Crystal M. Smith-Spangler, MD; Jessie L. Juusola, MS; Eva A. Enns, MS; Douglas K. Owens, MD, MS; Alan M. Garber, MD, PhD
Disclaimer: The content is solely the responsibility of the authors and does not necessarily represent the official views of the Department of Veterans Affairs.
Acknowledgment: The authors thank the reviewers for their thorough and helpful reviews and Sarah Adler, MBA, Katherine Steele, MS, and Iliana Jaatmaa Harrysson, BS, who contributed to earlier versions of this analysis.
Grant Support: By the Veterans Affairs Palo Alto Health Care System, Stanford University Management Science and Engineering Advisory Board Fellowship Fund, National Defense Science and Engineering Graduate Fellowship, National Science Foundation Graduate Fellowship, and Department of Veterans Affairs.
Potential Conflicts of Interest: Disclosures can be viewed at www.acponline.org/authors/icmje/ConflictOfInterestForms.do?msNum=M09-2224.
Reproducible Research Statement:Study protocol, statistical code, and data set: Available from Dr. Smith-Spangler (e-mail, firstname.lastname@example.org).
Requests for Single Reprints: Crystal M. Smith-Spangler, MD, Center for Health Policy, Stanford University, 117 Encina Commons, Stanford, CA 94305; e-mail, email@example.com.
Current Author Addresses: Drs. Smith-Spangler, Owens, and Garber and Ms. Enns: Stanford University, Center for Health Policy, 117 Encina Commons, Stanford, CA 94305.
Ms. Juusola: Stanford University, Department of Management Science and Engineering, 499 Terman Engineering Center, Stanford, CA 94305.
Author Contributions: Conception and design: C.M. Smith-Spangler, J.L. Juusola, D.K. Owens, A.M. Garber.
Analysis and interpretation of the data: C.M. Smith-Spangler, J.L. Juusola, E.A. Enns, D.K. Owens, A.M. Garber.
Drafting of the article: C.M. Smith-Spangler, J.L. Juusola, D.K. Owens.
Critical revision of the article for important intellectual content: C.M. Smith-Spangler, D.K. Owens, A.M. Garber.
Final approval of the article: C.M. Smith-Spangler, D.K. Owens, A.M. Garber.
Statistical expertise: C.M. Smith-Spangler, D.K. Owens, A.M. Garber.
Obtaining of funding: D.K. Owens.
Administrative, technical, or logistic support: A.M. Garber.
Collection and assembly of data: J.L. Juusola, E.A. Enns.
Smith-Spangler CM, Juusola JL, Enns EA, Owens DK, Garber AM. Population Strategies to Decrease Sodium Intake and the Burden of Cardiovascular Disease: A Cost-Effectiveness Analysis. Ann Intern Med. 2010;152:481-487. doi: 10.7326/0003-4819-152-8-201004200-00212
Download citation file:
Published: Ann Intern Med. 2010;152(8):481-487.
Sodium consumption raises blood pressure, increasing the risk for heart attack and stroke. Several countries, including the United States, are considering strategies to decrease population sodium intake.
To assess the cost-effectiveness of 2 population strategies to reduce sodium intake: government collaboration with food manufacturers to voluntarily cut sodium in processed foods, modeled on the United Kingdom experience, and a sodium tax.
A Markov model was constructed with 4 health states: well, acute myocardial infarction (MI), acute stroke, and history of MI or stroke.
Medical Panel Expenditure Survey (2006), Framingham Heart Study (1980 to 2003), Dietary Approaches to Stop Hypertension trial, and other published data.
U.S. adults aged 40 to 85 years.
Incremental costs (2008 U.S. dollars), quality-adjusted life-years (QALYs), and MIs and strokes averted.
Collaboration with industry that decreases mean population sodium intake by 9.5% averts 513Â 885 strokes and 480Â 358 MIs over the lifetime of adults aged 40 to 85 years who are alive today compared with the status quo, increasing QALYs by 2.1 million and saving $32.1 billion in medical costs. A tax on sodium that decreases population sodium intake by 6% increases QALYs by 1.3 million and saves $22.4 billion over the same period.
