Effect of longer‐term modest salt reduction on blood pressure

Abstract

Background

A reduction in salt intake lowers blood pressure (BP) and, thereby, reduces cardiovascular risk. A recent meta‐analysis by Graudal implied that salt reduction had adverse effects on hormones and lipids which might mitigate any benefit that occurs with BP reduction. However, Graudal's meta‐analysis included a large number of very short‐term trials with a large change in salt intake, and such studies are irrelevant to the public health recommendations for a longer‐term modest reduction in salt intake. We have updated our Cochrane meta‐analysis.  

Objectives

To assess (1) the effect of a longer‐term modest reduction in salt intake (i.e. of public health relevance) on BP and whether there was a dose‐response relationship; (2) the effect on BP by sex and ethnic group; (3) the effect on plasma renin activity, aldosterone, noradrenaline, adrenaline, cholesterol, low‐density lipoprotein (LDL), high‐density lipoprotein (HDL) and triglycerides.

Search methods

We searched MEDLINE, EMBASE, Cochrane Hypertension Group Specialised Register, Cochrane Central Register of Controlled Trials, and reference list of relevant articles.

Selection criteria

We included randomised trials with a modest reduction in salt intake and duration of at least 4 weeks.

Data collection and analysis

Data were extracted independently by two reviewers. Random effects meta‐analyses, subgroup analyses and meta‐regression were performed.

Main results

Thirty‐four trials (3230 participants) were included. Meta‐analysis showed that the mean change in urinary sodium (reduced salt vs usual salt) was ‐75 mmol/24‐h (equivalent to a reduction of 4.4 g/d salt), the mean change in BP was ‐4.18 mmHg (95% CI: ‐5.18 to ‐3.18, I²=75%) for systolic and ‐2.06 mmHg (95% CI: ‐2.67 to ‐1.45, I²=68%) for diastolic BP. Meta‐regression showed that age, ethnic group, BP status (hypertensive or normotensive) and the change in 24‐h urinary sodium were all significantly associated with the fall in systolic BP, explaining 68% of the variance between studies. A 100 mmol reduction in 24 hour urinary sodium (6 g/day salt) was associated with a fall in systolic BP of 5.8 mmHg (95%CI: 2.5 to 9.2, P=0.001) after adjusting for age, ethnic group and BP status. For diastolic BP, age, ethnic group, BP status and the change in 24‐h urinary sodium explained 41% of the variance between studies. Meta‐analysis by subgroup showed that, in hypertensives, the mean effect was ‐5.39 mmHg (95% CI: ‐6.62 to ‐4.15, I²=61%) for systolic and ‐2.82 mmHg (95% CI: ‐3.54 to ‐2.11, I²=52%) for diastolic BP. In normotensives, the mean effect was ‐2.42 mmHg (95% CI: ‐3.56 to ‐1.29, I²=66%) for systolic and ‐1.00 mmHg (95% CI: ‐1.85 to ‐0.15, I²=66%) for diastolic BP. Further subgroup analysis showed that the decrease in systolic BP was significant in both whites and blacks, men and women. Meta‐analysis of hormone and lipid data showed that the mean effect was 0.26 ng/ml/hr (95% CI: 0.17 to 0.36, I²=70%) for plasma renin activity, 73.20 pmol/l (95% CI: 44.92 to 101.48, I²=62%) for aldosterone, 31.67 pg/ml (95% CI: 6.57 to 56.77, I²=5%) for noradrenaline, 6.70 pg/ml (95% CI: ‐0.25 to 13.64, I²=12%) for adrenaline, 0.05 mmol/l (95% CI: ‐0.02 to 0.11, I²=0%) for cholesterol, 0.05 mmol/l (95% CI: ‐0.01 to 0.12, I²=0%) for LDL, ‐0.02 mmol/l (95% CI: ‐0.06 to 0.01, I²=16%) for HDL, and 0.04 mmol/l (95% CI: ‐0.02 to 0.09, I²=0%) for triglycerides.

Authors' conclusions

A modest reduction in salt intake for 4 or more weeks causes significant and, from a population viewpoint, important falls in BP in both hypertensive and normotensive individuals, irrespective of sex and ethnic group. With salt reduction, there is a small physiological increase in plasma renin activity, aldosterone and noradrenaline. There is no significant change in lipid levels. These results provide further strong support for a reduction in population salt intake. This will likely lower population BP and, thereby, reduce cardiovascular disease. Additionally, our analysis demonstrates a significant association between the reduction in 24‐h urinary sodium and the fall in systolic BP, indicating the greater the reduction in salt intake, the greater the fall in systolic BP. The current recommendations to reduce salt intake from 9‐12 to 5‐6 g/d will have a major effect on BP, but are not ideal. A further reduction to 3 g/d will have a greater effect and should become the long term target for population salt intake.

Plain language summary

Modest salt reduction lowers blood pressure in all ethnic groups at all levels of blood pressure without adverse consequences

The public health recommendations in most countries are to reduce salt intake from the current levels of approximately 9‐12 grams per day to less than 5‐6 grams per day. Our pooled analysis of randomised trials of 4 weeks or more in duration shows that such a reduction in salt intake lowers blood pressure both in individuals with raised blood pressure and in those with normal blood pressure. The fall in blood pressure is shown in both whites and blacks, men and women. Additionally, our results show that a longer‐term modest reduction in salt intake has no adverse effect on hormone and lipid levels. These findings provide further strong support for a reduction in population salt intake. This will likely lower population blood pressure and reduce strokes, heart attacks and heart failure. Furthermore, our results are consistent with the fact that the lower the salt intake, the lower the blood pressure. The current recommendations to reduce salt intake to 5‐6 grams per day will lower blood pressure, but a further reduction to 3 grams per day will lower blood pressure more. Therefore, 3 grams per day should become the long‐term target for population salt intake.

Summary of findings

Summary of findings for the main comparison. Change in systolic and diastolic blood pressure (SBP, DBP) from usual to reduced salt intake in hypertensive and normotensive individuals.

	Outcomes	Number of Trials	Number of Participants	Median BP on usual salt (mmHg)	Mean change in BP* (mmHg) with salt reduction [95% CI], P	Quality of the evidence (GRADE)
All trials together Duration of salt reduction: Median: 4 weeks (range 4 weeks to 3 years) Mean reduction in UNa: 75 mmol/24h (equivalent to 4.4 g/d salt)	SBP	33	3206	141	‐4.18 [‐5.18, ‐3.18], P<0.00001	⊕⊕⊕⊕ High
	DBP	34	3230	86	‐2.06 [‐2.67, ‐1.45], P<0.00001	⊕⊕⊕⊕ High
Hypertensives Duration of salt reduction: Median 5 weeks (range 4 weeks to 1 year) Mean reduction in UNa: 75 mmol/24h (equivalent to 4.4 g/d salt)	SBP	21	966	148	‐5.39 [‐6.62, ‐4.15], P<0.00001	⊕⊕⊕⊕ High
	DBP	22	990	93	‐2.82 [‐3.54, ‐2.11], P<0.00001	⊕⊕⊕⊕ High
Normotensives Duration of salt reduction: Median 4 weeks (range 4 weeks to 3 years) Mean reduction in UNa: 75 mmol/24h (equivalent to 4.4 g/d salt)	SBP	12	2240	127	‐2.42 [‐3.56, ‐1.29], P<0.0001	⊕⊕⊕⊕ High
	DBP	12	2240	77	‐1.00 [‐1.85, ‐0.15], P=0.02	⊕⊕⊕⊕ High

* Negative value indicates that the effect favours reduced salt. UNa: urinary sodium.

Summary of findings 2. Change in 24h urinary sodium (UNa) and blood pressure (BP) in hypertensive and normotensive individuals by ethnic group.

	Number of Trials	Number of Participants	Mean effect [95% CI] mmHg	P value
Hypertensive whites
SBP (mmHg)	16	599	‐5.12 [‐6.27, ‐3.96]	<0.00001
DBP (mmHg)	17	623	‐2.66 [‐3.37, ‐1.95]	<0.00001
24h UNa (mmol)	17	623	‐77.44 [‐85.22, ‐69.66]	<0.00001
Hypertensive blacks
SBP (mmHg)	5	171	‐7.83 [‐10.96, ‐4.71]	<0.00001
DBP (mmHg)	5	171	‐4.08 [‐5.90, ‐2.26]	<0.0001
24h UNa (mmol)	5	171	‐66.87 [‐82.79, ‐50.95]	<0.00001
Hypertensive Asians
SBP (mmHg)	1	29	‐5.41 [‐9.27, ‐1.56]	0.008
DBP (mmHg)	1	29	‐2.17 [‐4.31, ‐0.03]	0.047
24h UNa (mmol)	1	29	‐68.42 [‐89.19, ‐47.64]	<0.001
Normotensive whites
SBP (mmHg)	12	1901	‐2.11 [‐3.03, ‐1.19]	<0.00001
DBP (mmHg)	12	1901	‐0.88 [‐1.68, ‐0.08]	0.03
24h UNa (mmol)	12	1901	‐76.45 [‐89.52, ‐63.38]	<0.00001
Normotensive blacks
SBP (mmHg)	3	412	‐4.02 [‐7.44, ‐0.61]	0.02
DBP (mmHg)	3	412	‐1.98 [‐4.45, 0.49]	0.12
24h UNa (mmol)	3	412	‐40.31 [‐97.16, 16.55]	0.16

SBP: Systolic blood pressure. DBP: Diastolic blood pressure. UNa: Urinary sodium.

Summary of findings 3. Change in 24h urinary sodium (UNa) and blood pressure (BP) in hypertensive and normotensive individuals by sex.

	Number of Trials	Number of Participants	Mean effect [95% CI] mmHg	P value
Hypertensive men
SBP (mmHg)	9	227	‐6.40 [‐8.00, ‐4.80]	<0.00001
DBP (mmHg)	10	239	‐3.96 [‐5.47, ‐2.46]	<0.00001
24h UNa (mmol)	10	239	‐86.07 [‐100.17, ‐71.97]	<0.00001
Hypertensive women
SBP (mmHg)	9	181	‐7.11 [‐8.81, ‐5.41]	<0.00001
DBP (mmHg)	10	193	‐3.41 [‐4.29, ‐2.53]	<0.00001
24h UNa	10	193	‐69.56 [‐77.56, ‐61.55]	<0.00001
Normotensive men
SBP (mmHg)	6	1391	‐3.39 [‐5.63, ‐1.16]	0.003
DBP (mmHg)	6	1391	‐1.78 [‐3.01, ‐0.55]	0.005
24h UNa (mmol)	6	1391	‐67.26 [‐81.90, ‐52.62]	<0.00001
Normotensive women
SBP (mmHg)	6	691	‐4.26 [‐6.20, ‐2.31]	<0.0001
DBP (mmHg)	6	691	‐2.18 [‐2.95, ‐1.41]	<0.00001
24h UNa (mmol)	6	691	‐62.98 [‐88.59, ‐37.37]	<0.00001

SBP: Systolic blood pressure. DBP: Diastolic blood pressure. UNa: Urinary sodium.

Summary of findings 4. Change in plasma renin activity, aldosterone, noradrenaline and adrenaline.

	Number of Trials	Number of Participants	Median value on usual salt	Mean change with salt reduction [95% CI]	P value
Plasma renin activity (ng/ml/hr)	14	455	1.07	0.26 [0.17, 0.36]	<0.00001
Aldosterone (pmol/l)	9	340	299	73.20 [44.92, 101.48]	<0.00001
Noradrenaline (pg/ml)	6	129	351	31.67 [6.57, 56.77]	0.01
Adrenaline (pg/ml)	4	84	64	6.70 [‐0.25, 13.64]	0.06

Summary of findings 5. Change in plasma lipids.

	Number of Trials	Number of Participants	Median value on usual salt	Mean change with salt reduction [95% CI]	P value
Cholesterol (mmol/l)	8	365	5.3	0.05 [‐0.02, 0.11]	0.18
Low‐density lipoprotein (LDL) (mmol/l)	5	262	3.2	0.05 [‐0.01, 0.12]	0.11
How‐density lipoprotein (HDL) (mmol/l)	6	278	1.3	‐0.02 [‐0.06, 0.01]	0.19
Triglycerides (mmol/l)	6	309	1.3	0.04 [‐0.02, 0.09]	0.22

Background

The current public health recommendations in most countries are to reduce salt intake from approximately 9‐12 g/d to 5‐6 g/d (WHO 2003; SACN 2003). There is much evidence demonstrating that such a reduction in salt intake lowers blood pressure (BP). The evidence comes from different types of studies including epidemiological, migration, population‐based intervention, genetic and animal studies, as well as treatment trials (Elliott 1996; Poulter 1990; Forte 1989; Lifton 1996; Denton 1995; He 2002). As raised BP throughout its range is a major cause of cardiovascular disease, a reduction in salt intake lowers BP and, therefore, would reduce cardiovascular risk. Indeed, both prospective cohort studies and outcome trials have demonstrated that a lower salt intake is related to a reduced risk of cardiovascular disease (Strazzullo 2009;He 2011).

