Three salt substitution strategies using potassium chloride were investigated to reduce sodium in three target products in Vietnam (salt, fish sauce and bot canh [a popular seasoning in Vietnam]). All strategies were detailed and costed both in terms of program costs and salt reformulation costs. Population impact of the strategies was also estimated based on the proportion of products reformulated and the adoption of the low salt products by consumers. Inputs including program costs, reformulation costs, program impact and other variables were then modelled to understand the cost-effectiveness of the salt substitution strategies (Figure 1).
Definition of salt substitution strategies
The three salt substitution strategies included voluntary, subsidised and regulatory approaches as proxies for low, medium and high impact interventions respectively (see Table 1). These would lead to the consumers replacing a proportion of the target product consumption with potassium-enriched, low sodium options.
Briefly, the voluntary scenario assumes companies will voluntarily provide new reformulated low sodium products. This requires no government support (low cost) but is assumed to have a lower uptake by consumers and therefore lower impact on population level sodium consumption (see Impact of salt substitution strategies). Both the subsidised and regulatory scenarios require government intervention to either incentivise or mandate food companies to reformulate existing target products with potassium chloride. In the subsidised scenario, the Vietnam Government is assumed to fund reformulation of a proportion of the target products, whereas in the regulatory scenario, the Vietnam Government will mandate that all target products are reformulated with potassium chloride.
As voluntary strategy assumes that initiatives are left to the market and food industry with no involvement or coordination from Government, no coordinated mass media campaign was included. In contrast, the subsidised strategy included a communications and media campaign to drive uptake, as stakeholder research with manufacturers indicated this would be essential to justify R&D costs [28]. Finally, the regulatory strategy included no media campaign as compliance was assured through regulation. This is consistent with a similar scenario modelled for China [4].
Calculation of impact of salt substitution strategies
The decrease in sodium chloride content in food products due to substitution with potassium chloride for each modelled strategy is based on the coverage, efficacy and impact formula described by Gillespie et al [29]. The method takes into consideration the proportion of daily sodium intake from the low sodium target products, the effect of reformulation, the proportion of targeted products to be reformulated and the expected uptake of the low sodium products. This provides an estimated daily reduction in sodium consumption which can be applied to the average daily sodium intake (Table 2).
Recognising the uncertainty of these assumptions and their impact on the model result, a threshold analysis was conducted (see Sensitivity and Threshold Analysis). The model does not consider any beneficial or harmful effects stemming from potassium chloride.
All strategies started from the average salt intake for the Vietnamese population, which was estimated to be 9.40g/day [23]. The salt reduction in target products due to reformulation was consistent throughout the strategies (60% based on previous research [30]).
For the voluntary strategy, it was assumed that manufacturers would choose to reformulate 30% of the target products (salt, fish sauce and bot canh). However due to reformulation costs being passed onto consumers, it was assumed uptake of these products would be low (5%). To ascertain the impact of this strategy, the daily salt intake (9.4g) was multiplied by the following variables per Gillespie et al [29]: the proportion of dietary salt from the target products (70%); the salt reduction in target products due to reformulation (60%); the proportion of products reformulated (30%); the product uptake (5%). This resulted in a modest total dietary salt reduction of 0.06g/day (Table 2).
In contrast, the regulatory strategy was assumed to have maximum uptake as legislation would guarantee all target products are reformulated with potassium chloride, which would lead to 100% of people having to use the reformulated products. The latter assumption takes into account the high adoption rate (>90%) seen in the Vietnam National Iodisation Program [31]. The impact of the regulatory strategy on dietary salt intake was calculated by multiplying the baseline daily salt intake (9.4g/day) by the following variables: the proportion of dietary salt from the target products (70%); the salt reduction in target products due to reformulation (60%); the proportion of products reformulated (100%); the product uptake (100%). This resulted in a total dietary salt reduction of 3.95g/day (Table 2).
For the subsidised intervention, it was assumed 50% of the target products would be reformulated with potassium chloride. It was assumed a government subsidy would be used to supply potassium chloride at the same price or cheaper than sodium chloride. This would lead to reformulated products being sold at the same price or cheaper than regular products. Product uptake of the reformulated products was based on a Vietnamese population survey that found approximately 44% of respondents would limit adding salt or sauces when cooking when given the option [32]. Furthermore, the media and health promotion campaign accompanying the subsidised programme is expected to have an additive effect on the intake of high-sodium products and is reflected as an extra 5% reduction from baseline [13].
The impact of the subsidised strategy on dietary salt intake was calculated by multiplying the baseline daily salt intake (9.4g/day) by the following variables: the proportion of dietary salt from the target products (70%); the salt reduction in target products due to reformulation (60%); the proportion of products reformulated (50%); the product uptake (44%). This resulted in a total dietary salt reduction of 0.87g/day (Table 2).