Results are sensitive to the assumption that consumers have no disutility with modest reductions in sodium intake.
Efforts to reduce population sodium intake could result in other dietary changes that are difficult to predict.
Strategies to reduce sodium intake on a population level in the United States are likely to substantially reduce stroke and MI incidence, which would save billions of dollars in medical expenses.
Department of Veterans Affairs, Stanford University, and National Science Foundation.
Sodium restriction lowers blood pressure, which in turn reduces heart disease, but most persons cannot voluntarily reduce their sodium intake. Population strategies, such as voluntary reduction of food sodium content by manufacturers or taxing sodium, may be more effective.
This cost-effectiveness analysis suggests that either population strategy would lead to dramatic decreases in the number of strokes and myocardial infarctions and would save the health system billions of dollars.
The analysis does not account for unintended consequences of reducing sodium, such as a compensatory increase in consumption of calories.
Population strategies to reduce sodium intake could lead to dramatic improvements in health and could save billions of dollars.
Adults in the United States consume an estimated 3900 mg of sodium per day (75% of which comes from processed foods ), much more than the maximum recommended intake of 2300 mg/d (2, 3). Persons with hypertension, African Americans, and persons older than 40 years, or about 70% of U.S. adults, should consume no more than 1500 mg of sodium daily (4). Moderate reductions in sodium intake decrease systolic blood pressure (SBP) by about 2 to 5 mm Hg (5). Increased blood pressure is associated with increased mortality from ischemic heart disease and stroke (6).
The success of recent efforts in the United Kingdom to reduce population sodium consumption has sparked interest worldwide in policies to decrease sodium intake (7). The United Kingdom Food Standards Agency, an independent government department that monitors food safety, began working with manufacturers in 2003 to cut sodium in processed foods by developing voluntary maximum sodium targets for specific foods. Manufacturers have successfully decreased the sodium content of many products (8), leading to an estimated 9.5% decrease in population sodium intake (9–11).
Although no country has implemented a tax to decrease sodium consumption, economic incentives affect consumer behavior (12, 13), and taxes have been successful in reducing tobacco and alcohol consumption (14, 15). Persons often ignore large future costs when offered smaller, more immediate benefits (16), and a tax on sodium could theoretically increase awareness of the long-term costs of sodium consumption.
Two recent analyses (17, 18) estimated the number of cases of hypertension that could be averted if U.S. adults consumed less sodium, finding substantial gains in quality-adjusted life-years (QALYs) and cost savings. However, the studies did not examine specific policies that might achieve such changes. We examine the health benefits and costs, in terms of risk for myocardial infarction (MI) and stroke, of 2 governmental strategies to reduce sodium intake in the United States by using a mathematical model that projects cardiovascular event rates and costs. The 2 strategies are in collaboration with industry to set voluntary maximum sodium targets for processed foods, modeled on the United Kingdom experience, and a tax on sodium used for food production. We aim to inform decision makers about the potential effect of these strategies in the United States. This will complement the efforts of the Institute of Medicine (19), which is preparing to issue recommendations for decreasing sodium intake later this year.
We created a computer-simulated, state-transition (Markov cohort) model of incidence, prevalence, mortality, and direct costs associated with stroke and MI in U.S. adults aged 40 to 85 years. We assessed costs and benefits from a societal perspective and discounted them at 3% per year. We used the decision model to estimate length and quality of life and medical expenditures over the lifetime of all U.S. adults aged 40 to 85 years; outside this age range, estimates of the effects of sodium reduction on SBP and on the incidence of stroke and MI are either not available or less reliable. We evaluated our assumptions extensively in sensitivity analyses, including the use of 2 alternate relationships between sodium intake and blood pressure. We created the model and performed analyses by using TreeAge Pro 2008 (TreeAge Software, Williamstown, Massachusetts) and Matlab (MathWorks, Natick, Massachusetts) to check for input errors and to take advantage of different software capabilities. Our cost-effectiveness analysis adheres to the recommendations of the Panel on Cost-effectiveness in Health and Medicine (20). The Appendix contains additional details.
