The Science of Salt: A global review on changes in sodium levels in foods

Abstract

This review aims to summarize and synthesize studies reporting on changes in sodium levels in packaged food products, restaurant foods, and hospital or school meals, as a result of salt reduction interventions. Studies were extracted from those published in the Science of Salt Weekly between June 2013 and February 2018. Twenty‐four studies were identified: 17 assessed the changes in packaged foods, four in restaurant foods, two in hospital or school meals, and one in both packaged and restaurant foods. Three types of interventions were evaluated as part of the studies: voluntary reductions (including targets), labeling, and interventions in institutional settings. Decreases in sodium were observed in all studies (n = 8) that included the same packaged foods matched at two time points, and in the studies carried out in hospitals and schools. However, there was little to no change in mean sodium levels in restaurant foods. The pooled analysis of change in sodium levels in packaged foods showed a decrease in sodium in unmatched food products (−36 mg/100 g, 95% CI −51 to −20 mg/100 g) and in five food categories—breakfast cereals, breads, processed meats, crisps and snacks, and soups. Twenty‐two of the 24 studies were from high‐income countries, limiting the applicability of the findings to lower resource settings.

1. INTRODUCTION

Excess salt intake is a worldwide public health problem,1, 2 with an increasing amount of high‐quality research showing the relationship between salt and raised blood pressure.3, 4, 5, 6, 7, 8 In turn, raised blood pressure is a well‐established risk factor for cardiovascular diseases (CVDs),2 the leading cause of death worldwide.9 The recent Global Burden of Disease (GBD) study reported that high systolic blood pressure was the leading cause of overall mortality and accounted for 218 million disability‐adjusted life years (DALYs) globally in 2017.2 The same study showed that of all dietary risk factors, a diet high in sodium was accountable for the largest number of all‐cause deaths (3.2 million deaths) and for about 70 million DALYs.2 Most populations consume more salt than the World Health Organization's (WHO) maximum daily salt intake recommendation of <5 g per day (g/d), with an estimated global mean salt consumption of about 10 g/d.10

Since 2013, Member States of the WHO have committed to reducing population salt intake by 30% by 2025.11 Many countries have initiated salt reduction interventions, and a number of reviews have shown some progress in reducing salt intake in populations.12, 13, 14 To further support countries in implementing salt reduction interventions and achieving a decrease in population salt intake, the WHO published the SHAKE Technical Package for Salt Reduction in 2016, consisting of implementation strategies in five key areas including surveillance, reformulation of products to contain less salt, labeling, consumer education, and working in settings.15 Ongoing monitoring and evaluation of these efforts are essential to strengthen policies and actions around salt.

The Science of Salt Weekly initiative (available at https://www.whoccsaltreduction.org/portfolio/science-of-salt-weekly/) was established in 2013 with the aim of providing an up‐to‐date account of the growing body of literature around salt. As part of this initiative, regularly updated systematic reviews on the relationship between salt and health outcomes, and the implementation of salt reduction interventions are being published in the Journal of Clinical Hypertension.16 To date, 11 studies (six on the relationship of salt and health outcomes17, 18, 19, 20, 21, 22 and five related to the implementation of salt reduction interventions23, 24, 25, 26, 27) have been published. The last implementation review was the first to employ a thematic approach focusing on studies that reported knowledge, attitudes, and behaviors relating to salt.27 Similarly, the current review adopts a thematic focus on studies that assessed changes in sodium levels in foods.

In many high‐income countries, prepared or packaged foods contribute about 75% of dietary sodium28; and there has been an increasing trend in consumption of processed foods in low‐ and middle‐income countries.29 Most countries have included interventions to reduce sodium in processed foods in their national salt reduction strategy, for example, engaging with the food industry, front‐of‐pack nutrition labeling, and setting sodium reduction targets.12 The objectives of this review are as follows: (a) to summarize and synthesize the literature around salt reduction interventions aimed at reducing sodium levels in foods and meals; and (b) to determine whether there were changes in sodium levels in packaged foods, restaurant foods, and hospital or school meals following these interventions.

2. METHODS

Articles identified and published in the Science of Salt Weekly newsletter related to the implementation of salt reduction interventions were included in this review. Studies were identified through a MEDLINE literature search using a comprehensive search strategy as previously reported.16 Articles published since the start of the Science of Salt (SOS) initiative in June 2013 to February 2018 were screened for relevance and eligibility by two review authors (JAS and SRT). Studies were included if the primary or secondary outcomes provided data related to changes in sodium levels in foods, in response to the implementation of salt reduction interventions directly (ie, interventions that were primarily focused on salt reduction) or indirectly (ie, nutrition interventions that included salt reduction as a component). All study designs and all types of foods and food categories were included. Studies that only reported sodium levels at one time point (eg, surveillance or baseline measurement studies) or those that estimated the effects of the interventions through modeling were excluded.

