having (during the time frame of available data) single‐component initiatives, which in all cases consisted of public information/education campaigns.
Initiatives in all 15 countries included public information/education campaigns. Other intervention types, by frequency of occurrence, consisted of food product reformulation (this included structured voluntary efforts; see Appendix 1) (Austria ‐ Austria country questionnaire 2014‐2015, Denmark ‐ Denmark country questionnaire 2014‐2015, Finland ‐ Laatikainen 2006, France ‐ Dubuisson 2010, Ireland ‐ Perry 2010, Netherlands ‐ Hendriksen 2013, New Zealand ‐ New Zealand Ministry for Primary Industries, Switzerland ‐ Chappuis 2011, Thailand ‐ Supornsilaphachai 2013, Turkey ‐ Erdem 2010, United Kingdom ‐ Scottish Centre for Social Research 2011; Millett 2012; Wyness 2012); food procurement policy in specific settings (Austria ‐ Austria country questionnaire 2014‐2015, Denmark ‐ Denmark country questionnaire 2014‐2015, Finland ‐ Laatikainen 2006, France ‐ Dubuisson 2010, Netherlands ‐ Hendriksen 2013, Switzerland ‐ Chappuis 2011, Thailand ‐ Supornsilaphachai 2013, Turkey ‐ Erdem 2010, United Kingdom ‐ Scottish Centre for Social Research 2011; Millett 2012; Wyness 2012); on‐package nutrition labelling (Denmark ‐ Denmark country questionnaire 2014‐2015, Finland ‐ Laatikainen 2006, Ireland ‐ Perry 2010, Netherlands ‐ Hendriksen 2013, New Zealand ‐ New Zealand Ministry for Primary Industries, Thailand ‐ Supornsilaphachai 2013, United Kingdom ‐ Scottish Centre for Social Research 2011; Millett 2012; Wyness 2012, United States of America ‐ Pfeiffer 2014); and restrictions on marketing to children (United Kingdom ‐ Scottish Centre for Social Research 2011; Millett 2012; Wyness 2012). No initiatives included pricing strategies (e.g. taxation, subsidisation).
Overall, 11 of the 15 initiatives could be considered as multi‐component and incorporating activities of a structural nature; the remaining four initiatives (Canada ‐ McLaren 2014, China ‐ Du 2014 , Japan ‐ Udagawa 2008, United States of America ‐ Pfeiffer 2014) were less structural and included fewer (one or two) activities.
In terms of the start year of initiatives, we must note that identifying a clear start date was not always a straightforward task because the activities involved in developing and implementing a national sodium reduction initiative are complex and evolve over time. From the best information we could gather, we determined that the start year of the initiatives in our review ranged from 1979 (Finland ‐ Laatikainen 2006) to 2011 (Austria ‐ Austria country questionnaire 2014‐2015, Turkey ‐ Erdem 2010). Most initiatives started relatively recently; only three started before the year 2000 (Canada ‐ McLaren 2014, Finland ‐ Laatikainen 2006, United States of America ‐ Pfeiffer 2014), and 12 started in the year 2000 or later (France ‐ Dubuisson 2010 and Japan ‐ Udagawa 2008 (2001), Ireland ‐ Perry 2010 and United Kingdom ‐ Scottish Centre for Social Research 2011; Millett 2012; Wyness 2012 (2003), New Zealand ‐ New Zealand Ministry for Primary Industries (2005), China ‐ Du 2014 and Thailand ‐ Supornsilaphachai 2013 (2006), Denmark ‐ Denmark country questionnaire 2014‐2015 and Switzerland ‐ Chappuis 2011 (2008), Austria ‐ Austria country questionnaire 2014‐2015 and Turkey ‐ Erdem 2010 (2011)).
We must note that in some countries, where efforts related to sodium reduction had been ongoing for several years, we applied discretion in selecting a prominent event/effort that could be conceptualised as the start of the initiative. For example, in the United States (Pfeiffer 2014), advice to reduce salt intake to less than 6 grams/d has been provided consistently since the 1980s, and the Nutrition Labeling and Education Act of 1994 specified inclusion of the daily value of < 2400 mg of sodium as part of the new nutrition label. Thus, we assigned a start year of the United States intervention as late 1980s to early 1990s (Pfeiffer 2014). Other efforts in the United States, at national, state and municipal levels, have been made since the time that we defined to represent the start of the intervention. Similarly, in Canada (McLaren 2014), we identified the beginning of that country's national sodium reduction initiative as a major revision to the Food Guide, which appeared in 1982 and included a moderation statement about salt, even though other activities have occurred since then, such as establishment in 2007 of the national Sodium Working Group (SWG 2010) and the requirement that pre‐packaged foods contain nutrition labelling (http://www.hc‐sc.gc.ca/fn‐an/label‐etiquet/nutrition/index‐eng.php). Data are not currently available to permit evaluation of the national impact of these latter initiatives in Canada (i.e. no national post‐intervention data point), but we expect that these data will become available for inclusion in a later update of this review.
The study designs represented in our review include uncontrolled pre‐post design and open cohort design. Most (n = 14/15) initiatives were evaluated via an uncontrolled pre‐post design (Austria ‐ Austria country questionnaire 2014‐2015, Canada ‐ McLaren 2014, Denmark ‐ Denmark country questionnaire 2014‐2015, Finland ‐ Laatikainen 2006, France ‐ Dubuisson 2010, Ireland ‐ Perry 2010, Japan ‐ Udagawa 2008, Netherlands ‐ Hendriksen 2013, New Zealand ‐ New Zealand Ministry for Primary Industries, Switzerland ‐ Chappuis 2011, Thailand ‐ Supornsilaphachai 2013, Turkey ‐ Erdem 2010, United Kingdom ‐ Scottish Centre for Social Research 2011; Millett 2012; Wyness 2012, United States of America ‐ Pfeiffer 2014). One initiative (China ‐ Du 2014) was evaluated via an open cohort design. Some initiatives had multiple pre‐intervention and post‐intervention data points (thereby fitting the definition of an interrupted time series design) but were classified as uncontrolled pre‐post because, once comparability of jurisdiction and measurement method was taken into account, only one useable pre‐intervention and/or post‐intervention data point remained.
Our primary, and only, outcome variable was dietary sodium consumption, assessed by any measure of salt/sodium intake.
Across all data points identified for each initiative (see Characteristics of included studies and Table 1 through Table 2), we observed a variety of measures: 24‐hour dietary recall (Austria ‐ Austria country questionnaire 2014‐2015, Canada ‐ McLaren 2014, China ‐ Du 2014, Netherlands ‐ Hendriksen 2013, New Zealand ‐ New Zealand Ministry for Primary Industries), 48‐hour dietary recall (Finland ‐ Laatikainen 2006), seven‐day food record or diet diary (Denmark ‐ Denmark country questionnaire 2014‐2015, France ‐ Dubuisson 2010, Ireland ‐ Perry 2010, United Kingdom ‐ Scottish Centre for Social Research 2011; Millett 2012; Wyness 2012), four‐day food record (Ireland ‐ Perry 2010), food frequency questionnaire (Ireland ‐ Perry 2010), unspecified "nutritional intake survey" (Japan ‐ Udagawa 2008) and "dietary survey" (Thailand ‐ Supornsilaphachai 2013), spot urine (Denmark ‐ Denmark country questionnaire 2014‐2015, Ireland ‐ Perry 2010, New Zealand ‐ New Zealand Ministry for Primary Industries, Switzerland ‐ Chappuis 2011, Turkey ‐ Erdem 2010, United Kingdom ‐ Scottish Centre for Social Research 2011; Millett 2012; Wyness 2012, United States of America ‐ Pfeiffer 2014), 24‐hour urine (Finland ‐ Laatikainen 2006, Ireland ‐ Perry 2010, Netherlands ‐ Hendriksen 2013, Switzerland ‐ Chappuis 2011, Turkey ‐ Erdem 2010, United Kingdom ‐ Scottish Centre for Social Research 2011; Millett 2012; Wyness 2012). New Zealand used a Total Diet Study method (see references under New Zealand), which involved identifying and purchasing retail foods commonly consumed by the population and preparing them as for usual consumption on the basis of which population estimates of nutrients (including sodium) were calculated.
| Data point 1* | Data point 2* | |
| Data source | 2008 Austrian Study on Nutritional Status | 2012 Austrian Study on Nutritional Status |
| Overall estimate – as originally reported | Mean salt intake: 8.3 grams/d, SD = 3.53 | Mean salt intake: 8.15 grams/d, SD = 2.99 |
| Overall estimate – revised for comparability | See above | See above |
| Measurement tool | 24‐Hour dietary recall | Two 24‐hour dietary recalls |
| Sample size | n = 2123 | n = 380† |
| Progress indicators available | Gender | Gender |
underlined text = computed by review authors
*Data points on which primary calculation of mean difference (pre‐intervention vs post‐intervention) is based (see Data and analyses ‐ "overall")
†This number is based only on the number of adults in the sample, so it is comparable with the estimate from 2008 (i.e. kids and elderly were not considered)
| Data point 1* | Data point 2 | Data point 3* | |
| Data source | 1988‐1994 NHANES | 2003‐2006 NHANES | 2010 NHANES |
| Overall estimate – as originally reported | Crude geometric mean sodium intake: 3280 milligrams/d, SD = 83.3 (95% CI 3277.64 to 3282.34) | Crude geometric mean sodium intake: 3270 milligrams/d, SD = 101 (95% CI 3267.13 to 3272.87) | Crude geometric mean sodium intake: 3400 milligrams/d, SD = 87.6 (95% CI 3396.18 to 3403.82) |
| Overall estimate – revised for comparability | Crude geometric mean salt intake: 8.34 grams/d | Crude geometric mean salt intake: 8.31 grams/d | Crude geometric mean salt intake: 8.64 grams/d |
| Measurement tool | Estimated 24‐hour urine | Estimated 24‐hour urine | Estimated 24‐hour urine |
| Sample size | n = 1249 | n = 1235 | n = 525 |
| Progress indicators available | Gender, race‐ethnicity | Gender, race‐ethnicity | Gender, race‐ethnicity |
underlined text = computed by review authors
*Data points on which primary calculation of mean difference (pre‐intervention vs post‐intervention) is based (see Data and analyses ‐ "overall")
Included studies also varied in terms of reporting of information on study funding sources. Of the 15 included studies, seven provided information on study funding sources for all data points. Of those seven studies, six reported non‐industry funding (Canada ‐ McLaren 2014, France ‐ Dubuisson 2010, Japan ‐ Udagawa 2008, Netherlands ‐ Hendriksen 2013, New Zealand ‐ New Zealand Ministry for Primary Industries, United States ‐ Pfeiffer 2014), and one reported industry and non‐industry funding (Ireland ‐ Perry 2010). The remaining eight studies did not provide information on funding sources for one or more data point(s). Amongst those eight studies, authors of two (China ‐ Du 2014, Switzerland ‐ Chappuis 2011) declared no conflicts of interest, and authors of the other six (Austria ‐ Austria country questionnaire 2014‐2015, Denmark ‐ Denmark country questionnaire 2014‐2015, Finland ‐ Laatikainen 2006, Thailand ‐ Supornsilaphachai 2013, Turkey ‐ Erdem 2010, United Kingdom ‐ Scottish Centre for Social Research 2011; Millett 2012; Wyness 2012) reported no conflicts of interest for one or more data point(s). The limited amount of information on funding and conflicts of interest reflects, at least in part, the fact that we drew much of our material from grey literature rather than from peer‐reviewed literature.
Description of information on differential impact of initiatives
We recorded axes of stratification (inequity) considered in all initiatives, using PROGRESS as a guide (place of residence (e.g. urban/rural), race/ethnicity, occupation, gender, religion, education, socio‐economic position, social capital).
Across all data points for all initiatives, a variety of axes of inequity were considered: place of residence (China ‐ Du 2014, Finland ‐ Laatikainen 2006, France ‐ Dubuisson 2010, Switzerland ‐ Chappuis 2011, Turkey ‐ Erdem 2010, United Kingdom ‐ Scottish Centre for Social Research 2011; Millett 2012; Wyness 2012), race/ethnicity (United Kingdom ‐ Scottish Centre for Social Research 2011; Millett 2012; Wyness 2012, United States of America ‐ Pfeiffer 2014), occupation (France ‐ Dubuisson 2010), gender (all except Thailand ‐ Supornsilaphachai 2013), education (Canada ‐ McLaren 2014, China ‐ Du 2014, Finland ‐ Laatikainen 2006, Turkey ‐ Erdem 2010, United Kingdom ‐ Scottish Centre for Social Research 2011; Millett 2012; Wyness 2012), social class (Ireland ‐ Perry 2010, United Kingdom ‐ Scottish Centre for Social Research 2011; Millett 2012; Wyness 2012). Income was considered in the initiatives for Canada (McLaren 2014) and China (Du 2014). All initiatives except Thailand (Supornsilaphachai 2013) considered one or more axes of stratification at one or more time points.
Unfortunately, for the most part, axes of stratification were not considered consistently across surveys (i.e. they were considered at one data point but not another). As a result, assessment of equity of impact was possible for only five initiatives: Canada ‐ McLaren 2014 (education, income); China ‐ Du 2014 (place of residence, education, income); Finland ‐ Laatikainen 2006 (place of residence, education); United Kingdom ‐ Scottish Centre for Social Research 2011; Millett 2012; Wyness 2012 (race/ethnicity, social class), United States of America ‐ Pfeiffer 2014 (race/ethnicity). Further, because these indicators have different meanings and are measured differently in different countries, they do not lend themselves to quantitative synthesis across countries. Therefore, we have considered them individually and qualitatively in the review text that follows.
The only exception is sex ("g"ender in the PROGRESS framework), for which comparable pre‐initiative and post‐initiative information was available for nine countries: Austria ‐ Austria country questionnaire 2014‐2015, Canada ‐ McLaren 2014, China ‐ Du 2014, Finland ‐ Laatikainen 2006, France ‐ Dubuisson 2010, Netherlands ‐ Hendriksen 2013, Switzerland ‐ Chappuis 2011, United Kingdom (England ‐ Millett 2012, Great Britian ‐ Wyness 2012, Scotland ‐ Scottish Centre for Social Research 2011), United States of America ‐ Pfeiffer 2014.
Excluded studies
Of the 45 initiatives that contained at least two data points on the basis of our companion review (Trieu 2015) and were assessed closely for eligibility, we excluded 18 (see Characteristics of excluded studies table). Reasons for exclusion included the following: no useable pre‐intervention data point (Argentina, Australia, Barbados, Indonesia, Italy, Portugal, Slovakia); lack of clarity regarding start year of the initiative, despite efforts to clarify (Bangladesh, Iceland, Israel, Singapore, Uruguay); and data points based on non‐comparable jurisdiction and/or measurement methods (Croatia, Malaysia, Poland, Slovenia, Sri Lanka, Vietnam). In cases for which more than one reason led to exclusion, we have listed reasons judged to be the most prominent.
