Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
2 Applying the Guiding Principles Report This chapter describes how the committee applied the recommenda- tions from the Guiding Principles for Developing Dietary Reference Intakes Based on Chronic Disease (Guiding Principles Report) (NASEM, 2017a) to its review of the Dietary Reference Intakes (DRIs) for potassium and sodium. The committeeâs interpretation of the Guiding Principles Report described in this chapter sets the stage for the evidence review, methodologi- cal details, and rationale presented for DRIs based on chronic disease for potassium and sodium (see Chapters 6 and 10, respectively).1 This chapter also describes the committeeâs approach to the new DRI category in context of the DRIs for adequacy and toxicity for potassium and sodium. BACKGROUND Historically, undernutrition and nutritional deficiencies were preva- lent in the population, contributing to high rates of diet-related disease. Although the standardization of food fortification and enrichment along with dietary guidance to the public contributed to reducing the prevalence of nutrition deficiencies, there was a subsequent rise in the prevalence of obesity and related chronic diseases. As the public health burden in the 1â Throughout this report, DRIs based on chronic disease is used when broadly describing the category, such as when referring to the guidance in the Guiding Principles Report (NASEM, 2017a). This aligns with the committeeâs use of the phrases DRIs for adequacy, which broadly refers to the Estimated Average Requirement, Recommended Dietary Allowance, and Adequate Intake, and DRIs for toxicity, which refers to the Tolerable Upper Intake Level. 37
38 DIETARY REFERENCE INTAKES FOR SODIUM AND POTASSIUM United States and Canada shifted toward risk of chronic disease, nutri- tion science has increasingly focused on the effect of dietary determinants, including nutrients and other food components, as potential modifiers of chronic disease risk. The public health significance of chronic disease warrants concerted efforts to understand the relationships between diet and chronic disease risk, but such efforts must navigate methodological challenges. UnderÂ standing dietary determinants of chronic disease often requires different kinds of conceptual approaches and evidence than are needed for the evaluation of nutrient deficiencies and toxicities. Dietary intake patterns are multidimensional, dynamic, and change over the course of a lifespan. Chronic diseases are complex, multifaceted, and develop over time. These complexities make identifying the relationship between nutrient intake and chronic disease difficult, especially when longitudinal data are limited or unavailable. Additionally, the extended time between exposure and out- come often precludes the use of randomized controlled trials to establish a causal relationship. Since its inception, the DRIs were intended to consider chronic disease risk (IOM, 1994), but available evidence on chronic disease outcomes was typically too limited to inform the derivation of specific reference values. Furthermore, the DRI process lacked a mechanism for evaluating the evidence for causal and intakeâresponse relationships between nutrient intake and chronic disease riskâtwo components of the DRI organizing framework (see Chapter 1, Box 1-2). As described in Chapter 1, efforts to overcome these challenges ultimately led to the Guiding Principles Report (NASEM, 2017a), which expanded the DRI model to include a new DRI category based on chronic disease. THE COMMITTEEâS INTERPRETATION OF THE GUIDING PRINCIPLES REPORT The Guiding Principles Report provides recommendations on meth- odological approaches to establishing DRIs based on chronic disease (see Box 2-1). Pursuant to its task (see Chapter 1, Box 1-1), the committee applied those recommendations in context of the available evidence on potassium and sodium. The following sections not only summarize the committeeâs interpretation of recommendations in the Guiding Principles Report that were central to its approach to the new DRI category, but they also describe a few instances in which the committee considered it important and necessary to adapt some of the guidance. The committee notes that adaptations made for potassium and sodium do not invalidate potential applications of the concepts articulated in the Guiding Principles Report to future DRI reviews.
APPLYING THE GUIDING PRINCIPLES REPORT 39 BOX 2-1 Recommendations Excerpted from the Guiding Principles Report The recommendations listed below reflect the consensus of a separate National Academies committee, as presented in the Guiding Principles Report (NASEM, 2017a). Recommendation 1: Until better intake assessment methodologies are devel- oped and applied widely, Dietary Reference Intake (DRI) committees should strive to ensure that random and systematic errors and biases of nutrient or other food substance exposure assessment methodologies are considered in their evidence review. In the long term, research agendas should include accelerated efforts to improve nutrient or other food substance exposure assessments for application in studies of chronic disease risk. Recommendation 2: The ideal outcome used to establish chronic disease DRIs should be the chronic disease of interest, as defined by accepted diagnostic crite- ria, including composite endpoints, when applicable. Surrogate markers could be considered with the goal of using the findings as supporting information of results based on the chronic disease of interest. To be considered, surrogate markers should meet the qualification criteria for their purpose. Qualification of surrogate markers must be specific to each nutrient or other food substance, although some surrogates will be applicable to more than one causal pathway. Recommendation 3: The committee recommends that DRI committees use Grad- ing of Recommendations Assessment, Development and Evaluation (GRADE) in assessing the certainty of the evidence related to the causal association between nutrient or other food substances and chronic diseases. Using GRADE, the committee recommends that a decision to proceed with development of chronic disease DRIs be based on at least moderate certainty that a causal relationship exists and on the existence of an intakeâresponse relationship. Recommendation 4: The committee recommends the use of a single outcome indicator on the causal pathway. However, when a single food substance reduces the risk of more than one chronic disease, reference values could be developed for each chronic disease. The committee, however, does not recommend the use of âmultiple indicators of a chronic diseaseâ or âmultiple indicators for multiple diseasesâ unless there is sufficient experience with the use of algorithms or other strong evidence suggesting that multiple indicators point to risk of a chronic dis- ease, due to potential lack of reliability or consistency in the results. Recommendation 5: The committee recommends extrapolation of intakeâÂ response data for chronic disease DRIs only to populations that are similar to studied populations in the underlying factors related to the chronic disease of interest. continued