Using a data extraction tool designed for this review, information on study year, country of study, study design, type of foods and food categories assessed, sample size, source of sodium data, intervention and setting, duration of intervention, and changes in sodium levels in foods (by food category and/or overall, if reported) were extracted. For consistency purposes, sodium content in mg/100 g was preferred over mg/serve if both were reported, while sales‐weighted sodium was utilized if both unweighted and sales‐weighted estimates were available. For studies that assessed sodium levels at more than two time points, only the change from the first to the last measurement was considered in the main analysis.

Included studies were described and summarized. Results were presented according to the type of food assessed: packaged foods available in supermarket or grocery stores, restaurant foods, and school and hospital meals. For packaged food products, some studies collected information and compared the same food products at two different time points (matched), while others included all products available in the shop (unmatched). Hence, the results for unmatched and matched products were presented separately. Since the majority of studies reported the results for unmatched packaged foods, it was only possible to conduct subgroup analyses according to the intervention type and duration of intervention for unmatched products. Furthermore, changes in sodium levels in unmatched packaged foods in ten product categories (ie, bread, breakfast cereal, cheese, biscuits, ready meals, processed meat, fish and fish products, crisps and snacks, soup, and vegetables) were investigated.

To examine the effects of the interventions on sodium levels by the type of food (packaged foods, restaurant foods, and hospital or school meals), we applied two data analysis strategies depending on the availability of data. The mean differences in sodium levels over time of unmatched packaged food products were combined in a meta‐analysis, guided by the steps below.

• For studies that only reported mean sodium levels at different time points, but not change over time, we derived the change using the immediate form of two‐sample mean‐comparison test in STATA SE version 12.0 for Windows (StataCorp LP), based on the method of Gosset.30 This was used if the number of products, mean, and standard deviation (SD) were reported at two time points (before and after intervention).

• For studies that only reported the median sodium level and range, the equations proposed by Hozo et al31 were used to convert median to mean, and range to SD.

• Conversion of standard error (SE) and confidence interval (CI) of group means or mean differences to SD were performed using the equations in the Cochrane Handbook for Systematic Reviews of Interventions.32

• For studies that reported sodium levels of foods in multiple groups (eg, sodium content of branded and private‐label products), where appropriate, groups were combined to produce a single pairwise comparison, using the equations defined in the Cochrane Handbook.32

• Pooled mean differences within‐study and between‐studies were derived using the generic inverse‐variance method for continuous outcomes in Review Manager (RevMan) version 5.3 software.33 A random‐effects model was applied across all analyses.

On the contrary, for matched packaged food products, restaurant foods, and school or hospital meals, a narrative synthesis of findings was undertaken due to the small number of studies and/or the heterogeneous nature of the results reported in the studies.

3. RESULTS

3.1. Characteristics of studies

A total of 703 studies were identified, of which 86 were deemed potentially eligible for full review (Figure 1). Of these, 62 were excluded: 42 studies reported sodium levels at one time point (baseline measurements) and 20 were modeling studies. Ultimately, 24 studies were included in this review. All studies (n = 24) were serial cross‐sectional in design and came from nine countries: seven high‐income and two middle‐income countries (Table 1). There were nine studies in the United States (US),34, 35, 36, 37, 38, 39, 40, 41, 42 three each in New Zealand,43, 44, 45 Australia,46, 47, 48 and the United Kingdom (UK),49, 50, 51 two in Canada,52, 53 and one each in Slovenia,54 The Netherlands,55 Brazil,56 and India.57 There were three types of salt reduction interventions evaluated: voluntary reductions (including targets), labeling, and interventions in institutional settings such as hospitals or schools. Study duration ranged from 8 months to 14 years.

Figure 1.

Studies included in the current review, June 2013 to February 2018

Table 1.

