Strengthen Methods to Measure Potassium and Sodium Intake

Collection of multiple 24-hour urinary sodium excretions with quality control methods is considered the most accurate method for quantifying usual sodium intake in an individual (Holbrook et al., 1984; Lerchl et al., 2015). Even when conducted correctly, this method has limitations. Like all urinary sodium assessments, 24-hour urinary excretion does not capture all modes of sodium loss (e.g., sweat, fecal), and therefore underestimates the true quantity of sodium consumed. Furthermore, collection of 24-hour urine samples is burdensome for participants, which often makes this assessment method challenging to use in a large, free-living population.

Spot urine samples collected at a single point in time are more convenient to collect than 24-hour urinary excretions, but they are subject to multiple errors owing to intra- and inter-day variations in urine osmolality, sodium excretion, meal timing, fluid intake, ambient temperature, physical activity, and diuretic use (Ji et al., 2012). Spot urine samples have also demonstrated systematic bias as compared to 24-hour urine samples—underestimating 24-hour urinary sodium excretion at high intake levels, and overestimating 24-hour urinary sodium excretion at low intake levels—both in the general population and in individuals with chronic kidney disease (Dougher et al., 2016; He et al., 2018; Huang et al., 2016; Mente et al., 2014).

The alternative to determining sodium intake from urinary samples is to use self-reported dietary intake assessment methods, such as 24-hour dietary recalls or food frequency questionnaires. These methods also have limitations. Sodium intake is highly correlated with energy intake, so underreporting energy intake is likely to result in underreporting of sodium intake (Bailey et al., 2007). The majority of sodium is consumed from sources prepared outside the home (Harnack et al., 2017). Unless specific brands of foods and beverages are reported and updated food composition databases are used to analyze the data, the results are subject to inaccuracies. Furthermore, methods are lacking for quantifying sodium added at home during food preparation or during consumption (Anderson et al., 2010).

Estimating potassium intake is also challenging. Compared to sodium, a lower proportion of potassium consumed is recovered in urine (Aburto et al., 2013; Tasevska et al., 2006). This incomplete recovery may systematically differ across population groups and may depend on intake of other nutrients in the diet (Turban et al., 2013; Weaver et al., 2016), making it difficult to control for these differences. Short-term, self-reported assessment methods, particularly multiple 24-hour dietary recalls, provide reasonably accurate estimates of usual potassium intake because potassium content in foods is generally naturally occurring and not added during food preparation or consumption, and therefore is not as variable and brand specific as sodium. As discussed in Chapter 3, one of the challenges of attributing health effects to potassium is its collinearity with other nutrients in the diet.

To strengthen methodological approaches and improve the accuracy for estimating potassium and sodium intake, especially if the two nutrients are considered jointly, future research is needed to do the following:

• Identify or develop methods to minimize systematic biases between spot and 24-hour urine collections (e.g., nonlinear modeling). If successful, these data could be used to derive new equations to predict 24-hour potassium and sodium exposure from spot urine samples, by groups defined by sex, age, and ethnicity, if appropriate.
• Collect multiple 24-hour urine samples from subsamples of large cohorts to permit calibration of spot urine samples collected from the majority of participants.
• Identify individual-level and environment-level attributes (e.g., age, sex, ethnicity, body mass index, physical activity, ambient temperature) that affect the proportion of consumed potassium and sodium excreted in the urine. Using this information, determine whether equations can be developed to adjust each individual's 24-hour urine excretion information.
• Determine whether and to what extent the use of multiple spot urine samples collected at various times on different days improves calibration over a single spot urine sample for both potassium and sodium.
• Develop methodological approaches for capturing, analyzing, and synthesizing data that can characterize complex dietary interactions, especially between potassium and other nutrients in the diet.