40 DIETARY REFERENCE INTAKES FOR SODIUM AND POTASSIUM BOX 2-1â Continued Recommendation 6: The committee recommends that DRIs for chronic disease risk take the form of a range, rather than a single number. Intakeâresponse rela- tionships should be defined as different ranges of the intakeâresponse relationship where risk is at minimum, is decreasing, and/or is increasing (i.e., slope = 0, nega- tive, or positive). When a nutrient or other food substance reduces the risk of more than one chronic disease, DRIs could be developed for each chronic disease, even if the confidence levels for each chronic disease are different. The magnitude of risk slope considered necessary to set a DRI should be decided based on clearly articulated public health goals, such as those previously identified by other authori- ties (e.g., Healthy People 2020). The committee does not recommend, however, developing a family of DRIs for any one nutrient or other food substance for differ- ent risk reduction targets for the same chronic disease. Recommendation 7: The committee recommends retaining Tolerable Upper In- take Levels (ULs) based on traditional toxicity endpoints. In addition, if increased intake of a substance has been shown to increase the risk of a chronic disease, such a relationship should be characterized as the range where a decreased intake is beneficial. If the increase in risk only occurs at intakes greater than the traditional UL, no chronic disease Dietary Reference Intake would be required, because avoiding intakes greater than the UL will avoid the chronic disease risk. Recommendation 8: The committee recommends that to develop a chronic dis- ease DRI, the level of certainty in the intakeâresponse relationship should generally be the same as the level of certainty for a determination of causality, that is, at least âmoderate,â using the Grading of Recommendations Assessment, Development and Evaluation (GRADE). However, in some cases, for example when a food sub- stance increases chronic disease risk, the level of certainty considered acceptable might be lower. In all cases, a thorough description of the scientific uncertainties is essential in describing quantitative intakeâresponse relationships. Requiring at least âmoderateâ certainty extends to cases where relationships between intake and a surrogate marker and between the same surrogate marker and the chronic disease are characterized separately, in a piecemeal (i.e., two-stage) approach. Recommendation 9: The committee recommends that, if possible, health risk/ benefit analyses be conducted and the method to characterize and decide on the balance be made explicit and transparent. Such a decision needs to consider the certainty of evidence for harms and benefits of changing intake and be based on clearly articulated public health goals. If DRI committees do not perform such risk/ benefit analyses, it is still necessary to describe the disease outcomes and their severities, the magnitudes of risk increases and decreases over various ranges of intakes, and other factors that would allow users to make informed decisions.
APPLYING THE GUIDING PRINCIPLES REPORT 41 Recommendation 10: Because of the need for close coordination and exchange of ideas when setting DRIs based on indicators of adequacy, toxicity, and chronic disease, one single National Academies of Sciences, Engineering, and Medicine parent committee should develop DRIs for the prevention of nutrient deficiencies and toxicities and for reducing the risk of chronic disease. Due to the need for different expertise and different methodological considerations, two subcommit- tees could be established at the discretion of the parent committee, for reviewing evidence on (1) adequacy and toxicity and (2) chronic disease, respectively. Recommendation 11: When sufficient evidence exists to develop chronic disease DRIs for one or more nutrient or other food substances that are interrelated in their causal relationships with one or more chronic diseases, a committee should be convened to review the evidence of their association with all selected diseases. Using a chronic disease as the starting point for the review is not recommended because balancing health risks and benefits for multiple nutrient or other food sub- stances that are related to a single chronic disease endpoint will be a challenge in cases where the same nutrient or other food substances might be associated with more than one chronic disease. SOURCE: NASEM, 2017a. Nomenclature and Conceptual Underpinnings Guidance from the Guiding Principles Report Nutrient deficiency diseases from inadequate intake and adverse effects from excess intake are well established for many essential nutrients. These relationships are based on the concept of a threshold effect. Intake of an essential nutrient below a certain threshold inevitably leads to deficiencies. For some nutrients, intakes above a certain upper threshold increase risk for adverse effects. As described above, the relationship between intake of a nutrient and risk of chronic disease is more complex; it is unlikely that there is a threshold intake level for which zero risk of chronic disease exists. The Guiding Principles Report presented a conceptual diagram of poten- tial intakeâresponse relationships that show variable types of relationships including curvilinear, linear, or U-shaped curves (see Figure 2-1). The dif- ferent possible intakeâresponse relationships set chronic disease risk apart from the threshold concepts of adequacy and toxicity. The Guiding Principles Report described possible complications in translating the evidence for a chronic disease intakeâresponse relationship into a reference value. For instance, such relationships are often continuous, and benefits of increasing or decreasing intakes could exist across a broad
42 DIETARY REFERENCE INTAKES FOR SODIUM AND POTASSIUM FIGURE 2-1 Possible DRI ranges for a single chronic disease, depending on the shape of the intakeâresponse relationship, as presented in the Guiding Principles Report.