Characteristics of included studies and summary of findings

First author, year, country of study	Data collection year, sample size	Source of sodium info	Intervention and type	Summary of findings
Packaged foods
Pravst, 2017,54 Slovenia	2011: 1374 products from four stores 2015: 5759 products from five stores 22 food categories	Food labels	National action plan to reduce salt Voluntary efforts and/or targets	There was a significant increase in the average sodium content of plain pasta (15 to 61 mg/100 g), ready meals (480 to 510 mg/100 g), yoghurt products (46 to 48 mg/100 g), and pasta sauces (386 to 601 mg/100 g), and a significant reduction in sodium content of soft drinks (20 to 13 mg/100 mL). No changes were observed for other food categories
Ni Mhurchu, 2017,43 New Zealand	2014: 14 368 2015: 14 415 2016: 15 358 438 matched products 14 food categories	Food labels	Health Star Rating front‐of‐pack nutrition labeling system Labeling	In both 2015 and 2016, products displaying HSR labels had significantly lower mean sodium content (−302 and −218 mg/100, respectively) compared to non‐HSR products. Furthermore, of the 438 products that had HSR labels in 2016 and were also available in 2014, there was a small but significant decline in the mean sodium content (−49 mg/100 g) over time
Johnson, 2017,57 India	2010: 4166 products (38% were sodium‐labeled) 2012‐2014: 5686 products from 11 stores (32% were sodium‐labeled) 14 food groups, 33 food categories, 90 subcategories	Food labels	National nutrition labeling regulations Labeling	There was an increase in the mean sodium content of cereal and grain products (327 to 477 mg/100 g), dairy and dairy alternatives (257 to 513 mg/100 g), non‐alcoholic beverages (59 to 119 mg/100 g), and sauces and spreads (1473 to 2217 mg/100 g). No changes were observed for the other food groups
Brooks, 2017,35 United States	2013: 4347 2015: 4996 Packaged food facings in vending machines and cafeterias or kiosks	Food labels	Community‐level intervention in hospitals, health centers, recreational organizations, and organizations serving homeless populations Intervention in institutional setting	Overall, there was a significant decrease in the proportion of food facings with greater than 200 mg per serving of sodium (29.0% to 21.5%; P = 0.003). The percentage of products with >200 mg of sodium significantly decreased in hospital cafeterias and kiosks (30.7% to 20.9%; P = 0.003), but not in vending machines
Ning, 2017,44 New Zealand	52 products from five stores 19 matched products Four categories	Nutrition analysis provided by the companies, verified and approved by the National Heart Foundation	Tick labeling program Labeling	There was a decrease in the mean sodium content of reformulated and newly formulated breakfast cereals (−68 mg and −125 mg/100) and edible spreads (−320 mg and −206 mg/100). There were also reductions in the mean sodium content of reformulated cooking sauces (−168 mg/100 g) and newly formulated processed poultry (−231 mg/100 g)
Temme, 2017,55 the Netherlands	2011: 436 2016: 2272 Eight food groups, 27 subgroups	Food labels and chemical analyses	Private‐public agreement on improvement of product composition: salt, saturated fat, and sugar Voluntary efforts and/or targets	Overall, there was a significant reduction in the average salt content of bread (−19%), cut potato crisps (−26%), tomato/vegetable sauces (−15%), meal sauces with a binder (−19%), and soups sold as liquid (−12%). There was a significant increase in the mean salt content of biscuits (36%). Among the 14 subgroups with maximum salt targets, there was a significant reduction in six subgroups which ranged from −9% to −42%
Nilson, 2017,56 Brazil	2011: 1067 2013‐2014: 1288 2017: 981 20 food categories	Food labels from company websites and customer services	Regional targets for several food categories by the Pan American Health Organization Voluntary efforts and/or targets	There was a significant 8% to 34% reduction in mean sodium content in 13 food categories from 2011 to 2017. The products analyzed in 2017 showed that 100% of products in over half of the categories met the regional targets and only one category had <85% of the products meeting the targets
Poti, 2017,36 United States	Number of packaged foods analyzed not reported Nutrient data from a total of 172 042 households 10 food categories	Food labels of food and drink purchases brought to homes (household consumer panel data)	Food industry efforts and call from governments and organizations to voluntary reduce sodium levels Voluntary efforts and/or targets	There was a significant reduction in the mean sodium content of purchased packaged foods from 2000 (411 mg/100 g) to 2014 (361 mg/100 g). Also, there were significant decreases in the mean sodium content of all food categories, ranging from −3% (processed meat) to −18% (soups) reduction over time
Pombo‐Rodrigues, 2017,49 United Kingdom	2004: 295 2006: 246 2009: 94 2012: 49 2015: 270 22 matched products Breakfast cereals	Food labels	National salt reduction strategy with a primary focus on reformulation Voluntary efforts and/or targets	Overall, there was a gradual reduction in the mean salt content of breakfast cereals from 2004 (0.96 g/100 g) to 2015 (0.46 g/100 g). The 22 products consistently surveyed over the 5 y showed a significant reduction of 47% from 2004 (1.57 g/100 g) to 2015 (0.83 g/100 g)
Curtis, 2016,37 United States	2009: 6336 2012: 6898 2014: 7396 61 food categories, grouped into 15 metacategories	NSRI packaged food database (combines sales data and nutrition information)	National Salt Reduction Initiative Voluntary efforts and/or targets	The overall sales‐weighted mean sodium density significantly declined by 6.8% from 2009 to 2014. There were significant reductions in the sales‐weighted mean sodium density in 26 categories from 2009 to 2014, with no significant increases observed. The proportion of products meeting the 2012 NSRI targets also increased from 2009 (33%) to 2012 (42%) to 2014 (45%)
Arcand, 2016,52 Canada	2010‐2011: 6918 2013: 9199 12 major food categories, 105 major and minor subcategories	Food labels	Health Canada's sodium reduction benchmark targets Voluntary efforts and/or targets	There were significant decreases in mean sodium content in 16.2% of food categories, significant increases in 1.9% of food categories, and no changes in sodium content in 81.9% of food categories. There was an increase in the proportion of products meeting at least one of the sodium benchmark targets from 51.4% to 58.2%
Trevena, 2015,46 Australia	2011: 4501 2012: 5184 2013: 5995 2792 matched products 15 food categories	Food labels	Australian Food and Health Dialogue Voluntary efforts and/or targets	There was no change observed over time in both private‐label and branded products. Among products present in both 2011 and 2013, there were small but significant reductions in mean sodium content of both private‐label (439 to 430 mg/100 g) and branded products (551 to 544 mg/100 g), with no difference in reformulation between the two groups (P = 0.73)