Assumptions for strategy implementation
To reflect the real-world planning and management of a population health intervention, each salt substitution strategy included progressive phases of implementation, modelled on Webb 2017 [26]. Specifically, the first phase includes two years of project management, training and meetings, advocacy and law enforcement prior to deploying the programme. These years are dedicated to the planning and development of the intervention and therefore no health effects are assumed to take place. Following this stage, the voluntary and subsidised strategy included a partial implementation phase, resulting in 50% of the total salt reduction effect of each respective strategy, to account for progressive uptake. The regulatory strategy assumes that from year three the total effect of the programme would be realised, as once legislation is implemented the programme should be 100% effective (leveraging experience from the Vietnam National Iodisation Program [31]). From years six onwards, all programmes are assumed to be at full implementation.
Cost-effectiveness model structure and perspective
A Markov cohort model was developed in TreeAge Pro based on approaches adopted in previous publications [13, 17, 26]. The model included four health states, namely healthy, post-stroke, post-IHD and death. Annual transitions captured the incidence of stroke and IHD events, mortality due to IHD or stoke events and natural mortality.
The model starts with people aged 30 years being assigned to live normally in the current environment (no intervention), or alternately, live with one of three salt substitution strategies (voluntary, subsidised or regulatory). A base case age of 30 years was selected as being representative of the median age of the Vietnam population and as it had been used previously to assess cost-effectiveness of salt reduction in Vietnam [13]. At the beginning of the model, the average SBP is calculated according to the Vietnamese population characteristics [13, 33]. The model links sodium intake to SBP based on the linear regression model published by Law et al. 1991 [34]. Secondly, SBP is linked to the probability of IHD or stroke based on Cobiac et al. 2012 [17].
Patients that experience a stroke or IHD event either die as a result of the event or progress to a “post event” health state. In the post-stroke health state, patients have an increased mortality risk compared to the healthy cohort for the lifetime of the model, whereas in the post-IHD cohort there is an increased mortality risk for three years post event, after which mortality reverts to the natural mortality risk of that age (see input parameters). As the model has a yearly cycle length, the acute phase (initial, short-term event) for IHD and stroke is captured in a state transition, and the chronic phase (post event, long term) of the disease is captured in the respective health states. Patients cannot transition from the post-stroke or post-IHD state to the healthy state, and conservatively it was assumed patients cannot experience more than one stroke or IHD event.
The time horizon of the economic model spans the lifetime of the Vietnamese population (capped at 100 years of age). A lifetime horizon (which follows a cohort from age 30 until death) was used to fully capture the benefits and costs associated with the salt substitution strategy. The model uses a discount rate of 3% for both benefits and costs; rates of 0% and 5% were used in sensitivity analysis. Given the aims of the study, this population wide model evaluates each scenario from a Vietnam Government perspective and excludes costs borne by industry or individuals.
Cost-effectiveness model input parameters
A summary of key model inputs is presented in Table 3.
Clinical events and epidemiology
Normative data for blood pressure and the incidence of stroke and IHD were sourced from Ha 2011 [13] (see Supplementary Table 1 and 2). This paper provided the most applicable data characterising the relationship between blood pressure and incidence of IHD/stroke for the Vietnamese population. The impact of sodium intake on blood pressure was estimated based on published research by Law 1991 [34], which aligns with previous economic evaluations. The relationship between blood pressure and stroke/IHD was calculated based on the percentage relative risk reduction of stroke (6.3% per 1% SBP reduction) and IHD (3.4% per 1% SBP reduction) as published by Cobiac 2012 [17]. The resulting relative risk reduction for each sodium reduction program is provided in the Supplementary Material (see Supplementary Table 3 and 4).
At any point in the model, patients are assumed to be at risk of death. This represents the probability that a person of a specific age will die before their next birthday. All-cause mortality was sourced from the Global Health Observatory data repository Vietnam life tables, stratified by age and sex [35].
The incidence of mortality following stroke was sourced from Trishwell 2012 [36] and Kiyohara 2003 [37]. The former provided short-term mortality following the event (estimated to be 37%) the latter provided the long term risk of mortality relative to the general population, stratified by year. The risk of mortality following IHD was sourced from the Southeast Asian NCD impact module dataset through the WHO-CHOICE OneHealth tool [38] as well as Tang 2007 [39]. The former provided the age specific mortality risk following IHD and the latter provided long term mortality risk, stratified by year. It was assumed after Year 3 patients have same mortality risk as the rest of the “healthy” population.
