Background
Setting and reference | Interventions | Main results | Comment e.g., on key features and potential limitations |
---|---|---|---|
26 European countriesb, Murray et al 2003 [52] | (i) Cooperation between government and the food industry for a stepwise decrease in salt content of processed foods and for labelling; (ii) Legislation to decrease sodium content in processed foods and appropriate labelling (a combined package); (iii) Health education focusing on body mass index and cholesterol concentrations; (iv) A combined package of (ii) and (iii). | 0.7-1.3 million DALYs averted per year (in European countries group). Very cost-effective at US$14-37 per DALY averted (DR = 3 %). Legislation reported to be more cost-effective than voluntary agreements. | This study did not consider cost-savings from preventing CVD. It was reliant on the relatively simplistic WHO Choice methodology for costing intervention programmes. |
US, Palar & Sturm 2009 [53] | Reducing average population sodium intake to 1200-2300 mg/day [d]. | Large annual QALY gains (312,000) and large annual savings in health costs (US$18 billion). | QALYs were also valued as part of a societal perspective. No specific intervention or intervention costs. |
US, Smith-Spangler et al 2010 [54] | (i) Collaboration with industry to reduce sodium by 9.5 %, and (ii) a sodium tax to reduce sodium by 6 %. | Both interventions achieved large QALY gains (2.1 million and 1.3 million respectively over the cohort’s lifetime). Cost-savings at $US32.1 and 22.4 billion respectively. | A high quality study but the cost of implementing the tax intervention was not considered. |
US, Bibbins-Domingo et al 2010 [18] | A regulatory intervention to reduce the level of salt intake by 3 g/d. | Large annual QALY gains (194,000 to 392,000) and annual cost-savings at $US10 to 24 billion. Salt reduction was more cost-effective than treating hypertension with medications. The anticipated relative benefits in blacks were greater than those for non-blacks across all age and sex groups. | A high quality study but no specific intervention or intervention costs detailed. This is only one of two studies in this table to consider ethnic inequalities. |
Australia, Cobiac et al 2010 [55] | Voluntary and mandatory reduction of salt content in breads, margarine, and cereals; dietary advice; and labelling programme. | Both salt content interventions were cost-saving (e.g., $A3.3 billion for the mandatory one over the cohorts lifetime) but health gain was much greater for the mandatory vs voluntary intervention (e.g., 110,000 vs 5300 lifetime DALYs averted). The labelling programme was cost-effective but not the dietary advice. | Included a useful comparison between a voluntary and mandatory intervention. Used WHO Choice methods rather than more country-specific intervention costing data. |
England & Wales, Barton et al 2011 [56] | Legislative means (unspecified) to reduce salt intake by 3 g/d | Any salt-reduction intervention costing up to £40 million a year was estimated to be cost-saving. For a 3 g/d reduction over 10 years the total QALY gain was 131,000. | No specific intervention was modelled. See also comments in a review [57]. |
Finland, Martikainen et al 2011 [17] | A population-wide 1 g/d salt reduction (by unspecified means). | Large QALY gains (26,100 by the year 2030). Cost-savings were 150–225 million Euros by 2030 (but when combined with the saturated fat reduction intervention). | Also considered reductions in productivity losses. No specific intervention was modelled. See also comments in a review [57]. |
Australia, Cobiac et al 2012 [58] | Mandatory reduction of salt content in breads, margarine, and cereals; and Community Heart Health Programme (CHHP). | Large number of DALYs averted per year (80,000) for the mandatory intervention (vs 3000 in the CHHP) and cost-saving. (See also a similar study listed above by these authors). | This study allowed for a comparison of the mandatory salt reduction with various CVD treatment interventions (the former being more cost-effective). |
England, Dodhia et al 2012 [59] | Included: (i) reductions of salt leading to 2 mmHg and (ii) 5 mmHg reductions in blood pressure; (iii) reduced intake down to 6 g/d via assumed food industry agreement; (iv) advice for DASH-sodium diets. | Large number of DALYs averted for (i) to (iv) in the 200,000 to 900,000 range (DR = 3.5 %). Salt reduction in the population was always reported to be cost-saving except for dietary advice in some age-groups (but here it was still cost-effective). The maximum saving for an intervention was £1.9 billion (over 10y). | High quality study which allowed comparisons with CVD treatment interventions. |