Despite the evidence above, a recent meta‐analysis by Graudal et al (Graudal 2011; Graudal 2012) implied that salt reduction had adverse effects on hormones and lipids which might mitigate any benefit that occurs with the reduction in BP. However, Graudal et al's meta‐analysis (Graudal 2011; Graudal 2012) is flawed from a public health perspective, as they included a large number of very short‐term trials with a large change in salt intake, e.g. from 20 to less than 1 g/d for only 4‐5 days, and such metabolic studies are irrelevant to the current public health recommendations for a modest reduction in salt intake for a long period of time. We have updated our Cochrane meta‐analysis to determine the effects of a longer‐term modest reduction in salt intake (i.e. of public health relevance) on BP, plasma renin activity, aldosterone, noradrenaline, adrenaline, cholesterol, low‐density lipoprotein (LDL), high‐density lipoprotein (HDL) and triglycerides, as well as further sub‐group analyses to study the effects of salt reduction on BP by ethnic group and sex.    

Objectives

Our systematic review aimed to determine the effect of a longer‐term modest reduction in salt intake on BP in both hypertensive and normotensive individuals and to assess whether there was a dose‐response to salt reduction. We also assessed the effect of salt reduction on BP by ethnic group and sex. Furthermore, we aimed to study the effect of salt reduction on plasma renin activity, aldosterone, noradrenaline, adrenaline, cholesterol, LDL, HDL and triglycerides.

Methods

Criteria for considering studies for this review

Types of studies

For inclusion, trials needed to satisfy the following criteria:

1. Random allocation either to a modestly reduced salt intake or usual salt intake (i.e. control).

2. No concomitant interventions (i.e. nonpharmacologic interventions, antihypertensive or other medications) in either group.

3. The reduction in 24‐h urinary sodium must be within the range of 40 to 120 mmol (i.e. 2.3 to 7.0 g/d salt). The reduction in 24‐h urinary sodium was calculated as UNa (Post) ‐ UNa (Pre) for crossover trials, where UNa (Post) designated to the average 24‐h urinary sodium at the end of the reduced salt intake period and UNa (Pre) designated to the average 24‐h urinary sodium at the end of the usual salt intake period (i.e. control period). In parallel trials the change in urinary sodium was calculated as {[UNa (Post) ‐ UNa (Pre)] reduced salt group} ‐ {[UNa (Post) ‐ UNa (Pre)] usual salt group}, where UNa (Post) designated to the average 24‐h urinary sodium at the end of follow‐up and UNa (Pre) designated to the average 24‐h urinary sodium at baseline.

4. Duration of salt reduction must have been for 4 or more weeks.

Types of participants

Studies of adults (18 years or older) with normal or raised BP, irrespective of gender and ethnicity, were included. Trials in children, pregnant women, or patients with other diseases rather than hypertension, such as diabetes, heart failure, were excluded.

Types of interventions

The intervention included was to reduce salt intake. Studies with concomitant interventions (i.e. nonpharmacologic interventions, antihypertensive or other medications) were excluded. One trial with factorial design (i.e. the Trials of Hypertension Prevention, Phase II) was included (TOHP II 1997), however, in this trial the low salt arm (without weight intervention) was compared to the control group (without salt and without weight intervention).

Types of outcome measures

The main outcome measures extracted from each individual trial were the changes in systolic and diastolic BP, and 24h urinary sodium excretion. These were calculated as the differences between the reduced salt and the usual salt groups for mean change from baseline for parallel trials. For crossover trials, the changes were calculated as the mean differences between the end of reduced salt and the usual salt period. Other outcome measures included plasma renin activity, aldosterone, noradrenaline, adrenaline, cholesterol, LDL, HDL and triglycerides.

Search methods for identification of studies

In our first meta‐analysis (He 2002), we developed a search strategy to search for randomised salt reduction trials. In this current update, our original search strategy was modified by Douglas Salzwedel, Trials Search Coordinator at the Cochrane Hypertension Group. Using the updated strategy, the following electronic databases were searched:

• The Cochrane Hypertension Group Specialised Register 1948 to November 2012;

• The Cochrane Central Register of Controlled Trials (CENTRAL) Issue 11, 2012;

• Ovid MEDLINE(R) 1946 to November 2012;

• Ovid Embase 1974 to November 2012

Additionally, we reviewed reference list of relevant original and review articles to search for more trials. There were no language restrictions. Electronic databases were searched using a strategy combining the Cochrane Highly Sensitive Search Strategy for identifying randomised trials in MEDLINE with selected MeSH and free text terms for salt and blood pressure. The MEDLINE search strategy (Appendix 1) was translated into Embase (Appendix 2), CENTRAL (Appendix 3), and the Hypertension Group Specialised Register (Appendix 4) using the appropriate controlled vocabulary as applicable.

Data collection and analysis

Data Extraction: Data were extracted independently by two persons (F.J. He and J.F. Li) using a standard form and differences were resolved by discussion with a third reviewer (G.A. MacGregor). Relevant data recorded were characteristics of the study, design (parallel or crossover), type of the study (open, single‐blind or double‐blind), method of randomisation, method of blinding (use of placebo, random‐zero or automated sphygmomanometers, or BP observer‐blind), study duration, pre‐ and post‐intervention results. For the purpose of pooled analyses, statistics that could be used to estimate the variance of the outcome measures were also recorded.

Statistical Analyses: For each trial, we calculated the treatment effect for systolic and diastolic BP, and other outcome measures. For crossover trials, the treatment effect was the difference in outcomes between the end of reduced salt period and the end of usual salt (i.e. control) period. For parallel trials, the treatment effect was the difference between the two treatment groups in the change in outcomes from baseline to the end of follow‐up.

For each trial, we also calculated the variance of the treatment effect for outcomes. This was derived from standard deviations or standard errors of paired differences between baseline and the end of follow‐up for each group in a parallel trial (Cappuccio 1991) or between the two treatment periods in a crossover trial, or if these statistics were not given, from confidence intervals, exact t or P values. If the exact variance of paired difference was not derivable, it was imputed either by inverting a boundary P value (e.g. P<0.05 became P=0.05) or assuming a correlation coefficient of 0.5 between the initial and final measurement (Follmann 1992).

To assess the mean effect sizes, we pooled the data by the inverse variance method in random‐effects meta‐analysis. We used the I² test to examine heterogeneity, with I²>50% considered to be important (Higgins 2003). To explore the source of heterogeneity, we performed meta‐regression analyses (multiple regression models) weighted by the inverse variance of the change in systolic or diastolic BP. The meta‐regression analysis was also used to examine whether there was a dose‐response relationship between the change in 24‐h urinary sodium and the change in BP. We used funnel plot asymmetry to detect whether there was publication bias and Egger's regression test to measure funnel plot asymmetry (Egger 1997; Sterne 2001).

Prespecified subgroupings included BP status (i.e. hypertensive or normotensive) and further subgroupings by ethnic group and sex. The purpose of the subgroup analysis was to determine whether there was a significant effect of salt reduction on BP in each group itself rather than identifying difference in the effect between groups. For the analysis stratified by ethnic group, trials were included in the group of "white" if ≥85% of participants were white. If the information on ethnic group was not available, the trial was excluded from this subgroup analysis. For hormone and lipid data, subgroup analyses were not performed because of the small number of trials that reported such outcomes.

Statistical analyses were performed using Cochrane Collaboration Review Manager 5.2 software and the Statistical Package for the Social Sciences (SPSS).

Results

Description of studies

The search strategy identified 3252 citations, of which we excluded 2995 on the basis of abstract and title. A detailed assessment was given to 257 papers, of which 227 were excluded and the reasons for exclusion were summarised in PRISMA Flow Diagram (Figure 1).

1.

PRISMA  Flow  Diagram

A total of 30 papers met our inclusion criteria and were included in our meta‐analysis. Among these 30 papers, 4 included both hypertensive and normotensive individuals (Sacks 2001 (H); Sacks 2001 (N); Puska 1983 (H); Puska 1983 (N); Cappuccio 1997 (H); Cappuccio 1997 (N); Melander 2007 (H); Melander 2007 (N)). In our meta‐analysis, the main outcome data (i.e. BP) were recorded for hypertensives and normotensives separately. To avoid confusion in counting the number of trials, each of these 4 papers was counted as 2 trials. Therefore, our meta‐analysis included a total of 34 trials, of which 22 were in hypertensive individuals and 12 in normotensive individuals.

In 3 studies (Morgan 1981 (F); Morgan 1981 (M); Nestel 1993 (F); Nestel 1993 (M); Watt 1985 (HH) offspring of two parents with high BP; Watt 1985 (LL) offspring of two parents with low BP) where subgroup data were reported only, they were entered for subgroups separately. For 3 trials that included both hypertensives and normotensives (Sacks 2001 (H); Sacks 2001 (N); Cappuccio 1997 (H); Cappuccio 1997 (N); Melander 2007 (H); Melander 2007 (N)), each had an additional entry for all participants, i.e. hypertensives and normotensives combined, where the data for lipids or hormones were reported. Therefore, there were a total of 40 entries in the table of Characteristics of included studies.

For 2 papers (MacGregor 1989; Sacks 2001 (H); Sacks 2001 (N)) where 3 levels of salt intakes were studied, we included the high and intermediate levels (i.e. urinary sodium reduced from 190 to 108 mmol/24h) in one trial (MacGregor 1989) and in the other (Sacks 2001 (H); Sacks 2001 (N)) we included the high and low levels (i.e. urinary sodium reduced from 145 to 65 mmol/24h in hypertensive individuals and from 139 to 64 mmol/24h in normotensive individuals on the normal American diet). The characteristics of the trials included in the meta‐analysis are summarised in Table: "Characteristics of included studies".

Risk of bias in included studies

Criteria for the assessment of study quality were as follows.

Allocation concealment

• Adequate: The randomisation method described did not allow participants and investigators to foresee assignment, e.g. a prior numbered or coded tablet containers of identical appearance prepared by an independent pharmacy, central randomisation.

• Unclear: Randomisation was stated, but no detailed information was provided on the method used to generate the random allocation sequence and the mechanism used to implement the random allocation sequence.

• Inadequate: Method of randomisation allowed participants or investigators to foresee assignment, e.g. unsealed envelopes or alternate medical record numbers.

Blinding

• Blinding of the investigator: yes/no/not stated.

• Blinding of the participant: yes/no/not stated.

• Blinding of the outcome assessor: yes/no/not stated.

Incomplete outcome data addressed?

• Yes: Intention to treat analysis was undertaken, or all participants who were randomised, completed the study, or detailed information was reported on the number of participants who were lost of follow‐up after randomisation as well as reasons.

• Unclear: No information provided.

• No: Incomplete outcome data were not adequately addressed.

The risk of bias assessments for the trials included in the meta‐analysis are shown in Table: Characteristics of included studies.

Effects of interventions

See: Table 1; Table 2; Table 3; Table 4; Table 5

Effect on BP

Trials in all individuals

A total of 34 trials with 3230 participants were included. The characteristics of the trials included in the meta‐analysis are summarised in Table (Characteristics of included studies). The median age was 50 years (ranging from 22 to 73 years). Of the 34 trials, 23 used crossover design and 11 used paralleled comparisons. Twenty‐two of the 34 trials were double blind, 11 were BP observer blind, and 1 did not report any blinding procedure. The study duration varied from 4 weeks to 3 years (median: 4 weeks). The median BP on the usual salt intake was 141/86 mmHg. The median 24‐h urinary sodium on the usual salt intake was 160 mmol (9.4 g/d salt), ranging from 125 to 200 mmol (7.3 to 11.7 g/d salt). The pooled estimate of the change in 24‐h urinary sodium from the usual to the reduced salt intake was ‐75 mmol (range ‐40 to ‐118 mmol), equivalent to a reduction in salt intake of 4.4 g/d (range 2.3 to 6.9 g/d). This average reduction in salt intake is similar to that of the current public health recommendations.

Analysis 1.1; Analysis 1.2 show the change in BP in individual trials included in the meta‐analysis and the mean effect size. The pooled estimates of changes in BP were ‐4.18 mmHg (95% CI: ‐5.18 to ‐3.18, P<0.00001, I²=75%) for systolic and ‐2.06 mmHg (95% CI: ‐2.67 to ‐1.45, P<0.00001, I²=68%) for diastolic BP (Table 1).