Monro, 2015,45 New Zealand	2003: 323 2013: 885 182 matched products 9 food groups	Food labels	Tick labeling program and HeartSAFE Labeling and voluntary target	Overall, there were no significant reductions in the mean sodium content of both matched and unmatched products. Among matched products, there were significant reductions in the mean sodium content of both private‐label (−69 mg/100 g) and branded (−50 mg/100 g) products (both P's < 0.001)
Taillie, 2015,39 United States	Number of packaged foods analyzed not reported	Food labels of purchased packaged foods	Walmart's Healthier Food Initiative: front‐of‐pack labeling, price reductions on healthier items, and reducing sodium by 25% Voluntary efforts and/or targets	There was a decrease in the sodium density of purchased packaged foods from 2000 (231 mg/100 g) to 2013 (198 mg/100 g). However, this reduction did not differ to what would have been observed had pre‐implementation trends continued
Trevena, 2014,47 Australia	2010: 400 2011: 437 2012: 464 2013: 548 Three food groups	Food labels	Australian Food and Health Dialogue Voluntary efforts and/or targets	There were significant reductions in the mean sodium content of bread products (454 to 415 mg/100 g), breakfast cereals (316 to 237 mg/100 g), and bacon/ham/cured meats (1215 to 1114 mg/100 g) from 2010 to 2013. The percentage of bread products (42% to 67%) and bacon/ham/cured meats (28% to 47%) meeting the targets also increased, and 56% of breakfast cereals had reduced sodium levels by at least 15%
Eyles, 2013,50 United Kingdom	2006: 47 337 2011: 49 714 16 858 matched products 14 food groups	Food labels of food and drink purchases brought to homes (household consumer panel data)	UK Food Standards Agency voluntary sodium reduction targets Voluntary efforts and/or targets	There was a significant decrease in the overall mean sodium content of processed foods from 2006 to 2011 (−26 mg/100 g; P < 0.001). Among matched products, the decline in mean sodium content was also significant (−23 mg/100 g; P < 0.001). Both reductions were equivalent to a 7% fall in sodium levels in foods
Jacobson, 2013,42 United States	402 matched packaged foods grouped into 11 categories 78 matched restaurant foods grouped into eight categories	Food labels of packaged foods Nutrition data from websites of restaurants	Food industry efforts and call from governments and organizations to voluntary reduce sodium levels Voluntary efforts and/or targets	There was a 3.5% reduction in the mean sodium content of packaged foods from 2005 (575 mg/100 g) to 2011 (555 mg/100 g), while there was a 2.6% increase in the mean sodium content of restaurant foods during the same period (493 to 505 mg/100 g). Overall, there was a lack of substantial changes in sodium levels in foods across the study period
Brinsden, 2013,51 United Kingdom	2001: 39 2006: 138 2011: 203 18 matched products Breads	Food labels	National salt reduction strategy with a primary focus on reformulation Voluntary efforts and/or targets	There was a gradual decrease in the mean salt content of breads from 2001 (1.2 g/100 g) to 2006 (1.1 g/100 g) to 2011 (1.0 g/100 g), equivalent to a 20% reduction over the 10‐y period. Among matched products, there was a 17% fall in the mean salt content from 2001 to 2011. The proportion of products meeting the 2012 bread target increased from 28% in 2001 to 52% in 2006 to 71% in 2011
Restaurant foods
Wolfson, 2018,34 United States	21 577 menu items from 66 restaurants from 2012 to 2016	Nutrition data from websites of restaurants	Food industry efforts and call from governments and organizations to voluntary reduce sodium levels Voluntary efforts and/or targets	Overall, there was no change in the mean sodium content of menu items from 2012 (753 mg/item) to 2016 (758 mg/item). However, items newly introduced in 2016 had significantly lower sodium (1053 mg/item) compared to food items present in 2012 only (1157 mg/item). Changes in the mean sodium content varied by menu category and restaurant type
Scourboutakos, 2014,53 Canada	2198 foods reported in 2010 and 2013 from 61 restaurants	Nutrition data from websites of restaurants	Restaurant industry efforts to voluntarily reduce sodium levels Voluntary efforts and/or targets	Overall, there was a small but significant reduction in the mean sodium content of restaurant foods (401 to 390 mg/100 g) from 2010 to 2013. There were decreases in sodium content in 30.1% of foods (481 to 407 mg/100), increases in 16.3% of foods (353 to 431 mg/100 g), and no changes in 53.6% of foods (371 to 368 mg/100 g). The change in sodium content varied by restaurant and food category
Garcia, 2014,48 Australia	2009: 302 2010: 348 2011: 381 2012: 379 products from six fast‐food chains	Nutrition data from websites of restaurants	Restaurant industry efforts to voluntary reduce sodium levels Voluntary efforts and/or targets	There was a decrease in the overall mean sodium content of fast foods (514 to 471 mg/100 g) from 2009 to 2012. Across the food categories examined, there were small reductions in the mean sodium content over the 4 y, apart from side menu items
Rudelt, 2014,41 United States	1997‐1998: 450 1999‐2000: 457 2001‐2002: 472 2003‐2004: 509 2005‐2006: 574 2007‐2008: 590 2009‐2010: 695 lunch or dinner menu items from eight fast‐food restaurants	Food and nutrition database updated from the websites of restaurants	Restaurant industry efforts to voluntary reduce sodium levels Voluntary efforts and/or targets	Overall, there was an increase in the mean sodium content of menu offerings from 1997‐1998 (624 mg/item) to 2009‐2010 (770 mg/item). Mean sodium content of entrees (934 to 1095 mg/item) and condiments (249 to 314 mg/item) increased, while mean sodium content of side dishes (624 to 583 mg/item) decreased during the same period. Of the eight restaurants, seven showed increases in mean sodium content of menu offerings
Hospital and school meals
Moran, 2015,38 United States	Nutrient data from 7‐d breakfast, lunch, and dinner menus from each hospital Regular‐diet meals from eight hospitals	Food labels of products used; or info from vendors; or info from similar products from a nutrient database	Healthy Hospital Food Initiative which includes nutrition standards Intervention in institutional setting	There was a significant 19% reduction in the median sodium content of meals from baseline (2636 mg) to endline (2149 mg). Post‐intervention, the eight hospitals met the patient meal standard for sodium (≤2300 mg), compared to only two hospitals meeting the target pre‐intervention
Cummings, 2014,40 United States	Nutrient data from 683 and 728 schools for SY 2010‐2011 and 2011‐2012 School meals	Chemical analyses and food production records	Strategy‐focused menu planning including sodium limits, environment modifications, and awareness campaign Intervention in institutional setting	There were decreases in the mean sodium content of elementary (747 to 494 mg/meal) and secondary breakfast (1064 to 494 mg/meal), and secondary lunch (1137 to 1090 mg/meal), while an increase was observed in the mean sodium content of elementary lunch (1011 to 1114 mg/meal)