New Zealand, Nghiem et al 2015 [5] | Eight interventions (mix of mandatory and voluntary interventions – see Table 5 for findings combined with the results of this particular study). | All interventions (except dietary counselling) were cost-saving. The largest gain was from a “sinking lid” intervention (211,000 QALYs over the cohort’s lifetime; $US0.7 billion in savings). The interventions were estimated to produce relatively greater health gain for indigenous people (Māori). | The study had some limitations including around price elasticity data (for the salt tax) and the hypothetical nature of some interventions (e.g., sinking lid). See the Discussion where we compare these results with the current analysis. |
Methods
Model structure and perspective
Input parameters
Variable | Sources and key details | Key values and uncertainty |
---|---|---|
Baseline variables in 2011 | ||
Sodium intake | Source: New Zealand (NZ) nutrition survey data [60], with significant variation by sex, but not by ethnicity or age (for adults). No trend under business-as-usual (BAU) specified, given no notable trend since the 1980s [61]. Although these values are based on spot urine tests, such tests are a reasonable means for studying populations as per this systematic review [62]. The values are also similar to previous NZ studies which have used 24-h urine collections [63]. | 4013 mg/d for men and 3115 mg/d for women (nil uncertainty; rather uncertainty around the intervention associated reduction was considered – see below) |
Incidence, prevalence and case-fatality data for CHD and stroke | Calculated using linked Health Tracker data, with coherency checks using DisMod II and smoothing with regression as required. Future annual percentage change (APC) in incidence and CFR were both set at -2.0 % each as per the NZBDS. | |
Morbidity (disability weights [DW]) | DW for CHD = 0.081 (average) DW for stroke = 0.226 (average). Uncertainty: e.g., for non-Māori males, 95%CI: 0.05–0.11 for CHD and 0.11–0.23 for stroke. (For more details uncertainty see Nghiem et al [19]). | |
Baseline health system costs for CHD and stroke states, and non-diseased states | Calculated from Health Tracker data by sex and age in 2011 for people: (a) without either CHD or stroke; (b) with CHD only, and excess to (a); (c) with stroke only, and excess to (a). (See an online report [19] for details). | Examples for 60 year old women (gamma distribution with SD = 10 % of mean): (a) NZ$2381; (b) NZ$16,258 for the first year, NZ$5,395 for second and subsequent years; (c) NZ$20,553 and NZ$5991 for stroke. |
Epidemiological associations | ||
Change in systolic blood pressure (sBP) (in mm Hg) for each 100 mmol/d change in sodium intake | Derived from the regressions models developed by Law et al [35]. The small differences in BP by ethnic group did not justify separate modelling by ethnicity (higher in Māori by 3 mm Hg for systolic BP and 4 mm Hg for diastolic BP in both sexes compared to non-Māori [64]). Also of note is that no trend in BP into the future was considered given the unclear picture in NZ (of a downward trend in population BP levels from 1982 to 2002 and then an upward trend from then 2008/09) [64]. | For men and women: |
Age-group; sBP (mm Hg) change | ||
30–39: 5.5 | ||
40–49: 6.6 | ||
50–59: 9.2 | ||
60–69: 10.3 | ||
Relationship between blood pressure and CVD risks | The hazard ratio for a 20 mm Hg reduction in systolic BP ranged from 0.49 to 0.67 for CHD and from 0.38 to 0.67 for stroke (depending on age). For uncertainty: SD = +/- 10 % of the point estimate for each age group. |
Incidence, prevalence and case-fatality
Morbidity and disability weights
Intervention specification and parameters
Intervention | Sources and extra details | Key values and uncertainty (average adult)a
|
---|---|---|