APPLYING THE GUIDING PRINCIPLES REPORT 43 FIGURE 2-1 Continued NOTES: These relationships, and their confidence intervals, are âidealizedâ and meant for illustration, and are likely to be more complicated (e.g., less smoothly changing) in practice. The different scenarios are qualitatively the same whether ab- solute or relative risk is considered. However, to estimate the significance of the effect on the population of the different choices of ranges, absolute risks are also needed. Panels A and Aâ² represent strictly monotonically changing intakeâresponse relation- ships; panels B, Bâ², Bâ³, and Bâ²â³ represent different âJ-shapedâ relationships, where there is a plateau at one end of the intake range. Panel C represents a âU-shapedâ relationship, where there is an intake level that minimizes risk. RDA = Recommended Dietary Allowance; UL = Tolerable Upper Intake Level; solid line = best estimate of intakeâresponse; dashed lines = confidence intervals of intakeâresponse. SOURCE: NASEM, 2017a. range of intakes. Another potential challenge is that individuals within a population may have different baseline risks for chronic disease, owing to factors other than dietary intake (e.g., genetics, other environmental exposures). Such considerations were anticipated to hinder DRI commit- teesâ ability to identify a single value to characterize the complexities of the intakeâresponse relationship. The Guiding Principles Report therefore described three possible ways to define a DRI based on chronic disease (see Table 2-1). Committeeâs Application of the Guiding Principles Report Although the scope of its work was limited to potassium and sodium, the committee was mindful that its application of the Guiding Principles Report might have implications for future DRI reviews, particularly in assessing nonessential nutrients and food substances. One such consider- ation was the nomenclature the committee used for the new DRI category. In an effort to promote consistency with future DRI reviews, the committee sought to use terminology that would be broadly applicable, yet sufficiently descriptive. The committee acknowledges, however, that the nomenclature used in this report may be reevaluated in future DRI reviews. The committee considered the use of distinct terms to describe the different types of intakeâresponse relationships that could exist within a new DRI category based on chronic disease (see Figure 2-1 and Table 2-1). However, introducing multiple names and acronyms may be confusing for DRI users. For example, unique terminology for each intakeâresponse relationship has the potential to subdivide the new DRI category in a way that may make it difficult for users to understand the relationship between the DRI values and chronic disease risk. By contrast, a single DRI category that allows flexibility in characterizing the different types
44 DIETARY REFERENCE INTAKES FOR SODIUM AND POTASSIUM TABLE 2-1 Three Possible DRIs Based on Chronic Disease, as Identified in the Guiding Principles Report Possible DRI for Single Chronic Disease Relationshipa Description Region of IntakeâResponse Acceptable Range Range of usual intakes of a Region where slope is flat, of Intakes food substance without outside of which there is increased risk of chronic increased risk of chronic disease, disease deficiency, or toxicity Range of Beneficial Range of usual intakes of a Region where slope is negative, Increased Intakes food substance where outside of which slope is non- increasing intake can reduce negative, or there is increased risk risk of chronic disease of deficiency or toxicity Range of Beneficial Range of usual intakes of a Region where slope is positive, Decreased Intakes food substance where outside of which slope is non- decreasing intake can reduce negative, or there is increased risk risk of chronic disease of deficiency or toxicity aIn each case, defining the region of the intakeâresponse relationship corresponding to the DRI requires judgment as to what âslopeâ is small or large enough, and at what confidence level to consider flat, negative, or positive. SOURCE: Adapted from NASEM, 2017a. of intakeâresponse relationships for chronic diseases risk would provide greater simplicity and conceptual unity. The committee determined that, in the context of this DRI review, the three terms and descriptions presented in the Guiding Principles Report (see Figure 2-1 and Table 2-1) should be consolidated into a single DRI category called the Chronic Disease Risk Reduction Intake (CDRR). The committee considered several options for the new DRI category before selecting the CDRR. Because the DRIs comprise a set of different reference value categories, labeling the new category itself the chronic dis- ease DRIs or DRI based on chronic disease had the potential of dividing the DRIs into âthe adequacy and toxicity DRIsâ and âthe chronic disease DRIs.â Such a distinction would appear to counter the Guiding Principles Report recommendation that a single DRI committee be convened to estab- lish the adequacy, toxicity, and chronic disease reference values for a specific nutrient (see Box 2-1, Guiding Principles Report Recommendation 10). Accordingly, the committee considered it important to use nomenclature that positioned the new category as one of several DRI categories. The committee also considered how to align the nomenclature with the naming convention used for the other DRI categories, which include descriptions such as âlevel,â ârequirement,â and others. Although the Guiding Princi-
APPLYING THE GUIDING PRINCIPLES REPORT 45 ples Report conceptualized this new category to be expressed as a range, the committeeâs experience suggested that there may be circumstances in which a range may not be a sufficiently clear, effective, or appropriate expression of the CDRR. One option considered was using âtargetâ or âgoal,â but such descriptions had the potential to convey a threshold between risk and no risk for chronic disease. Ultimately, the committee determined that âintakeâ was sufficiently descriptive and would likely be broadly adapt- able to different scenarios. The omission of the âIâ in the acronym for this category is for simplicity, similar to the abbreviation of UL (for Tolerable Upper Intake Level). Although its approach to reviewing the evidence to establish DRIs based on chronic disease and deriving the sodium CDRRs was conceptually aligned with the Guiding Principles Report, the committee further consid- ered issues of implementation and clarity of communication in the expres- sion of values. The sodium CDRR values established in this report were informed by the shape and strength of evidence2 for the intakeâresponse relationship over the studied range of intakes. Defining the upper end of the range for sodium posed challenges. Had the committee established the sodium CDRR as a range and required moderate strength of evidence for an intakeâresponse relationship to do so, the upper bound would be a sodium intake level that is exceeded by a portion of the population (see Chapter 11, Tables 11-4 and 11-6). Such a range would be subject to pos- sible misinterpretation. First, it could be incorrectly viewed as a desirable range of intakes (akin to the concept of the Acceptable Macronutrient Dis- tribution Range), rather than a range of intakes over which reductions in sodium intake are expected to reduce chronic disease risk. Second, it could be incorrectly interpreted as suggesting that high intakes are not associated with chronic disease risks, whereas intakes above this range are likely to pose a continuing risk. The committee was further challenged by the lack of evidence suitable for deriving a sodium UL based on toxicological adverse effects. As shown in Figure 2-1, the Guiding Principles Report had concep- tualized the UL as intake level above which the CDRR would not need to be characterized, because the potential for toxicological risk would be increas- ing. Without a UL for sodium, this principle could not be applied. This situ- ation fit the scenario anticipated by the Guiding Principles Report in which a lower strength of evidence of the intakeâresponse relationship could be used to support a DRI based on chronic disease (see Box 2-1, Guiding Principles Report Recommendation 8). Thus, the committee extrapolated 2â For consistency throughout this report and in alignment with the terminology used in the AHRQ Systematic Review, the committee uses the term strength of the evidence instead of quality of the evidence or certainty of the evidence when describing the grading of the evidence used to derive DRIs based on chronic disease.