3.2. Interventions in packaged foods, restaurant foods, and hospital or school meals

Eighteen studies evaluated changes in sodium levels in packaged food products, 10 of these reported on unmatched packaged foods, one was based on matched food products, and seven included both unmatched and matched products. For studies of unmatched food products, sample sizes ranged from 39 to 47 337 at baseline (median: 1374; IQR: 323‐6336) and from 33 to 49 714 at follow‐up (median: 4996; IQR: 548‐7396). For studies of matched food products, sample sizes ranged from 18 to 16 858 (median: 292; IQR: 21‐1615). The sodium content in packaged foods was measured in several ways: 17 studies utilized the food labels (nutrition information panels),35, 36, 37, 39, 42, 43, 45, 46, 47, 49, 50, 51, 52, 54, 55, 56, 57 seven of these combined food labels with sales data or household purchases data to derive sales‐weighted sodium density,36, 37, 39, 43, 44, 50, 54 and one used both food labels and chemical analysis.55 One study sourced information from a nutrient analysis given by companies.44 Most studies investigated sodium levels across a range of food categories (ranged from 3 to 106 categories), while two studies focused on just one major food category (breakfast cereals49 and bread51) and its subcategories, one study included packaged food facings in vending machines, cafeterias, or kiosks,35 while one did not specify the food categories included in the sodium analyses.39 In terms of the type of salt reduction intervention, thirteen studies assessed the effects of voluntary reductions (including targets) on packaged foods,36, 37, 39, 42, 46, 47, 49, 50, 51, 52, 54, 55, 56 four assessed the effect of labeling,43, 44, 45, 57 and one study was carried out in institutional settings to reduce the proportion of packaged foods with more than 200 mg of sodium per serving.35

Changes in sodium levels in restaurant foods were assessed in five studies—three in the United States34, 41, 42 and one each from Australia48 and Canada.53 All studies involved voluntary reductions, and sodium data were collected from the websites of restaurants. On the other hand, two studies in the United States evaluated the changes in sodium level in meals in institutional settings—one in hospitals38 and one in schools.40 Sodium levels were quantified by analyzing the food offerings in the menus.