Salt substitution at 59 %: In all processed food the NaCl is legally required to be largely substituted with other salts at the level of 59 % (mix of potassium and magnesium salts). | This 59 % substitution level was that used in a randomised trial in the Netherlands [27]. This trial reported that all the foods were rated equally by both groups for appearance and palatability-except that the bread and table salt were considered “less salty” by those in the intervention group. It was assumed that the intervention would apply to processed foods (72 % of sodium intake in NZ) and table salt (15 % of intake), but not to milk, seafood, fruit, vegetables and fresh meat (13 % of intake) [66]. Of note, is that another study found that consumption of bread was not affected at 52 % reduction in sodium and up to 67 % when KCl and yeast extract was used for flavour compensation [46]. We assumed a phased-in implementation over a five year period (i.e., five equal steps beginning in the baseline year of 2011). | Reduction of 51.5 % in daily sodium intake from the reduced intake of processed foods and table salt (or 1824 mg [79.3 mmol] sodium per day for an average adult). For uncertainty we used +/- 10 % of the point estimate (normal distribution). |
Salt substitution at 25 %: As per the intervention above but at a lower level. | This substitution level is as per the majority of the salt substitution studies in the meta-analysis by Peng et al [16]. The same assumptions about the phase-in period and application to just processed foods and to table salt (as per the intervention directly above) applied. | Reduction of 21.8 % in daily sodium intake from the reduced intake of processed foods and table salt (or 773 mg [33.6 mmol] sodium per day for an average adult). For uncertainty we used +/- 10 % of the point estimate (normal distribution). |
Tight limits on sodium in bread: A legal requirement for commercial bread to have a sodium level that is ≤280 mg/100 g. | This is the level used in Finnish law for the labelling of low salt bread (i.e., 0.7 % salt) [28]. This level is equivalent to 0.28 % sodium which is equivalent to 0.28 g (or 280 mg) sodium per 100 g. For the baseline year we used the mean value of 439 mg/100 g for NZ bread [29]. We also assumed that to ensure ready compliance with the law, the manufacturers aim for an average level in bread that is slightly lower at 270 mg per 100 g. So this would shift the contribution of dietary sodium from bread from 20.6 % [66] to 12.7 % (270/439 x 20.6), i.e., a 7.9 % absolute reduction. That is a reduction of 280 mg/day (out of the baseline consumption of 3544 mg/day [60]). Of note is that this level is still higher than some breads currently on the NZ market e.g., 186 mg/100 g for one multigrain bread [30]. We assumed a phased-in implementation over a five year period (i.e., five equal steps beginning in the baseline year of 2011). | Reduction of 7.9 % (or 280 mg [12.2 mmol] sodium per day for an average adult). For uncertainty we used +/- 10 % of the point estimate (normal distribution). |
Modest limits on sodium in bread: As per the intervention above but for a less stringent limit of ≤400 mg/100 g (i.e., as per a target value for Australia [31]) | As per the row above except that we assumed that to ensure ready compliance with the law, the manufacturers aimed for the 390 mg per 100 g level. So this would shift the dietary contribution of sodium from bread from 20.6 to 18.3 % (390/439 × 20.6), i.e., a 2.3 % absolute reduction. That is a reduction of 81.5 mg/day (out of the baseline consumption of 3544 mg/day). We assumed full implementation in the baseline year. | Reduction of 2.3 % (or 81.5 mg [3.5 mmol] sodium per day for an average adult). For uncertainty we used +/- 10 % of the point estimate (normal distribution). |
-
Salt substitution at 59 % was at a level that was successfully used in a randomised trial in the Netherlands [27]. For the New Zealand population this was estimated to involve a reduction of 51.5 % in daily sodium intake from the reduced intake of processed foods and table salt (or 1824 mg [79.3 mmol] sodium per day for an average adult).
-
Salt substitution at 25 % was at a level used in the majority of the salt substitution studies in a meta-analysis [16]. This was estimated to involve a reduction of 21.8 % in daily sodium intake from the reduced intake of processed foods and table salt (or 773 mg [33.6 mmol] sodium per day for an average adult).
-
Tight limits on sodium in bread (≤280 mg/100 g) were based on the level used in Finnish law for the labelling of low salt bread (i.e., 0.7 % salt) [28]. This was estimated to involve a reduction of 7.9 % (or 280 mg [12.2 mmol] sodium per day for an average adult). For the baseline year, we used the estimated mean value of sodium in New Zealand bread of 439 mg/100 g [29], and we note that some commercial breads are already well below the intervention level (e.g., 186 mg/100 g for one multigrain bread [30]).
-
Modest limits on sodium in bread (of ≤400 mg/100 g) were based on a target value for Australia [31]. This was estimated to involve a reduction of 2.3 % (or 81.5 mg [3.5 mmol] sodium per day for an average adult).