46 DIETARY REFERENCE INTAKES FOR SODIUM AND POTASSIUM the intakeâresponse relationship for sodium and chronic disease risk above the intake range where the strength of evidence was at least moderate. As detailed in Chapter 10, the committee expressed the sodium CDRR as the lowest intake level of intake for which there was sufficient evidence to characterize chronic disease risk reduction (see Box 2-2). Strength of the Evidence Guidance from the Guiding Principles Report One of the general assumptions underpinning the DRI model is that available data are often insufficient to draw conclusions, and scientific judgment and transparent documentation must be used when assessing scientific uncertainties (Taylor, 2008). Another general assumption is that failure to derive a reference value is often not a viable public health option (Taylor, 2008). In the case of essential nutrients, there is an obligation for a DRI committee to determine DRIs for adequacy (i.e., Estimated Average Requirements [EARs] and Recommended Dietary Allowances [RDAs], or Adequate Intakes [AIs] when an EAR and an RDA cannot be derived). Accordingly, a DRI committee uses the best available evidence to do so. Similarly, if there is evidence of adverse effects from high levels of intake, a DRI committee uses its expert judgment and best available evidence to determine a level of intake after which risk increases to establish a UL. In contrast, the Guiding Principles Report described the new DRI category as being established only when the body of evidence on the relationship between a nutrient and chronic disease risk is sufficient and when an intakeâresponse relationship can be characterized. The conceptual distinc- tion between these DRI categories is summarized in Table 2-2. BOX 2-2 Chronic Disease Risk Reduction Intake for Sodium Context: The sodium Chronic Disease Risk Reduction Intake (CDRR) is the lowest level of intake for which there was sufficient strength of evidence to character- ize a chronic disease risk reduction. The concept of a range is embedded in the expression of the sodium CDRR in that for intakes above the CDRR, reduction in sodium intake is expected to reduce chronic disease risk. For sodium, the CDRR is the intake above which intake reduction is expected to reduce chronic disease risk within an apparently healthy population.
APPLYING THE GUIDING PRINCIPLES REPORT 47 TABLE 2-2 Conceptual Distinction Between DRIs for Adequacy and Toxicity and DRIs Based on Chronic Disease DRIs for Adequacy and Toxicity DRIs Based on Chronic Disease Needed because deficiencies Are not warranted unless sufficient (of essential nutrients) and toxicities: evidence exists because: â¢ Will affect everyone, if intake is â¢ Risk to acquire chronic diseases varies by inadequate or excessive individual â¢ Are caused by a single nutrient â¢ Chronic diseases are often related to many â¢ Are prevented by nutritional risk factors (e.g., genetic, environmental) interventions â¢ Nutritional interventions will only partly ameliorate the risk of chronic disease SOURCE: Adapted from NASEM, 2017b. Recognizing the potential for misinterpretation from the results of individual studies, various tools have been developed to assess the strength of scientific evidence that examines a specific health-related question. These tools support a more objective and transparent process, although expert interpretation and judgment are still needed. To determine the strength of the body of evidence for a relationship between intake and chronic disease risk, the Guiding Principles Report (NASEM, 2017a) recommended using the Grading of Recommendations Assessment, Development and Evalua- tion (GRADE) system. GRADE, which was developed in the health care context (see Figure 2-2), rates a body of evidence by assessing five domains that may reduce the strength of the evidence and three domains that may increase the strength (see Box 2-3). This assessment leads to one of four ratingsâhigh, moderate, low, or very lowâto describe the certainty in how close the estimated effect is to the true effect (Balshem et al., 2011). The Guiding Principles Report recommended a GRADE rating of at least mod- erate strength for both the causal relationship and the intakeâresponse rela- tionship for the DRI based on chronic disease to be established, although it was also noted that âwhen a food substance increases chronic disease risk, the level of certainty considered acceptable might be lowerâ (NASEM, 2017a, p. 220). Committeeâs Application of the Guiding Principles Report The DRI organizing framework guides DRI committees to establish reference values based on the strength of the evidence. The DRI organizing framework provides flexibility to accommodate different evidentiary sce- narios that DRI committees may encounter and allows committees to factor in public health ramifications. To that end, the processes for assessing the
48 DIETARY REFERENCE INTAKES FOR SODIUM AND POTASSIUM FIGURE 2-2 Schematic view of GRADEâs process for developing recommendations. NOTE: GRADE = Grading of Recommendations Assessment, Development and Evaluation; OC = outcomes; PICO = population, intervention, comparator, and outcome; RCT = randomized controlled trial; S = studies. *Also labeled âconditionalâ or âdiscretionary.â SOURCE: Reprinted from Guyatt et al., 2011a, with permission from Elsevier. strength of evidence and integrating such an assessment into the decision- making process for the DRIs for adequacy and the DRIs for Â oxicity are not t yet standardized. Recommendations in the Guiding Principles Report intro- duce a more formal strength-of-evidence assessment to the DRI process, specifically for informing decision making related to DRIs based on chronic disease. The fundamental conceptual differences outlined in Table 2-2 call for a more standardized approach to assessing and applying strength of