3.3. Changes in sodium levels in foods

3.3.1. Packaged food products

Unmatched packaged food products

Of the 17 studies of unmatched packaged foods, six were excluded from the pooled analysis to determine the change in mean sodium levels over time. Three studies did not report any measure of variability,39, 44, 47 one study reported the outcomes in terms of the proportion of products with sodium >200 mg per serving,35 and two studies focused on just one major food category.49, 51 The meta‐analysis of change from the 11 studies showed a significant reduction in the mean sodium levels of unmatched packaged food products over time (−36 mg/100 g, 95% CI −51 to −20 mg/100 g; Figure 2). Subgroup analyses showed that apart from one subgroup (labeling; Table 2), there was a significant reduction in mean sodium levels within all subgroups (all P‐value's <0.05). There was no differential effect by the type and duration of the intervention on sodium levels (P > 0.05). Pooled mean differences in sodium levels over time in ten food categories are presented in Table 3. Details on the food categories included in the analyses from each study are shown in Table S1. Significant reductions in average sodium levels were observed in breads (−61 mg/100 g), breakfast cereals (−109 mg/100 g), processed meats (−76 mg/100 g), crisps and snacks (−77 mg/100 g), and soups (−60 mg/100 g). No reduction was noted for cheeses, biscuits, ready meals, fish and fish products, and vegetables.

Figure 2.

Random‐effects meta‐analysis of the change in sodium levels in packaged food products

Table 2.

Changes in sodium levels in packaged food products over time by subgroup

Subgroup	Mean difference (95% CI), mg/100 g	P‐value for within‐subgroup difference	P‐value for between‐subgroup differences
By type of intervention
Voluntary reductions (including targets)	−38 (−56 to −21)	<0.001	0.49
Labeling	−25 (−58 to 8)	0.14
By length of follow‐up
5 y or less	−27 (−44 to −11)	0.001	0.26
More than 5 y	−51 (−90 to −12)	0.010

Table 3.

Changes in sodium levels in different food categories

Food category	No. of studies	Mean difference (95% CI), mg/100 g	P‐value for overall effect
Bread	10	−61 (−81 to −42)	<0.001
Breakfast cereal	10	−109 (−140 to −77)	<0.001
Cheese	10	−19 (−72 to 34)	0.48
Biscuits	9	−4 (−34 to 26)	0.79
Ready meals	4	−6 (−75 to 62)	0.86
Processed meat	9	−76 (−125 to −27)	0.002
Fish and fish products	7	−4 (−37 to 29)	0.82
Crisps and snacks	8	−77 (−126 to −28)	0.002
Soup	7	−60 (−101 to −19)	0.004
Vegetables	7	−27 (−54 to 0)	0.05

Matched food products

Of the eight studies of matched packaged foods, five studies reported the overall change in mean sodium over time: Australia (−8 mg/100 g, n = 2792),46 the UK (−23 mg/100 g, n = 16 858),50 the United States (−20 mg/100 g, n = 402),42 and New Zealand (−55 mg/100, n = 182; −49 mg/100; n = 438).43, 45 Two studies from the UK, one including breakfast cereals49 and the other including breads,51 both reported a significant reduction in sodium over time, −740 mg/100 g (n = 22) and −210 mg/100 g (n = 18), respectively. The other study from New Zealand reported reductions in sodium in three food categories: breakfast cereals (−68 mg/100 g, n = 2), cooking sauces (−168 mg/100 g, n = 14), and edible spreads (−320 mg/100 g, n = 3).44

3.3.2. Restaurant foods

One study showed that between 1997 and 2010 in the United States, mean sodium levels in foods in eight leading fast‐food restaurants increased from 624 mg/item (n = 450) to 770 mg/item (n = 695), equivalent to a 23% increase.41 Another US study showed a 3% increase in the sodium content of matched fast‐food restaurant products (n = 78) between 2005 and 2011.42 A third US study found no change in the mean sodium content of fast foods (n = 2795) and other types of restaurant foods from 2012 to 2016.34 On the contrary, significant reductions in sodium levels in restaurant foods were reported in Canada and Australia.48, 53 In Canada, one study showed average sodium content of foods (n = 2198) from major sit‐down and fast‐food restaurants decreased from 401 mg/100 g in 2010 to 390 mg/100 g in 2013 (P < 0.01).53 In Australia, one study showed mean sodium levels declined by 43 mg/100 g (95% CI −66 to −20 mg/100; n = 379) between 2009 and 2012, across six large fast‐food chains.48