Costing of intervention scenarios and health system costs
Intervention | Intervention costs |
---|---|
Salt substitution at 59 % | The cost was that of a new law for NZ, which was based on the average cost of new act [67] at NZ$ 3,680,000 in 2011 dollars (these NZ$ dollar values are detailed in an online report [68]). Salt substitutes can cost around 50 % more than normal salt [16]. However, as salt is currently extremely cheap (e.g., wholesale prices of $0.7 per kg in NZ) the extra cost for salt from processed food would be under 1 cent per day (for the average NZ intake of 9 g salt per day) and so was ignored. |
It was considered out-of-scope given our health system perspective to consider reformulation costs and costs associated with package labelling changes. Our approach is also in accord with past NZ laws relating to food labelling, alcohol labelling and tobacco labelling in that manufacturers are not compensated for the costs imposed by the new law. For example, the NZ law requiring pictorial health warnings on tobacco packaging did not compensate industry for printing costs or lost sales. Furthermore, we assumed no additional costs from the existing routine evaluation efforts by the NZ Government (nutrition surveys and food surveys) and negligible enforcement and legal costs associated with non-compliance (owing to the relatively low levels of corruption in the NZ setting and the high compliance with laws e.g., the law banning smoking in bars and restaurants [69]). | |
Salt substitution at 25 % | As per the row above, i.e., the cost of a new law. |
Tight limits on sodium in bread (280 mg/100 g) | The cost is just that of the cost of a new law for NZ (see above). As per the arguments above, there was no consideration of reformulation costs and package labelling costs. The technology exists to manufacture lower sodium bread as per examples already present on the NZ market e.g., 186 mg/100 g for one multigrain bread [30]. |
Modest limits on sodium in bread (400 mg/100 g) | As per the row above. |
Validation
Analysis
Results
Modelled intervention | Health gain (QALYs for remainder of the cohort’s life) | Health system cost (NZ$; millions) for remainder of the cohort’s life | Incremental cost-effectiveness ratio (ICER) |
---|---|---|---|
“Do nothing” comparatorb
| 33,200,000 (33,000,000 to 33,500,000) | 162,000 (146,000 to 180,000) | - |
1) Salt substitution at the 59 % level (processed food) | 294,000 (238,000 to 359,000) | −1500 (−1980 to − 1090) | Dominant |
2) “Sinking lid” for salt supply to the marketc
| 211,000 (170,000 to 255,000) | −1110 (−1460 to − 830) | Dominant |
3) “Salt tax”c
| 195,000 (159,000 to 237,000) | −1000 (−1320 to − 740) | Dominant |
4) Salt substitution at 25 % | 121,000 (97,300 to 147,000) | −620 (−820 to − 450) | Dominant |
5) Mandatory 25 % reduction of sodium in all processed foods (“Mandatory-All”c) | 110,000 (87,500 to 135,000) | −600 (−800 to − 440) | Dominant |
6) UK Package (media campaign and voluntary action by industry)c
| 85,100 (69,600 to 102,000) | −440 (−570 to − 320) | Dominant |
7) Mandatory 25 % reduction of sodium in bread, processed meats and sauces (“Mandatory-3G” c) | 61,700 (49,700 to 74,900) | −340 (−440 to − 240) | Dominant |
8) Tight limits on sodium in bread (280 mg/100 g) | 43,500 (34,700 to 52,800) | −220 (−290 to − 160) | Dominant |
9) UK style “Mass Media Campaign”c
| 25,200 (14,200 to 36,700) | −120 (−200 to − 62) | Dominant |
10) Modest limits on sodium in bread (400 mg/100 g) | 15,600 (12,600 to 18,900) | −83 (−110 to − 61) | Dominant |
11) Endorsement Label Programmec(current practice in NZ) | 7900 (5500 to 10,400) | −34 (−52 to − 19) | Dominant |
12) Dietary counselling by dietitiansc(current practice in NZ) | 200 (100 to 330) | 6.90 (4.20 to 10.2) | NZ$36,900 per QALY gained |
Intervention/population group | Incremental (to “do nothing”) cost per adult in NZ$ | QALYs gained per adult |
---|---|---|
Salt substitution at 59 % | ||
Age < 65 years (starting in 2011) | −835 | 0.146 |
Age 65+ years (starting in 2011) | −112 | 0.073 |
Women | −550 | 0.115 |
Men | −760 | 0.141 |
Māori | −533 | 0.163 |
Non-Māori | −658 | 0.123 |
Salt substitution at 25 % | ||
Age < 65 years (starting in 2011) | −348 | 0.060 |
Age 65+ years (starting in 2011) | −46 | 0.030 |