Determine Potassium and Sodium Requirements and Toxicological Outcomes

Potassium and sodium are physiologically essential nutrients, and their concentrations in the blood are tightly regulated through homeostatic controls. This regulation, coupled with the pervasiveness of these nutrients in the food supply, makes it challenging to characterize the distribution of potassium and sodium requirements in the population. Without such information, the DRI for adequacy for each nutrient will remain an Adequate Intake (AI) rather than an Estimated Average Requirement (EAR) and a Recommended Dietary Allowance (RDA). Future DRI committees' efforts to derive EARs and RDAs for potassium and sodium would be facilitated by rigorously designed balance studies that assess intake levels of potassium and sodium needed to achieve balance across the lifespan. These balance studies would do the following:

• Determine an optimal study duration to accurately assess potassium and sodium balance, accounting for both infradian rhythm (i.e., lasting longer than 1 day) and potentially high intra-individual variability in urinary excretion.
• Provide participants with intake levels that have been chemically determined so that intake can be accurately assessed.
• Measure all excretion modes, including urinary, fecal, and sweat.
• Characterize sequestration of sodium in the skin and muscle and its relationship to intake, as well as to characteristics such as life stage and degree of adiposity.

The applicability of the potassium AI is uncertain for subpopulations prone to hyperkalemia or hypokalemia, such as individuals taking certain medications (e.g., angiotensin-converting-enzyme inhibitors [ACE-Is], angiotensin II receptor blockers [ARBs], diuretics) or individuals with adrenal insufficiency, chronic kidney disease, or type 2 diabetes. Health care providers may need to individualize potassium intake recommendations for these individuals, given the clinical context. The committee's inability to determine the applicability of the potassium AI to such subpopulations is due in part to a lack of data on how key biological and drug effects influence (1) potassium balance and (2) the effect of potassium intake on serum potassium concentrations. Furthermore, individuals with such conditions are typically excluded from potassium supplementation trials. To better characterize the potassium intake needs in these at-risk subpopulations, future research would do the following:

• Evaluate the effect of potassium supplementation on balance, serum potassium concentrations, blood pressure, and cardiovascular disease in individuals taking common medications that influence potassium homeostasis (e.g., ACE-Is, ARBs, diuretics), and in individuals with chronic kidney disease, diabetes, and heart failure.

In the expanded DRI model, the Tolerable Upper Intake Level (UL) is intended to characterize toxicological risk. In an effort to identify a toxicological indicator, the committee reviewed adverse effects from trials and case reports. Ethical considerations preclude human studies that are designed to produce toxic effects as the primary outcome. Thus, secondary reports of adverse events that are observed in studies of beneficial effects are likely to be a critical source of information for future DRI committees. Detailed and systematic collection of such information may elucidiate patterns and trends, and provide evidence for an indicator not previously considered. To further characterize the effects of high intakes of potassium and sodium, future research would do the following:

• Support the use of animal and in vitro studies for evaluating potential toxicological effects.
• Collect and report adverse effects in human studies in a systematic manner.
• Report and thoroughly describe case studies related to potassium and sodium overconsumption.
• Identify reliable indicators of excess potassium and sodium intake.

Strengthen the Evidence on the Relationship Between Potassium Intake and Chronic Disease Risk

There is moderate strength of evidence for a causal relationship between increased potassium intake (achieved by potassium supplementation) and decreased blood pressure among adults with hypertension. However, a lack of an intake–response relationship, coupled with a lack of direct evidence of a causal relationship between potassium intake and cardiovascular disease, precluded the committee from establishing a Chronic Disease Risk Reduction Intake (CDRR) for potassium. To strengthen the evidence on the relationship between potassium and chronic disease risk, future research would accomplish the following:

• Assess the effect of different doses and forms of potassium (e.g., dietary potassium, potassium chloride supplements, potassium bicarbonate supplements) on blood pressure so that an intake–response relationship can be established.
• Conduct an adequately powered trial of sufficient duration to study the effect of potassium supplementation on cardiovascular disease with concurrent measures of blood pressure, particularly among populations in which there is minimal risk of adverse effects from potassium supplementation.
• Conduct pooled meta-analyses that combine individual-level data to identify subgroup differences and comprehensively evaluate intake–response relationships.
• Conduct trials to further evaluate the interrelationships of potassium and other nutrients (e.g., calcium, magnesium) to determine the independent effects of each on blood pressure and cardiovascular disease.