3.3.3. Hospital or school meals

One study showed that implementation of menu planning, environment modifications, and awareness campaigns in schools resulted in decreased sodium levels from 2010 to 2011 in elementary and secondary school breakfasts by 253 and 569 mg/meal, respectively, and in secondary school lunches by 47 mg/meal.40 However, sodium content of elementary school lunches increased by 103 mg/meal.40 One study showed that implementation of the Healthy Hospital Food Initiative (HHFI) resulted in median sodium levels in meals decreasing by 19%, from 2636 to 2149 mg/d (P < 0.05).38

4. DISCUSSION

This review identified 24 studies from June 2013 to February 2018 that measured changes in sodium levels in packaged food products, restaurant foods, and hospital or school meals. Our findings suggest that there have been modest reductions in sodium levels in packaged food products, both unmatched and matched, and in certain food categories, as well as decreases in sodium in hospital or school meals. However, there has been little to no change in the mean sodium levels of restaurant foods, and in one US study, there was a substantial increase in sodium content of foods. Most studies (n = 22) came from just seven high‐income countries with two from two middle‐income countries. The low number of studies from low‐ and middle‐income countries is in line with our previous reviews23, 24, 25, 26, 27 and means that the applicability of findings generated from the current review may be limited in these settings. Further, compared to the 2015 review of salt reduction initiatives around the world which reported that 38 countries (28 high‐income and 10 middle‐income countries) have voluntary or mandatory sodium targets for foods,12 the current review shows that only eight of these countries (seven high‐income and one middle‐income countries) have available data on change to measure the food industry's progress to reduce the sodium content of products. These findings suggest (a) the need for further interventions to reduce sodium levels in foods in low‐ and middle‐income countries and (b) the need for monitoring and evaluation studies in countries where sodium targets for foods had been established.

This review identified three types of interventions to reduce sodium levels in foods—voluntary reductions, labeling, and interventions in institutional settings. Voluntary reductions (including targets and/or government engagement with the food industry to voluntarily reduce sodium levels in their products) were by far the most common type of intervention used in evaluated studies, including for 13 packaged food interventions and all five restaurant interventions. This is not surprising, as previous work has shown that about 80% of countries with a salt reduction strategy that involves food industry engagement have set voluntary targets for foods.12 A 2014 study showed that there is an increasing trend toward setting mandatory sodium targets for different categories of food58; however, we did not find any evaluation studies on change in sodium levels in foods from these countries. Only four studies identified in this review were based on labeling interventions and only three on interventions in institutional settings. This is despite the fact that the 2015 review identified 43 countries with interventions in institutional settings, and 31 with front‐of‐pack‐labeling schemes to reduce sodium in foods.12 While this discrepancy in the number of known interventions and published evaluation studies might suggest a lack of monitoring and evaluation of these interventions, it is also possible that sodium levels in foods data are available, but unpublished (hence beyond the scope of our search), as systematic approaches to monitoring and evaluation have been documented.59 Regardless, it is important that interventions to reduce sodium through labeling and institutional settings are evaluated both to demonstrate progress and to ensure that lessons can be shared.

Measuring changes in sodium levels in packaged food products was the focus of the majority of studies from this review. Specifically, most studies involved unmatched packaged foods, meaning, all products available in the supermarket or grocery stores at the time of the survey were included. The large number of studies reporting changes in sodium levels in unmatched packaged foods provided an opportunity to combine the results in a meta‐analysis. Our pooled analysis showed that there was a modest but significant reduction in mean sodium levels over time, by 36 mg/100 g. Consistent reductions were shown in our subgroup analyses according to the type and duration of intervention. However, it is still difficult to draw firm conclusions (ie, whether this reflects true reductions in sodium levels in foods), since the studies reported on a variety of food categories, and there were large differences in the number of products within each study. Further examination of changes in sodium levels pooled from different countries in individual food categories found that there were significant modest reductions in sodium levels in breads, breakfast cereals, processed meats, crisps and snacks, and soups, with the largest difference found in breakfast cereals, which was a 109 mg/100 g reduction. As these are some of the most targeted product categories for reformulation,12 the reductions found in these categories are likely reflective of the effect of the interventions.

It was not possible to pool the changes in sodium levels in matched packaged food products (same set of products followed up over time) due to the heterogeneous nature of the outcomes reported, and the absence of correlation coefficient between baseline and follow‐up measurements. Nonetheless, individual examination of the eight studies with available data on matched packaged foods revealed that all studies showed a decrease in mean sodium over time, overall or in at least one food category. This likely reflects the effects of reformulation carried out by manufacturers. Again, it is not possible to draw firm conclusions, since these reductions could be due to the differences in sampling approaches across the studies, and the number and type of products covered, among others. However, the fact that both matched and unmatched packaged foods showed a declining trend in sodium is noteworthy, as this suggests that while companies are reducing sodium in their existing products, they are also producing new products that have lower sodium content.