APPLYING THE GUIDING PRINCIPLES REPORT 49 BOX 2-3 Grading of Recommendations Assessment, Development and Evaluation (GRADE) Systemâ Domains Used to Rate the Strength of the Evidence Domains That May Reduce the Strength of Evidence* â¢ Risk of bias is systematic error attributable to limitations in the study design or execution. â¢ Imprecision is random error that occurs when studies have a small sample size and the number of events is also small. â¢ Inconsistency is unexplained heterogeneity or variability of study results. â¢ Indirectness occurs when a study does not compare the interventions of interest, apply the intervention to the population of interest, or measure the outcomes that are important to patients. â¢ Publication bias is a systematic underestimation or overestimation of the underlying beneficial or harmful effect caused by the selective publication of studies. Domains That May Increase the Strength of Evidence â¢ Large magnitude of effect, with consideration for both the magnitude and preci- sion of the estimate. â¢ Intakeâresponse gradient. â¢ Plausible residual confounding, which under certain circumstances can in- crease confidence in an estimate. *These definitions are direct quotes from NASEM, 2017a (p. 9), which are adapted from Schunemann et al., 2013. evidence to the derivation of DRIs based on chronic disease compared to the other DRI categories. In its application of the Guiding Principles Report guidance, the com- mittee explored the body of evidence provided in 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). Although the tool that was used in the AHRQ Systematic Review was not GRADE, it is conceptually similar (Berkman et al., 2013). One of the noted differences is terminology. For example, the AHRQ Systematic Review referred to the assessment of the body of evidence as âstrength of the evidence,â whereas GRADE refers to âquality (or certainty) of the evidence.â Furthermore, where GRADE uses the ratings of high, moderate, low, and very low, the AHRQ Systematic Review used high, moderate, low, and insufficient. Given that the two
50 DIETARY REFERENCE INTAKES FOR SODIUM AND POTASSIUM approaches are similar, the committee elected to use the AHRQ Systematic Review terminology throughout this report. To effectively use the strength of evidence ratings in the AHRQ System- atic Review, the committee first evaluated the methodological approaches taken (see Appendix C). From this evaluation, the committee identified two components of the strength-of-evidence assessment that merited further consideration: risk of bias and inconsistency. Risk of bias was considered an important domain requiring further investigation because of its use in determining the validity of individual study results. The domain of incon- sistency assesses the comparability of results in a body of evidence. The committee found that the AHRQ Systematic Review did not thoroughly investigate and explain causes of heterogeneity in the results when high levels of inconsistency were found; in some cases, such an investigation was needed to interpret the results of meta-analyses. Accordingly, the commit- tee also investigated the domain of inconsistency. These additional analyses informed and clarified the committeeâs approach. Risk of biasâ Before the strength of a body of evidence can be determined, the individual studies are assessed. The quality of an individual study can vary depending on its specific design features and conduct. To account for this in rating the strength of evidence, consideration is given to risk of bias (see Box 2-4). The AHRQ Systematic Review assessed risk of bias for all studies meet- ing the inclusion criteria. The committee reviewed the risk-of-bias criteria for both randomized controlled trials and observational studies (for the risk-of-bias criteria, see Appendix C, Annex C-1). One of the risk-of-bias domains was considered by the committee to be of particular importanceâ methods of potassium and sodium intake assessment. The method used to assess potassium and sodium intake can affect the strength of dietâindicator relationships, the strength of intakeâresponse relationships, and the estima- tion of usual intake distribution for a population (see Chapter 3). The com- mittee reviewed this and the other domains and concurred with the tools that the AHRQ Systematic Review used to assess risk of bias. The committee also considered the inclusion of observational studies in determining the strength of the evidence for the relationship between potas- sium or sodium intake and each indicator selected for establishing a CDRR. According to GRADE (Guyatt et al., 2011b), although evidence that relies only on observational studies can be upgraded in rating, such evidence is generally classified as low strength of evidence because observational studies have an inherently weaker design for evaluating evidence on causal effects. Only when there is a large effect size, an intakeâresponse relation- ship is observed, or plausible residual confounding increases confidence in the estimates can the strength of evidence be upgraded. In addition, in the
APPLYING THE GUIDING PRINCIPLES REPORT 51 BOX 2-4 Risk of Bias (Validity of a Study) In the design of a study, two types of validity are considered: external (or generalizability) and internal (or comparability). The latter is concerned with the truth of the result of the study, and the risk of a systematic deviation from the truth is termed risk of bias. Flaws in the design, conduct, and reporting of a study can lead to the under- or overestimation of effect of an intervention. This is distinct from the precision of the study results, which is concerned with the extent to which the study result is free from random error. In evaluating the risk of bias, key components of the design related to inter- nal validity must be assessed. For randomized controlled trials, this includes the sequence generation for allocation of participants to interventions, concealment of the allocation, blinding of participants, personnel, and outcome assessors to the allocation of the intervention, incomplete outcome data or attrition of partici- pants from the study, and selective reporting of outcomes (Higgins and Green, 2011). For observational studies, domains through which bias might be introduced include confounding or residual confounding, selection and rate of participation or dropout of participant subgroups, measurement of interventions (systematic over- or underreporting), departures from intended interventions (secular trends over time), missing data, measurement of outcomes, and selection of the reported result (Sterne et al., 2016). Tools for assessing the risk of bias are applied to the studies included in a systematic review, and the studies are then classified according to the level of their risk of bias, such as high, moderate, or low. Sensitivity analysis can then be conducted by considering, for example, the meta-analysis including only studies rated as having low risk of bias. case of sodium, the majority of the observational studies were rated as high risk of bias, mainly because of the biases in the sodium intake ascertainment methods used in observational study designs (for strengths and weaknesses of these methods, see Chapter 3). For these reasons, the committee primar- ily relied on randomized controlled trials to inform its decision making regarding establishing DRIs based on chronic disease for potassium and sodium. The committee considered observational studies rated as having low risk of bias to supplement the decisions from randomized control trials, particularly when randomized controlled trial data were few or unavailable. Inconsistency âMeta-analyses use statistical methods to compare results from different studies, as a means to identify a consistent pattern across studies. Heterogeneity across the studies in a meta-analysis can arise for a variety of reasons, including variability in the participant characteristics, interventions, and outcomes evaluated; there can also be trial-level variability in study