Our model consists of 5 health states (Appendix Figure 1). The “well” state includes persons without a history of MI or stroke; persons with histories of either condition are assigned to the “cardiovascular disease” (CVD) state. Each month, persons in the well and CVD states were at risk for MI, stroke, or death from other causes. Persons remained in the “acute MI or stroke” state for 1 month and then progressed to either the “dead” state or the CVD state. Once in the CVD state, persons could not return to the well state but could have recurrent MI or stroke.
The model represents the clinical events that can occur at 1-month intervals. Persons in the well state have never had a heart attack or stroke, whereas persons in the CVD state have a history of heart attack or stroke. During each 1-month period, a person in the well or CVD state is at risk for acute MI, acute stroke, or noncardiac death. Persons remain in the acute MI or acute stroke state for up to 1 month and either die of MI or stroke or move to the CVD state. Persons in the CVD state may not return to the well state. CVD = cardiovascular disease; MI = myocardial infarction.
Population data came from U.S. Census estimates (21), and we estimated non-CVD mortality risk from cause-specific adult mortality data (22). We estimated monthly risks for first MI and stroke and mortality after MI by fitting exponential curves to Framingham Heart Study data (both cohorts, 1980 to 2003) (23). We estimated mortality from stroke by fitting exponential curves to data from the Cardiovascular Health Study (23) (Appendix Table 1). We assumed the risk for MI or stroke, given history of MI or stroke, to be twice that of a person without history of MI or stroke (range of 1.5 to 5, based on estimates from the literature) (24, 25). Because persons with history of MI or stroke have a higher risk for complications, such as congestive heart failure, aspiration pneumonia, and falls, we assumed the risk for death from non-MI and nonstroke causes in the CVD state to be 1.5 times that of a person without a history of MI or stroke (range in sensitivity analysis, 1 to 2.2) (26).
Appendix Table 1.
We calculated baseline medical costs incurred by patients with and without a history of MI or stroke from the household component of 2006 Medical Expenditure Panel Survey (27) by using SAS 9.1.3 statistical software (SAS, Cary, North Carolina) (Appendix). We obtained acute event costs (28, 29) and health state utilities (30, 31) from the medical literature. We adjusted all costs to 2008 U.S. dollars by using the overall Consumer Price Index (32).
To estimate the prevalence of CVD at each age, we tracked for 45 years or until death a cohort of men and women aged 40 years without history of MI or stroke. Our model's calculated prevalence of CVD was similar to estimates of CVD from the 1999 to 2002 National Health and Nutrition Examination Survey (23) (Appendix).
The effect of modest reductions in sodium intake on quality of life is unknown, although some evidence suggests that many adults find these reductions acceptable (33–36) and that consumers' preferred level of salt in foods decreases after a period of decreased sodium intake (37). Therefore, we assumed no loss of quality of life due to decreased sodium intake. In sensitivity analyses, we allowed the utility of a lower-sodium diet to vary from 0.990 to 1 and allowed the decrement to last a lifetime. We performed an additional probabilistic sensitivity analysis that assigned the quality-of-life impairment from reduced sodium intake to the first 2 years of the intervention only.
We estimated that collaboration with industry would decrease sodium intake by the same amount as in the United Kingdom (9.5%), with a range of 5% to 40% (the United Kingdom goal is to decrease sodium intake by 37%) (9–11). For the tax strategy, we envisioned a national excise tax on sodium used for food production at the industrial level that would increase the price of salty foods by 40%, a level similar to the cigarette tax (38). A tax on sodium at the industrial level simplifies the administration of the tax, and taxing foods exceeding a threshold might simply induce manufacturers to change the serving size. We assumed that producers would pass the cost of the tax on to consumers in the price of foods but not change the sodium content of food products. We estimated that an increase in price of 40% would result in a decrease in sodium intake of 6.0% (range, 1.2% to 21.6%) (price elasticity of demand [the percentage change in quantity demanded divided by the percentage change in price, which is a measure of the sensitivity of demand to price], −0.15; range, −0.03 to −0.54) (39, 40), similar to other authors' (41) estimates of the effect of a tax on sodium consumption.