Ongoing studies
We classified initiatives in 12 countries (Belgium, Brazil, Bulgaria, Chile, Costa Rica, Fiji, Hungary, Lithuania, Mongolia, Norway, Republic of South Korea, Sweden) as ongoing (see Characteristics of ongoing studies table) because, although a start date was clearly defined and one or more pre‐intervention data points were given, no post‐intervention data point was available at the time of writing.
Risk of bias in included studies
We have presented under Characteristics of included studies our judgements on each risk of bias domain for each included initiative and have provided a summary in Figure 4 (note that the United Kingdom is counted as one initiative despite three available sets of evaluations). We characterised all 15 initiatives as having high risk of bias overall because they included one or more domains characterised as presenting high risk of bias. The overall high risk of bias reflects the study designs used, which in all cases were uncontrolled. Although use of uncontrolled designs is understandable given the nature of the intervention (government initiated; evaluated as a natural experiment), we have scored those designs as presenting high risk of bias because of the difficulty involved in ruling out alternative explanations for observed effects. Accordingly, we rated all studies as having high risk of bias on the basis of the confounding domain.
Risk of bias summary: review authors' judgements about each risk of bias item for each included initiative.
As all of our included studies had a summary rating of high risk of bias overall, we sought a means of capturing variations in methodological quality amongst them. We elected to characterise each initiative in terms of the proportion of bias domains scored as 'high' or 'uncertain', out of a total of seven domains (sampling, confounding, reliability/validity of outcome measure, blinding of outcome assessment, representativeness of sample of underlying population, risk of selective outcome reporting, other bias). Using this metric, we determined that the best studies in our sample were Finland (Laatikainen 2006) and France (Dubuisson 2010), which we scored as having 'high' or 'uncertain' risk of bias in only one of seven bias domains. We assigned high quality to the United Kingdom (Scottish Centre for Social Research 2011; Millett 2012; Wyness 2012) on the basis of the proportion metric, which we scored as presenting 'high' or 'uncertain' risk of bias on one of seven domains according to evaluations in England (Millett 2012) and Scotland (Scottish Centre for Social Research 2011), and on two of seven domains according to evaluations of Great Britain as a whole (Wyness 2012) (for an overall metric of 2/7). The worst studies in our sample were Japan (Udagawa 2008), New Zealand (New Zealand Ministry for Primary Industries), Thailand (Supornsilaphachai 2013) and Turkey (Erdem 2010), all of which we scored as having 'high' or 'uncertain' risk of bias on five of seven domains (we scored no studies as having 'high' or 'uncertain' risk of bias on six or seven of seven domains). The remaining countries fell in between, scoring 'high' or 'uncertain' risk of bias on two of seven bias domains (Canada ‐ McLaren 2014, China ‐ Du 2014, Ireland ‐ Perry 2010, United States of America ‐ Pfeiffer 2014), or on three of seven (Denmark ‐ Denmark country questionnaire 2014‐2015) or four of seven bias domains (Austria ‐ Austria country questionnaire 2014‐2015, Netherlands ‐ Hendriksen 2013, Switzerland ‐ Chappuis 2011).
We next present a summary of risk of bias by domain. See Characteristics of included studies and Included studies sections for details and sources. We must acknowledge that many of our studies had important methodological strengths, including large, nationally representative samples, rigorous measurement of dietary sodium intake and use of existing survey or administrative data (not collected for the purpose of evaluating the initiative), which reduces the likelihood that outcomes are biased by knowledge of the initiative.
Sampling
We judged that seven of the 15 studies had low risk of bias on sampling, given their use of a random (simple or complex) sampling strategy (Canada ‐ McLaren 2014, China ‐ Du 2014, Denmark ‐ Denmark country questionnaire 2014‐2015, Finland ‐ Laatikainen 2006, France ‐ Dubuisson 2010, Ireland ‐ Perry 2010, United Kingdom ‐ Scottish Centre for Social Research 2011; Millett 2012; Wyness 2012). We scored five studies (Austria ‐ Austria country questionnaire 2014‐2015, Netherlands ‐ Hendriksen 2013, Switzerland ‐ Chappuis 2011, Thailand ‐ Supornsilaphachai 2013, United States of America ‐ Pfeiffer 2014) as having high risk of bias on sampling because they reported full or partial use of non‐random or convenience sampling. For the remaining three studies, information was insufficient to reveal the sampling method used for one or more data points (Japan ‐ Udagawa 2008, Turkey ‐ Erdem 2010), or, in the case of New Zealand (New Zealand Ministry for Primary Industries), this domain was not applicable because no sampling per se was performed (see New Zealand and Characteristics of included studies table).
Confounding
We scored all 15 studies as having high risk of bias on confounding because they used an uncontrolled study design.
Reliability/Validity of outcome measure
Methods of outcome measurement included dietary surveys (e.g. 24‐hour recall) and urinary estimates (spot urine samples, 24‐hour urine samples). In the light of significant expense and burden associated with 24‐hour urine estimates for large population‐based national samples, particularly in lower‐income and middle‐income countries, we accommodated spot urine and survey‐based methods in this literature.
Although 24‐hour urine is widely viewed as the most accurate method, all methods have both strengths and limitations (McLean 2014). For example, although some dietary surveys (e.g. 24‐hour recall using an automated multi‐pass method) have shown strong positive correlation with 24‐hour urine estimates, dietary surveys may mis‐estimate or under‐estimate sodium consumption as the result of mis‐reporting or under‐reporting and challenges associated with accurately quantifying sodium in recipes. Dietary surveys are unique amongst measurement methods in their ability to identify sources of sodium, which may be useful for informing interventions (e.g. identifying groups of foods for a possible reformulation policy). Spot urinary estimates offer pragmatic advantages (collected during a single encounter) but may not accurately reflect 24‐hour urinary sodium excretion. However, methods (e.g. formulae) have been developed to improve the correspondence between spot and 24‐hour samples, and these methods may suffice for producing reasonably accurate population‐level (but not individual‐level) estimates. Finally, although it is widely viewed as the most accurate method (approximately 90% of ingested sodium is excreted in the urine over a 24‐hour period, under 'normal' conditions of climate and physical activity), 24‐hour urine suffers from potential bias associated with low response rates (due to burden), under‐collection and limitations of methods available to accurately identify incomplete samples (McLean 2014). Accordingly, our risk of bias judgement for this domain was not based solely on the method used. Rather, all methods in studies reviewed could potentially score as introducing low risk of bias in this domain; to do so, convincing text was needed to show that the tool was administered in a careful and systematic manner and/or that reliability and validity of the tool were known from other cited literature or had been assessed within the included study.
On this basis, we judged nine of 15 studies as having low risk of bias for the reliability/validity of outcome measures (China ‐ Du 2014, Denmark ‐ Denmark country questionnaire 2014‐2015, Finland ‐ Laatikainen 2006, France ‐ Dubuisson 2010, Ireland ‐ Perry 2010, Netherlands ‐ Hendriksen 2013, Switzerland ‐ Chappuis 2011, United Kingdom ‐ Scottish Centre for Social Research 2011; Millett 2012; Wyness 2012, and United States of America ‐ Pfeiffer 2014). Two countries (Canada ‐ McLaren 2014, New Zealand ‐ New Zealand Ministry for Primary Industries) were scored as having high risk of bias in this domain. In Canada (McLaren 2014), although a dietary survey was used at both time points, study authors mentioned that these surveys were not directly comparable because of changes to survey content and administration over time. New Zealand (New Zealand Ministry for Primary Industries) also received a score of high in this domain because the reliability/validity of the tools used to determine salt intake at both time points was not discussed. In the remaining four studies (Austria ‐ Austria country questionnaire 2014‐2015, Japan ‐ Udagawa 2008, Thailand ‐ Supornsilaphachai 2013, Turkey ‐ Erdem 2010), information provided on the measurement tool was insufficient to permit review authors to discern reliability, validity and comparability over time.
Blinding of outcome assessment
We scored studies for which pre‐existing data were used to evaluate the intervention as having low risk of bias in this domain on the basis of reasoning that absence of a direct connection between the initiative and the data constituted a form of blinding. We also scored studies in which data were collected for the purpose of evaluating the intervention but blinding was explicitly incorporated as having low risk of bias in this domain. Twelve studies satisfied the criteria for low risk of bias (Austria ‐ Austria country questionnaire 2014‐2015, Canada ‐ McLaren 2014, China ‐ Du 2014, Denmark ‐ Denmark country questionnaire 2014‐2015, Finland ‐ Laatikainen 2006, France ‐ Dubuisson 2010, Ireland ‐ Perry 2010, Japan ‐ Udagawa 2008, New Zealand ‐ New Zealand Ministry for Primary Industries, Thailand ‐ Supornsilaphachai 2013, Turkey ‐ Erdem 2010, United States of America ‐ Pfeiffer 2014). For the Netherlands (Hendriksen 2013), information about the data was insufficient to reveal risk of bias. We scored Switzerland (Chappuis 2011) as having high risk of bias in this domain. For the United Kingdom, evaluations for England (Millett 2012) and Scotland (Scottish Centre for Social Research 2011) were scored as showing low risk of bias in this domain, but the evaluation for Great Britain (Wyness 2012) was scored high (because although 24‐hour urine was used, investigators gave no explicit indication that blinding was incorporated) for an overall score of high risk of bias for the United Kingdom (Scottish Centre for Social Research 2011; Millett 2012; Wyness 2012).
Representativeness of sample of the underlying population
We judged five initiatives (Canada ‐ McLaren 2014, Finland ‐ Laatikainen 2006, France ‐ Dubuisson 2010, United Kingdom ‐ Scottish Centre for Social Research 2011; Millett 2012; Wyness 2012, United States of America ‐ Pfeiffer 2014) as having low risk of bias in this domain on the basis of efforts, such as application of sampling weights or demonstrated similarity between sample data and census data, to show that the sample resembled the underlying population. We scored Ireland (Perry 2010), Netherlands (Hendriksen 2013) and Switzerland (Chappuis 2011) as having high risk of bias in this domain on the basis of demonstrated differences between the sample and the underlying population. The remaining seven countries provided insufficient information to permit a judgement (Austria ‐ Austria country questionnaire 2014‐2015, China ‐ Du 2014, Denmark ‐ Denmark country questionnaire 2014‐2015, Japan ‐ Udagawa 2008, Thailand ‐ Supornsilaphachai 2013, Turkey ‐ Erdem 2010) or, in the case of New Zealand (New Zealand Ministry for Primary Industries), this domain was not applicable because no sampling per se was involved.
Risk of selective outcome reporting
We scored 10 initiatives (Austria ‐ Austria country questionnaire 2014‐2015, Canada ‐ McLaren 2014, China ‐ Du 2014, Finland ‐ Laatikainen 2006, France ‐ Dubuisson 2010, Ireland ‐ Perry 2010, Netherlands ‐ Hendriksen 2013, Switzerland ‐ Chappuis 2011, United Kingdom ‐ Scottish Centre for Social Research 2011; Millett 2012; Wyness 2012, United States of America ‐ Pfeiffer 2014) as having low risk of bias in this domain because investigators reported outcome data in a useable format (i.e. means along with estimates of variance (standard deviation, standard error or 95% confidence interval)). We scored the remaining five initiatives (Denmark ‐ Denmark country questionnaire 2014‐2015, Japan ‐ Udagawa 2008, New Zealand ‐ New Zealand Ministry for Primary Industries, Thailand ‐ Supornsilaphachai 2013, Turkey ‐ Erdem 2010) as having high risk of bias in this domain because they did not provide estimates of variance for one or more data points, and we were unable to obtain that information through other sources or through contact with study authors.
Other sources of bias
We scored all 15 initiatives as having low risk of bias in this domain, largely because many potentially significant sources of bias are captured in the 'confounding' domain, in which all initiatives were scored as having high risk of bias.
Effects of interventions
See: Summary of findings for the main comparison
A total of 15 initiatives met the inclusion criteria. Summary statistics for all data points (reported values, as well as values converted to sodium in grams per day) are shown in Table 1; Table 3; Table 4; Table 5; Table 6; Table 7; Table 8; Table 9; Table 10; Table 11; Table 12; Table 13; Table 14; Table 15; Table 16; Table 17; and Table 2. Note that we have provided 17 tables for 15 initiatives because separate evaluations of the United Kingdom initiative were available for Great Britain as a whole (Wyness 2012), for England only (Millett 2012) and for Scotland only (Scottish Centre for Social Research 2011).