1.1. Analysis.

Comparison 1 Change in BP, Outcome 1 Change in Systolic BP.

1.2. Analysis.

Comparison 1 Change in BP, Outcome 2 Change in Diastolic BP.

To explore the source of heterogeneity, meta‐regression analysis was performed with the change of BP (systolic or diastolic) as dependent variable and the independent variables included age (mean age of the participants in individual trials), BP status (hypertensive=1; normotensive=0), ethnic group (i.e. the proportion of whites as a continuous variable), and the change in 24‐h urinary sodium. The results showed that the change in 24‐h urinary sodium, age, BP status and ethnic group were all significantly associated with the change in systolic BP. The regression coefficients indicated that a 100 mmol reduction in 24 hour urinary sodium (6 g/d salt) was associated with a decrease of 5.8 mmHg (95% CI: 2.5 to 9.2, P=0.001) in systolic BP, a one‐year increase in age was associated with a 0.06 mmHg (95% CI: 0.006 to 0.116, P=0.030) greater decrease in systolic BP with salt reduction, being hypertensive was associated with a greater fall in systolic BP (P=0.042) compared with normotensives, and a larger proportion of whites (or a smaller proportion of blacks) was associated with a smaller fall in systolic BP (P=0.001). These 4 variables together explained 68% of the variance between studies. In a separate regression model, sex (i.e. the proportion of men as a continuous variable) was added to the independent variable list, and there was little change to the adjusted R² (R²=0.68 without sex in the regression model and R²=0.70 with sex added to the regression model).  Sex was not significantly associated with the change in systolic BP. For diastolic BP, age, ethnic group, BP status and 24‐h urinary sodium together explained 41% of the variance between studies. Among these 4 variables, only ethnic group was significant (P=0.021) and the other 3 variables were not significantly associated with the change in diastolic BP. When sex was added to the regression model, there was little change in the adjusted R² (R²=0.41 and 0.44 for the regression model with and without sex respectively). Sex was not significantly associated with the change in diastolic BP.

Trials in hypertensive individuals

Nine hundred and ninety hypertensive individuals were studied in 22 trials (Table: Characteristics of included studies). Median age was 50 years (ranging from 24 to 73 years). Of the 22 trials, 16 used crossover design and 6 used paralleled comparisons. Fourteen of the 22 trials were double blind, 7 were BP observer blind, and 1 did not report any blinding procedure. The study duration varied from 4 weeks to 1 year (median: 5 weeks). The median BP on usual salt intake was 148/93 mmHg. The median 24‐h urinary sodium on the usual salt intake was 162 mmol (9.5 g/d salt), ranging from 125 to 191 mmol (7.3 to 11.2 g/d salt). The pooled estimate of the change in 24‐h urinary sodium from the usual to the reduced salt intake was ‐75 mmol (range ‐53 to ‐117 mmol), equivalent to a reduction in salt intake of 4.4 g/d (range 3.1 to 6.8 g/d).

Analysis 1.1; Analysis 1.2 show the change in BP in individual trials included in the meta‐analysis and the mean effect size. The pooled estimates of changes in BP were ‐5.39 mmHg (95% CI: ‐6.62 to ‐4.15, P<0.00001, I²=61%) for systolic and ‐2.82 mmHg (95% CI: ‐3.54 to ‐2.11, P<0.00001, I²=52%) for diastolic BP (Table 1).

Meta‐regression with the change in BP as dependent variable and age, ethnic group and the change in 24‐h urinary sodium as independent variables, showed that the change in 24‐h urinary sodium and ethnic group were significantly associated with the fall in systolic BP, whereas age was not significantly associated with the fall in systolic BP. A 100 mmol reduction in 24 hour urinary sodium (6 g/day salt) was associated with a fall in systolic BP of 10.8 mmHg (95CI: 3.5 to 18.2, P<0.01) after adjusting for age and ethnic group. All 3 variables together explained 46% of the variance between studies. For diastolic BP, the 3 variables together explained 11% of the variance between studies and none of the 3 variables was significantly associated with the fall in diastolic BP.

Trials in normotensive individuals

Two thousand two hundred and forty individuals with normal BP were studied in 12 trials (Table: Characteristics of included studies). Median age was 50 years (ranging from 22 to 67 years). Of the 12 trials, 7 used crossover design and 5 used paralleled comparisons. Eight of the 12 trials were double blind and 4 were BP observer blind. The study duration varied from 4 weeks to 3 years (median: 4 weeks). The median BP on usual salt intake was 127/77 mmHg. The median 24‐h urinary sodium on the usual salt intake was 153 mmol (8.9 g/d salt), ranging from 128 to 200 mmol (7.5 to 11.7 g/d salt). The pooled estimate of the change in 24‐h urinary sodium from the usual to the reduced salt intake was ‐75 mmol (range ‐40 to ‐118 mmol), equivalent to a reduction in salt intake of 4.4 g/d (range 2.3 to 6.9 g/d).

Analysis 1.1; Analysis 1.2 show the change in BP in individual trials included in the meta‐analysis and the mean effect size. The pooled estimates of changes in BP were ‐2.42 mmHg (95% CI: ‐3.56 to ‐1.29, P<0.0001, I²=66%) for systolic and ‐1.00 mmHg (95% CI: ‐1.85 to ‐0.15, P=0.02, I²=66%) for diastolic BP (Table 1).

Meta‐regression with the change in BP as dependent variable and age, ethnic group and the change in 24‐h urinary sodium as independent variables, showed that the change in 24‐h urinary sodium was significantly associated with the fall in systolic BP, whereas age and ethnic group were not significantly associated with the fall in systolic BP. A 100 mmol reduction in 24 hour urinary sodium (6 g/day salt) was associated with a fall in systolic BP of 4.3 mmHg (95% CI: 0.1 to 8.5, P<0.05) after adjusting for age and ethnic group. All 3 variables together explained 51% of the variance between studies. For diastolic BP, the 3 variables together explained 43% of the variance between studies. Among these 3 variables, only ethnic group was significant (P=0.042) and the other 2 variables were not significantly associated with the change in diastolic BP.

Further sub‐group analysis

Table 2 and Table 3 show the pooled results of 24‐h urinary sodium and BP by ethnic group and sex for hypertensives and normotensives separately. There was a significant fall in systolic BP in both whites and blacks, men and women. The fall in diastolic BP was significant in most of the subgroups. There was only 1 trial in Asians (He 2009). Most of the participants in this trial were of South Asian origin (i.e. originating from the Indian subcontinent) and all participants had raised BP. The study showed that there was a significant fall in both systolic and diastolic BP with a modest reduction in salt intake (Table 2).

Effect on hormones and lipids

Plasma renin activity

Of the 34 trials, 14 reported the data of plasma renin activity. One study reported plasma renin concentration (Melander 2007) and it was excluded from the analysis for plasma renin activity. The median plasma renin activity was 1.07 ng/ml/hr on the usual salt intake. The pooled estimate of the change in plasma renin activity was 0.26 ng/ml/hr (95% CI: 0.17 to 0.36, P<0.00001, I²=70%) (Table 4).

Aldosterone

Of the 34 trials, 9 had plasma aldosterone measured. One trial (Benetos 1992) was excluded from the aldosterone analysis as the plasma aldosterone was extremely high after the unit conversion (235277.8 pmol/l on the usual salt and 269166.7 pmol/l on the reduced salt intake). The median plasma aldosterone was 299 pmol/l on the usual salt intake. The pooled estimate of the change in aldosterone was 73.20 pmol/l (95% CI: 44.92 to 101.48, P<0.00001, I²=62%) (Table 4).

Noradrenaline

Of the 34 trials, 6 reported the data of plasma noradrenaline. The median plasma noradrenaline was 351 pg/ml on the usual salt intake. The pooled estimate of the change in plasma noradrenaline was 31.67 pg/ml (95% CI: 6.57 to 56.77, P=0.01, I²=5%) (Table 4).

Adrenaline

Of the 34 trials, only 4 trials reported the data of plasma adrenaline. The median plasma adrenaline was 64 pg/ml on the usual salt intake. The pooled estimate of the change in plasma adrenaline was 6.70 pg/ml (95% CI: ‐0.25 to 13.64, P=0.06, I²=12%) (Table 4).

In addition to the above, one other trial reported that "no changes were observed in plasma catecholamines” (Schorr 1996), but data were not provided. Therefore, this trial was excluded from the above pooled analysis for noradrenaline and adrenaline.

Cholesterol

Of the 34 trials, 8 reported the data of plasma cholesterol. The median plasma cholesterol was 5.3 mmol/l on the usual salt intake. The pooled estimate of the change in cholesterol was 0.05 mmol/l (95% CI: ‐0.02 to 0.11, P=0.18, I²=0%) (Table 5). Two other trials reported no significant change in cholesterol (Chalmers 1986; Erwteman 1984), however, no data were provided for pooled analysis.

LDL

Of the 34 trials, 5 reported the data of plasma LDL. The median plasma LDL was 3.2 mmol/l on the usual salt intake. The pooled estimate of the change in LDL was 0.05 mmol/l (95% CI: ‐0.01 to 0.12, P=0.11, I²=0%) (Table 5).

HDL

Of the 34 trials, 6 reported the data of plasma HDL. The median plasma HDL was 1.3 mmol/l on the usual salt intake. The pooled estimate of the change in HDL was ‐0.02 mmol/l (95% CI: ‐0.06 to 0.01, P=0.19, 16%) (Table 5). One other trial reported no significant change in HDL, but data were not provided (Erwteman 1984), therefore, this trial was excluded from the pooled analysis.

Triglycerides

Of the 34 trials, 6 reported the data of plasma triglycerides. The median plasma triglycerides was 1.3 mmol/l on the usual salt intake. The pooled estimate of the change in triglycerides was 0.04 mmol/l (95% CI: ‐0.02 to 0.09, P=0.22, I²=0%) (Table 5).

Study quality

The risk of bias graph is shown in Figure 2. Among the 34 trials included in our meta‐analysis, 26 were judged to have adequate concealment of allocation of treatments (Table:Characteristics of included studies). In 8 trials the information on concealment of allocation was not available. Despite the fact that only 7 out of 34 trials performed intention‐to‐treat analysis, the percentage of participants who were lost of follow‐up after randomisation was small (6.7% on average).

2.

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

We included double‐blind, BP observer‐blind, and open studies due to the fact that 1) some trials e.g. the DASH (Dietary Approaches to Stop Hypertension)‐Sodium (Sacks 2001 (H); Sacks 2001 (N)), although non‐double‐blinded, were well conducted with good compliance to different diets; and 2) it is very difficult to make any dietary intervention study double‐blind. In relation to salt, this can only be done by the use of salt tablets (Slow Sodium and placebo). Among the 34 trials included in our meta‐analysis, 22 were double‐blind, 11 were BP observer‐blind and only one small trial in hypertensives was non‐blind. Re‐analysing the data by excluding the non‐blind study (Parijs 1973) showed that the results were unchanged. The mean net change in BP for hypertensive individuals was ‐5.35 mmHg (95%CI: ‐6.62 to ‐4.09) for systolic and ‐2.88 mmHg (95%CI: ‐3.58 to ‐2.18) for diastolic BP after the non‐blinded study was excluded.

Publication bias

We created the funnel plots by plotting the treatment effect against the reciprocal of the standard error of the treatment effect (Figure 3; Figure 4). For diastolic BP the funnel plots were symmetrical around the mean effect size line (asymmetry test: P=0.416) (Egger 1997). For systolic BP, the graphic plot was suggestive of bias (asymmetry test: P=0.025). This asymmetry of funnel plot might be because smaller studies showing no effect were under‐reported in the literature. However, in our meta‐analysis it is more likely to be due to the smaller effects of two larger and longer‐term trials (TOHP I 1992; TOHP II 1997). The smaller effects in these two trials are attributable to the smaller reduction of salt intake achieved in the longer‐term trials. When these two trials were removed from the analysis, the asymmetry test was not significant (P=0.247).

3.

Funnel plot to explore publication bias (systolic BP). The vertical line is at the mean effect size. Precision is the reciprocal of the standard error of the change in systolic BP.

4.

Funnel plot to explore publication bias (diastolic BP). The vertical line is at the mean effect size. Precision is the reciprocal of the standard error of the change in diastolic BP.