To predict the change in SBP under each strategy, we applied the predicted percentage change in sodium intake to sex-specific estimates of average U.S. sodium intake (42) and inputted this estimate into sodium–SBP dose–response curves derived from the Dietary Approaches to Stop Hypertension trial data (43), adjusted for age (44) (Appendix Table 2 lists predicted change in SBP and the Appendix lists the equations). In separate sensitivity analyses, we assumed that decreases in SBP were independent of age, and we used an alternate relationship between sodium and SBP (45). We assumed that strategies to decrease sodium intake decrease the incidence of first stroke or MI but do not change survival after MI or stroke has occurred. To simplify the model, we assumed that the full effect of each strategy on sodium consumption would be achieved immediately and last a lifetime.
Appendix Table 2.
We used data from a meta-analysis of prospective patient-level data on blood pressure and CVD mortality from more than 1 million adults without previous CVD (6) to estimate the relative risk for first stroke or MI on the basis of each intervention's calculated decrease in SBP (Appendix Table 3). We compared our model's predictions with a recent meta-analysis of prospective studies that examined sodium intake (46) and found the model's predictions of stroke risk to be similar (Appendix).
Appendix Table 3.
The per-person cost of collaborating with industry was estimated by dividing the annual budget of the United Kingdom Food Standards Agency in 2008 (47) by the number of adults aged 40 to 85 years in the United Kingdom (48), converting this to U.S. dollars by using the World Bank purchasing power parity coefficient (49) and assuming that 15% of the Food Standards Agency's budget was spent on the salt campaign (range in sensitivity analysis, 5% to 33%). By using this per-person cost in the United Kingdom and assigning it to each person aged 40 to 85 years in the United States, we allow for the total cost of collaboration with industry to be higher in the United States than in the United Kingdom. For example, if the United Kingdom spends the equivalent of $32 million annually ($1.53 per person aged 40 to 85 years), the U.S. government would spend $202 million annually. We did not include costs of reformulation incurred by manufacturers, because processors generally included sodium reduction in previously planned product reformulations, which usually happen at least every 3 years (8). We assumed that the costs of administering the tax strategy would be negligible. Tax revenue would subsidize the cost of lower-sodium foods for low-income persons. Therefore, the tax functions as a transfer of money between groups, so we do not include the revenue from the tax as a benefit of the strategy (Appendix). Table 1 lists selected model inputs.
The authors were supported by the Department of Veterans Affairs, Stanford University, and the National Science Foundation. The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; or preparation, review, and approval of the manuscript.
Collaboration with industry to achieve a 9.5% reduction in population sodium intake would result in a 1.25–mm Hg decrease in mean SBP of persons aged 40 to 85 years. This blood pressure reduction, in turn, would avert 513 885 strokes and 480 358 MIs and increase life-years lived by more than 1.3 million over the lifetime of U.S. adults aged 40 to 85 years alive today, saving $32.1 billion in direct medical costs (Table 2). Forty-four percent of the savings ($14 billion) comes from fewer hospitalizations for acute stroke and acute MI, and the remaining savings come from fewer persons incurring higher long-term medical costs due to MI or stroke. Sixty-six percent of the savings ($21.1 billion) accrue to persons aged 65 years or older. Larger reductions in sodium intake lead to greater benefits (Appendix Figure 2). A sodium tax achieving a 6% decrease in sodium intake and a 0.93–mm Hg decrease in mean SBP would avert 327 892 strokes and 306 137 MIs and increase life-years lived by 840 113, saving $22.4 billion over the lifetime of adults aged 40 to 85 years alive today.
For collaboration with industry, we plot the estimated effects of a 9.5% decrease in population sodium intake as well as the effects if greater decreases in population sodium intake are achieved (20% and 40%). QALY = quality-adjusted life-year.
One-way and 2-way sensitivity analyses reveal that either strategy is preferable to the status quo, unless consumption of less sodium causes lower quality of life. A net loss of QALYs may result when this utility is lower than 0.9975 to 0.9985. In all scenarios analyzed in sensitivity analysis (2 alternate relationships between sodium and SBP and removal of intervention effect during the first 2 years of implementation), both strategies remained cost saving and increased QALYs with only modest increases or decreases in the magnitude of the changes as long as reduced sodium consumption did not decrease quality of life (Appendix Table 4).