| Data point 1* | Data point 2* | |
| Data source | 1970‐1972 Nutrition Canada Survey | 2004 Canadian Community Health Survey |
| Overall estimate – as originally reported | Mean sodium intake: 2403.9 milligrams/d, SD = 1362.5 | Mean sodium intake: 3057 milligrams/d, SD = 606.2 |
| Overall estimate – revised for comparability | Mean salt intake: 6.11 grams/d, SD = 3.46 | Mean salt intake: 7.77 grams/d, SD = 1.54 |
| Measurement tool | 24‐Hour dietary recall | 24‐Hour dietary recall |
| Sample size | n = 4540 | n = 10,449 |
| Progress indicators available | Gender, education and income | Gender, education and income |
underlined text = computed by review authors
*Data points on which primary calculation of mean difference (pre‐intervention vs post‐intervention) is based (see Data and analyses ‐ "overall")
| Data point 1 | Data point 2 | Data point 3 | Data point 4 | Data point 5 | Data point 6* | Data point 7* | |
| Data source | 1991 China Health and Nutrition Survey | 1993 China Health and Nutrition Survey | 1997 China Health and Nutrition Survey | 2000 China Health and Nutrition Survey | 2004 China Health and Nutrition Survey | 2006 China Health and Nutrition Survey | 2009 China Health and Nutrition Survey |
| Overall estimate – as originally reported | Mean sodium intake: 6.6 grams/d, SD = 3.4 | Mean sodium intake: 6.6 grams/d, SD = 3.4 | Mean sodium intake: 6.2 grams/d, SD = 3.5 | Mean sodium intake: 6.0 grams/d, SD = 3.2 | Mean sodium intake: 5.2 grams/d, SD = 2.7 | Mean sodium intake: 5.0 grams/d, SD = 2.8 | Mean sodium intake: 4.7 grams/d, SD = 2.6 |
| Overall estimate – revised for comparability | Mean salt intake: 16.76 grams/d | Mean salt intake: 16.76 grams/d | Mean salt intake: 15.77 grams/d | Mean salt intake: 15.24 grams/d | Mean salt intake: 13.21 grams/d | Mean salt intake: 12.70 grams/d | Mean salt intake: 11.94 grams/d |
| Measurement tool | 3 consecutive 24‐hour dietary recalls | 3 consecutive 24‐hour dietary recalls | 3 consecutive 24‐hour dietary recalls | 3 consecutive 24‐hour dietary recalls | 3 consecutive 24‐hour dietary recalls | 3 consecutive 24‐hour dietary recalls | 3 consecutive 24‐hour dietary recalls |
| Sample size | n = 7337 | n = 6958 | n = 7241 | n = 7940 | n = 7250 | n = 6826 | n = 6932 |
| Progress indicators available | Gender, place of residence, education, income | Gender, place of residence, education, income | Gender, place of residence, education, income | Gender, place of residence, education, income | Gender, place of residence, education, income | Gender, place of residence, education, income | Gender, place of residence, education, income |
underlined text = computed by review authors
*Data points on which primary calculation of mean difference (pre‐intervention vs post‐intervention) is based (see Data and analyses ‐ "overall")
Note: Based on all data points, “sodium intake significantly decreased over time (P‐trend, 0.001, general linear regression models)”
| Data point 1 | Data point 2 | Data point 3 | |
| Data source | 2006 Danish Health | 2008 Danish National Survey of Dietary Habits and Physical Activity | 2010 Danish Health |
| Overall estimate – as originally reported | Median salt intake: 8.93 grams/d | Mean salt intake: 8.5 grams/d, SD = 2.9 (95% CI 8.41 to 8.59) | Median salt intake: 8.27 grams/d |
| Overall estimate – revised for comparability | Same as above | Same as above | Same as above |
| Measurement tool | Spot urine | Food record for 7 consecutive days | Spot urine |
| Sample size | n = 3294 | n = 4431 | n = 1478 |
| Progress indicators available | Gender | Gender | Gender |
underlined text = computed by review authors
Denmark was not included in quantitative synthesis because of missing or unuseable information (e.g. mean, SD, n) at 1 or more data points
| Data point 1* | Data point 2 | Data point 3* | Data point 4 | Data point 5 | Data point 6 | |
| Data source | 1979 North Karelia Kuopio area | 1982 North Karelia Kuopio area | 1987 North Karelia Kuopio area | 2002 North Karelia | 2007 5 regions in Finland | 2012 5 regions in Finland |
| Overall estimate – as originally reported | Mean salt intake (North Karelia): 11.65 grams/d, SD = 4.39 Mean salt intake (Kuopio area): 11.78 grams/d, SD = 4.72 | Mean salt intake (North Karelia): 11.94 grams/d ± 4.62 Mean salt intake (Kuopio area): 11.59 grams/d, SD = 4.50 | Mean salt intake (North Karelia): 10.36 grams/d, SD = 4.46 Mean salt intake (Kuopio area): 10.63 grams/d, SD = 4.10 | Mean salt intake (North Karelia): 8.43 grams/d, SD = 3.19 | Mean salt intake: 8.0 grams/d | Mean sodium intake: 3.01 grams/d, SD = 1.10 |
| Overall estimate – revised for comparability | Same as above | Same as above | Same as above | Same as above | Same as above | Mean salt intake: 7.65 grams/d |
| Measurement tool | 24‐Hour urine | 24‐Hour urine | 24‐Hour urine | 24‐Hour urine | 48‐Hour dietary recall | 48‐Hour dietary recall |
| Sample size | n = 536 (North Karelia) n = 670 (Kuopio area) | n = 484 (North Karelia) n = 428 (Kuopio area) | n = 409 (North Karelia) n = 400 (Kuopio area) | n = 342 (North Karelia) | n = 1576 | n = 1295 |
| Progress indicators available | Gender and education | Gender and education | Gender and education | Gender and education | Gender | Gender, place of residence and education |
underlined text = computed by review authors
*Data points on which primary calculation of mean difference (pre‐intervention vs post‐intervention) is based (see Data and analyses ‐ "overall")
| Data point 1a* | Data point 1b | Data point 2* | |
| Data source | 1998‐1999 Individual and National Food Consumption Surveys | 1998‐1999 Individual and National Food Consumption Surveys | 2006‐2007 Individual and National Food Consumption Surveys |
| Overall estimate – as originally reported | Mean sodium intake: 3145.7 milligrams/d, SD = 1016.4 | Mean sodium intake: 137.40 mmol/d, SD = 101.48 | Mean sodium intake: 2966.6 milligrams/d ± 922.4 |
| Overall estimate – revised for comparability | Mean salt intake: 8.00 grams/d | Mean salt intake: 7.97 grams/d | Mean salt intake: 7.54 grams/d |
| Measurement tool | 7‐Day open‐ended food record | 7‐Day open‐ended food record | 7‐Day open‐ended food record |
| Sample size | n = 1345 | n = 1474 | n = 1922 |
| Progress indicators available | Gender | Gender | Gender, place of residence and occupation |
underlined text = computed by review authors
*Data points on which primary calculation of mean difference (pre‐intervention vs post‐intervention) is based (see Data and analyses ‐ "overall")
| Data point 1 | Data point 2 | Data point 3* | Data point 4* | Data point 4a | Data point 4b | Data point 5 | |
| Data source | 1997‐1999 North‐South Ireland Food Consumption Survey | 1998 Survey of Lifestyle Attitudes and Nutrition in Ireland | 2002 Survey of Lifestyle Attitudes and Nutrition in Ireland | 2007 Survey of Lifestyle Attitudes and Nutrition in Ireland | 2007 Survey of Lifestyle Attitudes and Nutrition in Ireland | 2007 Survey of Lifestyle Attitudes and Nutrition in Ireland | 2008‐2010 National Adult Nutrition Survey |
| Overall estimate – as originally reported | Mean salt intake: 8.3 grams/d | Mean salt intake: 8.3 grams/d, SD = 3.8 | Mean salt intake: 8.2 grams/d, SD = 5.9 | Mean salt intake: 7.85 grams/d, SD = 3.70 | Mean salt intake: 8.9 grams/d | Mean salt intake: 9.3 grams/d | Mean salt intake: 7.4 grams/d |
| Overall estimate – revised for comparability | Same as above | Same as above | Same as above | Same as above | Same as above | Same as above | Same as above |
| Measurement tool | 7‐Day diet diaries | Food frequency questionnaire | Food frequency questionnaire | Food frequency questionnaire | Spot urine | 24‐Hour urine | 4‐Day semi weighted food record |
| Sample size | n = 1379 | n = 6539 | n = 5992 | n = 9172 | n = 1207 | n = 599 | n = 1500 |
| Progress indicators available | Gender | Gender | Gender | Gender and social class | Gender | Gender | Gender |
underlined text = computed by review authors
*Data points on which primary calculation of mean difference (pre‐intervention vs post‐intervention) is based (see Data and analyses ‐ "overall")
| Data point 1 | Data point 2 | Data point 3 | Data point 4 | Data point 5 | Data point 6 | Data point 7 | Data point 8 | Data point 9 | Data point 10 | |
| Data source | 1997 National Health and Nutrition Survey | 2003 National Health and Nutrition Survey | 2004 National Health and Nutrition Survey | 2005 National Health and Nutrition Survey | 2006 National Health and Nutrition Survey | 2007 National Health and Nutrition Survey | 2008 National Health and Nutrition Survey | 2009 National Health and Nutrition Survey | 2010 National Health and Nutrition Survey | 2012 National Health and Nutrition Survey |
| Overall estimate – as originally reported | Mean salt intake: 13.5 grams/d | Mean salt intake: 11.7 grams/d | Mean salt intake: 11.2 grams/d | Mean salt intake: 11.5 grams/d | Mean salt intake: 11.2 grams/d | Mean salt intake: 11.1 grams/d | Mean salt intake: 10.9 grams/d | Mean salt intake: 10.7 grams/d | Mean salt intake: 10.6 grams/d | Mean salt intake: 10.4 grams/d |
| Overall estimate – revised for comparability | Same as above | Same as above | Same as above | Same as above | Same as above | Same as above | Same as above | Same as above | Same as above | Same as above |
| Measurement tool | Nutritional intake survey | Nutritional intake survey | Nutritional intake survey | Nutritional intake survey | Nutritional intake survey | Nutritional intake survey | Nutritional intake survey | Nutritional intake survey | Nutritional intake survey | Nutritional intake survey |
| Sample size | n = 13,289 | Unknown | n = 8762 | Unknown | Unknown | n = 8885 | Unknown | Unknown | n = 8815 | n = 8247 |
| Progress indicators available | ‐‐ | Gender | Gender | Gender | Gender | Gender | Gender | Gender | Gender | ‐‐ |
underlined text = computed by review authors
Japan was not included in quantitative synthesis because of missing or unuseable information (e.g. mean, SD, n) at 1 or more data points
| Data point 1* | Data point 2 | Data point 3* | |
| Data source | 2006 Cross‐sectional study of adults in Doetinchem | 2007‐2010 National Food Consumption Survey | 2010 Cross‐sectional study of adults in Doetinchem |
| Overall estimate – as originally reported | Mean sodium excretion: 148 mmol/d, SD = 55 Median salt intake: 8.7 grams/d (IQR 6.7 to 11.0) | Mean salt intake: 8.45 grams/d | Mean sodium excretion: 148 mmol/d, SD = 58 Median salt intake: 8.5 grams/d (IQR 6.6 to 10.9) |
| Overall estimate – revised for comparability | Mean salt intake: 8.58 grams/d | Same as above | Mean salt intake: 8.58 grams/d |
| Measurement tool | 24‐Hour urine | 2 non‐consecutive 24‐hour recalls | 24‐Hour urine |
| Sample size | n = 317 | n = 2160 | n = 342 |
| Progress indicators available | Gender | Gender | Gender |
underlined text = computed by review authors
*Data points on which primary calculation of mean difference (pre‐intervention vs post‐intervention) is based (see Data and analyses ‐ "overall")
| Data point 1 | Data point 2a | Data point 2b | Data point 3 | |
| Data source | 2003‐2004 New Zealand Total Diet Study | 2008‐2009 New Zealand Adult Nutrition Survey | 2008‐2009 New Zealand Adult Nutrition Survey | 2009 New Zealand Total Diet Study |
| Overall estimate – as originally reported | N/A | Mean salt intake: 6.37 grams/d | Mean salt intake: 9 grams/d | N/A |
| Overall estimate – revised for comparability | Same as above | Same as above | ||
| Measurement tool | N/A | Multiple pass 24‐hour dietary recall | Spot urine | N/A |
| Sample size | N/A | n = 4721 | Unknown | N/A |
| Progress indicators available | Gender | ‐‐ | ‐‐ | Gender |
underlined text = computed by review authors
New Zealand was not included in quantitative synthesis because of missing or unuseable information (e.g. mean, SD, n) at 1 or more data points
| Data point 1a* | Data point 1b | Data point 2* | |
| Data source | 1984 Representative sample of the Swiss population | 1984 Representative sample of the Swiss population | 2011 Swiss survey on salt intake |
| Overall estimate – as originally reported | Mean salt intake: 8.4 grams/d, SD = 3.6 | Mean salt intake: 10.3 grams/d, SD = 6.2 | Mean salt intake: 9.2 grams/d, SD = 3.8 |
| Overall estimate – revised for comparability | Same as above | Same as above | Same as above |
| Measurement tool | 24‐Hour urine | Spot urine | 24‐Hour urine |
| Sample size | n = 147 | n = 966 | n = 1448 |
| Progress indicators available | None | None | Gender and place of residence |
underlined text = computed by review authors
*Data points on which primary calculation of mean difference (pre‐intervention vs post‐intervention) is based (see Data and analyses ‐ "overall")
| Data point 1 | Data point 2 | Data point 3 | Data point 4 | Data point 5 | |
| Data source | 1960 National Nutrition Survey | 1975 National Nutrition Survey | 2003 National Nutrition Survey | 2008 | 2008‐2009 Fourth Thai National Health Examination Survey |
| Overall estimate – as originally reported | Mean sodium intake: 2.4 grams/d | Mean sodium intake: 2.4 grams/d | Mean sodium intake: 4.0 grams/d | Mean sodium intake: 4.4 grams/d | Mean sodium intake: 3.3 grams/d |
| Overall estimate – revised for comparability | Mean salt intake: 6.10 grams/d | Mean salt intake: 6.10 grams/d | Mean salt intake: 10.16 grams/d | Mean salt intake: 11.18 grams/d | Mean salt intake: 8.38 grams/d |
| Measurement tool | Dietary survey | Dietary survey | Dietary survey (food list recall and food frequency checklist of foods/condiments) | Dietary survey (household survey) | Dietary survey (food list recall and food frequency checklist of foods/condiments) |
| Sample size | Unknown | Unknown | Unknown | Unknown | Unknown |
| Progress indicators available | ‐‐ | ‐‐ | ‐‐ | ‐‐ | ‐‐ |
Note: A 2007 data point was removed because we did not have enough information to suggest that it was national in scope (i.e. comparable with other data points)
underlined text = computed by review authors
Thailand was not included in quantitative synthesis because of missing or unuseable information (e.g. mean, SD, n) at 1 or more data points
| Data point 1a | Data point 1b | Data point 2 | |
| Data source | 2008 SALTURK 1 | 2008 SALTURK 1 | 2012 SALTURK 2 |
| Overall estimate – as originally reported | Mean salt intake: 16.6 grams/d, SD = 7.3 | Mean salt intake: 18.01 grams/d | Mean salt intake: 15.0 grams/d |
| Overall estimate – revised for comparability | Same as above | Same as above | Same as above |
| Measurement tool | 24‐Hour urine | Spot urine | 24‐Hour urine |
| Sample size | n = 816 | n = 1970 (total number of eligible participants) | n =925; “657 person (according to urine creatine levels)” |
| Progress indicators available | Gender, place of residence and education level | ‐‐ | ‐‐ |
Note: We believe that the difference between estimates 1a and 1b is that 1a is based on 24‐hour urine and 1b is based on spot urine. The overall conclusion for Turkey does not differ when 1a versus 1b is used
underlined text = computed by review authors
Turkey was not included in quantitative synthesis because of missing or unuseable information (e.g. mean, SD, n) at 1 or more data points