Strengthen the Evidence on the Relationship Between Sodium Intake and Chronic Disease Risk

The meta-analyses used in this report helped the committee consider the totality of the evidence, but were limited by their ecological nature—that is, the studies were the unit of analysis rather than pooled individual-level data. Without access to individual-level data, the committee was unable to conduct analyses to draw conclusions about the differential effects of sodium reductions among population subgroups. To leverage existing data and overcome this limitation, future research would do the following:

• Conduct meta-analyses combining individual-level data across various trials so that subgroup differences, particularly for sodium and blood pressure, can be more comprehensively evaluated (e.g., by age, sex, race/ethnicity, baseline blood pressure, genetics). Both mean effects and differences in the intake–response relationship are of interest.

Although the committee characterized the relationship between sodium intake and chronic disease risk as at least moderate, additional studies are still needed on the effects of sodium intake reductions on chronic disease risk. There are a number of design and execution challenges to such trials, especially those that evaluate long-term effects of behavioral modification of dietary intake on chronic disease endpoints. Given the limitations and uncertainties in the current evidence, future research would do the following:

• Explore the feasibility of conducting large, methodologically rigorous randomized controlled trials that study the effect of sodium intake levels (ideally a range) on chronic disease endpoints, with particular attention to subpopulations that may have different responses to sodium intake.

Blood pressure changes in response to a sodium intervention are variable, roughly following a bell-shaped curve (He et al., 2009). Defining individuals with moderate to high sodium sensitivity remains a challenge (Elijovich et al., 2016). Sodium sensitivity is most often defined as a proportional change (e.g., ≥ 3 percent, ≥ 10 percent) or an absolute change (e.g., ≥ 3 mm Hg, ≥ 10 mm Hg) in mean arterial pressure during sodium intervention. A common method to define sodium sensitivity is to assess acute blood pressure response to rapid sodium loading (e.g., intravenous administration of 2 L normal [0.9 percent] saline over 4 hours) and depletion (e.g., a 10 mmol sodium diet and three doses of oral furosemide over 24 hours). Another common method of assessing sodium sensitivity is to measure the blood pressure response to low (e.g., < 50 mmol/d) and high (e.g., > 250 mmol/d) sodium intake over 1–2 weeks. In general, the dietary approach is considered the most rigorous approach for the characterization of sodium sensitivity, and evidence suggests that the blood pressure response has long-term reproducibility and stability in the general population (Gu et al., 2013). Several genes have been identified as being associated with increased sodium sensitivity, both in rats and humans, which suggests a genetic basis for salt sensitivity (Kelly and He, 2012). The Genetic Epidemiology Network of Salt Sensitivity study estimated that heritabilities for blood pressure response to high sodium were 0.22, 0.33, and 0.33 for systolic, diastolic, and mean arterial pressure, respectively (Gu et al., 2007).

The ability to accurately characterize sodium sensitivity among individuals could inform future updates to the sodium CDRR. Given current limitations in characterizing sodium sensitivity, future research would do the following:

• Characterize how blood pressure response to changes in sodium intake varies by age, sex, race/ethnicity, adiposity, genotype, and clinical conditions such as hypertension, diabetes, and chronic kidney disease.
• Identify both rare and common genetic variants that will help identify individuals who are predisposed to sodium sensitivity.
• Improve methods for identifying sodium-sensitive individuals—including through discovery and validation of biomarkers in the blood or urine and use of proteomics and metabolomics—to increase accuracy and facilitate implementation in clinical and public health settings.

Several prospective cohort studies have reported a significant positive association between sodium intake and cardiovascular disease risk and all-cause mortality, independent of blood pressure (Cook et al., 2007; He et al., 1999; Mills et al., 2016). Mechanisms underlying this blood pressure–independent association are not well studied. Observational studies have reported significant positive associations between sodium intake and left ventricular hypertrophy, endothelial dysfunction, and arterial stiffness, independent of blood pressure (Avolio et al., 1986; DuPont et al., 2013; Rodriguez et al., 2011; Todd et al., 2010). Little evidence exists from randomized controlled trials on the effect of sodium intake on cardiovascular risk factors other than blood pressure. To better characterize the relationship between sodium intake and chronic disease, future research would do the following:

• Test the effects of different sodium intake levels on endothelial and vascular function.