Discussion

Our meta‐analysis demonstrates that a longer‐term modest reduction in salt intake of 4.4 g/d on average, causes significant and, from a population viewpoint, important falls in BP in individuals with both raised and normal BP. The BP falls, on average, by 5/3 mmHg in hypertensives and 2/1 mmHg in normotensives. Further subgroup analyses demonstrate that a modest reduction in salt intake leads to a significant fall in systolic BP in both whites and blacks, men and women. These results provide further strong support for a reduction in population salt intake which will result in a lower population BP and, thereby, a reduction in strokes, heart attacks and heart failure.

The effect of a chronic high salt intake is a gradual increase in BP throughout life.  The INTERSALT study (International Study of Salt and Blood Pressure) suggested a strong relationship between salt intake and a progressive increase in BP with age, i.e. 0.4 mmHg per year for a 6 g/d salt intake (Elliott 1996). A reduction in salt intake is, therefore, likely to attenuate the rise of BP with aging, in addition to the immediate BP‐lowering effect.

Dose‐response to salt reduction

Our meta‐regression analysis shows a significant dose‐response relationship between the reduction in salt intake and the fall in systolic BP, i.e. the greater the reduction in salt intake, the greater the fall in systolic BP. A reduction of 6 g/d in salt intake predicts a decrease of 5.8 mmHg in systolic BP after adjusting for age, ethnic group and BP status. 

It is acknowledged that the does‐response relationship from meta‐regression (i.e. between‐study investigation) should be viewed as exploratory and may be potentially prone to confounding. Meta‐analysis with individual participant data would have an advantage both statistically and clinically (Riley 2010; Berlin 2002) and, if available, should be used in the future to explore the dose‐response relationship further. Nevertheless, the dose‐response relationship found in our study is consistent with that observed from rigorously controlled trials with multi‐levels of salt intake which provided the most persuasive evidence. There have been two trials that studied three levels of salt intake. The first one was the randomised double‐blind cross‐over study in 20 individuals with untreated essential hypertension, where salt intake was reduced from 11.2 to 6.4 and to 2.9 g/d, each for one month (MacGregor 1989). BP was 163/100 mmHg with a salt intake of 11.2 g/d, and reduced to 155/95 mmHg when salt intake was decreased to 6.4 g/d (i.e. a decrease of 8/5 mmHg). BP fell further to 147/91 mmHg when salt intake was reduced to 2.9 g/d (i.e. a further fall of 8/4 mmHg). After the trial was completed, individuals continued the lowest salt intake. Among the 20 participants, 19 were followed up for 1 year. In 16 individuals, BP remained controlled without any antihypertensive medication and the average BP was 142/87 mmHg with a salt intake of 3.2 g/d (MacGregor 1989). The other trial that has studied the dose‐response relationship is the DASH‐Sodium study. Over 400 individuals with normal or mildly raised BP were randomised to receive either the normal American diet (control group) or the DASH diet which is rich in fruits, vegetables, and low‐fat dairy products. Within each group, participants were given 3 levels of salt intake (i.e. 8 to 6 and 4 g/d) in a randomised crossover manner, each for 4 weeks. The results demonstrated a clear dose‐response relationship both on the normal American diet and on the DASH diet. The fall in BP was greater at a lower level of salt intake, i.e. from 6 to 4 g/d compared with that from 8 to 6 g/d (Sacks 2001 (H); Sacks 2001 (N)).

From the evidence above, it is clear that the recommendations to reduce salt from the current levels of approximately 9‐12 g/d to 5‐6 g/d will have a significant effect on BP, but are not ideal. A further reduction to 3 g/d will have a much greater effect on BP. Therefore, 3 g/d should become the long‐term target for population salt intake. Indeed, the UK government's health advisory agency, the National Institute for Health and Clinical Excellence (NICE) has recommended a reduction in the population's salt consumption to 3 g/d by 2025 (NICE 2010). In USA, it is recommended that sodium intake should be reduced to less than 2.3 g/d (i.e. ≈6 g/d salt) for adults, with an even further reduction to 1.5 g/d (i.e. ≈4 g/d salt) for about half the population, including African Americans, all adults 51 and older, and those with hypertension, diabetes or chronic kidney disease (IOM 2010).

Study duration

In spite of including studies of 1 month or more, the median duration of salt reduction in our meta‐analysis was only 5 weeks in the hypertensives and 4 weeks in the normotensives. Whether salt reduction has exerted its maximum effect by 4‐5 weeks is not known, but much evidence would suggest that this is unlikely (Forte 1989). Among the 34 trials included in our meta‐analysis, two had duration of over 1 year (TOHP I 1992; TOHP II 1997) and both trials were in normotensive individuals. These two trials did not show a greater fall in BP compared with other trials in normotensive individuals. However, the reduction in salt intake achieved in these two trials was half that achieved in other trials. On average, salt intake was reduced by 2.4 g/d in these two longer‐term trials, whereas in the other trials in normotensives, salt intake was reduced by 4.8 g/d. These longer‐term studies clearly highlight the difficulty in keeping individuals on a lower salt intake due to the widespread presence of salt in nearly all processed, canteen and restaurant food.

Variations of BP response to salt reduction

Previous studies have shown that, for a given reduction in salt intake, the fall in BP was larger in individuals of African origin, in older people and in those with raised BP compared to whites, young people and individuals with normal BP respectively (Bray 2004; He 1998; He 2001). The results from our meta‐regression analyses are consistent with these observations.

The term "salt sensitivity" has been commonly used to describe the variations of BP response to salt reduction. However, almost all of the studies on "salt sensitivity" have used a protocol of very large and sudden changes in salt intake. Such studies are irrelevant to the public health recommendations of more modest reduction in salt intake for a prolonged period of time. Our meta‐analysis demonstrates that a longer‐term modest reduction in salt intake has a significant effect on BP in both hypertensive and normotensive individuals, men and women, whites and blacks; although there is a variation in the extent of the fall in BP. These results in conjunction with other evidence (He 2010), particularly that a reduction in salt intake also lowers blood pressure in children (He 2006), provide strong support that  salt reduction should be carried out in the whole population. A reduction in population salt intake lowers population BP. Even a small reduction of BP across the entire population would have a large impact on reducing the burden of cardiovascular disease (Whelton 2002).

Effect of salt reduction on hormones and lipids

A recent meta‐analysis by Graudal et al (Graudal 2011; Graudal 2012) implied that salt reduction had adverse effects on plasma hormone and lipid levels which might mitigate any benefit that occurs with a long‐term fall in BP. However, Graudal et al’s meta‐analysis included a large number of very short‐term trials with a large change in salt intake, e.g. from 20 to less than 1 g/d for only 4‐5 days, and such metabolic studies are irrelevant to the current public health recommendations for a modest reduction in salt intake for a long period of time. Our meta‐analysis demonstrates that, with a longer‐term modest reduction in salt intake, there is no significant change in plasma cholesterol, LDL, HDL or triglycerides. Indeed, in Graudal et al's own meta‐analysis, the changes in lipids only occurred with short term trials, and a sub‐group analysis including  trials with a duration of 4 or more weeks showed no significant change in lipid levels (Graudal 2011; Graudal 2012).

When salt intake is reduced, there is a fall in extracellular volume and physiological stimulation of the renin‐angiotensin‐aldosterone system, as well as the sympathetic nervous system. These compensatory responses are bigger with sudden and large decreases in salt intake, and much smaller or minimal with a longer‐term modest salt reduction. Our meta‐analysis shows that, with a longer‐term modest reduction in salt intake, there is only a small physiological increase in plasma renin activity, aldosterone and noradrenaline. It is worth noting that all of the studies that were included in our meta‐analysis with these hormones measured, had a duration of only 4‐6 weeks (median duration: 4 weeks). It is likely that such effects may attenuate over time. Indeed, a study by Beckmann et al demonstrated that a modest reduction in salt intake, along with a reduction in body weight and saturated fat for one year, significantly reduced arterial plasma noradrenaline and adrenaline in hypertensive individuals (Beckmann 1995).

Salt reduction lowers BP by a similar mechanism to that of thiazide diuretics. Both stimulate the renin‐angiotensin system and, in the short term, the sympathetic nervous system. However, outcome trials have demonstrated that long‐term treatment with thiazide diuretics significantly reduced cardiovascular morbidity and mortality in hypertensive individuals (ALLHAT 2002).

Effect of salt reduction on cardiovascular risk

There is much evidence that raised BP throughout its range starting at 115/75 mmHg is a major cause of cardiovascular disease (PSC 2002). A modest reduction in salt intake lowers BP and, therefore, would reduce cardiovascular risk. It was estimated that a reduction of 6 g/d in salt intake would reduce stroke by 24% and coronary heart disease by 18% (He 2003). This would prevent ≈35,000 stroke and coronary heart disease deaths a year in the UK and ≈2.5 million deaths worldwide.

Both prospective cohort studies and outcome trials have shown that a lower salt intake is related to a reduced risk of cardiovascular disease (Strazzullo 2009;He 2011). Two recent papers in JAMA (Journal of the American Medical Association), however, claimed that a lower salt intake was associated with higher cardiovascular mortality (Stolarz‐Skrzypek 2011) or a J‐shaped association existed between salt intake and cardiovascular risk (O'Donnell 2011). These two papers have many methodological flaws, e.g. measurement error in assessing daily salt intake, confounding factors not controlled for, and reverse causality (i.e. the low salt intake is the result rather than the cause of participants' illness) (He 2011a; He 2012). Therefore, the results from these studies should be interpreted with great caution. A meta‐analysis of 12 cohort studies showed that an increase of 5 g/d in salt intake was associated with a 23% increase in the risk of stroke and a 17% increase in the risk of cardiovascular disease (Strazzullo 2009).

Evidence from outcome trials of long term salt reduction is very limited due to the innate difficulty in conducting such trials. A recent meta‐analysis of 7 randomised trials by Taylor et al, published simultaneously in The Cochrane Library (Taylor 2011a) and the American Journal of Hypertension (Taylor 2011), claimed that "Cutting down on the amount of salt has no clear benefits in terms of likelihood of dying or experiencing cardiovascular disease"  and The Cochrane Library’s press release headline stated “Cutting down on salt does not reduce your chance of dying” (Cochrane 2011). Both of these statements are incorrect. Despite this, these headline grabbing statements received very misleading worldwide media publicity.

Among the 7 trials included in Taylor et al’s meta‐analysis, one in heart failure should not have been included as the participants were severely salt and water depleted due to aggressive diuretic therapy (Paterna 2008). Additionally, the findings in patients with severe heart failure on multiple drug treatments are not generalisable to the general population. In the remaining 6 trials, there is a reduction in all clinical outcomes (all‐cause mortality, cardiovascular mortality and events), although none of these are statistically significant. The non‐significant findings are most likely due to a lack of statistical power, particularly as Taylor et al analysed the trials for hypertensives and normotensives separately. A re‐analysis of the data by combining hypertensives and normotensives together shows that there is a significant reduction in cardiovascular events by 20% (P<0.05) (Figure 5) and a non‐significant reduction in all‐cause mortality (5‐7%), in spite of the small reduction in salt intake of 2.0‐2.3 g/d (He 2011).  These results add strongly to the evidence that salt reduction has a major impact on reducing strokes, heart attacks and heart failure.

5.

Cardiovascular disease (CVD) events in a meta‐analysis of randomised salt reduction trials using fixed effect model with normotensives and hypertensives combined. TOHP I: Trial of Hypertension Prevention, phase 1. TOHP II: Trial of Hypertension Prevention, phase 2. TONE: Trial of Nonpharmacologic Interventions in Elderly.

Salt reduction is one of the most cost‐effective public health measures to reduce cardiovascular disease

Several studies have shown that a reduction in salt intake is one of the most cost‐effective interventions to reduce cardiovascular disease in both developed and developing countries. For instance, a recent study in the US showed that even a very modest reduction in salt intake of only 10% which could be easily achieved, as demonstrated in the UK, would prevent hundreds of thousands of strokes and heart attacks over the lifetimes of adults aged 40‐85 years who are alive today, and could save more than $32 billion in medical expenses in the US alone (Smith‐Spangler 2010). A larger decrease in salt intake would result in a larger health improvement and greater cost savings (Bibbins‐Domingo 2010).

The UK salt reduction campaigns which started in 2003/2004 have been successful and the average salt intake, as measured by 24‐hour urinary sodium, has fallen gradually from 9.5 to 8.1 g/d by 2011 (i.e. a 15% reduction, P<0.05 for the downward trend) (FSA 2008). A cost‐effective analysis by NICE showed that the UK salt reduction campaigns cost £15 million and a 0.9 g/d reduction in salt intake that was achieved by 2008, led to ≈6000 fewer CVD deaths per year, saving the UK economy ≈£1.5 billion per annum (NICE 2010). Based on NICE’s estimation, the further reduction of 0.5 g/d from 2008 to 2011, would prevent approximately additional 3000 CVD deaths per year and result in even greater cost savings to the UK economy.   