| Data point 1a | Data point 1b* | Data point 2a | Data point 2b | Data point 3a | Data point 3b | Data point 4 | Data point 5a | Data point 5b | Data point 6a | Data point 6b* | Data point 7 | |
| Data source | 2003 Health Survey for England | 2003 Health Survey for England | 2004 Health Survey for England | 2004 Health Survey for England | 2005 Health Survey for England | 2005 Health Survey for England | 2005‐2006 National Diet and Nutrition Survey | 2006 Health Survey for England | 2006 Health Survey for England | 2007 Health Survey for England | 2007 Health Survey for England | 2011 National Diet and Nutrition Survey |
| Overall estimate – as originally reported | Geometric mean salt intake: 5.29 grams/d | Urinary sodium excretion: 109.03 mmol of sodium/d, SD = 20.41 (95% CI 108.05 to 110.01) | Geometric mean salt intake: 5.99 grams/d | Urinary sodium excretion: 111.28 mmol of sodium/d, SD = 19.82 (95% CI 110.55 to 112.01) | Geometric mean salt intake: 4.80 grams/d | Urinary sodium excretion: 99.94 mmol of sodium/d, SD = 22.17 (95% CI 99.30 to 100.58) | Mean salt intake: 9.0 grams/d, SD = 3.7 (95% CI 8.66 to 9.34) | Geometric mean salt intake: 4.73 grams/d | Urinary sodium excretion: 97.49 mmol of sodium/d, SD = 20.00 (95% CI 97.07 to 97.91) | Geometric mean salt intake: 4.55 grams/d | Urinary sodium excretion: 94.16 mmol of sodium/d, SD =19.39 (95% CI 93.58 to 94.74) | Mean salt intake: 8.1 grams/d, SD = 5.79 (95% CI 7.85 to 8.35) |
| Overall estimate – revised for comparability | Same as above | Mean salt intake:6.32 grams/d | Same as above | Mean salt intake: 6.45 grams/d | Same as above | Mean salt intake: 5.80 grams/d | Same as above | Same as above | Mean salt intake: 5.65 grams/d | Same as above | Mean salt intake: 5.46 grams/d | Same as above |
| Measurement tool | Spot urine | Spot urine | Spot urine | Spot urine | Spot urine | Spot urine | 24‐Hour urine | Spot urine | Spot urine | Spot urine | Spot urine | 24‐Hour urine |
| Sample Size | n = 1668 | n = 1668 | n = 2840 | n = 2840 | n = 4643 | n = 4643 | n = 448 | n = 8844 | n = 8844 | n = 4269 | n = 4269 | n = 547 |
| Progress indicators available | Gender, ethnicity and social class | Gender and place of residence | Gender, ethnicity and social class | Gender and place of residence | Gender, ethnicity and social class | Gender and place of residence | Gender | Gender, ethnicity and social class | Gender and place of residence | Gender, ethnicity and social class | Gender and place of residence | Gender |
underlined text = computed by review authors
*Data points on which primary calculation of mean difference (pre‐intervention vs post‐intervention) is based (see Data and analyses ‐ "overall")
| Data point 1a | Data point 1b* | Data point 1c | Data point 1d | Data point 2* | |
| Data source | 2000‐2001 National Diet and Nutrition Survey | 2000‐2001 National Diet and Nutrition Survey | 2000‐2001 National Diet and Nutrition Survey | 2000‐2001 National Diet and Nutrition Survey | 2008 Survey by National Centre for Social Research |
| Overall estimate – as originally reported | Mean sodium intake: 2794.39 milligrams/d, SD = 861.44 | Mean salt intake: 9.53 grams/d, SD = 4.48 | Median dietary sodium intake: 2611 milligrams/d (IQR = 1243) | Median 24‐hour sodium excretion: 140 mmol/d (IQR = 99.4) | Mean salt intake: 8.64 grams/d, SD = 4.39 |
| Overall estimate – revised for comparability | Mean salt intake: 7.01 grams/d (95% CI 7.00 to 7.21) | Same as above | Estimated salt intake: 6.5 grams/d | Estimated salt intake: 8.2 grams/d | Same as above |
| Measurement tool | 7‐Day consecutive weighted dietary record | 24‐Hour urine | 7‐Day consecutive weighted dietary record | 24‐Hour urine | 24‐Hour urine |
| Sample size | n = 1724 | n = 1147 | n = 2150 | n = 692 | |
| Progress indicators available | Gender | Gender | Gender, place of residence, education and social class | Gender, place of residence, education and social class | Gender |
underlined text = computed by review authors
*Data points on which primary calculation of mean difference (pre‐intervention vs post‐intervention) is based (see Data and analyses ‐ "overall")
| Data point 1* | Data point 2 | Data point 3 | Data point 4* | Data point 5 | |
| Data source | 2003 Scottish Health Survey | 2006 | 2008 Scottish Health Survey | 2009 Scottish Health Survey | 2009 |
| Overall estimate – as originally reported | Mean salt intake: 6.8 grams/d | Mean salt intake: 9.1 grams/d, SD = 4.1 | Mean salt intake: 6.4 grams/d | Mean salt intake: 6.8 grams/d | Mean salt intake: 8.8 grams/d, SD = 3.7 |
| Overall estimate – revised for comparability | Same as above | Same as above | Same as above | Same as above | Same as above |
| Measurement tool | Spot urine | 24‐Hour urine | Spot urine | Spot urine | 24‐Hour urine |
| Sample size | n = 1148 | n = 442 | n = 1041 | n = 1045 | n = 702 |
| Progress indicators available | Gender | Gender | Gender | Gender | Gender |
underlined text = computed by review authors
*Data points on which primary calculation of mean difference (pre‐intervention vs post‐intervention) is based (see Data and analyses ‐ "overall")
Of these 15 initiatives, 10 provided sufficient data for quantitative analysis of overall impact, and nine provided sufficient data for quantitative analysis of differential impact by sex. We omitted five initiatives from the quantitative synthesis of overall impact (Denmark ‐ Denmark country questionnaire 2014‐2015, Japan ‐ Udagawa 2008, New Zealand ‐ New Zealand Ministry for Primary Industries, Thailand ‐ Supornsilaphachai 2013, Turkey ‐ Erdem 2010) because variance estimates to accompany means were missing, and this could not be remedied by using other reported data or by contacting study authors.
We have used an asterisk to indicate data points that were used for analysis, for each country, Table 1; Table 3; Table 4; Table 5; Table 6; Table 7; Table 8; Table 9; Table 10; Table 11; Table 12; Table 13; Table 14; Table 15; Table 16; Table 17; and Table 2, and we have shown these in Analysis 1.1. Countries were approximately equally split in terms of those evaluated on the basis of a dietary survey (Austria ‐ Austria country questionnaire 2014‐2015, Canada ‐ McLaren 2014, China ‐ Du 2014, France ‐ Dubuisson 2010, Ireland ‐ Perry 2010, Japan ‐ Udagawa 2008, Thailand ‐ Supornsilaphachai 2013) and those evaluated on the basis of urine samples (Denmark ‐ Denmark country questionnaire 2014‐2015, Finland ‐ Laatikainen 2006, Netherlands ‐ Hendriksen 2013, Switzerland ‐ Chappuis 2011, Turkey ‐ Erdem 2010, United Kingdom ‐ Scottish Centre for Social Research 2011; Millett 2012; Wyness 2012, United States of America ‐ Pfeiffer 2014). The New Zealand Total Diet Survey approach does not fit into either category (New Zealand Ministry for Primary Industries).
We identified no adverse effects amongst these initiatives.
Comparison 1. Mean change in salt intake (grams/d) from pre‐intervention to post‐intervention ‐ OVERALL
Amongst the 10 countries (64,798 participants) included in the quantitative analysis (see Analysis 1.1), five showed a statistically significant mean decrease in salt intake from pre‐intervention to post‐intervention, ranging from Finland ‐ Laatikainen 2006 (mean decrease of 1.15, 95% confidence interval (CI) ‐1.69 to ‐0.61 grams/d) to Ireland ‐ Perry 2010 (mean decrease of 0.35, 95% CI ‐0.52 to ‐0.18 grams/d). Two initiatives (Canada ‐ McLaren 2014, Switzerland ‐ Chappuis 2011) showed a statistically significant mean increase in salt intake from pre‐intervention to post‐intervention (Canada: 1.66, 95% CI 1.56 to 1.76 grams/d; Switzerland: 0.80, 95% CI 0.19 to 1.41 grams/d).
The I2 measure of heterogeneity (between‐study variation), which ranges from 0 to 100, with higher values indicating greater heterogeneity, shows that heterogeneity for this group of 10 studies was very high (> 90%). Therefore, we did not focus on the pooled result.
Comparison 2. Mean change in salt intake (grams/d) from pre‐intervention to post‐intervention ‐ SUBSET OF MULTI‐COMPONENT INITIATIVES THAT INCLUDE STRUCTURAL ACTIVITIES
When we focused on the subset (n = 7 countries) of multi‐component initiatives that incorporated activities of a structural nature (e.g. food product reformulation, food procurement policy in specific settings; see Analysis 2.1, Figure 2 and Appendix 1), we found that most (four of seven) showed a statistically significant mean decrease in salt intake (grams/d) from pre‐intervention to post‐intervention. The significant decrease ranged from ‐1.15 (95% CI ‐1.69 to 0‐.61) grams/d in Finland (Laatikainen 2006) to ‐0.35 (95% CI ‐0.52 to ‐0.18) grams/d in Ireland (Perry 2010). One initiative (Switzerland ‐ Chappuis 2011) showed a statistically significant mean increase in salt intake (grams/d) from pre‐intervention to post‐intervention at 0.80 (95% CI 0.19 to 1.41) grams/d. Austria (Austria country questionnaire 2014‐2015) and Netherlands (Hendriksen 2013) did not show a statistically significant change in salt intake (grams/d) from pre‐intervention to post‐intervention.
The I2 measure revealed that heterogeneity for this group of seven studies was very high (> 90%). Therefore, we did not focus on the pooled result.
Comparisons 3 and 4. Mean change in salt intake (grams/d) from pre‐intervention to post‐intervention ‐ WOMEN ONLY (Analysis 3.1) and MEN ONLY (Analysis 4.1)
Nine countries (Austria ‐ Austria country questionnaire 2014‐2015, Canada ‐ McLaren 2014, China ‐ Du 2014, Finland ‐ Laatikainen 2006, France ‐ Dubuisson 2010, Netherlands ‐ Hendriksen 2013, Switzerland ‐ Chappuis 2011, United Kingdom ‐ Scottish Centre for Social Research 2011; Millett 2012; Wyness 2012, United States of America ‐ Pfeiffer 2014) provided data that permitted quantitative analysis of differential impact by sex; that is, whether mean change in salt intake (grams/d) from pre‐intervention to post‐intervention differed between men and women (see Analysis 3.1 and Analysis 4.1). Gender‐stratified estimates are shown in Table 18.
| Country | Pre‐intervention | Post‐intervention |
| AUSTRIA | ||
| Males | Data point 1 (2008) Mean salt intake = 9.4 grams/d SD = 3.56; n = 778 | Data point 2 (2012): Mean salt intake: 8.7 grams/d |
| Females | Data point 1 (2008) Mean salt intake = 7.6 grams/d SD = 2.81; n = 1345 | Data point 2 (2012): Mean salt intake: 7.6 grams/d SD = 3.11; n = 232 |
| CANADA | ||
| Males | Data point 1 (1970‐1972): Mean salt intake = 7.32 grams/d SD = 3.8; n = 1974 | Data point 2 (2004): Mean salt intake: 8.19 grams/d SD = 1.65; n = 4837 |
| Females | Data point 1 (1970‐1972): Mean salt intake = 5.00 grams/d SD = 2.71; n = 2566 | Data point 2 (2004): Mean salt intake: 7.41 grams/d |
| CHINA | ||
| Males | Data point 6 (2006): Mean salt intake = 13.21 grams/d SD = 7.37; n = 3242 | Data point 7 (2009): Mean salt intake = 12.45 grams/d |
| Females | Data point 6 (2006): Mean salt intake = 12.19 grams/d SD = 6.86; n = 3584 | Data point 7 (2009): Mean salt intake = 11.43 grams/d |
| FINLAND – Kuopio only | ||
| Males | Data point 1 (Kuopio only 1979): Mean salt intake = 13.1 grams/d SD = 5.20; n = 343 | Data point 3 (Kuopio only 1987): Mean salt intake = 12.0 grams/d SD = 4.45; n=180 |
| Females | Data point 1 (Kuopio only 1979): Mean salt intake = 10.4 grams/d SD = 4.15; n = 327 | Data point 3 (Kuopio only 1987): Mean salt intake = 9.5 grams/d SD = 3.78; n=220 |
| FRANCE | ||
| Males | Data point 1 (1998‐1999): Mean salt intake = 9.26 grams/d SD = 3.01; n = 613 | Data point 2 (2006‐2007): Mean salt intake = 8.69 grams/d SD = 2.83; n = 840 |
| Females | Data point 1 (1998‐1999): Mean salt intake = 6.93 grams/d SD = 2.16; n = 732 | Data point 2 (2006‐2007): Mean salt intake = 6.65 grams/d SD = 1.88; n = 1082 |
| NETHERLANDS | ||
| Males | Data point 1 (2006): Mean salt intake = 9.51 grams/d SD = 3.19; n = 137 | Data point 3 (2010): Mean salt intake = 10.09 grams/d SD = 3.65; n = 154 |
| Females | Data point 1 (2006): Mean salt intake = 7.89 grams/d SD = 2.96; n = 180 | Data point 3 (2010): Mean salt intake = 7.42 grams/d SD = 2.49; n = 188 |
| SWITZERLAND | ||
| Males | Data point 1a (1984): Salt intake = 10.4 grams/d | Data point 2 (2011): Salt intake = 10.6 grams/d SD = 4.2; n = 706 |
| Females | Data point 1a (1984): Salt intake = 7.3 grams/d | Data point 2 (2011): Salt intake = 7.8 grams/d |
| UNITED KINGDOM (Great Britain) | ||
| Males | Data point 1b (2000‐2001) Mean salt intake: 11.0 grams/d SD = 5.02; n = 833 | Data point 2 (2008): Mean salt intake: 9.68 grams/d SD = 4.10; n = 294 |
| Females | Data point 1b (2000‐2001) Mean salt intake: 8.1 grams/d SD = 3.88; n = 891 | Data point 2 (2008): Mean salt intake: 7.66 grams/d SD = 4.77; n = 398 |
| UK (England) – geometric mean, salt intake | ||
| Males | Data point 1a (2003): Geometric mean salt intake = 6.10 grams/d SD = 3.46; n = 735 | Data point 6a (2007): Geometric mean salt intake = 5.16 grams/d SD = 3.36; n = 1926 |
| Females | Data point 1a (2003): Geometric mean salt intake = 4.73 grams/d SD = 3.12; n = 933 | Data point 6a (2007): Geometric mean salt intake = 4.12 grams/d SD = 2.47; n = 2343 |
| UK (Scotland) | ||
| Males | Data point 1 (2003): Mean salt intake = 7.6 grams/d SD = 5.17; n = 508 | Data point 4 (2009): Mean salt intake = 7.3 grams/d SD = 3.78; n = 447 |
| Females | Data point 1 (2003): Mean salt intake = 6.1 grams/d SD = 3.87; n = 640 | Data point 4 (2009): Mean salt intake = 5.7 grams/d SD = 4.37; n = 598 |
| UNITED STATES | ||
| Males | Data point 1 (1988‐1994): Crude geometric mean salt intake = 10.04 grams/d SD = 7.81; n = 645 | Data point 3 (2010): Crude geometric mean salt intake = 10.02 grams/d SD = 6.25; n = 258 |
| Females | Data point 1 (1988‐1994): Crude geometric mean salt intake = 6.79 grams/d SD = 6.18; n = 604 | Data point 3 (2010): Crude geometric mean salt intake = 7.55 grams/d SD = 3.84; n = 267 |
For women (see Analysis 3.1), findings were split, with three initiatives showing a statistically significant mean decrease in salt intake (grams/d) from pre‐intervention to post‐intervention (China ‐ Du 2014: ‐0.76, 95% CI ‐1.07 to ‐0.45, Finland ‐ Laatikainen 2006: ‐0.90, 95% CI ‐1.57 to ‐0.23, France ‐ Dubuisson 2010: ‐0.28, 95% CI ‐0.47 to ‐0.09); two initiatives showing a statistically significant mean increase in salt intake (grams/d) from pre‐intervention to post‐cessation (Canada ‐ McLaren 2014: 2.41, 95% CI 2.3 to 2.52, United States ‐ Pfeiffer 2014: 0.76, 95% CI 0.09 to 1.43); and the remaining four countries (Austria ‐ Austria country questionnaire 2014‐2015, Netherlands ‐ Hendriksen 2013, Switzerland ‐ Chappuis 2011, United Kingdom ‐ Scottish Centre for Social Research 2011; Millett 2012; Wyness 2012) showing a non‐statistically significant mean change in salt intake from pre‐cessation to post‐cessation. Please note that, for women, we considered the United Kingdom as non‐statistically significant overall, because data from Great Britain as a whole (Wyness 2012) and from Scotland only (Scottish Centre for Social Research 2011) showed a non‐significant mean change in salt intake from pre‐cessation to post‐cessation. However, data from England only (Millett 2012) showed a statistically significant mean decrease in salt intake from pre‐cessation to post‐cessation (‐0.61, 95% CI ‐0.83 to ‐0.39).