Explore Potassium and Sodium in Relation to Each Other and Other Dietary Components

Evaluate Developmental Origins of Health and Disease Related to Potassium and Sodium

The developmental origins of health and disease (DoHAD) posits that the in utero and early infant environment, including nutrition or stress, alters the long-term risk for chronic disease of the adult offspring. Studies in both human and animal models demonstrate that it is not only the immediate in utero environment for a fetus, but also the in utero environment of the fetus's parents that can result in DoHAD-enhanced risk for chronic disease (Mandy and Nyirenda, 2018). Different potassium and sodium intake levels during early life may affect chronic disease risk as one ages. At present, however, the longitudinal effects of potassium and sodium intake are not well characterized. One study reported an increased risk of hypertension in adult offspring of rats fed either a low-sodium or high-sodium diet during pregnancy and lactation (Koleganova et al., 2011). To clarify the long-term effects of potassium and sodium exposure, future research would accomplish the following:

• Evaluate the first- and second-generation effects of high and low potassium and sodium intakes of both the mother and father in order to identify optimal intakes of these nutrients during reproductive stages of life and characterize the variability of risk for chronic disease in the population.

Explore Opportunities in the Food Supply

Defining Healthy Populations Relative to the DRI Model When Prevalence of Chronic Disease Is High

The DRI model focuses on healthy populations, but an ongoing challenge in both the United States and Canada is the high prevalence of chronic disease, including obesity, type 2 diabetes, hypertension, and cardiovascular disease in adults, and the increasing prevalence of obesity and type 2 diabetes in children. In adults, the healthy population is the minority, rather than majority, of the total. Describing the focus of the DRIs as “apparently healthy population” does not fully resolve this challenge; high prevalence of these chronic diseases is accompanied by high prevalence of medication and medical nutrition therapy to manage them.

The expansion of the DRI model to include DRIs based on chronic disease magnifies the challenge of high chronic disease prevalence. DRIs based on chronic disease focus on primary prevention of chronic disease, but issues of secondary prevention and disease management are also pressing public health problems. An ongoing challenge and knowledge gap is how to assess if the DRI specified is relevant to the substantive proportion of the population that might be considered “apparently healthy” with appropriate medical management. A related challenge is discerning when and how to include these populations in the DRI framework of assessment of indicators and intake–response relationships. Future DRIs would benefit from the following:

• Further research and evaluation to determine whether the population focus of the DRI for each nutrient should be redefined and, if so, how.

Integrating an Externally Conducted Systematic Review into the DRI Process

To enhance the effectiveness of an externally conducted systematic review, alignment of the needs of the DRI committee (evidence users) with the information from evidence providers (e.g., Agency for Healthcare Research and Quality) is critical. This synchronization requires that the information in the systematic review meet the needs of the DRI committee and that the process minimizes the likelihood that bias or conflicts of interest adversely affect the usefulness and integrity of the systematic review. Involving the DRI committee in the development of the systematic evidence review, to the extent that best practices allow, has the potential to create a more efficient DRI process. This interaction is often handled with protocols that are developed prior to initiation of data collection for the review. Such protocols describe appropriate interactions among the contributors—in this case, between the DRI committee, the sponsors, and the scientists conducting the systematic review. Appropriate interactions may include finalizing key questions related to DRI decisions, identifying inclusion/exclusion criteria, reaching a general agreement on table formats and information to be provided, and discussing the overall approach and methodologies to be applied. Once the appropriate interactions have been completed and protocols agreed upon, the evidence providers would then conduct the review independent of the DRI committee. This independence is necessary to minimize the perception that the evidence users and evidence generators exerted bias in the conduct of the review. To create greater efficiency in the process and usability of the final systematic reviews for DRI applications, future DRI reviews would do the following:

• Develop a priori protocols for how and when the DRI committee, the sponsors, and those who design and execute the systematic review coordinate at key points in the process, as guided by systematic review best practices.