52 DIETARY REFERENCE INTAKES FOR SODIUM AND POTASSIUM BOX 2-5 Identifying and Explaining Sources of Heterogeneity in Meta-Analyses A meta-analysis of studies identified through a systematic review employs statistical methods with a focus on comparing and contrasting study results with the goal of identifying consistent patterns or sources of disagreements among these results. When there are important inconsistencies in the results (direction, magnitude, significance) that cannot be explained, the strength of evidence (i.e., confidence in the estimate of effect) for that outcome decreases (e.g., from high to moderate). Interpretation of a meta-analysis, therefore, is complicated by the presence of this heterogeneity among the studies because the observed differences in the intervention effect among the studies could be attributable to the true intervention effect or to variability in the population, interventions, and outcomes being studied (clinical diversity), and/or study design and risk of bias (methodological diversity). Heterogeneity âmanifests itself in the observed inter- vention effects being more different from each other than one would expect due to random error (chance) aloneâ (Higgins and Green, 2011, p. 9.27). To properly interpret the results of a meta-analysis, heterogeneity must be assessed and factored into interpretation of the findings. Heterogeneity is actually expected and potential sources can be identified from the formulation of the population, intervention, comparison, and outcome (PICO) statements. For example, the condition(s) of interest in the population are predefined but may still differ from one study to another. Similarly, the char- acteristics of the participants of interest likely differ among studies. In addition, specific aspects of the intervention may differ (e.g., type of supplement, diet) and different co-interventions may be permitted among studies, which can affect the results. In addition, different surrogate measures and composite outcomes are often considered. The types of study design, the methodological quality of the studies, and the duration of the study might be sources of heterogeneity. As a first step in assessing the heterogeneity of the outcome of interest, the results of the comparisons for the interventions are displayed in a forest plot of each treatment comparison and reviewed for patterns and/or outliers. To help determine the extent of heterogeneity for the outcome of interest, statistical measures for heterogeneity, such as Iâ2, are considered. As noted by Higgins et al. (2003, p. 558): Iâ2 = 100% Ã (Q â df)/Q where Q is Cochranâs heterogeneity statistic and df the degrees of freedom. Negative values of Iâ2 are put equal to zero so design and risk of bias. As described in Box 2-5, unexplained heterogeneity can affect the interpretation of results in meta-analyses. As such, the strength- of-evidence domain of inconsistency, which characterizes heterogeneity, can play an important role in synthesizing and interpreting a body of evidence.
APPLYING THE GUIDING PRINCIPLES REPORT 53 that Iâ2 lies between 0% and 100%. A value of 0% indicates no observed heterogeneity, and larger values show increasing heterogeneity. Different thresholds for identifying the extent of heterogeneity have been proposed. For example: â¢ Iâ2 â¤ 0.25 heterogeneity is not an issue â¢ 0.25 < Iâ2 < 0.50 heterogeneity exists but is not an issue â¢ 0.50 â¤ Iâ2 < 0.75 heterogeneity exists and its causes should be explained â¢ 0.75 â¤ Iâ2 the causes of heterogeneity must be explained One method for attempting to explain the cause of heterogeneity is influence analysis. For each pair of treatments, a study is removed and a meta-analysis of the remaining studies performed. This is repeated for each study that is part of the direct evidence. The results of the meta-analyses (such as the point and confidence interval estimates, and Iâ2) are then assessed to identify the studies having the greatest effect on heterogeneity. Conceptually similar, cross validation can be performed to help explain the cause of heterogeneity. This method evaluates heterogeneity by removing a study considered to be an outlier and deriving a predictive distribution from the remaining studies. To determine whether heterogeneity exists, the observed treatment effect for the outlier study is compared to the predicted treatment effect for this study based on the predictive distribution. After identifying studies substantively contrib- uting to heterogeneity, the characteristics (as per the PICO statement) and meth- odological quality of these studies are assessed. Comparing these characteristics and quality indicators with the main body of studies in the evidence base may help identify source(s) of the heterogeneity. Once identified, there are two approaches for determining whether a subgroup effect interacts with the treatment effect: 1. Perform a subgroup analysis, which consists of a separate analysis at each level of the subgroup. 2. Perform a meta-regression analysis, which contains a common between- trial heterogeneity estimate and an interaction term Î² with the treatment effect. Assessment of heterogeneity is an essential aspect of synthesizing results from the studies in a systematic review. Such an assessment can be informative in identifying characteristics of the population that yield different results, which in turn can lead to a better understanding of the efficacy of the intervention under study. The AHRQ Systematic Review performed meta-analyses for key ques- tions and subquestions when randomized controlled trials were available, but it did not explore the potential sources of heterogeneity. Recognizing the importance of explaining the inconsistencies in order to have confi-