For men (see Analysis 4.1), most initiatives (five of nine) showed a statistically significant mean decrease in salt intake (grams/d) from pre‐intervention to post‐intervention, ranging from United Kingdom (based on data from Great Britain as a whole ‐ Wyness 2012: ‐1.32, 95% CI ‐1.9 to ‐0.74, to France ‐ Dubuisson 2010: ‐0.57, 95% CI ‐0.88 to ‐0.26). One initiative showed a statistically significant mean increase in salt intake (grams/d) from pre‐intervention to post‐cessation (Canada: 0.87, 95% CI 0.7 to 1.04), and the remaining three countries (Netherlands ‐ Hendriksen 2013, Switzerland ‐ Chappuis 2011, United States of America ‐ Pfeiffer 2014) showed a non‐statistically significant mean change in salt intake from pre‐cessation to post‐cessation. Please note that for men, we considered United Kingdom as having a statistically significant decrease because data from Great Britain as a whole (Wyness 2012) and from England only (Millett 2012) showed a statistically significant mean decrease in salt intake from pre‐cessation to post‐cessation. However, data from Scotland only (Scottish Centre for Social Research 2011) showed a non‐statistically significant change.
The I2 measure showed that heterogeneity for this group of nine studies was very high (> 90%) for both men and women. Therefore, we did not focus on the pooled results.
Although Denmark (Denmark country questionnaire 2014‐2015) did not provide data that could be included in our analysis of differential impact by sex, information on differential impact by sex was reported within our documentation for Denmark (Denmark country questionnaire 2014‐2015); specifically, it was reported that with adjustments for body mass index (BMI) and age, a statistically significant (P value < .0001) mean decrease in salt intake from pre‐intervention to post‐intervention was found for both men (beta = ‐0.91) and women (beta = ‐0.52).
Mean change in salt intake (grams/d) from pre‐intervention to post‐intervention ‐ differential by other axes of inequity
We recorded axes of stratification (inequity) considered in all initiatives, using PROGRESS as a guide (place of residence (e.g. urban/rural), race/ethnicity, occupation, gender, religion, education, socio‐economic position, social capital). See the Characteristics of included studies table and Table 1; Table 3; Table 4; Table 5; Table 6; Table 7; Table 8; Table 9; Table 10; Table 11; Table 12; Table 13; Table 14; Table 15; Table 16; Table 17; and Table 2.
Information was insufficient to permit a quantitative analysis of differential impact by other axes of stratification (e.g. education, place of residence), in other words, to assess whether mean change in salt intake (grams/d) from pre‐intervention to post‐intervention differed across other axes of inequity. However, a subset of countries (Canada ‐ McLaren 2014, China ‐ Du 2014, Finland ‐ Laatikainen 2006, United Kingdom ‐ England ‐ Millett 2012, United States of America ‐ Pfeiffer 2014) incorporated an analysis of differential impact across different axes. Given that different methods were used across these studies, their equity analysis is not amenable to quantitative analysis (synthesis). Therefore, we have summarise those efforts in the following sections.
For Canada, McLaren 2014 examined inequities (by family income and highest educational attainment) in dietary sodium consumption in 1970‐1972 and in 2004. Overall, adjusted models revealed few statistically significant associations between income or education and sodium consumption. For men, an emerging inequity in use of table salt was indicated, whereby men of higher income were less likely than men of lower income to report using table salt in 2004, but no association was found in 1970‐72. For women, a negative association between education and sodium consumption (mg/d) was noted in 1970‐72, but no association in 2004. Study authors concluded that emerging inequity in reported use of table salt amongst men could reflect modest information‐based sodium reduction efforts that were implemented during the time frame considered (i.e. revision to the Food Guide in 1982). For sodium consumption in milligrams/d, no evidence of inequity was found in 2004, and in fact, in women a positive effect was observed (i.e. higher education associated with higher sodium consumption), which might reflect the very high prevalence of excess consumption.
For China, the China Health and Nutrition Survey cohort (Du 2014) permitted analysis of trends over time in sodium intake (1991‐2009) by place of residence (urban or rural), education (below high school vs high school or higher) and income (tertiles of inflation‐adjusted per capita household income). Study authors reported that sodium intake (grams/d) decreased significantly (1991‐2009) in all groups including those with higher and lower education; high, middle and low income; and urban and rural places of residence (Du 2014 ‐ Supplementary Table S1). The national sodium reduction initiative in China began in 2006.
For Finland, Laatikainen 2006 examined trends in urinary sodium excretion using surveys from 1979, 1982, 1987 and 2002 in four geographic areas: North Karelia, the Kuopio area, Southwestern Finland and the Helsinki area. Investigators observed different trends by geographical area (place of residence), such that urinary sodium excretion decreased significantly in eastern Finland between 1982 and 1987 but remained stable in southwestern Finland. Between 1982 and 2002, a significant decrease was seen in North Karelia and in southwestern Finland (Laatikainen 2006). In terms of education, the overall trend showed lowest salt intake in the highest education categories across the study period, but this effect was statistically significant only for men in North Karelia in 1979, and for women in southwestern Finland in 2002.
For the United Kingdom, Millett 2012 examined differential impact by social class (manual/non‐manual, based on the UK Registrar General's social class classification) and by ethnicity (white, black, South Asian) for the United Kingdom sodium reduction initiative, using 2003‐2007 data from the Health Survey for England. Mean salt intake decreased significantly from pre‐intervention to post‐intervention in all groups, and the reduction did not differ significantly between ethnic groups nor between social class groups, suggesting similar impact. Inequities in salt intake by social class and by ethnicity, which were noted in 2003, were thus still apparent in 2007. In other words, because reductions occurred in all social class and ethnic groups, inequities between groups remained. Another analysis of differential impact of the United Kingdom initiative was performed by Shankar in 2013 (see United Kingdom ‐ England); using econometric methods, investigators concluded that urinary sodium excretion levels were reduced following the United Kingdom initiative in almost all groups defined by birth cohort, gender and region. Stronger impact was observed amongst women than amongst men.
Finally for the United States of America, Pfeiffer 2014 examined trends in sodium intake (24‐hour urinary excretion, estimated from measured sodium concentrations in spot urine) between 1988 and 2010 using National Health and Nutrition Examination Survey (NHANES) data, including trends by race/ethnicity (Mexican‐American, Non‐Hispanic black, Non‐Hispanic white). Although cross‐sectional analysis showed differences in sodium intake between race/ethnic groups (e.g. mean estimated 24‐hour urine sodium excretion was highest in Mexican‐American and lowest in non‐Hispanic white groups in 1988‐1994 and 2003‐2006), no statistically significant changes in sodium intake were evident over time for any of these groups. In contrast, a statistically significant increase in sodium was observed across the sample as a whole.
Impact of studies that were not included in the quantitative synthesis
Initiatives in five countries could not be included in the quantitative synthesis because missing information (e.g. missing estimates of variance) could not be remedied by using other reported data or by contacting study authors. Available information from three of those countries (Japan ‐ Udagawa 2008, Denmark ‐ Denmark country questionnaire 2014‐2015, Turkey ‐ Erdem 2010) suggested a gradual decline in salt intake over the time period when the national dietary sodium reduction initiative was started. As the result of missing information, we do not know whether these apparent decreases in salt intake were statistically significant, with the exception of Denmark (Denmark country questionnaire 2014‐2015), which reported a statistically significant decrease over time in the text of their documentation. Of the other two omitted countries, the nature of change in salt intake over time in Thailand (Supornsilaphachai 2013) was mixed (appeared to be increasing on examination of one data point; appeared to be decreasing on examination of another), and data from New Zealand (New Zealand Ministry for Primary Industries) are not easily comparable because they were modelled rather than estimated, and therefore lacked key information such as sample size. Overall, it remains unknown what impact inclusion of these countries would have had on our results and conclusions.
Summary of main findings
Main findings are summarized in Table 19 .
| Country (initiative) | Overall impact | Impact for women only | Impact for men only |
| Austria | NS | NS | Decrease |
| Canada | Increase | Increase | Increase |
| China | Decrease | Decrease | Decrease |
| Finland | Decrease | Decrease | Decrease |
| France | Decrease | Decrease | Decrease |
| Ireland | Decrease | N/A | N/A |
| Netherlands | NS | NS | NS |
| Switzerland | Increase | NS | NS |
| United Kingdom ‐ England | Decrease | Decrease | Decrease |
| United Kingdom ‐ Great Britain | Decrease | NS | Decrease |
| United Kingdom ‐ Scotland | NS | NS | NS |
| United States | NS | Increase | NS |
Increase = Statistically significant increase in salt intake (grams/d) from pre‐intervention to post‐intervention
Decrease = Statistically significant decrease in salt intake (grams/d) from pre‐intervention to post‐intervention
NS = no statistically significant change in salt intake (grams/d) from pre‐intervention to post‐intervention
N/A = data did not permit separate analysis for males versus females
The United Kingdom is counted as one country/initiative but has three rows in the table because data were available for England only, Scotland only and Great Britain
Sensitivity analysis
Comparisons 5 and 6. Mean change in salt intake (grams/d) from pre‐intervention to post‐intervention ‐ using other available post‐intervention data points (Analysis 5.1) and using other available pre‐intervention data points (Analysis 6.1)
For initiatives with multiple available data points, analysis of impact was based on the pre‐intervention data point closest in time to the start year of the intervention, as well as the post‐intervention data point farthest away in time from the start year of the initiative. As a sensitivity check, we re‐ran analysis of impact using alternative post‐intervention data points, along with alternative pre‐intervention data points when available.
Of the 10 countries included in the overall analysis of impact, three (Finland ‐ Laatikainen 2006, United Kingdom ‐ Scottish Centre for Social Research 2011; Millett 2012; Wyness 2012, United States of America ‐ Pfeiffer 2014) had alternative post‐intervention data points (recall that for the United Kingdom we have data for UK/Great Britain as a whole ‐ Wyness 2012, for England only ‐ Millett 2012 and for Scotland only ‐ Scottish Centre for Social Research 2011). Finland (Laatikainen 2006), UK ‐ Scotland (Scottish Centre for Social Research 2011) and the United States of America (Pfeiffer 2014) each had one other available post‐intervention data point and UK ‐ England (Millett 2012) had three other available post‐intervention data points, as follows.
-
Finland: initiative started in 1979; main analysis based on post‐intervention data point from 1987; alternative post‐intervention data point available for 1982 (Laatikainen 2006).
-
UK ‐ Scotland: initiative started in 2003; main analysis based on post‐intervention data point from 2009; alternative post‐intervention data point available for 2006 (Scottish Centre for Social Research 2011).
-
United States of America: initiative started late 1980s to early 1990s; main analysis based on post‐intervention data point from 2010; alternative post‐intervention data point available for 2003‐2006 (Pfeiffer 2014).
-
UK‐England: initiative started in 2003; main analysis was based on post‐intervention data point from 2007; alternative post‐intervention data points available for 2004, 2005, and 2006 (Millett 2012).
We re‐ran the analysis using the only other post‐intervention data point available for Finland (Laatikainen 2006), UK ‐ Scotland (Scottish Centre for Social Research 2011) and the United States of America (Pfeiffer 2014), as well as the second most recent post‐intervention data point available for UK ‐ England (Millett 2012) (from 2006) (see Analysis 5.1). Use of these alternative post‐intervention data points did not change the effect observed in the main analysis (Analysis 1.1) for UK ‐ Scotland (Scottish Centre for Social Research 2011) (non‐significant effect based on both data points), United States of America (Pfeiffer 2014) (non‐significant effect based on both data points) or UK ‐ England (Millett 2012) (statistically significant mean decrease in salt intake from pre‐intervention to post‐intervention based on both data points). For Finland (Laatikainen 2006), the statistically significant mean decrease in salt intake (grams/d) from pre‐intervention to post‐intervention observed in Analysis 1.1 was reduced to non‐significance when the alternative post‐intervention data point was used (Analysis 5.1).
Of the 10 countries included in the overall analysis of impact, two (China ‐ Du 2014, Ireland ‐ Perry 2010) used alternative pre‐intervention data points. China (Du 2014) included five other available pre‐intervention data points, and Ireland (Perry 2010) included one other available pre‐intervention data point, as follows.
-
China: initiative started in 2006; main analysis based on pre‐intervention data point from 2006; alternative pre‐intervention data points available for 1991, 1993, 1997, 2000 and 2004 (Du 2014).
-
Ireland: initiative started in 2003; main analysis based on pre‐intervention data point from 2002; alternative pre‐intervention data point available from 1998 (Perry 2010).
We re‐ran the overall analysis using the only other available pre‐intervention data point for Ireland (Perry 2010), as well as the alternative pre‐intervention data point for China (Du 2014) that was second‐closest to the start of the initiative (from 2004) (see Analysis 6.1). Use of these alternative pre‐intervention data points did not change the effect observed in the main analysis (Analysis 1.1), specifically, the statistically significant mean decrease in salt intake (grams/d) from pre‐intervention to post‐intervention observed in Analysis 1.1 for both China (Du 2014) and Ireland (Perry 2010) remained in the analysis when the alternative pre‐intervention data point was used (Analysis 6.1).
Methodological quality of studies and publication bias
Across the 10 studies included in the assessment of overall effect, we found no clear indication of publication bias by examining the funnel plot (Figure 5).
Funnel plot of comparison 1.1: Mean change in salt intake (grams/d) from pre‐intervention to post‐intervention ‐ overall impact (men and women combined).