Women | −230 | 0.048 |
Men | −316 | 0.058 |
Māori | −224 | 0.067 |
Non-Māori | −275 | 0.051 |
Tight limits on sodium in bread (280 mg/100 g) | ||
Age < 65 years (starting in 2011) | −125 | 0.022 |
Age 65+ years (starting in 2011) | −16 | 0.011 |
Women | −82 | 0.017 |
Men | −113 | 0.021 |
Māori | −80 | 0.024 |
Non-Māori | −98 | 0.018 |
Modest limits on sodium in bread (400 mg/100 g) | ||
Age < 65 years (starting in 2011) | −45 | 0.0072 |
Age 65+ years (starting in 2011) | −10 | 0.0054 |
Women | −30 | 0.0060 |
Men | −43 | 0.0075 |
Māori | −31 | 0.0087 |
Non-Māori | −37 | 0.0066 |
Age and time distribution of the health gain
Age when the QALYs are gained (i.e., not age in 2011) | In first 10 year period (i.e., 2011 to 2020) | In second 10 year period (i.e., 2021 to 2030) | ||||
---|---|---|---|---|---|---|
QALYs gained | % of QALYs among 45+ year olds | % of QALYs among 55+ year olds | QALYs gained | % of QALYs among 45+ year olds | % of QALYs among 55+ year olds | |
35–44 | 51 | - | ||||
45–54 | 716 | 5.4 % | 1,207 | 2.0 % | ||
55–64 | 2,220 | 16.9 % | 17.8 % | 8,016 | 13.2 % | 13.5 % |
65–69b
| 1,928 | 14.6 % | 15.5 % | 8,978 | 14.8 % | 15.1 % |
70–74 | 1,673 | 12.7 % | 13.4 % | 7,259 | 12.0 % | 12.2 % |
75–84 | 3,787 | 28.7 % | 30.4 % | 20,322 | 33.5 % | 34.2 % |
85–94 | 2,316 | 17.6 % | 18.6 % | 12,180 | 20.1 % | 20.5 % |
95+ | 537 | 4.1 % | 4.3 % | 2,695 | 4.4 % | 4.5 % |
Sum 45+ | 13,177 | 100 % | 60,658 | 100 % | ||
Sum 55+ | 12,461 | 100 % | 59,451 | 100 % |
Discussion
Main findings and interpretation
Study limitations
-
Issues with model structure; and indeed our uncertainty estimates do not capture uncertainty arising from “model structure uncertainty”. For example, this model did not capture potential benefits of salt reduction on preventing stomach cancer [39] and renal disease [40]. Also the benefit of salt substitution was considered to be from the sodium reduction only and did not include any benefits to cardiovascular health arising from the addition of extra potassium (as might be expected as per this systematic review [41]). We also did not consider the benefit of salt substitution in terms of magnesium intake (e.g., the salt used in the trial in the Netherlands [27] used a natural salt from Iceland that was 17 % magnesium salts). Indeed, there is evidence that higher dietary intake of magnesium is associated with reduced CVD mortality according to a systematic review [42]. Similarly, for reduced risk of metabolic syndrome according to another systematic review [43]. Magnesium is also important for bone health [44] and yet median intakes of magnesium in North America are below the Recommended Daily Allowance [44]. Furthermore, a New Zealand study suggests that many women may also have suboptimal intakes given how higher magnesium intake is associated with having infants with higher birthweights [45]. But if any population wide salt substitute intervention did use magnesium-containing salts, then intakes of magnesium should ideally be monitored in ongoing nutrition surveys (to ensure that levels were in a safe range).
-
While there is little scientific uncertainty that high levels of sodium intake are hazardous to health, the point at which dietary sodium reduction ceases to be beneficial is still debated (hence the wide 1000 mg to 5000 mg range for the “theoretical minimum risk exposure level” used in Global Burden of Disease Study 2013 [1]). But as per the collected evidence noted in the Introduction, it still seems reasonable to model benefits of sodium reductions toward the World Health Organization (WHO) target of 2000 mg sodium per day (5 g salt) in adults [4].
-
It is possible that for some processed foods the 59 % sodium substitution level might impact on palatability (though this was not the case in a randomised trial in the Netherlands [27] and in a study that achieved 67 % sodium reduction in bread using potassium chloride and yeast extract for flavour compensation [46]). However, concerns over palatability could conceivably result in the food industry adding more sugar to some processed foods (suggesting the need for careful monitoring by government agencies after any such interventions). Similarly, some people might respond to changes in perceived saltiness of food by adding more salt at the table.