As described in Chapter 2 and Appendix C, the committee assessed the Agency for Healthcare Research and Quality systematic review, Sodium and Potassium Intake: Effects on Chronic Disease Outcomes and Risks (AHRQ Systematic Review) (Newberry et al., 2018), in a variety of ways prior to using and building on the evidence it provided. The committee determined that this step was necessary because the AHRQ Systematic Review was externally conducted and the committee was not involved in its design. The committee's independent assessment of the AHRQ Systematic Review included evaluating the fidelity of the methodology, the transparency of reporting, and the subjective decisions made. Through this process, the committee identified the need to further investigate sources of heterogeneity in meta-analyses and to reassess the strength-of-evidence ratings for selected indicators. Moreover, the Guiding Principles Report recommended that intake–response assessment also be evaluated using the Grading of Recommendations Assessment, Development and Evaluation system. Because key questions in the AHRQ Systematic Review focused only on issues of causality between nutrient intakes and chronic disease indicators, the committee performed additional analyses to evaluate intake–response relationships. For sodium, the CDRR was informed by the committee's intake–response analyses across multiple chronic disease indicators. Future DRI committees will likely need to critically evaluate the methods and conclusions of the systematic evidence review as they apply to both issues of causality and intake–response. The committee also anticipates that it is unlikely that future DRI committees will be able to avoid conducting some additional analyses, particularly with respect to intake–response assessment. Thus, future systematic reviews could in principle conduct some intake–response analyses, but future DRI committees may need to perform additional or alternative analyses and employ expert judgment. To better integrate externally conducted systematic reviews into the process, future DRI reviews would do the following:

• Develop protocols for conducting post-hoc analyses of systematic reviews to confirm the fidelity of the review, to update the externally conduced review, and to meet unanticipated committee needs in a manner that is transparent and justified.

Collecting Evidence to Supplement the Externally Conducted Systematic Reviews

In the first step of the DRI organizing framework, the committee assessed the current state of the evidence on potential indicators for the DRIs for adequacy, the DRIs for toxicity, and DRIs based on chronic disease. The AHRQ Systematic Review was limited to a review of potassium and sodium intakes on chronic disease outcomes and related risk factors. This meant that the committee had comprehensive data summaries for chronic disease (with the exception of a few potential chronic disease indicators not included in the review), but lacked similar systematic reviews or a sufficient context for evaluating indicators of adequacy and toxicity. The committee performed several scoping literature searches to determine if there were indicators not included in the AHRQ Systematic Review that could potentially inform any of the DRI categories for either potassium or sodium (see Appendix D). The literature scans led the committee to perform comprehensive literature searches for select indicators (see Appendix E). To better integrate different sources of information, future DRI reviews would do the following:

• Develop approaches for DRI committees to compile and evaluate available evidence for potential indicators not included in the independently conducted systematic review, to ensure the quality and completeness of the review while minimizing the potential for bias or conflicts of interest.

Integrating the Recommendations in the Guiding Principles Report into the DRI Process

The Guiding Principles Report was indispensable to the committee's derivation of the sodium CDRR values, as well as its decision not to establish a CDRR for potassium. Nonetheless, as the committee reviewed the evidence for these two nutrients, it needed to adapt some of the approaches and recommendations outlined by the Guiding Principles Report.

First, with respect to indicators of chronic disease, the Guiding Principles Report recommended selection of a single outcome indicator on the causal pathway, but acknowledged the possibility of using multiple indicators (NASEM, 2017). Rather than using any single outcome indicator as the basis for the sodium CDRR, the committee integrated evidence from multiple indicators (cardiovascular disease incidence, hypertension incidence, systolic blood pressure, and diastolic blood pressure), particularly because these indicators are known to be causally linked and evidence supports the relationship between sodium and each indicator. Future DRI committees may find it useful to use multiple, causally linked indicators for establishing DRIs based on chronic disease, to the extent that such evidence is available for the nutrient or food substance under investigation.

Second, the Guiding Principles Report provided guidance that is more applicable to situations in which observational data are used to assess intake–response relationships. For sodium, the strongest data were from randomized controlled trials, with the limitation that most trials only involved a single contrast. Thus, the committee needed to adapt existing intake–response meta-analysis methods to combine multiple trials with different sodium intake levels for the control and intervention groups. Future DRI committees will likely refine the approaches to assess intake–response relationships, depending on available evidence.

