Predicting the effectiveness of interventions on population‐level sodium reduction: A simulation modeling study

Abstract Background and Aims Interventions that significantly reduce dietary sodium intake are anticipated to decrease gastric cancer (GCa) burden. However, the optimal restriction strategies remain unknown at present. This study aims to understand where and to what extent policies modifying sodium consumption change the distribution of GCa burden, and the effects of potential salt reduction strategies in China. Methods The synthetic population in this microscopic simulation study is close to reality. We incorporated estimates of dietary patterns and GCa risk into the model of excess salt consumption. These estimates and simulated population were obtained from the China Health and Nutrition Survey, Global Burden of Disease Project, and the sixth census of China's National Bureau of Statistics, respectively. Results In the no intervention scenario, we estimated that disease burdens due to excess sodium intake would be at 472.9 million disability‐adjusted life years (DALYs) nationally between 2010 and 2030 (95% credible interval [CrI]: 371.1–567.7). The GCa burden caused by high sodium is projected to have a disproportionate impact on the central and southern provinces of China (9.2 and 4.5 million DALYs, respectively). Implementing a cooking salt substitute strategy would be expected to avoid a larger portion of GCa burden (about 67.2%, 95% CrI: 66.8%–67.6%) than the salt‐restriction spoon program (about 16.7%, 95% CrI: 16.1%–17.4%). Conclusion Dietary salt reduction policy is very powerfully effective in reducing the GCa burden overall. It is expected that proposed salt substitutes are more effective than traditional salt‐restriction spoons to avoid increased inequality.


| INTRODUCTION
China's National Nutrition Plan (2017-2030) 1 objectives include reducing the national average daily salt intake by at least 20%.
Although the high sodium diet is associated with an increased risk of gastric cancer (GCa), 2,3 adults in China, particularly those living in the central region, consume far more than the recommended amount in Chinese Dietary Guidelines. 4 Since the majority of sodium intake has come from the consumers' additional table salt during cooking, 5,6 which accounted for 67% of the total sodium intake, an effective strategy for the sodium intake reduction has not been to reformulate packaged and prepared foods, 7 as in the United States 8 and other developed countries. 9 To date, China has not formulated any sodium reduction strategy nationwide. 10 Studies in Chinese participants suggest two dietary salt reduction strategies special for China: saltrestriction spoons and salt substitutes. [11][12][13] The obvious diversity of dietary sodium intake exists among different geographic locations, gender, or age group in China. However, it is still unclear where and to what degree change sodium intake may increase inequality in GCa burden, and which interventions may best mitigate the increase.
Previous mathematical modeling studies have proposed that considerable reductions in sodium intake would be expected to decrease cardiovascular disease 14 and GCa. 15,16 Previous studies have examined the potential impact of dietary salt restriction strategies in adults over the age of 35 and particularly among hypertensive persons, but epidemiological evidence also shows an association between the longer-term dietary and eating behavior in childhood or adolescence and dietary pattern in adulthood in some studies. 17,18 Some successful salt reduction strategies, such as public awareness campaigns, food labeling, and reformulation of processed foods, have been implemented in select cities, counties, and even provinces in China. However, these interventions have not been systematically assessed at interprovincial and interregional. Yet three questions remain unsettled. The first question is whether the potential interventions could produce effect through changes in dietary patterns of various age groups. Diseases caused by excessive sodium intake among young adults under 30 are not as common as in the elderly, however, the prevalence of GCa has remained stable or even slightly increasing trend, and which restriction strategies may best mitigate the increase. 19 Second, how much effect would be expected if the salt reduction strategy currently being implemented in individual provinces of China were extended to more regions. The third unresolved question is whether the salt reduction policy can reduce the inequality of the GCa burden caused by the difference in sodium intake among provinces and regions. We sought to address these questions using individual-level data from health surveys, to examine the potential implications of interventions aimed at reducing salt consumption on reducing health inequalities ideally.
We developed a microsimulation model of salt consumption to

| METHODS
We constructed a discrete-time stochastic microsimulation model ( Figure 1) to estimate the impact and equality of China-specific cooking salt reduction interventions on GCa burdens. The model synthesizes information from the National Bureau of Statistics (NBS) 20 regarding China population structure by province and the F I G U R E 1 Model schematic. Data on salt consumption are used to calculate the change in sodium intake distributions and as inputs for microsimulation to project health outcomes due to sodium changes Health and Nutrition Survey for China (CHNS: 2004-2011) 21 regarding sodium intake estimation procedures to generate a closeto-reality simulated population. 22 This model simulates synthetic individuals rather than aggregate population averages (i.e., a Markov cohort model) and reflects the range of sodium intake given the disparities in dietary patterns in each province, and associated sodium excess status and attendant risk for GCa.

| Simulated population
We simulated 10 000 individuals resembling the 2010 China population to match province-specific gender, age (0-1 year, 1-4 year, and then by 5-year increments), and death trends. We evaluated all provinces subject to CHNS data availability (Supporting Information Text S1).
We calculated the results of each region based on geographic location and major dietary differences, including north (Beijing, Heilongjiang, and Each simulated individual was given probabilistic values for sodium consumption, GCa incidence risk, and mortality risk. The probability distributions of daily sodium intake were based on 24-h dietary recalls data of CHNS, and modified for within-person variations in consumption to estimate usual daily intake. 23 Table 1 highlights key model components. See Figure S1 for a map of modeled provinces.