54 DIETARY REFERENCE INTAKES FOR SODIUM AND POTASSIUM dence in the meta-analyses results, the committee carried out subgroup analyses and meta-regression analyses in instances in which heterogene- ity was judged to be high. Details of the committeeâs analyses to explore u Â nexplained heterogeneity are described in Chapters 6 and 10. Use of strength-of-evidence ratingâ Pursuant to the guidance provided in the Guiding Principles Report, the committee determined that it would establish a DRI based on chronic disease if there was at least moderate strength of evidence for both a causal and an intakeâresponse relation- ship between potassium or sodium intake and chronic disease risk. In this approach, situations can arise in which there is moderate or high strength of evidence of a causal relationship between intake of a nutrient and a chronic disease indicator, but insufficient or low strength of evidence of an intakeâresponse relationship. Pursuant to the guidance in the Guiding Principles Report, a DRI based on chronic disease would generally not be established in this case because of limitations in the evidence. The lack of a DRI based on chronic disease, however, does not necessarily mean that no benefit exists; rather, there is a lack of evidence of sufficient strength to characterize the intakeâresponse relationship and thereby establish a DRI based on chronic disease. The committee primarily used the strength-of-evidence grades provided in the AHRQ Systematic Review for causal relationships. In select instances in which the committee explored unexplained heterogeneity, the strength- of-evidence grading was reassessed. The AHRQ Systematic Review did not conduct intakeâresponse analyses. Accordingly, for chronic disease indica- tors with moderate strength of evidence selected to inform the derivation of the CDRRs, the committee sought to characterize the intakeâresponse relationship; details of the committeeâs additional analyses are provided in Chapters 6 and 10. Qualified Surrogate Markers Guidance from the Guiding Principles Report A surrogate marker is âa biomarker that is intended to substitute for a clinical endpointâ (Biomarkers Definitions Working Group, 2001, p. 91) by accurately predicting the effect of a measured intervention on an unÂ easured clinical outcome. Surrogate markers are particularly useful m when evaluating the effect of interventions on chronic disease relationships for which a long duration and large sample sizes are needed to evaluate chronic disease outcomes but are not feasible. The Guiding Principles Report recommended that if evidence on the relationship between intake and a qualified surrogate marker is to be used
APPLYING THE GUIDING PRINCIPLES REPORT 55 in establishing the DRI based on chronic disease, it ideally would be used as supporting evidence (NASEM, 2017a) (see Box 2-1, Guiding Principles Report Recommendation 2). Qualifying a surrogate marker involves âassessment of available evidence on associations between the biomarker and disease states, including data showing effects of interventions on both the biomarker and clinical outcomesâ (IOM, 2010, p. 2). The Guiding Prin- ciples Report further recommended that âqualification of surrogate markers must be specific to each nutrient or other food substance, although some surrogates will be applicable to more than one causal pathwayâ (NASEM, 2017a, p. 8). This suggests that, for a DRI committee to use a qualified surrogate marker for the purposes of informing a DRI based on chronic disease, fit for purpose needs to be demonstrated. This type of evaluation stemmed from the recognition that caution is needed when generalizing surrogate marker qualification status from one context to another (IOM, 2010; Yetley et al., 2017). Committeeâs Application of the Guiding Principles Report A 2010 Institute of Medicine report developed a conceptual framework for qualifying surrogate markers for specific uses (IOM, 2010). The two key components of the qualification framework are (1) an objective and rigorous evaluation of the available evidence, and (2) a scientific judgment that the potential surrogate marker is fit for the purpose for which it is intended (e.g., for setting DRIs for the apparently healthy population within a dietary context). The guidance described in the Guiding Principles Report (NASEM, 2017a) and the framework for surrogate markers (IOM, 2010) provided a conceptual foundation that the committee used in reviewing the evidence in support of establishing DRIs based on chronic disease for potassium and sodium. For the committee to consider whether a biomarker was a qualified surrogate marker and use it as supporting evidence to establish CDRRs, a moderate strength of evidence for both a causal relationship and an intakeâresponse relationship between potassium or sodium intake and the biomarker was deemed necessary. In its application of this guidance, the committee encountered two dif- ferent scenarios for blood pressure. The details of the evidence and the com- mitteeâs decision making are presented in Chapters 6 and 10, but because the two scenarios exemplify the concepts related to the use of qualified surrogate markers and fit for purpose, a brief description of the difference is provided here. For potassium, there was evidence of a significant reduction in both systolic and diastolic blood pressure with potassium supplementa- tion. However, an intakeâresponse relationship could not be discerned. Fur- thermore, there was insufficient evidence of an effect of potassium intake
56 DIETARY REFERENCE INTAKES FOR SODIUM AND POTASSIUM on cardiovascular disease outcomes. This lack of evidence prevented the committee from considering blood pressure a qualified surrogate marker for cardiovascular disease in the context of potassium intake and from using the evidence to support the derivation of potassium CDRR values. In contrast, the evidence on blood pressure in the context of sodium intake was more robust, and the committee was able to consider blood pressure as a qualitied surrogate marker for hypertension and cardiovascular disease (details of this assessment are provided in Chapter 10, Annex 10-2). Balancing Benefits and Harms Guidance from the Guiding Principles Report The Guiding Principles Report states, âdeficiency, toxicity, and multiple chronic diseases need to be considered when balancing benefits and harmsâ (NASEM, 2017a, p. 227). For example, when making decisions about establishing an adequate intake level, the committee may need to evaluate evidence as to whether nutrient intakes below the AI might increase the risk of a chronic disease. That is, DRI committees need to consider whether the benefits associated with the AI might be adversely affected by harms associated with a chronic disease at or below this intake level. Conversely, DRI committees would need to consider a similar evaluation as to whether intakes above a UL might confer benefits that needed to be balanced against the harms associated with intakes at this level. Committeeâs Application of the Guiding Principles Report The information gathered by the committee contained evidence on dif- ferent indicators related to potassium and sodium intakes that might result in benefits and harms. In deriving the DRIs for adequacy and toxicity, the committee made an effort to consider all types of benefits and harms, including potential chronic disease effects, and to be transparent about its rationale for the decisions made for each DRI category for both nutrients. THE CHRONIC DISEASE RISK REDUCTION INTAKE IN CONTEXT OF THE OTHER DRI CATEGORIES In its review of the evidence and application of the guidance in the Guiding Principles Report, the committee considered the conceptual inter- relationships among the DRI categories. The following sections briefly summarize how the committee applied its collective expert judgment to make the distinction between the CDRR and the other DRI categories for potassium and sodium. It was beyond its scope to determine how future