Changes in sodium levels in restaurant foods were evaluated in three high‐income countries (Canada, the United States, and Australia), while changes in hospital or school meals were only assessed in the United States. For restaurant foods, the three studies from the United States were consistent in showing that there was no decline in the mean sodium content of fast foods or even an increase, while decreases, although small, were reported in Australia (−43 mg/100 g) and Canada (−11 mg/100 g). This is surprising given that the United States has voluntary sodium targets for restaurant foods through the National Salt Reduction Initiative (NSRI)60 from 2009 to 2014, followed by the US Food and Drug Administration draft guidance for food industry to reduce sodium in processed and restaurant foods in 2016,61 while no targets for restaurant foods have been set in Canada and Australia. Nevertheless, all studies agree that the sodium level in restaurant foods (despite the declines observed in Australia and Canada) remains high, with some meals exceeding the recommended daily sodium intake.34, 41, 42, 48, 53 With the increasing trend of eating food away from home,62, 63 our findings suggest that there is a need for stronger actions to reduce sodium levels in foods eaten out of the home. On the other hand, the two studies done in institutional settings (one in schools40 and one in hospitals38) to reduce sodium in meals both showed positive results, suggesting that setting sodium standards in schools or hospitals is a plausible intervention, particularly when paired with complementary supportive structures, support and resources.64 However, it is important to note that these two studies were implemented at a small scale (city‐ or county‐wide), and these are settings in which little monitoring take place, hence, further studies to support the feasibility of wider implementation and effectiveness of these interventions are needed.

This study has strengths and limitations. To the best of our knowledge, this is the first study to report an overall change in sodium levels in unmatched packaged food products from a pooled analysis. We have taken an inclusive approach in our analyses by using all available data reported in the studies. This was done, for example, through following published Equations31, 32 to combine group‐level estimates, and convert medians and ranges to means and SDs. Our results, however, showed high between‐study variability, and although we conducted two subgroup analyses to explore possible sources, they were not able to explain the heterogeneity observed in the main analysis. This suggests that other factors not accounted for in our analyses are in play, so our results should be interpreted with caution. Nevertheless, we also investigated the changes in mean sodium at the food category level and in matched food products, which complemented the results of our pooled analysis. A strength of this review is the inclusion of restaurant foods, and hospital and school meals (in addition to packaged foods), which provided a complete picture of the changes in sodium levels in foods. Limitations of the review include the restriction by language and publication date. It is also likely that changes in sodium levels in foods are reported in the grey literature, which we have not included in our search. Lastly, we were unable to pool the results for matched food products due to the absence of correlation coefficient to compute change, or the difference in the way the outcomes were reported in the studies.

In conclusion, our findings demonstrate that many countries have implemented salt reduction interventions aimed at reducing mean sodium levels in foods; however, only few have published evaluation studies following these interventions. Our results suggest that it is possible to reduce sodium content of packaged foods through voluntary reductions in high‐income countries. This is important, given the high contribution of packaged foods to dietary sodium in this setting, and the rapidly increasing trend in consumption of these foods in lower resource settings. Lastly, this review highlights the need for stronger actions to reduce sodium levels in foods, coupled with adequate monitoring and evaluation system, particularly in low‐ and middle‐income countries, to reach the global target of a 30% reduction in population salt intake by 2025.

CONFLICT OF INTEREST

JW is Director of the World Health Organization Collaborating Centre on Population Salt Reduction and is supported by a National Heart Foundation Future Leaders Fellowship Level II (102039). JW has funding from the World Health Organization, the Victorian Health Promotion Foundation (20122), and the National Health and Medical Research Council of Australia for research on salt reduction (1052555 and 1111457). KT is supported by the National Health and Medical Research Council of Australia Early Career Fellowship (1161597). JAS is supported by the National Health and Medical Research Council of Australia Postgraduate Scholarship (1168948). NRCC was a paid consultant to the Novartis Foundation (2016‐2017) to support their program to improve hypertension control in low‐ to middle‐income countries which includes travel support for site visits and a contract to develop a survey. NRCC has provided paid consultative advice on accurate blood pressure assessment to Midway Corporation (2017) and is an unpaid member of World Action on Salt and Health (WASH). Of the studies reviewed, CJ, SRT, and JW were authors on Johnson C et al (2017), and JA was an author on Arcand J et al (2016). ES, BM, AH, and RM have no conflicts of interest to declare.

Supporting information

 

Santos JA, Sparks E, Thout SR, et al. The Science of Salt: A global review on changes in sodium levels in foods. J Clin Hypertens. 2019;21:1043–1056. 10.1111/jch.13628