Asaria et al estimated the effects and cost of strategies to reduce salt intake and control tobacco use for 23 low‐ and middle‐income countries that account for 80% of chronic disease burden in the developing world.  They demonstrated that a 15% reduction in mean population salt intake could avert 8.5 million cardiovascular deaths and a 20% reduction in smoking prevalence could avert 3.1 million cardiovascular deaths over 10 years (Asaria 2007). The modest reduction in salt intake could be achieved by a voluntary reduction in the salt content of processed foods and condiments by manufacturers combined with a sustained mass‐media campaign aimed to encourage dietary change within households and communities. The main costs of the strategy to reduce salt consumption would be awareness campaigns through mass‐media outlets and regulation of food products by public‐health officers, with an average cost estimated to be US$0.09 per person per year.  The cost for tobacco control, including both price and non‐price measures, was US$0.26 per person per year. These figures clearly suggest that a reduction in salt intake is more, or at the very least just, as cost‐effective as tobacco control in terms of reducing cardiovascular disease on its own, the leading cause of death and disability worldwide. 

Authors' conclusions

Implications for practice.

Our meta‐analysis demonstrates that a modest reduction in salt intake, as currently recommended, has a significant effect on BP both in individuals with raised BP and in those with normal BP. The fall in BP is observed in both whites and blacks, men and women. These findings provide further strong support for a reduction in population salt intake. This will likely lower population BP and, thereby, likely reduce strokes, heart attacks and heart failure. Furthermore, our analysis demonstrates a dose‐response relationship, i.e. the greater the reduction in salt intake, the greater the fall in BP. The current recommendations to reduce salt intake to 5‐6 g/d will have a major effect on BP, but are not ideal. A further reduction to 3 g/d will have a greater effect. Therefore, 3 g/d should become the long‐term target for population salt intake. Indeed, NICE has recommended a reduction in salt intake to 3 g/d by 2025 for UK adult population (NICE 2010).

Many developed countries are now adopting a policy of reducing salt intake, firstly by persuading the food industry to reformulate food with less salt, as is occurring successfully in the UK (FSA 2008) and Finland (Karppanen 2006), and also encouraging people to use less salt in their own cooking and at the table. The major challenge now is to spread this out to all other countries, particularly developing countries where often salt intake is high and ≈80% of the global BP‐related disease burden occurs. All countries should adopt a coherent and workable strategy to reduce salt intake. A reduction in population salt intake will likely have major beneficial effects on health along with major cost savings in all countries around the world.

Implications for research.

The evidence that relates salt intake to BP is very strong. The mechanisms whereby salt raises BP are not fully understood. The existing concepts focus on the tendency for an increase in extracellular fluid volume. Increasing evidence suggests that small increases in plasma sodium may have a direct effect on BP independent of extracellular volume (Friedman 1990; de Wardener 2004; He 2005). Further studies are needed to investigate the mechanisms, in particular, the role of plasma sodium in regulating BP.

What's new

Date	Event	Description
28 February 2013	New search has been performed	A new search was performed using a search strategy modified from our previous one. Two new trials (one in hypertensives and the other included both hypertensive and normotensive individuals) that met our inclusion criteria have been added to the current update. Meta‐regression (multiple regression model) was performed to explore the source of heterogeneity and also to examine whether there was a dose‐response relationship between the change in 24‐h urinary sodium and the change in blood pressure (BP). Further subgroup analyses were performed on BP, i.e. stratified by ethnic group and sex, for hypertensives and normotensives respectively. Pooled analyses were carried out for plasma renin activity, aldosterone, noradrenaline, adrenaline, cholesterol, low‐density lipoprotein (LDL), high‐density lipoprotein (HDL) and triglycerides. Summary results: Our updated review confirms that a longer‐term modest reduction in salt intake lowers blood pressure significantly in both hypertensive and normotensive individuals, and the greater the reduction in salt intake, the greater the fall in blood pressure. Compared with our previous review, our current update demonstrates that the effects of salt reduction on systolic blood pressure are significant in both whites and blacks, men and women. Furthermore, our updated review demonstrates that, with a longer‐term modest reduction in salt intake, there is only a small physiological increase in plasma renin activity, aldosterone and noradrenaline. There is no significant change in adrenaline, cholesterol, low‐density lipoprotein (LDL), high‐density lipoprotein (HDL) or triglycerides. These findings provide further strong support for a reduction in population salt intake. This will likely lower population blood pressure and reduce strokes, heart attacks and heart failure.
28 February 2013	New citation required and conclusions have changed	2013 update

History

Protocol first published: Issue 3, 2004
 Review first published: Issue 3, 2004

Date	Event	Description
30 March 2011	Amended	Converted to new review format.
24 May 2006	New search has been performed	Minor update
9 May 2005	New citation required but conclusions have not changed	A repeated search using the search strategy developed previously (Journal of Human Hypertension 2002) was carried out in April 2005. Three new trials met the inclusion criteria and have been added to the meta‐analysis.

Appendices

Appendix 1. MEDLINE search strategy

Database: Ovid MEDLINE(R) 1946 to Present with Daily Update 
 Search Date: 11 December 2012 
 ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ 
 1 sodium chloride, dietary/ 
 2 exp sodium, dietary/ 
 3 diet, sodium‐restricted/ 
 4 ((sodium or salt) adj3 (restrict$ or curb$ or limit$ or minimi$ or low$ or reduc$ or intake or diet$ or free)).tw. 
 5 or/1‐4 
 6 randomized controlled trial.pt. 
 7 controlled clinical trial.pt. 
 8 randomized.ab. 
 9 placebo.ab. 
 10 clinical trials as topic/ 
 11 randomly.ab. 
 12 trial.ti. 
 13 or/6‐12 
 14 animals/ not (humans/ and animals/) 
 15 13 not 14 
 16 5 and 15

Appendix 2. EMBASE search strategy

Database: Embase <1974 to 2012 Week 49> 
 Search Date: 11 December 2012 
 ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ 
 1 sodium chloride, dietary/ 
 2 sodium intake/ 
 3 sodium restriction/ 
 4 ((sodium or salt) adj3 (restrict$ or curb$ or limit$ or minimi$ or low$ or reduc$ or intake or diet$ or free)).tw. 
 5 or/1‐4 
 6 randomized controlled trial/ 
 7 crossover procedure/ 
 8 double‐blind procedure/ 
 9 random$.tw. 
 10 (crossover$ or cross‐over$).tw. 
 11 placebo$.tw. 
 12 (doubl$ adj blind$).tw. 
 13 assign$.tw. 
 14 allocat$.tw. 
 15 or/6‐14 
 16 (animal$ not (human$ and animal$)).mp. 
 17 15 not 16 
 18 5 and 17

Appendix 3. CENTRAL search strategy

Database: Cochrane Central Register of Controlled Trials on Wiley <Issue 11, 2012> 
 Search Date: 11 December 2012 
 ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ 
 #1 MeSH descriptor: [Sodium Chloride, Dietary] this term only 
 #2 MeSH descriptor: [Sodium, Dietary] explode all trees 
 #3 MeSH descriptor: [Diet, Sodium‐Restricted] this term only 
 #4 sodium near/3 (restrict* or curb* or limit* or minimi* or low* or reduc* or intake or diet* or free):ti,ab in Trials 
 #5 salt near/3 (restrict* or curb* or limit* or minimi* or low* or reduc* or intake or diet* or free):ti,ab in Trials 
 #6 #1 or #2 or #3 or #4 or #5 in Trials

Appendix 4. Hypertension Group Specialised Register search strategy

Database: Hypertension Group Specialised Register 
 Search Date: 11 December 2012 
 ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ 
 1 Salt[TI] and (curb* or diet* or free or intake or limit* or low* or minimi* or reduc* or restrict*)[All fields] 
 2 Salt[TI] and (curb* or diet* or free or intake or limit* or low* or minimi* or reduc* or restrict*)[All fields] 
 3 1 or 2

Data and analyses

Comparison 1. Change in BP.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Change in Systolic BP	35		Change in SBP (Random, 95% CI)	‐4.18 [‐5.18, ‐3.18]
1.1 Hypertensives	21		Change in SBP (Random, 95% CI)	‐5.39 [‐6.62, ‐4.15]
1.2 Normotensives	14		Change in SBP (Random, 95% CI)	‐2.42 [‐3.56, ‐1.29]
2 Change in Diastolic BP	37		Change in DBP (Random, 95% CI)	‐2.06 [‐2.67, ‐1.45]
2.1 Hypertensives	23		Change in DBP (Random, 95% CI)	‐2.82 [‐3.54, ‐2.11]
2.2 Normotensives	14		Change in DBP (Random, 95% CI)	‐1.00 [‐1.85, ‐0.15]
3 Change in Systolic BP by Ethnic Group	31		Change in SBP (Random, 95% CI)	‐4.41 [‐5.39, ‐3.44]
3.1 Hypertensive Whites	16		Change in SBP (Random, 95% CI)	‐5.12 [‐6.27, ‐3.96]
3.2 Normotensive Whites	14		Change in SBP (Random, 95% CI)	‐2.11 [‐3.03, ‐1.19]
3.3 Hypertensive Blacks	5		Change in SBP (Random, 95% CI)	‐7.83 [‐10.96, ‐4.71]
3.4 Normotensive Blacks	3		Change in SBP (Random, 95% CI)	‐4.02 [‐7.44, ‐0.61]
4 Change in Diastolic BP by Ethnic Group	33		Change in DBP (Random, 95% CI)	‐2.16 [‐2.75, ‐1.57]
4.1 Hypertensive Whites	18		Change in DBP (Random, 95% CI)	‐2.66 [‐3.37, ‐1.95]
4.2 Normotensive Whites	14		Change in DBP (Random, 95% CI)	‐0.88 [‐1.68, ‐0.08]
4.3 Hypertensive Blacks	5		Change in DBP (Random, 95% CI)	‐4.08 [‐5.90, ‐2.26]
4.4 Normotensive Blacks	3		Change in DBP (Random, 95% CI)	‐1.98 [‐4.45, 0.49]
5 Change in Systolic BP by Sex	16		Change in SBP (Random, 95% CI)	‐5.52 [‐6.82, ‐4.23]
5.1 Hypertensive Men	9		Change in SBP (Random, 95% CI)	‐6.40 [‐6.00, ‐4.80]
5.2 Hypertensive Women	9		Change in SBP (Random, 95% CI)	‐7.11 [‐8.81, ‐5.41]
5.3 Normotensive Men	6		Change in SBP (Random, 95% CI)	‐3.39 [‐5.63, ‐1.16]
5.4 Normotensive Women	6		Change in SBP (Random, 95% CI)	‐4.26 [‐6.20, ‐2.31]
6 Change in Diastolic BP by Sex	18		Change in DBP (Random, 95% CI)	‐2.87 [‐3.54, ‐2.20]
6.1 Hypertensive Men	10		Change in DBP (Random, 95% CI)	‐3.96 [‐5.47, ‐2.46]
6.2 Hypertensive Women	10		Change in DBP (Random, 95% CI)	‐3.41 [‐4.29, ‐2.53]
6.3 Normotensive Men	6		Change in DBP (Random, 95% CI)	‐1.78 [‐3.01, ‐0.55]
6.4 Normotensive Women	6		Change in DBP (Random, 95% CI)	‐2.18 [‐2.95, ‐1.41]

1.3. Analysis.

Comparison 1 Change in BP, Outcome 3 Change in Systolic BP by Ethnic Group.

1.4. Analysis.

Comparison 1 Change in BP, Outcome 4 Change in Diastolic BP by Ethnic Group.

1.5. Analysis.

Comparison 1 Change in BP, Outcome 5 Change in Systolic BP by Sex.

1.6. Analysis.

Comparison 1 Change in BP, Outcome 6 Change in Diastolic BP by Sex.

Comparison 2. Change in hormone & lipid.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Change in PRA	14		Change in PRA (Random, 95% CI)	0.26 [0.17, 0.36]
2 Change in Aldosterone	9		Change in Aldosterone (Random, 95% CI)	73.20 [44.92, 101.48]
3 Change in Noradrenaline	6		Change in Noradrenaline (Random, 95% CI)	31.67 [6.57, 56.77]
4 Change in Adrenaline	4		Change in Adrenaline (Random, 95% CI)	6.70 [‐0.25, 13.64]
5 Change in Cholesterol	8		Change in Cholesterol (Random, 95% CI)	0.05 [‐0.02, 0.11]
6 Change in LDL	5		Change in LDL (Random, 95% CI)	0.05 [‐0.01, 0.12]
7 Change in HDL	6		Change in HDL (Random, 95% CI)	‐0.02 [‐0.06, 0.01]
8 Change in Triglyceride	6		Change in Triglyceride (Random, 95% CI)	0.04 [‐0.02, 0.09]

2.1. Analysis.

Comparison 2 Change in hormone & lipid, Outcome 1 Change in PRA.