Third, in the case of sodium, the committee adapted the Guiding Principles Report recommendation that DRIs based on chronic disease be expressed as a range instead of a single number. This recommendation was based on the premise that chronic disease risk varies with intake, and that there is a range over which increasing or decreasing intakes will reduce risk. The Guiding Principles Report noted that expressing the DRI as a single number might be misconstrued as suggesting the existence of a sharp dividing line between risk and no risk. Conversely, expressing the sodium CDRR as a range might lead to the false impression that any intake within that range is acceptable, as opposed to being a range where reducing intakes is beneficial.

As it considered the guidance and recommendations in the Guiding Principles Report to develop the new DRI category—the CDRR—the committee saw the need to provide a description of how the CDRR would apply to sodium. For the reasons stated in Chapter 2, the committee adapted the guidance in the Guiding Principles Report and expressed the sodium CDRR as the lowest level of intake for which there was sufficient strength of evidence to characterize a chronic disease risk reduction. As more experience with establishing DRIs based on chronic disease is gained, future DRI efforts would do the following:

• Balance the need for consistency over time with the need for refinements as new and unanticipated challenges are encountered.
• Develop a general or standardized description of the DRI based on chronic disease category.

Providing Additional Guidance on the Expanded DRI Model as Experience Is Gained

The Guiding Principles Report provided limited comment on how DRIs based on chronic disease affect or interact with the other DRI categories. The committee's experience clarified that the review of the evidence and the decisions about DRIs based on chronic disease can have implications for the other DRI categories. For potassium, the evidence that was previously considered for the AI in the Dietary Reference Intakes for Water, Potassium, Sodium, Chloride, and Sulfate (2005 DRI Report) (IOM, 2005) was considered in this report for the purposes of establishing a DRI based on chronic disease. This shift, caused by the expansion of the DRI model, narrowed the evidence the committee considered for establishing the potassium DRIs for adequacy. Ultimately, the lack of a specific indicator of potassium adequacy or status led the committee to establish AIs for all DRI age, sex, and life-stage groups. For sodium, the committee considered evidence on chronic disease–related indicators as context for the adequacy DRI, to ensure the selected intake levels would not increase chronic disease risk. Pursuant to a recommendation in the Guiding Principles Report, the committee considered only toxicological adverse effects for establishing ULs. The meaning of UL in this report, therefore, is fundamentally different than the definition used in the 2005 DRI Report. Although it was imperative for the committee to use its collective expert judgment regarding the interrelationship between DRIs based on chronic disease and the other DRI categories for potassium and sodium, it is beyond the scope of this report to determine how future DRI committees can systematically make such decisions.

To create more conceptually consistent DRIs moving forward, future DRI efforts would do the following:

• Provide additional guidance on how to address the interrelationship between DRIs based on chronic disease and the other DRI categories, particularly the AI. A decision-making framework that can be consistently applied to various scenarios would have broad application for future DRI committees.

The committee's methodologies reflect the state of the evidence on potassium and sodium, and do not necessarily establish a definitive methodological blueprint for future DRIs. The type and quantity of evidence, as well as specific health outcomes, will vary for each nutrient and food substance considered under the expanded DRI model. Approaches taken for future DRIs will be determined on a case-by-case basis at the expense of consistency across nutrients and DRI committees. As future DRI committees gain experience with the expanded DRI model with other nutrients and food substances, there will likely be greater clarity on which approaches are broadly applicable, as well as which recommendations in the Guiding Principles Report need to be adapted. To this end, future DRI efforts would do the following:

• Update and revise the guiding principles, as experience establishing DRIs based on chronic disease is gained.

It is beyond the scope of this study to provide definitive guidance on the proper clinical, educational, research, and public health applications of the DRI values in the expanded model. As the sodium CDRR and future DRIs based on chronic disease are used, a greater understanding will emerge of the strengths, limitations, usefulness, and misapplications of this new DRI category. Translating this information for DRI users will help bolster the use and understanding of the expanded DRI model. As such, DRI users would likely benefit from receiving the following:

• Revised guidance on using the expanded DRI model for dietary assessment and dietary planning.