We rated overall quality of the evidence by using the GRADE (Grades of Recommendation, Assessment, Development and Evaluation Working Group) framework (GRADE n.d.), which is based on five considerations: risk of bias, imprecision, inconsistency, indirectness and publication bias. We first rated all of our studies as having 'low' quality because of their observational, uncontrolled nature. The relatively small number of studies and heterogeneity across them suggest the need to downgrade quality to 'very low'. See summary of findings Table for the main comparison.
Discussion
Summary of main results
Of the 15 initiatives that met our inclusion criteria, 10 provided sufficient data for quantitative analysis of overall impact.
Diversity across the included studies and their effects was notable. Across the 10 studies included in the quantitative synthesis, five showed a statistically significant reduction in dietary sodium intake. These included Finland (Laatikainen 2006), where multi‐component and structural efforts have been under way since the 1970s, as well as countries where the initiative is much more recent, such as United Kingdom (Scottish Centre for Social Research 2011; Millett 2012; Wyness 2012) and Ireland (Perry 2010) (where initiatives began in 2003). Two of the 10 studies (Canada ‐ McLaren 2014, Switzerland ‐ Chappuis 2011) showed a statistically significant increase in dietary sodium intake. In Switzerland (Chappuis 2011), although the initiative included activities of a structural nature (e.g. food product reformulation), those activities were characterised as "voluntary commitments" by companies, and thus their impact may be limited. Furthermore, the initiative in Switzerland (Chappuis 2011) is relatively new (started in 2008), and our post‐intervention data point was from 2011. At the time of data collection, Switzerland (Chappuis 2011) indicated engaging in "meetings with companies" around food product reformulation, so it is possible that those activities were in their early stages, and the 2011 data point was too early to permit capture of any impact of that initiative. In Canada (McLaren 2014), the intervention captured by available data points was very modest ‐ a single‐component initiative consisting of public education. The observed effect for Canada (McLaren 2014) may reflect an increase in salt intake that was already occurring as the result of increasing consumption of ultra‐processed food products (Moubarac 2014), which the modest intervention examined here was insufficient to offset. Because of high levels of heterogeneity across studies, we focused on individual rather than pooled study results.
When we focused on the seven multi‐component initiatives and incorporated intervention activities of a structural nature (e.g. food product reformulation, food procurement in specific settings), we found that a decrease in salt intake from pre‐intervention to post‐intervention was more apparent, with four of seven initiatives (Finland ‐ Laatikainen 2006, France ‐ Dubuisson 2010, Ireland ‐ Perry 2010, United Kingdom ‐ Scottish Centre for Social Research 2011; Millett 2012; Wyness 2012) showing such an effect. This is consistent with Rose's (Rose 1992) population‐level strategy of prevention and further emphasises the importance of distinguishing between population‐level interventions that are more agentic and those that are more structural (McLaren 2010; Sumar 2011). Our findings indicate that structural population‐level interventions are likely more impactful in this context of national dietary sodium reduction initiatives.
In terms of the differential impact of initiatives across social groups, nine initiatives provided sufficient data for quantitative analysis of differential impact by sex ("g"ender in the PROGRESS framework). Those nine studies revealed an apparent reduction in salt intake for men, with more than half (five of the nine) the initiatives showing a statistically significant mean decrease in salt intake from pre‐intervention to post‐intervention. Findings for women were more equivocal (three initiatives showed a statistically significant mean decrease in salt intake from pre‐intervention to post‐intervention, two studies showed a statistically significant mean increase and four showed a non‐statistically significant change).
Information was insufficient to reveal differential impact of these initiatives by other axes of stratification (equity). Our qualitative synthesis of findings from the few countries that incorporated analysis of differential impact (methods and indicators were too diverse to be quantitatively synthesised) showed that in some studies inequities did not necessarily increase (worsen) in the context of a national sodium reduction initiative. However, additional studies that consistently incorporate analysis of equity of impact are needed to confirm this. Furthermore, it is important to emphasise that although it is promising to observe that inequities are not worsening, it would be preferable to demonstrate that inequities are in fact narrowing.
Overall completeness and applicability of evidence
In our companion review (Trieu 2015), 75 countries were found to have national dietary sodium reduction efforts under way. We considered all of these for inclusion in our review and retained only 15, of which only a further subset provided data amenable to quantitative analysis. Although the growing number of sodium reduction initiatives worldwide is promising, limited available data that can be used to monitor present a challenge in building the evidence base. In some cases, data limitations reflect that the initiative is relatively new, activities (including monitoring) are at an early stage and data may be forthcoming. In other cases, however, the data infrastructure is limited, for example, some initiatives were started with no baseline data in place to permit rigorous evaluation. Although initiatives in the former category were classified as 'ongoing' and may be included in updates of this review, the latter initiatives had to be classified as 'excluded' because of the impossibility of going backwards to produce pre‐intervention data.
Although our companion review (Trieu 2015) identified national sodium reduction efforts that were under way in countries at all levels of economic development and in all World Health Organization (WHO) regions, this review did not include initiatives from lower‐middle‐income nor low‐income countries, nor in African or eastern Mediterranean regions. Therefore, it is not known to what extent our findings apply to those countries and regions.
In addition to indicators of sodium consumption, we intended (see our protocol, McLaren 2013) to examine indicators of health outcomes related to sodium consumption (e.g. stroke, blood pressure, hypertension, cardiovascular disease) as well as secondary outcomes (e.g. knowledge, awareness, other health/disease outcomes). However, we were unable to include all of these outcomes because in many instances, our review involved piecing together an evaluation from existing national data sources. Ensuring a comprehensive review required that we locate, for every included country, every data source that included any of these outcomes. The potential scope of data sources accordingly became very large, and because of the unwieldy nature of the task, we made the decision to focus on indicators of sodium consumption for feasibility purposes. We made this decision reluctantly ‐ we recognise the importance of including clinical outcomes in other reviews.
Quality of the evidence
As has been noted, we found substantial between‐study variation by examining the I2 measure, which in all cases exceeded 90%. For this reason, we did not focus on the overall pooled result and we downgraded the quality of the evidence to very low. It is likely that the main reasons for heterogeneity include context (i.e. "population" in PICO ‐ countries vary substantially in social, economic, political and cultural conditions) and the nature of the initiative (i.e. "intervention" in PICO ‐ interventions vary in type and extent, which in turn reflects country‐specific attributes such as stakeholder (e.g. government) level of concern with salt, investment in salt reduction, co‐ordination with other stakeholders and so on.
Across the 10 studies included in the assessment of overall effect, we found no clear indication of publication bias by examining the funnel plot (Figure 5), which is consistent with our comprehensive search and extensive use of grey literature resources.
By using specific tools (adapted version of the Cochrane risk of bias tool and the GRADE (Grades of Recommendation, Assessment, Development and Evaluation Working Group) method (GRADE n.d.)), review authors determined that the studies included in this review were at high risk of bias, and that overall quality of the body of evidence in terms of study limitations, consistency of effect, imprecision, indirectness and publication bias (GRADE n.d.) was very low. By examining the proportion of bias domains in which each initiative had high or uncertain risk of bias, we were able to capture variations in methodological quality across our included countries, which ranged from studies of highest methodological quality in France (Dubuisson 2010), Finland (Laatikainen 2006) and the United Kingdom (Scottish Centre for Social Research 2011; Millett 2012; Wyness 2012) (high or uncertain risk of bias in only one or two of seven bias domains) to studies of lowest methodological quality in Japan (Udagawa 2008), New Zealand (New Zealand Ministry for Primary Industries), Thailand (Supornsilaphachai 2013) and Turkey (Erdem 2010) (high or uncertain in five of seven bias domains). Notably, we included none of the four studies of lowest quality in our quantitative synthesis because investigators provided insufficient data.
There is a need to ensure that existing tools to appraise the methods used in observational studies adequately take into account the complexities associated with evaluation of national level interventions as described in this review. While the domains of GRADE assist in the identification of quality across study designs, further methodological work is needed to ensure that there is appropriate discrimination between high and low quality observational studies.
Potential biases in the review process
For feasibility reasons, we focused on a single primary outcome: dietary sodium intake. Thus we do not know ‐ on the basis of this review ‐ whether or the extent to which population‐level interventions in government jurisdictions for dietary sodium reduction impact health outcomes such as blood pressure or hypertension and stroke prevalence. Review authors need to include these outcomes in future reviews.
Although we undertook extensive efforts to accurately characterise intervention activities for each country, some uncertainty remains in terms of exposure, related to the jurisdictional administrative structure of a country. For example, a country (in the country questionnaire) may not have reported "food procurement policy in specific settings" as one of the activities within its national initiative. However, such activities may have been undertaken in smaller jurisdictions in that country such as provinces, states or municipalities, and the country contact speaking to national efforts would not necessarily be aware of activities occurring at a different level of government. We would not have captured such activities. Furthermore, to the extent that turnover of government staff is high, our method of gathering information from country contacts (see Trieu 2015) may be subject to limitations if the individual is new to the role and/or if institutional memory of the initiative is low. However, for the types of initiatives considered in this review, much information is not available in published form; thus information obtained from country contacts is extremely important, and we view its inclusion as a strength of our review.
We have no reason to believe that we missed information on serious and/or rare adverse events.
We undertook the last search in January 2015, and latency between the date of the search and publication of the review may introduce bias. Latency reflects the complexity of the review.
Although we included the Cochrane Public Health Group Specialised Register in our search, we did not search other trial registries, and this may have caused us to miss some trials (e.g. cluster‐randomised controlled designs), if indeed they existed.
We imposed no other restrictions.
Agreements and disagreements with other studies or reviews
No other comprehensive systematic reviews have examined the impact of population‐level dietary sodium reduction initiatives in government jurisdictions, including equity of impact.
Depiction of the hypothetical impact of a population‐level intervention, achieved by shifting the frequency distribution of the risk factor in a direction of lower risk (adapted from Rose 1992).
Risk of bias summary: review authors' judgements about each risk of bias item for each included initiative.
Funnel plot of comparison 1.1: Mean change in salt intake (grams/d) from pre‐intervention to post‐intervention ‐ overall impact (men and women combined).
Comparison 1 Mean change in salt intake (grams/d) from pre‐intervention to post‐intervention ‐ OVERALL, Outcome 1 Salt intake in grams per day.
Comparison 2 Mean change in salt intake (grams/d) from pre‐intervention to post‐intervention ‐ SUBSET OF MULTI‐COMPONENT INITIATIVES THAT INCLUDE STRUCTURAL ACTIVITIES, Outcome 1 Salt intake in grams per day.
Comparison 3 Mean change in salt intake (grams/d) from pre‐intervention to post‐intervention ‐ WOMEN ONLY, Outcome 1 Salt intake in grams per day.
Comparison 4 Mean change in salt intake (grams/d) from pre‐intervention to post‐intervention ‐ MEN ONLY, Outcome 1 Salt intake in grams per day.
Comparison 5 Mean change in salt intake (grams/d) from pre‐intervention to post‐intervention ‐ using other available post‐intervention data points, Outcome 1 Salt intake in grams per day.
Comparison 6 Mean change in salt intake (grams/d) from pre‐intervention to post‐intervention ‐ using other available pre‐intervention data points, Outcome 1 Salt intake in grams per day.