2.2. Analysis.

Comparison 2 Change in hormone & lipid, Outcome 2 Change in Aldosterone.

2.3. Analysis.

Comparison 2 Change in hormone & lipid, Outcome 3 Change in Noradrenaline.

2.4. Analysis.

Comparison 2 Change in hormone & lipid, Outcome 4 Change in Adrenaline.

2.5. Analysis.

Comparison 2 Change in hormone & lipid, Outcome 5 Change in Cholesterol.

2.6. Analysis.

Comparison 2 Change in hormone & lipid, Outcome 6 Change in LDL.

2.7. Analysis.

Comparison 2 Change in hormone & lipid, Outcome 7 Change in HDL.

2.8. Analysis.

Comparison 2 Change in hormone & lipid, Outcome 8 Change in Triglyceride.

Characteristics of studies

Characteristics of included studies [ordered by year of study]

Parijs 1973.

Methods	X
Participants	N=15 Age: 41 yr Male: 43% Hypertensive
Interventions	UNa: ‐98 mmol/24h Duration: 4 wks
Outcomes	SBP DBP
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	Not stated
Incomplete outcome data (attrition bias) All outcomes	Low risk	Properly addressed, reasons for withdrawal were reported.
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	High risk	Not blinded to investigator
Blinding of the participant	High risk	Not blinded to participant
Blinding of the outcome assessor	High risk	Not blinded to outcome assessor

Morgan 1981 (F).

Methods	BP obs P
Participants	Intervention: N=6 Control: N=6 Age: 38 yr Male: 0% Hypertensive White: 100%
Interventions	UNa: ‐78 mmol/24h Duration: 8 wks
Outcomes	DBP
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	Not stated
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants who were randomised completed the study.
Selective reporting (reporting bias)	Unclear risk	DBP was reported and was included in our meta‐analysis. SBP was also reported, however, it was reported in combination with another group of patients who were on BP treatment. Because our meta‐analysis included participants who were not on any treatment and, in this trial, we could not separate the effect on salt reduction on SBP for individuals who were not on treatment, we did not include SBP in our pooled analysis.
Blinding of the investigator	High risk	Not blinded to investigator
Blinding of the participant	High risk	Not blinded to participant
Blinding of the outcome assessor	Low risk	Blood pressure observer blind

Morgan 1981 (M).

Methods	BP obs P
Participants	Intervention: N=6 Control: N=6 Age: 40 yr Male: 100% Hypertensive White: 100%
Interventions	UNa: ‐98 mmol/24h Duration: 8 wks
Outcomes	DBP
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	Not stated
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants who were randomised completed the study.
Selective reporting (reporting bias)	Unclear risk	DBP was reported and was included in our meta‐analysis. SBP was also reported, however, it was reported in combination with another group of patients who were on BP treatment. Because our meta‐analysis included participants who were not on any treatment and, in this trial, we could not separate the effect on salt reduction on SBP for individuals who were not on treatment, we did not include SBP in our pooled analysis.
Blinding of the investigator	High risk	Not blinded to investigator
Blinding of the participant	High risk	Not blinded to participant
Blinding of the outcome assessor	Low risk	Blood pressure observer blind

MacGregor 1982.

Methods	DB X
Participants	N=19 Age: 49 (30‐66) yr Male: 59% Hypertensive White: 63%; Black: 37%
Interventions	UNa: ‐76 mmol/24h Duration: 4 wks
Outcomes	SBP DBP PRA Aldo
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Adequate, use of Slow Sodium and placebo tablets
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants who were randomised completed the study.
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	Low risk	Double blind
Blinding of the participant	Low risk	Double blind
Blinding of the outcome assessor	Low risk	Outcome assessor blind

Watt 1983.

Methods	DB X
Participants	N=18 Age: 52 (31‐64) yr Male: 33% Hypertensive White: 100%
Interventions	UNa: ‐56 mmol/24h Duration: 4 wks
Outcomes	SBP DBP PRA
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Adequate, use of Slow Sodium and placebo tablets.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Properly addressed, reasons for withdrawal were reported.
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	Low risk	Double blind
Blinding of the participant	Low risk	Double blind
Blinding of the outcome assessor	Low risk	Outcome assessor blind

Silman 1983.

Methods	BP obs (RZ) P
Participants	Intervention: N=10 Control: N=15 Age: 50‐64 yr Hypertensive
Interventions	UNa: ‐53 mmol/24h Duration: 12 months
Outcomes	SBP DBP
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	Not stated
Incomplete outcome data (attrition bias) All outcomes	Low risk	Properly addressed, reasons for withdrawal were reported.
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	High risk	Not blinded to investigator
Blinding of the participant	High risk	Not blinded to participant
Blinding of the outcome assessor	Low risk	Use of random zero sphygmomanometer

Puska 1983 (H).

Methods	BP obs P
Participants	Intervention: N=15 Control: N=19 Age: 30‐50 yr Hypertensive White: 100%
Interventions	UNa: ‐117 mmol/24h Duration: 6 wks
Outcomes	SBP DBP
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	Not stated
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants who were randomised completed the study.
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	High risk	Not blinded to investigator
Blinding of the participant	High risk	Not blinded to participant
Blinding of the outcome assessor	Low risk	Blood pressure observer blind

Puska 1983 (N).

Methods	BP obs P
Participants	Intervention: N=19 Control: N=19 Age: 30‐50 Normotensive White: 100%
Interventions	UNa: ‐117 mmol/24h Duration: 6 wks
Outcomes	SBP DBP
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	Not stated
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants who were randomised completed the study.
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	High risk	Not blinded to investigator
Blinding of the participant	High risk	Not blinded to participant
Blinding of the outcome assessor	Low risk	Blood pressure observer blinded

Richards 1984.

Methods	BP obs (A) X
Participants	N=12 Age: 19‐52 yr Male: 67% Hypertensive White: 100%
Interventions	UNa: ‐105 mmol/24h Duration: 4‐6 wks
Outcomes	SBP DBP PRA Aldo Noradrenaline
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	Not stated
Incomplete outcome data (attrition bias) All outcomes	High risk	Not stated
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	High risk	Not blinded to investigator
Blinding of the participant	High risk	Not blinded to participant
Blinding of the outcome assessor	Low risk	Use of an automated version of the London School of Hygiene sphygmomanometer.

Erwteman 1984.

Methods	BP obs (RZ) P
Participants	Intervention: N=44 Control: N=50 Age: 46 (20‐70) yr Male: 62% Hypertensive White: 76%; Black: 24%
Interventions	UNa: ‐58 mmol/24h Duration: 6 months
Outcomes	SBP DBP
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	Not stated
Incomplete outcome data (attrition bias) All outcomes	Low risk	Properly addressed, reasons for withdrawal were reported.
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	Low risk	Assignment to different salt intakes was single blind. Participants' dietary salt intake was supervised by a dietitian who was asked not to discuss participants' dietary assignment with the physician in charge or the technician.
Blinding of the participant	High risk	Not blinded to participant
Blinding of the outcome assessor	Low risk	Outcome assessor blind

Watt 1985 (HH).

Methods	DB X
Participants	N=35 Age: 22 yr Male: 37% Normotensive White: 100%
Interventions	UNa: ‐74 mmol/24h Duration: 4 wks
Outcomes	SBP DBP
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Adequate, use of Slow Sodium and placebo tablets.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Not stated
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	Low risk	Double blind
Blinding of the participant	Low risk	Double blind
Blinding of the outcome assessor	Low risk	Outcome assessor blind

Watt 1985 (LL).

Methods	DB X
Participants	N=31 Age: 23 yr Male: 45% Normotensive White: 100%
Interventions	UNa: ‐60 mmol/24h Duration: 4 wks
Outcomes	SBP DBP
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Adequate, use of Slow Sodium and placebo tablets.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Not stated
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	Low risk	Double blind
Blinding of the participant	Low risk	Double blind
Blinding of the outcome assessor	Low risk	Outcome assessor blind

Chalmers 1986.

Methods	BP obs (A) P
Participants	Intervention: N=48 Control: N=52 Age: 52 yr Male: 85% Hypertensive All participants were white.
Interventions	UNa: ‐54 mmol/24h Duration: 12 wks
Outcomes	SBP DBP
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	Not stated
Incomplete outcome data (attrition bias) All outcomes	Low risk	Intention to treat analysis was undertaken
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	High risk	Not blinded to investigator
Blinding of the participant	High risk	Not blinded to participant
Blinding of the outcome assessor	Low risk	Use of automated blood pressure machine

Grobbee 1987.

Methods	DB X
Participants	N=40 Age: 18‐28 yr Male: 85% Hypertensive
Interventions	UNa: ‐72 mmol/24h Duration: 6 wks
Outcomes	SBP DBP Noradrenaline Chol
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Adequate, use of Slow Sodium and placebo tablets.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Properly addressed, reasons for withdrawal were reported.
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	Low risk	Double blind
Blinding of the participant	Low risk	Double blind
Blinding of the outcome assessor	Low risk	Outcome assessor blind

ANHMRC 1989 (X).

Methods	DB X
Participants	N=88 Age: 59 yr Male: 83% Hypertensive All participants were white.
Interventions	UNa: ‐67 mmol/24h Duration: 8 wks
Outcomes	SBP DBP
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Adequate, use of Slow Sodium and placebo tablets
Incomplete outcome data (attrition bias) All outcomes	Low risk	Intention to treat analysis was undertaken
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	Low risk	Double blind
Blinding of the participant	Low risk	Double blind
Blinding of the outcome assessor	Low risk	Outcome assessor blind

ANHMRC 1989 (P).

Methods	DB P
Participants	Intervention: N=50 Control: N=53 Age: 58 yr Male: 83% Hypertensive All participants were white.
Interventions	UNa: ‐71 mmol/24h Duration: 8 wks
Outcomes	SBP DBP
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Adequate, use of Slow Sodium and placebo tablets
Incomplete outcome data (attrition bias) All outcomes	Low risk	Intention to treat analysis was undertaken
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	Low risk	Double blind
Blinding of the participant	Low risk	Double blind
Blinding of the outcome assessor	Low risk	Outcome assessor blind

MacGregor 1989.

Methods	DB X
Participants	N=20 Age: 56 (43‐73) yr Male: 55% Hypertensive White: 75%; Black: 25%
Interventions	UNa: ‐82 mmol/24h Duration: 4 wks
Outcomes	SBP DBP PRA Aldo Noradrenaline
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Adequate, use of Slow Sodium and placebo tablets
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants who were randomised completed the study.
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	Low risk	Double blind
Blinding of the participant	Low risk	Double blind
Blinding of the outcome assessor	Low risk	Outcome assessor blind

Mascioli 1991.

Methods	DB X
Participants	N=48 Age: 52 yr Male: 79% Normotensive White: 98%; Black: 2%
Interventions	UNa: ‐20 mmol/8h Duration: 4 wks
Outcomes	SBP DBP
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Adequate, use of sodium chloride and placebo capsules.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Properly addressed, reasons for withdrawal were reported.
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	Low risk	Double blind
Blinding of the participant	Low risk	Double blind
Blinding of the outcome assessor	Low risk	Outcome assessor blind

Benetos 1992.

Methods	DB X
Participants	N=20 Age: 42 (22‐55) yr Male: 45% Hypertensive White: 100%
Interventions	UNa: ‐78 mmol/24h Duration: 4 wks
Outcomes	SBP DBP Noradrenaline
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Adequate, use of sodium chloride and placebo capsules.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants who were randomised completed the study.
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	Low risk	Double blind
Blinding of the participant	Low risk	Double blind
Blinding of the outcome assessor	Low risk	Outcome assessor blind

Cobiac 1992.

Methods	DB P
Participants	Intervention: N=26 Control: N=28 Age: 67 (60‐80) yr Male: 67% Normotensive White: 100%
Interventions	UNa: ‐73 mmol/24h Duration: 4 wks
Outcomes	SBP DBP
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Adequate, use of Slow Sodium and placebo tablets.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Intention to treat analysis was undertaken
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	Low risk	Double blind
Blinding of the participant	Low risk	Double blind
Blinding of the outcome assessor	Low risk	Outcome assessor blind

TOHP I 1992.