| Risk factor modeling and model outputs
The focus of this study was primary prevention, hence only the first episode GCa was considered. Trends of GCa-associated risk factors were considered by projecting trends observed in CHNS from 2004 to 2011 (Supporting Information Text S1). The exposure of simulates individuals to sodium was informed by four health and nutrition surveys employing a combination of weighing and three consecutive 24-h recalls between 2004 and 2011. 23 28 We allow the risk-free optimal sodium intake to be 0.6-2 g/day, and the model is 1.5 g/day, following the WHO recommended reference intake, 29 a PERT distribution 30 and the newly released US DRIs. 31 Given the annual sodium intake level of each simulated individual, we determined whether an individual was above the ideal sodium intake level that would be marked "excess" and would make the individual vulnerable to disease. Excess salt consumption is associated with a heightened relative risk (RR) of morbidity and mortality from GCa among young adults aged 18-30, and there is a lag between exposure and GCa. 26 In the simulation, we assumed that a direct effect through excess sodium consumption on GCa incidence

T A B L E 1 Model parameters and sources
Parameters Source  Demographic-specific life expectancies and mortality rates were estimated from World Health Organization (WHO). 25 Prevalence rates, mortality rates, and trends over time for GCa were obtained from the Global Burden of Disease Project (GBD). 24 The prevalence and mortality trends of GCa assumed that the annual percentage changes in per capita mortality and prevalence rate from 1990 to 2008 obtained from GBD data were continuous. The model calculated disease burden in net present DALYs, which are the sum of years lived with disability and years of life lost, discounted at an annual rate of 3%. The disability weight of GCa was obtained from the GBD. 26 In principle, disability weights might vary by age, gender, and personal preference. However, the GBD reports disability weight that is not stratified by these factors. Thus, in keeping with the disability weight methodology set by the GBD, our model did not use stratified disability weights. The primary microsimulation model outcome was the total morbidity and mortality burden associated with excess sodium intake per 1000 people and the overall synthetic population.
We considered culturally tailored dietary salt restriction strategies to mitigate the disease burden from excess sodium consumption.
Specifically, we compared two salt reduction strategies to keep the daily sodium intake within the ideal state. The two public health strategies were (i) restrictions on added salt, including a series of interventions (e.g., promote the use of 2-g salt spoons, teach the measurement of cooking salt, and educate on the potentially harmful effects of excess salt consumption) 11 ; (ii) salt substitutes, in which is composed of 65% sodium chloride (NaCl), 25% potassium chloride, and 10% magnesium sulfate. 12,13 In the first strategy, we used 2-g salt spoons to control the addition of salt during food preparation. In the second strategy, we simulated scenarios that only a certain portion of the five types of high-sodium foods (added salt, soy sauce, MSG, fermented products, and pickled foods) were replaced by lowsodium salts rich in potassium and magnesium. Methods of calculating GBD country-level aggregate estimates used for validation were not involved simulation modeling predictions but rather contemporaneous Bayesian statistics (Supporting Information Text S1).

| Sensitivity analyses
First, we performed one-way sensitivity analysis in which parameters change by 10% at a time in all model inputs (Table S3). In specially, we varied the initial sodium intake distribution across the range to account for potential changes in dietary patterns and the uncertainty of current measurement quality.

| Sensitivity analyses
By using a wide range of literature-based alternative input parameters, the relative advantages of salt substitutes and salt-restriction spoons were unaltered (Tables 2, 3, and S7-S9). Specifically, the relative superiority of salt substitutes did not change qualitatively with alterations in baseline sodium intake, GCa prevalence, or mortality.

| DISCUSSION
We The overall health potential from salt restriction strategies is likely to be great, however, the benefits and potential risks may be heterogeneous among the different populations. The provinces in central China with the highest existing health burden would be expected to benefit most, while people in the south with near-ideal intakes may be subjected to higher risks from very low sodium levels.
Considering the food choice preferences in the South, as well as more health-conscious participation in awareness campaigns and attention to food labels, salt substitution is significantly more effective than limiting salt spoons. Furthermore, the huge differences in cumulative burden among provinces in various regions may be due to socioeconomic inequality. 43 The projected benefits of reducing the GCa burden would be expected to be largest among adults aged 40-59 years given their food consumption profiles and the expected reduction in sodium content in each food based on salt substitution targets. The present rate of reduction in the cumulative burden among these expected to benefit from interventions suggests a limited health effect for salt restriction strategies targeting adults only. 44 Due to the long-term unchanged diet and consumption of less food for adults over 60, the interventions did not significantly reduce the cumulative burden. Our study suggests that careful consideration should be given to how to address such large-scale population-wide sodium reduction strategies to avert the increased health inequalities due to dietary in high sodium.
We bring together the best available epidemiological data, methods, and the way consumers behave in terms of current food preferences across all age groups to calculate expected cumulative burden changes, enhancing the findings from previous sodium