Appendix Table 4.
Probabilistic sensitivity analysis assuming no loss of utility from consuming less sodium shows that the strategy of collaboration with industry always results in a gain in QALYs and, in 98% of simulations, cost savings compared with the status quo. Savings over the lifetime of U.S. adults aged 40 to 85 years were at least $10 billion in 78% of simulations. In 40% of the simulations, savings were $30 billion or more (Appendix Figure 3). In a second probabilistic analysis that included a lifetime loss of utility from a reduced-sodium diet, the likelihood of cost savings did not change for collaboration with industry, although QALYs were lost in 43% of the simulations (Appendix Figure 4 and Appendix Table 5). Only 5% of the simulations resulted in cost savings greater than $50 000 per QALY lost. In 54% of the simulations, collaboration with industry resulted in a gain in QALYs and cost savings. If the impairment of quality of life from a reduced-sodium diet lasts only 2 years instead of a lifetime, it was more likely that the policies both increased QALYs and decreased costs. In 91% of simulations, collaboration with industry resulted in cost savings and an increase in QALYs compared with the status quo. In 7% of the simulations, collaboration with industry resulted in cost savings but a loss of QALYs.
We have plotted the percentage of simulations from the probabilistic sensitivity analysis that achieved different levels of costs savings for collaboration with industry versus doing nothing. For example, there is a 98% chance of any cost savings with collaboration compared with doing nothing, a 78% chance of saving at least $10 billion, and a 40% chance of saving at least $30 billion. The probability of cost savings is the same with and without inclusion of a quality-of-life impairment due to the reduced-sodium diet.
Assume a lifetime quality-of-life impairment with the reduced-sodium diet. For each simulation, the computer selects a set of values from the distribution of each variable in the model and estimates the costs and QALYs for the intervention and the status quo. Incremental costs and QALYs are calculated by subtracting the intervention estimate from doing nothing. This method of sensitivity analysis allows all variables in the model to vary across their estimated distributions with each simulation. The mean reduction of sodium intake for collaboration with industry is 9.5%. The Appendix describes the distributions used. In 54% of simulations, collaboration with industry both saves money and gains QALYs (lower-right quadrant). QALY = quality-adjusted life-year.
Appendix Table 5.
Both collaboration with industry and a sodium tax would increase QALYs and reduce medical costs among middle-aged and elderly adults, according to our results. If the United States achieved reductions similar to those achieved in the United Kingdom by collaborating with industry, 513 885 strokes and 480 358 MIs would be averted, saving more than $32.1 billion in medical costs over the lifetime of adults aged 40 to 85 years. Our results were robust in sensitivity analyses unless reduction of dietary sodium causes significant disutility. Collaboration with industry achieves greater health benefits and lower costs than the sodium tax, because we have assumed that manufacturers do not reformulate their products in response to a tax and because demand for salty foods is currently relatively unresponsive to prices. The availability of acceptable alternatives would make the demand for salty foods more responsive to prices, increasing the effectiveness of a tax.
If demand for salty foods is relatively unresponsive to prices, a tax on industrial salt to increase the price of the saltiest foods by 40% would need to be very large, about 160 000% of the current price of food-grade salt. Such a tax would generate approximately $400 billion per year, or about $1300 per U.S. adult per year, but this is neither realistic nor necessarily advantageous. For example, it could lead to a disproportionately large increase in the cost of foods consumed by poor persons, and Lakdawalla and colleagues (50) have shown that higher food prices lead to greater prevalence of nutritional deficiencies in the United States.
If modest reductions in sodium intake decrease quality of life by 0.15% (for example, from 1 to 0.9985) annually over a lifetime, gains in QALYs from sodium reduction strategies are eliminated and a loss of QALYs could result. A utility of 0.9985 means that adults would be willing to give up 1 half-day of life per year for the rest of their life to consume a high-sodium diet. However, studies of modest reductions in sodium intake (33–37) suggest that consumer's preferred level of sodium consumption resets once sodium intake decreases to a new level and that most persons find modest reductions acceptable; therefore, we suspect that any loss of quality of life due to decreased sodium intake would be temporary.