| Population‐level interventions in national government jurisdictions to reduce sodium consumption | ||||||
| Patient or population: people of all ages Settings: government jurisdictions Intervention: population‐level intervention to reduce dietary sodium consumption Comparison: mean change in salt intake (grams/d) from pre‐intervention to post‐intervention | ||||||
| Outcomes | Illustrative comparative risks | Relative effect | Number of participants | Quality of the evidence | Comments | |
| Pre‐intervention estimate | Post‐intervention estimate | |||||
| Salt intake in grams per day ‐ all eligible studies that provided sufficient data for quantitative analysis of overall impact (n = 10 countries) | Mean salt intake at pre‐intervention ranged from 6.1 to 12.7 grams/d | Mean salt intake at post‐intervention ranged from 5.5 to 11.9 grams/d | Not shown owing to high study heterogeneity | 64,798 participants | ⊕⊝⊝⊝ | As the result of high study heterogeneity (I2 > 90% in all analyses), we focus on individual studies rather than pooled effects. We downgraded the quality of evidence from low to very low due to all studies having high risk of bias (confounding domain) and the substantial inconsistency of effect sizes and the direction of the effect, which varied across the studies. |
| Salt intake in grams per day ‐ subset of initiatives that are multi‐component and incorporate activities of a structural nature (n = 7 countries) | Mean salt intake at pre‐intervention ranged from 6.3 to 11.8 grams/d | Mean salt intake at post‐intervention ranged from 5.5 to 10.6 grams/d | Not shown owing to high study heterogeneity | 34,227 participants (9 studies)* | ⊕⊝⊝⊝ | |
| Salt intake in grams per day ‐ women only (n = 9 countries) | Mean salt intake at pre‐intervention ranged from 4.7 to 12.2 grams/d | Mean salt intake at post‐intervention ranged from 4.1 to 11.4 grams/d | Not shown owing to high study heterogeneity | 27,184 participants (11 studies)* | ⊕⊝⊝⊝ | |
| Salt intake in grams per day ‐ men only (n = 9 countries) | Mean salt intake at pre‐intervention ranged from 6.1 to 13.2 grams/d | Mean salt intake at post‐intervention ranged from 5.2 to 12.5 grams/d | Not shown owing to high study heterogeneity | 22,977 participants (11 studies)* | ⊕⊝⊝⊝ | |
| GRADE Working Group grades of evidence | ||||||
| *Includes evaluations of the United Kingdom for England only, Scotland only and Great Britain. Including only Great Britain does not change overall findings We rated the overall quality of evidence using the GRADE (Grades of Recommendation, Assessment, Development and Evaluation Working Group) framework (GRADE n.d.), which is based on 5 considerations: risk of bias, imprecision, inconsistency, indirectness and publication bias. All of our studies started at 'low' quality because of their observational, uncontrolled nature. | ||||||
| Data point 1* | Data point 2* | |
| Data source | 2008 Austrian Study on Nutritional Status | 2012 Austrian Study on Nutritional Status |
| Overall estimate – as originally reported | Mean salt intake: 8.3 grams/d, SD = 3.53 | Mean salt intake: 8.15 grams/d, SD = 2.99 |
| Overall estimate – revised for comparability | See above | See above |
| Measurement tool | 24‐Hour dietary recall | Two 24‐hour dietary recalls |
| Sample size | n = 2123 | n = 380† |
| Progress indicators available | Gender | Gender |
| underlined text = computed by review authors *Data points on which primary calculation of mean difference (pre‐intervention vs post‐intervention) is based (see Data and analyses ‐ "overall") †This number is based only on the number of adults in the sample, so it is comparable with the estimate from 2008 (i.e. kids and elderly were not considered) | ||
| Data point 1* | Data point 2 | Data point 3* | |
| Data source | 1988‐1994 NHANES | 2003‐2006 NHANES | 2010 NHANES |
| Overall estimate – as originally reported | Crude geometric mean sodium intake: 3280 milligrams/d, SD = 83.3 (95% CI 3277.64 to 3282.34) | Crude geometric mean sodium intake: 3270 milligrams/d, SD = 101 (95% CI 3267.13 to 3272.87) | Crude geometric mean sodium intake: 3400 milligrams/d, SD = 87.6 (95% CI 3396.18 to 3403.82) |
| Overall estimate – revised for comparability | Crude geometric mean salt intake: 8.34 grams/d | Crude geometric mean salt intake: 8.31 grams/d | Crude geometric mean salt intake: 8.64 grams/d |
| Measurement tool | Estimated 24‐hour urine | Estimated 24‐hour urine | Estimated 24‐hour urine |
| Sample size | n = 1249 | n = 1235 | n = 525 |
| Progress indicators available | Gender, race‐ethnicity | Gender, race‐ethnicity | Gender, race‐ethnicity |
| underlined text = computed by review authors *Data points on which primary calculation of mean difference (pre‐intervention vs post‐intervention) is based (see Data and analyses ‐ "overall") | |||
| Data point 1* | Data point 2* | |
| Data source | 1970‐1972 Nutrition Canada Survey | 2004 Canadian Community Health Survey |
| Overall estimate – as originally reported | Mean sodium intake: 2403.9 milligrams/d, SD = 1362.5 | Mean sodium intake: 3057 milligrams/d, SD = 606.2 |
| Overall estimate – revised for comparability | Mean salt intake: 6.11 grams/d, SD = 3.46 | Mean salt intake: 7.77 grams/d, SD = 1.54 |
| Measurement tool | 24‐Hour dietary recall | 24‐Hour dietary recall |
| Sample size | n = 4540 | n = 10,449 |
| Progress indicators available | Gender, education and income | Gender, education and income |
| underlined text = computed by review authors *Data points on which primary calculation of mean difference (pre‐intervention vs post‐intervention) is based (see Data and analyses ‐ "overall") | ||
| Data point 1 | Data point 2 | Data point 3 | Data point 4 | Data point 5 | Data point 6* | Data point 7* | |
| Data source | 1991 China Health and Nutrition Survey | 1993 China Health and Nutrition Survey | 1997 China Health and Nutrition Survey | 2000 China Health and Nutrition Survey | 2004 China Health and Nutrition Survey | 2006 China Health and Nutrition Survey | 2009 China Health and Nutrition Survey |
| Overall estimate – as originally reported | Mean sodium intake: 6.6 grams/d, SD = 3.4 | Mean sodium intake: 6.6 grams/d, SD = 3.4 | Mean sodium intake: 6.2 grams/d, SD = 3.5 | Mean sodium intake: 6.0 grams/d, SD = 3.2 | Mean sodium intake: 5.2 grams/d, SD = 2.7 | Mean sodium intake: 5.0 grams/d, SD = 2.8 | Mean sodium intake: 4.7 grams/d, SD = 2.6 |
| Overall estimate – revised for comparability | Mean salt intake: 16.76 grams/d | Mean salt intake: 16.76 grams/d | Mean salt intake: 15.77 grams/d | Mean salt intake: 15.24 grams/d | Mean salt intake: 13.21 grams/d | Mean salt intake: 12.70 grams/d | Mean salt intake: 11.94 grams/d |
| Measurement tool | 3 consecutive 24‐hour dietary recalls | 3 consecutive 24‐hour dietary recalls | 3 consecutive 24‐hour dietary recalls | 3 consecutive 24‐hour dietary recalls | 3 consecutive 24‐hour dietary recalls | 3 consecutive 24‐hour dietary recalls | 3 consecutive 24‐hour dietary recalls |
| Sample size | n = 7337 | n = 6958 | n = 7241 | n = 7940 | n = 7250 | n = 6826 | n = 6932 |
| Progress indicators available | Gender, place of residence, education, income | Gender, place of residence, education, income | Gender, place of residence, education, income | Gender, place of residence, education, income | Gender, place of residence, education, income | Gender, place of residence, education, income | Gender, place of residence, education, income |
| underlined text = computed by review authors *Data points on which primary calculation of mean difference (pre‐intervention vs post‐intervention) is based (see Data and analyses ‐ "overall") Note: Based on all data points, “sodium intake significantly decreased over time (P‐trend, 0.001, general linear regression models)” | |||||||
| Data point 1 | Data point 2 | Data point 3 | |
| Data source | 2006 Danish Health | 2008 Danish National Survey of Dietary Habits and Physical Activity | 2010 Danish Health |
| Overall estimate – as originally reported | Median salt intake: 8.93 grams/d | Mean salt intake: 8.5 grams/d, SD = 2.9 (95% CI 8.41 to 8.59) | Median salt intake: 8.27 grams/d |
| Overall estimate – revised for comparability | Same as above | Same as above | Same as above |
| Measurement tool | Spot urine | Food record for 7 consecutive days | Spot urine |
| Sample size | n = 3294 | n = 4431 | n = 1478 |
| Progress indicators available | Gender | Gender | Gender |
| underlined text = computed by review authors Denmark was not included in quantitative synthesis because of missing or unuseable information (e.g. mean, SD, n) at 1 or more data points | |||
| Data point 1* | Data point 2 | Data point 3* | Data point 4 | Data point 5 | Data point 6 | |
| Data source | 1979 North Karelia Kuopio area | 1982 North Karelia Kuopio area | 1987 North Karelia Kuopio area | 2002 North Karelia | 2007 5 regions in Finland | 2012 5 regions in Finland |
| Overall estimate – as originally reported | Mean salt intake (North Karelia): 11.65 grams/d, SD = 4.39 Mean salt intake (Kuopio area): 11.78 grams/d, SD = 4.72 | Mean salt intake (North Karelia): 11.94 grams/d ± 4.62 Mean salt intake (Kuopio area): 11.59 grams/d, SD = 4.50 | Mean salt intake (North Karelia): 10.36 grams/d, SD = 4.46 Mean salt intake (Kuopio area): 10.63 grams/d, SD = 4.10 | Mean salt intake (North Karelia): 8.43 grams/d, SD = 3.19 | Mean salt intake: 8.0 grams/d | Mean sodium intake: 3.01 grams/d, SD = 1.10 |
| Overall estimate – revised for comparability | Same as above | Same as above | Same as above | Same as above | Same as above | Mean salt intake: 7.65 grams/d |
| Measurement tool | 24‐Hour urine | 24‐Hour urine | 24‐Hour urine | 24‐Hour urine | 48‐Hour dietary recall | 48‐Hour dietary recall |
| Sample size | n = 536 (North Karelia) n = 670 (Kuopio area) | n = 484 (North Karelia) n = 428 (Kuopio area) | n = 409 (North Karelia) n = 400 (Kuopio area) | n = 342 (North Karelia) | n = 1576 | n = 1295 |
| Progress indicators available | Gender and education | Gender and education | Gender and education | Gender and education | Gender | Gender, place of residence and education |
| underlined text = computed by review authors *Data points on which primary calculation of mean difference (pre‐intervention vs post‐intervention) is based (see Data and analyses ‐ "overall") | ||||||
| Data point 1a* | Data point 1b | Data point 2* | |
| Data source | 1998‐1999 Individual and National Food Consumption Surveys | 1998‐1999 Individual and National Food Consumption Surveys | 2006‐2007 Individual and National Food Consumption Surveys |
| Overall estimate – as originally reported | Mean sodium intake: 3145.7 milligrams/d, SD = 1016.4 | Mean sodium intake: 137.40 mmol/d, SD = 101.48 | Mean sodium intake: 2966.6 milligrams/d ± 922.4 |
| Overall estimate – revised for comparability | Mean salt intake: 8.00 grams/d | Mean salt intake: 7.97 grams/d | Mean salt intake: 7.54 grams/d |
| Measurement tool | 7‐Day open‐ended food record | 7‐Day open‐ended food record | 7‐Day open‐ended food record |
| Sample size | n = 1345 | n = 1474 | n = 1922 |
| Progress indicators available | Gender | Gender | Gender, place of residence and occupation |
| underlined text = computed by review authors *Data points on which primary calculation of mean difference (pre‐intervention vs post‐intervention) is based (see Data and analyses ‐ "overall") | |||
| Data point 1 | Data point 2 | Data point 3* | Data point 4* | Data point 4a | Data point 4b | Data point 5 | |
| Data source | 1997‐1999 North‐South Ireland Food Consumption Survey | 1998 Survey of Lifestyle Attitudes and Nutrition in Ireland | 2002 Survey of Lifestyle Attitudes and Nutrition in Ireland | 2007 Survey of Lifestyle Attitudes and Nutrition in Ireland | 2007 Survey of Lifestyle Attitudes and Nutrition in Ireland | 2007 Survey of Lifestyle Attitudes and Nutrition in Ireland | 2008‐2010 National Adult Nutrition Survey |
| Overall estimate – as originally reported | Mean salt intake: 8.3 grams/d | Mean salt intake: 8.3 grams/d, SD = 3.8 | Mean salt intake: 8.2 grams/d, SD = 5.9 | Mean salt intake: 7.85 grams/d, SD = 3.70 | Mean salt intake: 8.9 grams/d | Mean salt intake: 9.3 grams/d | Mean salt intake: 7.4 grams/d |
| Overall estimate – revised for comparability | Same as above | Same as above | Same as above | Same as above | Same as above | Same as above | Same as above |
| Measurement tool | 7‐Day diet diaries | Food frequency questionnaire | Food frequency questionnaire | Food frequency questionnaire | Spot urine | 24‐Hour urine | 4‐Day semi weighted food record |
| Sample size | n = 1379 | n = 6539 | n = 5992 | n = 9172 | n = 1207 | n = 599 | n = 1500 |
| Progress indicators available | Gender | Gender | Gender | Gender and social class | Gender | Gender | Gender |
| underlined text = computed by review authors *Data points on which primary calculation of mean difference (pre‐intervention vs post‐intervention) is based (see Data and analyses ‐ "overall") | |||||||
| Data point 1 | Data point 2 | Data point 3 | Data point 4 | Data point 5 | Data point 6 | Data point 7 | Data point 8 | Data point 9 | Data point 10 | |
| Data source | 1997 National Health and Nutrition Survey | 2003 National Health and Nutrition Survey | 2004 National Health and Nutrition Survey | 2005 National Health and Nutrition Survey | 2006 National Health and Nutrition Survey | 2007 National Health and Nutrition Survey | 2008 National Health and Nutrition Survey | 2009 National Health and Nutrition Survey | 2010 National Health and Nutrition Survey | 2012 National Health and Nutrition Survey |
| Overall estimate – as originally reported | Mean salt intake: 13.5 grams/d | Mean salt intake: 11.7 grams/d | Mean salt intake: 11.2 grams/d | Mean salt intake: 11.5 grams/d | Mean salt intake: 11.2 grams/d | Mean salt intake: 11.1 grams/d | Mean salt intake: 10.9 grams/d | Mean salt intake: 10.7 grams/d | Mean salt intake: 10.6 grams/d | Mean salt intake: 10.4 grams/d |
| Overall estimate – revised for comparability | Same as above | Same as above | Same as above | Same as above | Same as above | Same as above | Same as above | Same as above | Same as above | Same as above |
| Measurement tool | Nutritional intake survey | Nutritional intake survey | Nutritional intake survey | Nutritional intake survey | Nutritional intake survey | Nutritional intake survey | Nutritional intake survey | Nutritional intake survey | Nutritional intake survey | Nutritional intake survey |
| Sample size | n = 13,289 | Unknown | n = 8762 | Unknown | Unknown | n = 8885 | Unknown | Unknown | n = 8815 | n = 8247 |
| Progress indicators available | ‐‐ | Gender | Gender | Gender | Gender | Gender | Gender | Gender | Gender | ‐‐ |
| underlined text = computed by review authors Japan was not included in quantitative synthesis because of missing or unuseable information (e.g. mean, SD, n) at 1 or more data points | ||||||||||
| Data point 1* | Data point 2 | Data point 3* | |
| Data source | 2006 Cross‐sectional study of adults in Doetinchem | 2007‐2010 National Food Consumption Survey | 2010 Cross‐sectional study of adults in Doetinchem |
| Overall estimate – as originally reported | Mean sodium excretion: 148 mmol/d, SD = 55 Median salt intake: 8.7 grams/d (IQR 6.7 to 11.0) | Mean salt intake: 8.45 grams/d | Mean sodium excretion: 148 mmol/d, SD = 58 Median salt intake: 8.5 grams/d (IQR 6.6 to 10.9) |
| Overall estimate – revised for comparability | Mean salt intake: 8.58 grams/d | Same as above | Mean salt intake: 8.58 grams/d |