Methods	BP obs (RZ) P
Participants	Intervention: N=327 Control: N=417 Age: 43 (30‐54) yr Male: 71% Normotensive White: 77%; Black: 23%
Interventions	UNa: ‐44 mmol/24h Duration: 18 months
Outcomes	SBP DBP
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Adequate, central randomisation.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Intention to treat analysis was undertaken
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	High risk	Not blinded to investigator
Blinding of the participant	High risk	Not blinded to participant
Blinding of the outcome assessor	Low risk	Use of Hawksley random‐zero sphygmomanometer.

Nestel 1993 (F).

Methods	DB P
Participants	Intervention: N=15 Control: N=15 Age: 65 (60‐79) yr Male: 0% Normotensive White: 100%
Interventions	UNa: ‐94 mmol/24h Duration: 6 wks
Outcomes	SBP DBP
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Adequate, use of Slow Sodium and placebo tablets.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Properly addressed, reasons for withdrawal were reported.
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	Low risk	Double blind
Blinding of the participant	Low risk	Double blind
Blinding of the outcome assessor	Low risk	Outcome assessor blind

Nestel 1993 (M).

Methods	DB P
Participants	Intervention: N=17 Control: N=19 Age: 66 (60‐79) yr Male: 100% Normotensive White: 100%
Interventions	UNa: ‐76 mmol/24h Duration: 6 wks
Outcomes	SBP DBP
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Adequate, use of Slow Sodium and placebo tablets.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Properly addressed, reasons for withdrawal were reported.
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	Low risk	Double blind
Blinding of the participant	Low risk	Double blind
Blinding of the outcome assessor	Low risk	Outcome assessor blind

Fotherby 1993.

Methods	DB X
Participants	N=17 Age: 73 (66‐79) yr Male: 24% Hypertensive White: 100%
Interventions	UNa: ‐79 mmol/24h Duration: 5 wks
Outcomes	SBP DBP PRA Aldo
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Adequate, use of Slow Sodium and placebo tablets.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Properly addressed, reason for exclusion was reported.
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	Low risk	Double blind
Blinding of the participant	Low risk	Double blind
Blinding of the outcome assessor	Low risk	Outcome assessor blind

Ruppert 1993.

Methods	DB X
Participants	N=25 Age: 47 (27‐75) yr Male: 40% Normotensive White: 100%
Interventions	UNa: ‐118 mmol/24h Duration: 4 wks
Outcomes	SBP DBP PRA Noradrenaline Chol Trig LDL HDL
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Adequate, use of salt and placebo capsules
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants who were randomised completed the study.
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	Low risk	Double blind
Blinding of the participant	Low risk	Double blind
Blinding of the outcome assessor	Low risk	Outcome assessor blind

Schorr 1996.

Methods	DB X
Participants	N=16 Age: 64 (60‐72) yr Male: 48% Normotensive White: 100%
Interventions	UNa: ‐71 mmol/24h Duration: 4 wks
Outcomes	SBP DBP PRA Aldo Chol Trig LDL HDL
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Adequate, use of low sodium and high sodium chloride mineral water.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Properly addressed, reasons for exclusion were reported.
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	Low risk	Double blind
Blinding of the participant	Low risk	Double blind
Blinding of the outcome assessor	Low risk	Outcome assessor blind

Cappuccio 1997 (H).

Methods	DB X
Participants	N=29 Age: 67 (60‐78) yr Male: 51% Hypertensive
Interventions	UNa: ‐87 mmol/24h Duration: 4 wks
Outcomes	SBP DBP PRA Aldo
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Adequate, use of Slow Sodium and placebo tablets
Incomplete outcome data (attrition bias) All outcomes	Low risk	Properly addressed, reason for withdrawal was reported.
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	Low risk	Double blind
Blinding of the participant	Low risk	Double blind
Blinding of the outcome assessor	Low risk	Outcome assessor blind

Meland 1997.

Methods	DB X
Participants	N=16 Age: 50 yr Male: 81% Hypertensive White: 100%
Interventions	UNa: ‐66 mmol/24h Duration: 8 wks
Outcomes	SBP DBP
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Adequate, use of salt and placebo capsules.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	One patient dropped out during the trial and was replaced by another patient according to the same inclusion and randomisation procedures.
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	Low risk	Double blind
Blinding of the participant	Low risk	Double blind
Blinding of the outcome assessor	Low risk	Outcome assessor blind

TOHP II 1997.

Methods	BP obs (RZ) P
Participants	Intervention: N=515 Control: N=514 Age: 44 (30‐54) yr Male: 67% Normotensive White: 82%; Black: 18%
Interventions	UNa: ‐40 mmol/24h Duration: 36 months
Outcomes	SBP DBP
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Adequate, central randomisation.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Intention to treat analysis was undertaken
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	High risk	Not blinded to investigator
Blinding of the participant	High risk	Not blinded to participant
Blinding of the outcome assessor	Low risk	Use of Hawksley random‐zero sphygmomanometer.

Cappuccio 1997.

Methods	DB X
Participants	N=47 Age: 67 (60‐78) yr Male: 51% Hypertensive and normotensive White: 89%; Black: 4%; Asian: 7%
Interventions	UNa: ‐87 mmol/24h Duration: 4 wks
Outcomes	SBP DBP PRA Aldo Cholesterol
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Adequate, use of Slow Sodium and placebo tablets
Incomplete outcome data (attrition bias) All outcomes	Low risk	Properly addressed, reason for withdrawal was reported.
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	Low risk	Double blind
Blinding of the participant	Low risk	Double blind
Blinding of the outcome assessor	Low risk	Outcome assessor blind

Cappuccio 1997 (N).

Methods	DB X
Participants	N=18 Age: 67 (60‐78) yr Male: 51% Normotensive
Interventions	UNa: ‐76 mmol/24h Duration: 4 wks
Outcomes	SBP DBP PRA Aldo
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Adequate, use of Slow Sodium and placebo tablets
Incomplete outcome data (attrition bias) All outcomes	Low risk	Properly addressed, reason for withdrawal was reported.
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	Low risk	Double blind
Blinding of the participant	Low risk	Double blind
Blinding of the outcome assessor	Low risk	Outcome assessor blind

Sacks 2001 (H).

Methods	BP obs (RZ) X
Participants	N=76 Age: 52 yr Male: 39% Hypertensive White: 40%; Black: 60%
Interventions	UNa: ‐80 mmol/24h Duration: 4 wks
Outcomes	SBP DBP
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Adequate
Incomplete outcome data (attrition bias) All outcomes	Low risk	Intention to treat analysis was undertaken.
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	Low risk	The personnel involved in the collection of the outcome data were unaware of participants' diet assignment.
Blinding of the participant	High risk	Not blinded to participant
Blinding of the outcome assessor	Low risk	Blood pressure observer blinded

Sacks 2001 (N).

Methods	BP obs (RZ) X
Participants	N=116 Age: 47 yr Male: 50% Normotensive White: 46%; Black: 54%
Interventions	UNa: ‐75 mmol/24h Duration: 4 wks
Outcomes	SBP DBP
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Adequate
Incomplete outcome data (attrition bias) All outcomes	Low risk	Intention to treat analysis was undertaken.
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	Low risk	The personnel involved in the collection of the outcome data were unaware of participants' diet assignment.
Blinding of the participant	High risk	Not blinded to participant
Blinding of the outcome assessor	Low risk	Blood pressure observer blinded

Gates 2004.

Methods	DB X
Participants	N=12 Age: 64 yr Male: 50% Hypertensive White: 100%
Interventions	UNa: ‐89 mmol/24h Duration: 4 wks
Outcomes	SBP DBP PRA Chol Trig LDL HDL Noradrenaline
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Adequate, use of Slow Sodium and placebo tablets.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants who were randomised completed the study.
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	Low risk	Double blind
Blinding of the participant	Low risk	Double blind
Blinding of the outcome assessor	Low risk	Outcome assessor blind

Harsha 2004.

Methods	BP obs (RZ) X
Participants	N=193 Age: 49 yr Male: 46% Hypertensive and normotensive
Interventions	UNa: ‐77 mmol/24h Duration: 4 wks
Outcomes	Cholesterol LDL HDL Triglycerides
Notes	This paper reported the lipid data from the DASH‐Na trial.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Adequate
Incomplete outcome data (attrition bias) All outcomes	Low risk	Intention to treat analysis was undertaken.
Selective reporting (reporting bias)	Low risk	All lipid data from participants who gave fasting blood samples were analysed and reported.
Blinding of the investigator	Low risk	The personnel involved in the collection of the outcome data were unaware of participants' diet assignment.
Blinding of the participant	High risk	Not blinded to participant
Blinding of the outcome assessor	Low risk	The personnel involved in the measurements of lipids were blinded to participants' dietary assignment.

Swift 2005.

Methods	DB X
Participants	N=40 Age: 50 yr Male: 43% Hypertensive White: 0%; Black: 100%
Interventions	UNa: ‐78 mmol/24h Duration: 4 wks
Outcomes	SBP DBP PRA Aldo
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Adequate, use of Slow Sodium and placebo tablets
Incomplete outcome data (attrition bias) All outcomes	Low risk	Properly addressed
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	Low risk	Double blind
Blinding of the participant	Low risk	Double blind
Blinding of the outcome assessor	Low risk	Outcome assessor blind

Melander 2007.

Methods	DB X
Participants	N=39 Age: 53 (42‐64) yr Male: 51% Hypertensive and normotensive White: 100%
Interventions	UNa: ‐89 mmol/24h Duration: 4 wks
Outcomes	SBP DBP
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Adequate, use of salt and placebo capsules.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Properly addressed, reasons for withdrawal were reported.
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	Low risk	Double blind
Blinding of the participant	Low risk	Double blind
Blinding of the outcome assessor	Low risk	Outcome assessor blind

Melander 2007 (H).

Methods	DB X
Participants	N=19 Age: 53 (42‐64) yr Male: 53% Hypertensive White: 100%
Interventions	UNa: ‐93 mmol/24h Duration: 4 wks
Outcomes	SBP DBP
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Adequate, use of salt and placebo capsules.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Properly addressed, reasons for withdrawal were reported.
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	Low risk	Double blind
Blinding of the participant	Low risk	Double blind
Blinding of the outcome assessor	Low risk	Outcome assessor blind

Melander 2007 (N).

Methods	DB X
Participants	N=20 Age: 53 (42‐64) yr Male: 50% Normotensive White: 100%
Interventions	UNa: ‐85 mmol/24h Duration: 4 wks
Outcomes	SBP DBP
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Adequate, use of salt and placebo capsules.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Properly addressed, reasons for withdrawal were reported.
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	Low risk	Double blind
Blinding of the participant	Low risk	Double blind
Blinding of the outcome assessor	Low risk	Outcome assessor blind

He 2009.

Methods	DB X
Participants	N=169 Age: 50 yr Male: 67% Hypertensive White: 42%; Black: 41%: Asian: 17%
Interventions	UNa: ‐55 mmol/24h Duration: 6 wks
Outcomes	SBP DBP PRA Aldo
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Adequate, use of Slow Sodium and placebo tablets
Incomplete outcome data (attrition bias) All outcomes	Low risk	Properly addressed
Selective reporting (reporting bias)	Low risk	Both SBP and DBP were reported.
Blinding of the investigator	Low risk	Double blind
Blinding of the participant	Low risk	Double blind
Blinding of the outcome assessor	Low risk	Outcome assessor blind

UNa: urinary sodium; BP: blood pressure; SBP: systolic blood pressure; DBP: diastolic blood pressure; X: crossover; P: parallel; DB: Double blind; BP obs: blood pressure observer blinded; RZ: random zero manometer; A: automated sphygmomanometer; F: Female; M: Male; H: Hypertensive; N: Normotensive; HH: offspring of two parents with high blood pressure; LL: offspring of two parents with low blood pressure; PRA: plasma renin activity; Aldo: aldosterone; Chol: cholesterol; Trig: triglyceride; LDL: low‐density lipoprotein; HDL: high‐density lipoprotein.