The United Kingdom has reduced population sodium intake by 9.5% over the past 5 years, but we do not know whether further reductions in sodium intake can be achieved, whether these reductions are sustainable, or whether they have resulted in decreased rates of MI or stroke. Consumers could add table salt back to their food (51), or technological difficulties could limit the magnitude of sodium reduction in processed foods. If sodium consumption returns to previous levels, the benefits of sodium reduction strategies would be eliminated.
We did not formally study other unintended consequences of population sodium reduction, because a recent review (45) concluded that modest reductions in sodium intake do not impose substantial health risks. Although iodized salt is a principal source of dietary iodine, suggesting that reduced salt intake could lead to iodine deficiency, the salt in most processed foods in the United States is not iodized (52).
Sodium reductions may lead persons to consume more fats and sugars or simply more calories, leading to other health risks. Analysis of 2000 United Kingdom food survey data (41) suggests that fat and salt can function as substitutes and that, although increases in the price of salty foods can decrease salt intake, they might also lead to small, compensatory increases in fat intake, which could in turn diminish the health benefits and savings from sodium reduction strategies. However, these data may not apply to processed foods that are reformulated to contain less sodium while maintaining their appeal to consumers. Furthermore, just as salt reduction may lead to increased consumption of substitute nutrients, it may reduce consumption of unhealthy complements. Carbonated beverages, for example, complement salt intake, so consumption of “empty calories” in this form may decrease with salt consumption (53). It will be important to monitor the unintended consequences, both positive and negative, of any strategy for decreasing sodium intake.
Our analysis does not incorporate the cost savings from treating fewer cases of hypertension, end-stage kidney disease, or hypertensive (nonischemic) heart failure, although costs associated with ischemic heart failure are implicitly included in our cost estimates. Inclusion of these conditions would make strategies to reduce sodium intake even more cost saving. In addition, minority populations are underrepresented in the mostly white Framingham Heart Study, which informed our model's predictions of heart attack and stroke. African Americans with a higher incidence of heart attack and stroke may benefit more than the general population from efforts to reduce blood pressure through population dietary sodium reduction.
Finally, other dietary and lifestyle interventions, such as increasing intake of fresh fruits and vegetables, decreasing calorie and fat intake, and increasing physical activity, may be effective at reducing morbidity and mortality (18, 54, 55). Although few countries have developed population strategies to address these modifiable risk factors for disease, they represent complementary approaches that may also be considered.
This analysis shows that efforts to reduce population sodium intake through U.S. government policies are likely to avert acute strokes and MIs, increase QALYs, and save billions of dollars in medical costs. Collaboration with industry to establish voluntary sodium targets in processed foods is likely to be more effective than a sodium tax and seems to be an appropriate first step toward reducing population sodium intake and the burden of CVD.
The In the Clinic® slide sets are owned and copyrighted by the American College of Physicians (ACP). All text, graphics, trademarks, and other intellectual property incorporated into the slide sets remain the sole and exclusive property of the ACP. The slide sets may be used only by the person who downloads or purchases them and only for the purpose of presenting them during not-for-profit educational activities. Users may incorporate the entire slide set or selected individual slides into their own teaching presentations but may not alter the content of the slides in any way or remove the ACP copyright notice. Users may make print copies for use as hand-outs for the audience the user is personally addressing but may not otherwise reproduce or distribute the slides by any means or media, including but not limited to sending them as e-mail attachments, posting them on Internet or Intranet sites, publishing them in meeting proceedings, or making them available for sale or distribution in any unauthorized form, without the express written permission of the ACP. Unauthorized use of the In the Clinic slide sets will constitute copyright infringement.
Cardiology, High Value Care.
Results provided by:
Copyright © 2016 American College of Physicians. All Rights Reserved.
Print ISSN: 0003-4819 | Online ISSN: 1539-3704
Conditions of Use
This PDF is available to Subscribers Only