| Measurement tool | 24‐Hour urine | 2 non‐consecutive 24‐hour recalls | 24‐Hour urine |
| Sample size | n = 317 | n = 2160 | n = 342 |
| Progress indicators available | Gender | Gender | Gender |
| underlined text = computed by review authors *Data points on which primary calculation of mean difference (pre‐intervention vs post‐intervention) is based (see Data and analyses ‐ "overall") | |||
| Data point 1 | Data point 2a | Data point 2b | Data point 3 | |
| Data source | 2003‐2004 New Zealand Total Diet Study | 2008‐2009 New Zealand Adult Nutrition Survey | 2008‐2009 New Zealand Adult Nutrition Survey | 2009 New Zealand Total Diet Study |
| Overall estimate – as originally reported | N/A | Mean salt intake: 6.37 grams/d | Mean salt intake: 9 grams/d | N/A |
| Overall estimate – revised for comparability | Same as above | Same as above | ||
| Measurement tool | N/A | Multiple pass 24‐hour dietary recall | Spot urine | N/A |
| Sample size | N/A | n = 4721 | Unknown | N/A |
| Progress indicators available | Gender | ‐‐ | ‐‐ | Gender |
| underlined text = computed by review authors New Zealand was not included in quantitative synthesis because of missing or unuseable information (e.g. mean, SD, n) at 1 or more data points | ||||
| Data point 1a* | Data point 1b | Data point 2* | |
| Data source | 1984 Representative sample of the Swiss population | 1984 Representative sample of the Swiss population | 2011 Swiss survey on salt intake |
| Overall estimate – as originally reported | Mean salt intake: 8.4 grams/d, SD = 3.6 | Mean salt intake: 10.3 grams/d, SD = 6.2 | Mean salt intake: 9.2 grams/d, SD = 3.8 |
| Overall estimate – revised for comparability | Same as above | Same as above | Same as above |
| Measurement tool | 24‐Hour urine | Spot urine | 24‐Hour urine |
| Sample size | n = 147 | n = 966 | n = 1448 |
| Progress indicators available | None | None | Gender and place of residence |
| underlined text = computed by review authors *Data points on which primary calculation of mean difference (pre‐intervention vs post‐intervention) is based (see Data and analyses ‐ "overall") | |||
| Data point 1 | Data point 2 | Data point 3 | Data point 4 | Data point 5 | |
| Data source | 1960 National Nutrition Survey | 1975 National Nutrition Survey | 2003 National Nutrition Survey | 2008 | 2008‐2009 Fourth Thai National Health Examination Survey |
| Overall estimate – as originally reported | Mean sodium intake: 2.4 grams/d | Mean sodium intake: 2.4 grams/d | Mean sodium intake: 4.0 grams/d | Mean sodium intake: 4.4 grams/d | Mean sodium intake: 3.3 grams/d |
| Overall estimate – revised for comparability | Mean salt intake: 6.10 grams/d | Mean salt intake: 6.10 grams/d | Mean salt intake: 10.16 grams/d | Mean salt intake: 11.18 grams/d | Mean salt intake: 8.38 grams/d |
| Measurement tool | Dietary survey | Dietary survey | Dietary survey (food list recall and food frequency checklist of foods/condiments) | Dietary survey (household survey) | Dietary survey (food list recall and food frequency checklist of foods/condiments) |
| Sample size | Unknown | Unknown | Unknown | Unknown | Unknown |
| Progress indicators available | ‐‐ | ‐‐ | ‐‐ | ‐‐ | ‐‐ |
| Note: A 2007 data point was removed because we did not have enough information to suggest that it was national in scope (i.e. comparable with other data points) underlined text = computed by review authors Thailand was not included in quantitative synthesis because of missing or unuseable information (e.g. mean, SD, n) at 1 or more data points | |||||
| Data point 1a | Data point 1b | Data point 2 | |
| Data source | 2008 SALTURK 1 | 2008 SALTURK 1 | 2012 SALTURK 2 |
| Overall estimate – as originally reported | Mean salt intake: 16.6 grams/d, SD = 7.3 | Mean salt intake: 18.01 grams/d | Mean salt intake: 15.0 grams/d |
| Overall estimate – revised for comparability | Same as above | Same as above | Same as above |
| Measurement tool | 24‐Hour urine | Spot urine | 24‐Hour urine |
| Sample size | n = 816 | n = 1970 (total number of eligible participants) | n =925; “657 person (according to urine creatine levels)” |
| Progress indicators available | Gender, place of residence and education level | ‐‐ | ‐‐ |
| Note: We believe that the difference between estimates 1a and 1b is that 1a is based on 24‐hour urine and 1b is based on spot urine. The overall conclusion for Turkey does not differ when 1a versus 1b is used underlined text = computed by review authors Turkey was not included in quantitative synthesis because of missing or unuseable information (e.g. mean, SD, n) at 1 or more data points | |||
| Data point 1a | Data point 1b* | Data point 2a | Data point 2b | Data point 3a | Data point 3b | Data point 4 | Data point 5a | Data point 5b | Data point 6a | Data point 6b* | Data point 7 | |
| Data source | 2003 Health Survey for England | 2003 Health Survey for England | 2004 Health Survey for England | 2004 Health Survey for England | 2005 Health Survey for England | 2005 Health Survey for England | 2005‐2006 National Diet and Nutrition Survey | 2006 Health Survey for England | 2006 Health Survey for England | 2007 Health Survey for England | 2007 Health Survey for England | 2011 National Diet and Nutrition Survey |
| Overall estimate – as originally reported | Geometric mean salt intake: 5.29 grams/d | Urinary sodium excretion: 109.03 mmol of sodium/d, SD = 20.41 (95% CI 108.05 to 110.01) | Geometric mean salt intake: 5.99 grams/d | Urinary sodium excretion: 111.28 mmol of sodium/d, SD = 19.82 (95% CI 110.55 to 112.01) | Geometric mean salt intake: 4.80 grams/d | Urinary sodium excretion: 99.94 mmol of sodium/d, SD = 22.17 (95% CI 99.30 to 100.58) | Mean salt intake: 9.0 grams/d, SD = 3.7 (95% CI 8.66 to 9.34) | Geometric mean salt intake: 4.73 grams/d | Urinary sodium excretion: 97.49 mmol of sodium/d, SD = 20.00 (95% CI 97.07 to 97.91) | Geometric mean salt intake: 4.55 grams/d | Urinary sodium excretion: 94.16 mmol of sodium/d, SD =19.39 (95% CI 93.58 to 94.74) | Mean salt intake: 8.1 grams/d, SD = 5.79 (95% CI 7.85 to 8.35) |
| Overall estimate – revised for comparability | Same as above | Mean salt intake:6.32 grams/d | Same as above | Mean salt intake: 6.45 grams/d | Same as above | Mean salt intake: 5.80 grams/d | Same as above | Same as above | Mean salt intake: 5.65 grams/d | Same as above | Mean salt intake: 5.46 grams/d | Same as above |
| Measurement tool | Spot urine | Spot urine | Spot urine | Spot urine | Spot urine | Spot urine | 24‐Hour urine | Spot urine | Spot urine | Spot urine | Spot urine | 24‐Hour urine |
| Sample Size | n = 1668 | n = 1668 | n = 2840 | n = 2840 | n = 4643 | n = 4643 | n = 448 | n = 8844 | n = 8844 | n = 4269 | n = 4269 | n = 547 |
| Progress indicators available | Gender, ethnicity and social class | Gender and place of residence | Gender, ethnicity and social class | Gender and place of residence | Gender, ethnicity and social class | Gender and place of residence | Gender | Gender, ethnicity and social class | Gender and place of residence | Gender, ethnicity and social class | Gender and place of residence | Gender |
| underlined text = computed by review authors *Data points on which primary calculation of mean difference (pre‐intervention vs post‐intervention) is based (see Data and analyses ‐ "overall") | ||||||||||||
| Data point 1a | Data point 1b* | Data point 1c | Data point 1d | Data point 2* | |
| Data source | 2000‐2001 National Diet and Nutrition Survey | 2000‐2001 National Diet and Nutrition Survey | 2000‐2001 National Diet and Nutrition Survey | 2000‐2001 National Diet and Nutrition Survey | 2008 Survey by National Centre for Social Research |
| Overall estimate – as originally reported | Mean sodium intake: 2794.39 milligrams/d, SD = 861.44 | Mean salt intake: 9.53 grams/d, SD = 4.48 | Median dietary sodium intake: 2611 milligrams/d (IQR = 1243) | Median 24‐hour sodium excretion: 140 mmol/d (IQR = 99.4) | Mean salt intake: 8.64 grams/d, SD = 4.39 |
| Overall estimate – revised for comparability | Mean salt intake: 7.01 grams/d (95% CI 7.00 to 7.21) | Same as above | Estimated salt intake: 6.5 grams/d | Estimated salt intake: 8.2 grams/d | Same as above |
| Measurement tool | 7‐Day consecutive weighted dietary record | 24‐Hour urine | 7‐Day consecutive weighted dietary record | 24‐Hour urine | 24‐Hour urine |
| Sample size | n = 1724 | n = 1147 | n = 2150 | n = 692 | |
| Progress indicators available | Gender | Gender | Gender, place of residence, education and social class | Gender, place of residence, education and social class | Gender |
| underlined text = computed by review authors *Data points on which primary calculation of mean difference (pre‐intervention vs post‐intervention) is based (see Data and analyses ‐ "overall") | |||||
| Data point 1* | Data point 2 | Data point 3 | Data point 4* | Data point 5 | |
| Data source | 2003 Scottish Health Survey | 2006 | 2008 Scottish Health Survey | 2009 Scottish Health Survey | 2009 |
| Overall estimate – as originally reported | Mean salt intake: 6.8 grams/d | Mean salt intake: 9.1 grams/d, SD = 4.1 | Mean salt intake: 6.4 grams/d | Mean salt intake: 6.8 grams/d | Mean salt intake: 8.8 grams/d, SD = 3.7 |
| Overall estimate – revised for comparability | Same as above | Same as above | Same as above | Same as above | Same as above |
| Measurement tool | Spot urine | 24‐Hour urine | Spot urine | Spot urine | 24‐Hour urine |
| Sample size | n = 1148 | n = 442 | n = 1041 | n = 1045 | n = 702 |
| Progress indicators available | Gender | Gender | Gender | Gender | Gender |
| underlined text = computed by review authors *Data points on which primary calculation of mean difference (pre‐intervention vs post‐intervention) is based (see Data and analyses ‐ "overall") | |||||
| Country | Pre‐intervention | Post‐intervention |
| AUSTRIA | ||
| Males | Data point 1 (2008) Mean salt intake = 9.4 grams/d SD = 3.56; n = 778 | Data point 2 (2012): Mean salt intake: 8.7 grams/d |
| Females | Data point 1 (2008) Mean salt intake = 7.6 grams/d SD = 2.81; n = 1345 | Data point 2 (2012): Mean salt intake: 7.6 grams/d SD = 3.11; n = 232 |
| CANADA | ||
| Males | Data point 1 (1970‐1972): Mean salt intake = 7.32 grams/d SD = 3.8; n = 1974 | Data point 2 (2004): Mean salt intake: 8.19 grams/d SD = 1.65; n = 4837 |
| Females | Data point 1 (1970‐1972): Mean salt intake = 5.00 grams/d SD = 2.71; n = 2566 | Data point 2 (2004): Mean salt intake: 7.41 grams/d |
| CHINA | ||
| Males | Data point 6 (2006): Mean salt intake = 13.21 grams/d SD = 7.37; n = 3242 | Data point 7 (2009): Mean salt intake = 12.45 grams/d |
| Females | Data point 6 (2006): Mean salt intake = 12.19 grams/d SD = 6.86; n = 3584 | Data point 7 (2009): Mean salt intake = 11.43 grams/d |
| FINLAND – Kuopio only | ||
| Males | Data point 1 (Kuopio only 1979): Mean salt intake = 13.1 grams/d SD = 5.20; n = 343 | Data point 3 (Kuopio only 1987): Mean salt intake = 12.0 grams/d SD = 4.45; n=180 |
| Females | Data point 1 (Kuopio only 1979): Mean salt intake = 10.4 grams/d SD = 4.15; n = 327 | Data point 3 (Kuopio only 1987): Mean salt intake = 9.5 grams/d SD = 3.78; n=220 |
| FRANCE | ||
| Males | Data point 1 (1998‐1999): Mean salt intake = 9.26 grams/d SD = 3.01; n = 613 | Data point 2 (2006‐2007): Mean salt intake = 8.69 grams/d SD = 2.83; n = 840 |
| Females | Data point 1 (1998‐1999): Mean salt intake = 6.93 grams/d SD = 2.16; n = 732 | Data point 2 (2006‐2007): Mean salt intake = 6.65 grams/d SD = 1.88; n = 1082 |
| NETHERLANDS | ||
| Males | Data point 1 (2006): Mean salt intake = 9.51 grams/d SD = 3.19; n = 137 | Data point 3 (2010): Mean salt intake = 10.09 grams/d SD = 3.65; n = 154 |
| Females | Data point 1 (2006): Mean salt intake = 7.89 grams/d SD = 2.96; n = 180 | Data point 3 (2010): Mean salt intake = 7.42 grams/d SD = 2.49; n = 188 |
| SWITZERLAND | ||
| Males | Data point 1a (1984): Salt intake = 10.4 grams/d | Data point 2 (2011): Salt intake = 10.6 grams/d SD = 4.2; n = 706 |
| Females | Data point 1a (1984): Salt intake = 7.3 grams/d | Data point 2 (2011): Salt intake = 7.8 grams/d |
| UNITED KINGDOM (Great Britain) | ||
| Males | Data point 1b (2000‐2001) Mean salt intake: 11.0 grams/d SD = 5.02; n = 833 | Data point 2 (2008): Mean salt intake: 9.68 grams/d SD = 4.10; n = 294 |
| Females | Data point 1b (2000‐2001) Mean salt intake: 8.1 grams/d SD = 3.88; n = 891 | Data point 2 (2008): Mean salt intake: 7.66 grams/d SD = 4.77; n = 398 |
| UK (England) – geometric mean, salt intake | ||
| Males | Data point 1a (2003): Geometric mean salt intake = 6.10 grams/d SD = 3.46; n = 735 | Data point 6a (2007): Geometric mean salt intake = 5.16 grams/d SD = 3.36; n = 1926 |
| Females | Data point 1a (2003): Geometric mean salt intake = 4.73 grams/d SD = 3.12; n = 933 | Data point 6a (2007): Geometric mean salt intake = 4.12 grams/d SD = 2.47; n = 2343 |
| UK (Scotland) | ||
| Males | Data point 1 (2003): Mean salt intake = 7.6 grams/d SD = 5.17; n = 508 | Data point 4 (2009): Mean salt intake = 7.3 grams/d SD = 3.78; n = 447 |
| Females | Data point 1 (2003): Mean salt intake = 6.1 grams/d SD = 3.87; n = 640 | Data point 4 (2009): Mean salt intake = 5.7 grams/d SD = 4.37; n = 598 |
| UNITED STATES | ||
| Males | Data point 1 (1988‐1994): Crude geometric mean salt intake = 10.04 grams/d SD = 7.81; n = 645 | Data point 3 (2010): Crude geometric mean salt intake = 10.02 grams/d SD = 6.25; n = 258 |
| Females | Data point 1 (1988‐1994): Crude geometric mean salt intake = 6.79 grams/d SD = 6.18; n = 604 | Data point 3 (2010): Crude geometric mean salt intake = 7.55 grams/d SD = 3.84; n = 267 |
| Country (initiative) | Overall impact | Impact for women only | Impact for men only |
| Austria | NS | NS | Decrease |
| Canada | Increase | Increase | Increase |
| China | Decrease | Decrease | Decrease |
| Finland | Decrease | Decrease | Decrease |
| France | Decrease | Decrease | Decrease |
| Ireland | Decrease | N/A | N/A |
| Netherlands | NS | NS | NS |
| Switzerland | Increase | NS | NS |
| United Kingdom ‐ England | Decrease | Decrease | Decrease |
| United Kingdom ‐ Great Britain | Decrease | NS | Decrease |
| United Kingdom ‐ Scotland | NS | NS | NS |
| United States | NS | Increase | NS |
| Increase = Statistically significant increase in salt intake (grams/d) from pre‐intervention to post‐intervention Decrease = Statistically significant decrease in salt intake (grams/d) from pre‐intervention to post‐intervention NS = no statistically significant change in salt intake (grams/d) from pre‐intervention to post‐intervention N/A = data did not permit separate analysis for males versus females The United Kingdom is counted as one country/initiative but has three rows in the table because data were available for England only, Scotland only and Great Britain | |||
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Salt intake in grams per day Show forest plot | 12 | Std. Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Salt intake in grams per day Show forest plot | 9 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Salt intake in grams per day Show forest plot | 11 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Salt intake in grams per day Show forest plot | 11 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Salt intake in grams per day Show forest plot | 12 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Salt intake in grams per day Show forest plot | 12 | Mean Difference (IV, Random, 95% CI) | Totals not selected | |