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Akita 2003	Sub‐study of DASH‐Na trial (Sacks 2001)
Alam 1999	Duration of salt reduction less than 4 weeks
Alli 1992	Reduction in 24h urinary sodium less than 40 mmol
Ambrosioni 1982	24h urinary sodium not measured
Ames 2001	With concomitant intervention
Andersson 1984	With concomitant intervention
Andersson 1986	Duration of salt reduction less than 4 weeks
Appel 2001	Sub‐study of TONE trial (Whelton 1998)
Appel 2003	With concomitant intervention
Applegate 1992	With concomitant intervention
Arroll 1995	With concomitant intervention
Barba 2000	Duration of salt reduction less than 4 weeks
Beard 1982	With concomitant intervention
Beckmann 1995	With concomitant intervention
Berglund 1989	With concomitant intervention
Boero 2000	Duration of salt reduction less than 4 weeks
Bompiani 1988	With concomitant intervention
Bray 2004	Sub‐study of DASH‐Na trial (Sacks 2001)
Bruun 1990	Duration of salt reduction less than 4 weeks
Buckley 1994	Duration of salt reduction less than 4 weeks
Bulpitt 1984	With concomitant intervention
Burke 2005	With concomitant intervention
Burke 2007	With concomitant intervention
Burnier 1993	Duration of salt reduction less than 4 weeks
Buyck 2009	24h urinary sodium not measured
Calabrese 1985	Studies in children
Cappuccio 2006	Reduction in 24h urinary sodium less than 40 mmol
Carney 1991	With concomitant intervention
Charlton 2008	With concomitant intervention
Chen 2008	Duration of salt reduction less than 4 weeks
Chrysant 2000	With concomitant intervention
Cook 2005	Sub‐study of TOHP II (TOHP II 1997)
Cook 2007	Follow‐up study of TOHP (TOHP 1992, TOHP II 1997)
Cook 2009	Sub‐study of TOHP (TOHP 1992, TOHP II 1997)
Cooper 1984	Studies in children
Costa 1981	24h urinary sodium not measured
CSSS 2007	24h urinary sodium not measured
Cuzzola 2001	Duration of salt reduction less than 4 weeks
Damasceno 1999	Duration of salt reduction less than 4 weeks
Damasceno 2000	Duration of salt reduction less than 4 weeks
Davis 1994	With concomitant intervention
Davrath 1999	Duration of salt reduction less than 4 weeks
Del Rio 1990	With concomitant intervention
Del Rio 1993	Duration of salt reduction less than 4 weeks
Delemarre 2000	Studies in pregnant women
Dickinson 2009	Duration of salt reduction less than 4 weeks
Dimsdale 1990	Duration of salt reduction less than 4 weeks
Dodson 1984	Study in patients with diabetes
Dodson 1989	Study in patients with diabetes
Donovan 1993	Duration of salt reduction less than 4 weeks
Dubbert 1995	With concomitant intervention
Egan 1991	Duration of salt reduction less than 4 weeks
Egan 1991(b)	Duration of salt reduction less than 4 weeks
Ekinci 2009	Study in patients with diabetes
Ekinci 2010	Study in patients with diabetes
el Ashry 1987	Duration of salt reduction less than 4 weeks
Elmer 1995	With concomitant intervention
Elmer 2006	With concomitant intervention
Fagerberg 1984	With concomitant intervention
Fagerberg 1985	With concomitant intervention
Feldman 1996	Duration of salt reduction less than 4 weeks
Ferri 1993	Duration of salt reduction less than 4 weeks
Ferri 1994	Duration of salt reduction less than 4 weeks
Ferri 1996	Duration of salt reduction less than 4 weeks
Fliser 1993(a)	Duration of salt reduction less than 4 weeks
Fliser 1993(b)	With concomitant intervention
Forrester 2005	Duration of salt reduction less than 4 weeks
Fotherby 1997	Sub‐study of Fotherby 1993
Friberg 1990	Duration of salt reduction less than 4 weeks
Fuchs 1987	Duration of salt reduction less than 4 weeks
Gillies 1984	With concomitant intervention
Gillum 1981	Studies in children
Gomi 1998	Duration of salt reduction less than 4 weeks
Gow 1992	Duration of salt reduction less than 4 weeks
Grey 1996	Duration of salt reduction less than 4 weeks
Hargreaves 1989	Duration of salt reduction less than 4 weeks
Haythornthwaite 1992	Duration of salt reduction less than 4 weeks
He 2000	Follow‐up study of TOHP I (TOHP I 1992)
He 2005(a)	Sub‐study of previous trials
He 2005(b)	Sub‐study of previous trials
He, J 2009	Duration of salt reduction less than 4 weeks
Heagerty 1986	Duration of salt reduction less than 4 weeks
Herlitz 1998	Duration of salt reduction less than 4 weeks
Hofman 1983	Studies in children
Houlihan 2002(a)	Study in patients with diabetes
Houlihan 2002(b)	Study in patients with diabetes
Howe 1991	Studies in children
Howe1994	With concomitant intervention
HPTRG 1990	Reduction in 24‐h urinary sodium less than 40 mmol
Hu 2009	24h urinary sodium not measured
Inoue 1996	Duration of salt reduction less than 4 weeks
Ireland 2010	Reduction in 24h urinary sodium less than 40 mmol
Ishimitsu 1996	Duration of salt reduction less than 4 weeks
Iso 1996	With concomitant intervention
Iwaoka 1994	Duration of salt reduction less than 4 weeks
Jessani 2008	Duration of salt reduction less than 4 weeks
Johnson 2001	Duration of salt reduction less than 4 weeks
Jula 1990	With concomitant intervention
Jula 1992(a)	With concomitant intervention
Jula 1992(b)	With concomitant intervention
Kawasaki 1998	With concomitant intervention
Keven 2006	With concomitant intervention
Kirpizidis 2005	With concomitant intervention
Kojuri 2007	Total 24h urinary sodium excretion not reported.
Koolen 1984 (a)	Duration of salt reduction less than 4 weeks
Koolen 1984 (b)	Duration of salt reduction less than 4 weeks
Koopman 1990 (a)	With concomitant intervention
Koopman 1990 (b)	With concomitant intervention
Koopman 1997	With concomitant intervention
Kostis 2002	With concomitant intervention
Kristinsson 1988	With concomitant intervention
Kumanyika 1993	Sub‐study of TOHP I (TOHP I 1992)
Kumanyika 2005	Sub‐study of TOHP II
Kurtz 1987	Duration of salt reduction less than 4 weeks
Lawton 1988	Duration of salt reduction less than 4 weeks
Logan 1986	Reduction in 24‐h urinary sodium less than 40 mmol
Luft 1990	Duration of salt reduction less than 4 weeks
Macgregor 1982	Duration of salt reduction less than 4 weeks
MacGregor 1987	With concomitant intervention
Mallamaci 1996	Duration of salt reduction less than 4 weeks
Manunta 2001	Duration of salt reduction less than 4 weeks
Mark 1975	Duration of salt reduction less than 4 weeks
Mattila 2003	With concomitant intervention
Maxwell 1984	With concomitant intervention
McCarron 1997	With concomitant intervention
Meland 2009	With concomitant intervention
Miller 1988	Studies in children
Miller 1997	Duration of salt reduction less than 4 weeks
Morgan 1978	Reduction in 24‐h urinary sodium less than 40 mmol
Morgan 1988	Duration of salt reduction less than 4 weeks
Mtabaji 1990	Duration of salt reduction less than 4 weeks
Mu 2009	With concomitant intervention
Mufunda 1992	Duration of salt reduction less than 4 weeks
Muhlhauser 1996	With concomitant intervention
Myers 1983	Duration of salt reduction less than 4 weeks
Mülhauser 1996	Study in patients with diabetes
Nakamura 2003	24h urinary sodium not measured
Nowson 1988	With concomitant intervention
Nowson 2003	With concomitant intervention
Nowson 2004	With concomitant intervention
Nowson 2009	With concomitant intervention
Omland 2001	With concomitant intervention
Omvik 1995	With concomitant intervention
Overlack 1993	Duration of salt reduction less than 4 weeks
Overlack 1995	Duration of salt reduction less than 4 weeks
Palacios 2004	Study in children
Palmer 1989	24h urinary sodium not measured
Parfrey 1981	24h urinary sodium not measured
Parker 1990	With concomitant intervention
Pechere‐Bertschi 2008	Duration of salt reduction less than 4 weeks
Pedersen 1986	Duration of salt reduction less than 4 weeks
Perry 2003	Duration of salt reduction less than 4 weeks
Petrie 1998	Duration of salt reduction less than 4 weeks
Pimenta 2009	Duration of salt reduction less than 4 weeks
Pogson 2009	Blood pressure was not measured
Pomeranz 2002	Studies in neonates
Redon 1995	With concomitant intervention
Redon‐Mas 1993	With concomitant intervention
Richards 1986	Duration of salt reduction less than 4 weeks
Ruilope 1992	Duration of salt reduction less than 4 weeks
Ruilope 1993	Duration of salt reduction less than 4 weeks
Ruppert 1991	Duration of salt reduction less than 4 weeks
Ruppert 1994	Duration of salt reduction less than 4 weeks
Salvetti 1988	With concomitant intervention
Santello 1997	Duration of salt reduction less than 4 weeks
Saptharishi 2009	With concomitant intervention
Schmid 1990	Duration of salt reduction less than 4 weeks
Schorr 1997(a)	Duration of salt reduction less than 4 weeks
Schorr 1997(b)	Duration of salt reduction less than 4 weeks
Sciarrone 1990	With concomitant intervention
Sciarrone 1992	With concomitant intervention
Sharma 1990	Duration of salt reduction less than 4 weeks
Sharma 1991	Duration of salt reduction less than 4 weeks
Sharma 1993	Duration of salt reduction less than 4 weeks
Shore 1988	Duration of salt reduction less than 4 weeks
Sinaiko 1993	Studies in children
Singer 1991	With concomitant intervention
Singer 1994	Duration of salt reduction less than 4 weeks
Singer 1995	With concomitant intervention
Skrabal 1980	Duration of salt reduction less than 4 weeks
Skrabal 1981	Duration of salt reduction less than 4 weeks
Skrabal 1984(a)	Duration of salt reduction less than 4 weeks
Skrabal 1984(b)	Duration of salt reduction less than 4 weeks
Skrabal 1985	Duration of salt reduction less than 4 weeks
Staessen 1988	Reduction in 24h urinary sodium less than 40 mmol
Starmans‐Kool 2011	Duration of salt reduction less than 4 weeks
Steegers 1991	Studies in pregnant women
Sullivan 1980	Duration of salt reduction less than 4 weeks
Suppa 1988	With concomitant intervention
Suzuki 2000	Duration of salt reduction less than 4 weeks
Takahashi 2006	With concomitant intervention
Takashashi 2003	With concomitant intervention
Teow 1985	Duration of salt reduction less than 4 weeks
Thaler 1982	With concomitant intervention
TMHRG 1991	With concomitant intervention
Todd 2010	With concomitant intervention
Todd 2012	24h urinary sodium not measured
Townsend 2007	Duration of salt reduction less than 4 weeks
Tzemos 2008	Duration of salt reduction less than 4 weeks
Uusitupa 1996	With concomitant intervention
Uzu 1999	Duration of salt reduction less than 4 weeks
Uzu 2009	With concomitant intervention
Vaidya 2009	Duration of salt reduction less than 4 weeks
van BergeLandry 2004	Extreme change in salt intake, urinary sodium was changed from 309 to 24 mmol/24h
van Paassen 1996	Duration of salt reduction less than 4 weeks
Vedovato 2004	Study in patients with diabetes
Visser 2008	Duration of salt reduction less than 4 weeks
Vollmer 2001	Sub‐study of the DASH‐Na trial
Weder 1991	Duration of salt reduction less than 4 weeks
Wedler 1992	Duration of salt reduction less than 4 weeks
Weir 1995	Duration of salt reduction less than 4 weeks
Weir 1997	With concomitant intervention
Weir 1998	With concomitant intervention
Weir 2001	With concomitant intervention
Weir 2010	With concomitant intervention
Whelton 1998	With concomitant intervention
Whitten 1980	Studies in children
Wing 1984	With concomitant intervention
Wing 1998	With concomitant intervention
Yamakoshi 2006	With concomitant intervention
Yamamoto 1997	With concomitant intervention
Yazici 2009	With concomitant intervention
Zhao 2009	Duration of salt reduction less than 4 weeks
Zoccali 1993	Duration of salt reduction less than 4 weeks
Zoccali 1994(a)	Duration of salt reduction less than 4 weeks
Zoccali 1994(b)	Duration of salt reduction less than 4 weeks
Zoccali 1996	Duration of salt reduction less than 4 weeks

Contributions of authors

FH and JL screened the titles and abstracts, assessed trials for inclusion and trial quality, and extracted data. FH performed statistical analyses and wrote the draft manuscript. FH, JL and GM contributed to the revision and final version of the paper.

Sources of support

Internal sources

• Queen Mary University of London, UK.

  While working on this review, Feng He and Graham MacGregor's salaries were paid by Queen Mary University of London

• China Scholarship Council, China.

  While working on this review Jiafu Li was supported by China Scholarship Council.

External sources

• No sources of support supplied

Declarations of interest

None known.

New search for studies and content updated (conclusions changed)