Cost inputs
All costs were estimated from a Government perspective in 2019 Vietnamese Dong (VND) at an exchange rate of US$1 = 23,210 ₫ [40]. The purchasing power parity (PPP) of 2019 VND from 2015 US$ was calculated to be 7,792 using the CCEMG – EPPI-Centre Cost Converter [40].
Given the Government’s minimum involvement in the voluntary strategy, it was assumed no programme implementation costs would be accrued. It is recognised that these costs would be borne by consumers, however this was excluded from the analysis.
Programme costs for the regulatory and subsidised strategies include resources required in the planning, development, and implementation of a population-based health intervention as described by the WHO-CHOICE methodology [38]. This includes the estimated unit price of human resources, training, meetings, supplies, equipment and mass media campaigns (for subsidised strategy only). The resource needs for the regulatory and subsidised interventions were assessed at both a national and provincial level, to reflect regional nuances in cultural and dietary behaviours between provinces. These needs were based on Webb 2017 [26]. Personnel payment norms were based on unified cost norms as issued by United Nations Agencies in Vietnam, European Union (UN-EU) and the Ministry of Planning and Investment of Vietnam [41], inflated to 2019 VND. A per diem daily subsistence allowance of US$173 for attendees of meetings and training was applied, accounting for a travel allowance [42]. The cost of media and communications for the subsidised scenario was incorporated at both a national and provincial level, based on research by Ha 2011 [13].
In addition to programme management costs, the cost of a government subsidy for potassium chloride was included in the subsidised strategy. According to the Vietnamese Ministry of Agriculture and Rural Development, Vietnam produces approximately 534,798 tonnes of salt for human consumption each year [43]. Previous research indicated approximately 70% of salt intake comes from bot canh, fish sauce and salt added when cooking [24]. The literature pertaining to the cost of substituting sodium chloride with potassium chloride is scarce, and as a result, the cost of salt iodisation ($0.04USD per kg) was used as a proxy to estimate the cost of salt reformulation per kilogram [43]. The resulting cost estimate was 1,791₫ per capita, reflecting a proxy for the cost of providing potassium chloride for manufacturers at a similar price to regular salt. As the process of salt substitution with potassium chloride is more complex than salt iodisation this assumption is assessed in a sensitivity analysis.
Healthcare costs were derived from Vietnamese specific publications [44, 45] with the exception of the cost of acute stroke event which was not available from Vietnamese sources and therefore obtained from a Thai study [46]. To align with incidence data, the cost of IHD was calculated for the acute event, and for the long term recurring cost of chronic treatment post initial health event [44]. There was no long term cost of stroke applied in the model as rehabilitation and long term care is commonly done by family members at home [44].
According to the WHO, the Vietnamese Government pays for approximately 54% of total healthcare expenditure which was verified by a local source [47]. To reflect a Government perspective, all healthcare costs were calculated accordingly.
Quality of life
Quality of life values for healthy, post-stroke and post-IHD health states were applied in the model, and disutility values for stroke and IHD events were applied at health state transitions.
Quality of life values for the Vietnamese population were gender specific for the whole cohort according to blood pressure status, including patients having “ideal” SBP (<130 mmHg) or Stage 1 SBP (>130 mmHg) [48].
The long term stroke utility values were sourced from a UK disease-specific population study of the quality of life of patients post-stroke [49]. The quality of life of patients in the post-IHD health state was calculated using a Vietnamese population specific odds ratio comparing the utility of patients with and without history of previous cerebrovascular event [48], which has previously been used as a proxy for the long term quality of life in patients with stable cardiovascular disease [44].
Disutility values for acute stroke and IHD events were sourced from the Global Burden of Disease study [50]. Disutilities represent the decrement in quality of life due to symptoms or events associated with stroke and IHD events. In the model, disutilities are applied once at the time of the event, as well as a recurring disutility that signifies the long term quality of life lost postacute event. When someone experiences a stroke, their immediate quality of life markedly lowers compared to someone who is healthy, thus they receive a disutility of -0.312 [50]. Similarly for IHD, a person is assigned a one off quality of life decrement of -0.186 when experiencing the acute IHD event [50]. Those who suffer a stroke or IHD event will often experience long term impacts on their quality of life, referred to in the model as the post-IHD and post-stroke health states.
Sensitivity and threshold analyses
A number of one-way sensitivity analyses were run to identify key model drivers and assess any uncertainties. The parameters tested and each upper and lower variable are listed in Table 4.
In addition, the uncertainty surrounding the impact of the individual salt substitution strategies was recognised and tested in a threshold analysis. Specifically, the average salt reduction needed for average costs to equal cost savings was tested at varying time horizons.
Validation
In order to validate the model results, a comparison was undertaken between the model assumptions and results in relation to previous publications including Ha 2011 [13], Webb 2017 [26] and Cobiac 2010 [18].