APPLYING THE GUIDING PRINCIPLES REPORT 57 DRI committees can systematically make such decisions moving forward. The committee acknowledges the challenges that the CDRR might present for DRI users as they attempt to interpret it in the context of the DRI model that existed prior to the Guiding Principles Report. The need for additional guidance on the expanded DRI model, for both DRI committees and DRI users, is described as a future direction in Chapter 12. The CDRR and the DRIs for Adequacy The committee interpreted the CDRR as distinct from the DRIs for ade- quacy (i.e., EAR and RDA, or AI). In its approach, the committee attempted to make a delineation between the evidence it reviewed for establishing the DRIs for adequacy (which ultimately remained AIs for both potassium and sodium) and the evidence it reviewed for establishing the CDRRs. AIs are established when evidence is insufficient to establish EARs and RDAs. The AI is âa recommended average daily nutrient intake level based on observed or experimentally determined approximations or estimates of nutrient intake by a group (or groups) of apparently healthy people who are assumed to be maintaining an adequate nutritional stateâ (IOM, 2006, p. 11). An adequate nutritional state is defined in various ways, including normal growth, maintenance of normal plasma levels of nutri- ents, and other features of general health (IOM, 2006). Before DRIs based on chronic disease were included in the DRI model, evidence on chronic diseaseârelated indicators had been considered, and in some cases used to inform the derivation of an AI. The AI for total fiber, for instance, was established based on evidence of its relationship to coronary heart disease (IOM, 2002/2005). The expanded DRI model allows for a more nuanced characteriza- tion of the relationship between nutrient intake and chronic disease risk reduction. Although an important step forward, the expansion of the DRI model created challenges, particularly once the committee determined there was insufficient evidence to establish EARs and RDAs for potassium and sodium. For instance, in the 2005 DRI Report, the âadequate nutritional stateâ for potassium encompassed indicators that the committee considered in context for establishing a CDRR. The approach taken to the evidence in support of establishing the potassium DRIs for adequacy is therefore Â arkedly different than that taken in the 2005 DRI Report. m Despite the conceptual delineation, the review of the evidence indica- tors was context specific. For instance, the committee reviewed evidence of potential harmful health effects of a range of sodium intakes that was likely to extend below an AI. The range of potential harmful health effects included indicators related to chronic disease. In this context, the evidence was reviewed to ensure that the selected AI values did not potentially lead
58 DIETARY REFERENCE INTAKES FOR SODIUM AND POTASSIUM to detrimental effects. This use is different than using such evidence as an indicator to establish the sodium CDRRs. In the case of sodium, failure to identify chronic disease risk reduction at intakes below the CDRR reflects a lack of evidence rather than a lack of effect. This distinction is important from a practical perspective. In the past, the range of intakes between the RDA or AI and the UL has often been characterized as âsafe and adequate.â The committee cautions against interpreting the gap between the sodium AI and CDRR in such a manner, because intake levels below the sodium CDRR do not reflect a known absence of chronic disease risk. Moreover, as discussed in Chapter 10, there is evidence of benefits with respect to blood pressure with reducing intakes below the CDRR, but the evidence alone was not of sufficient strength to support chronic disease risk reduction. The CDRR and the UL The UL is âthe highest average daily nutrient intake level likely to pose no risk of adverse health effects for nearly all people in a particular groupâ (IOM, 2006, p. 11). The Guiding Principles Report recommended that the UL be retained in the expanded DRI model, but that it characterize toxico- logical risk (NASEM, 2017a). This recommendation narrows what would qualify as an adverse effect for a UL. In the expanded DRI model, consideration of both a UL and the CDRR is necessary because the meanings of both are different and valuable. The UL connotes an intake level after which toxicological risk increases with increasing intakes. For sodium, the CDRR reflects the lowest level of intake for which there was sufficient strength of evidence to characterize a chronic disease risk reduction. According to the Guiding Principles Report, if increases in chronic disease risk only occur at intakes greater than the UL, then no CDRR would be necessary. SUMMARY The Guiding Principles Report served as a foundation as the com- mittee considered the evidence to support DRIs based on chronic dis- ease for potassium and sodium. As the first to implement an expanded DRI model, the committee recognized opportunities to adapt some of the guidanceâparticularly related to nomenclatureâto ensure concepts were clearly and concisely conveyed. The committee followed the guidance on using strength-of-evidence grading in its decision making regarding the potassium and sodium CDRRs. To do so, it relied on evidence in the AHRQ Systematic Review. The committee concurred with the risk-of-bias tool that was used in the AHRQ Systematic Review and expanded the strength-of-
APPLYING THE GUIDING PRINCIPLES REPORT 59 evidence assessment to explore unexplained heterogeneity. To satisfy the criteria for establishing a CDRRâmoderate strength of evidence for both a causal and an intakeâresponse relationshipâthe committee also conducted intakeâresponse analyses for selected indicators. The committee assessed whether select biomarkers with at least moderate strength of evidence for a causal and an intakeâresponse relationship could serve as a qualified surrogate marker and be used as evidence to support the derivation of a CDRR, in the context of potassium and sodium intake. The committee also considered benefits and harms in its derivation of the potassium and sodium DRIs. The committee interpreted the Guiding Principles Report as creating a new DRI category, termed in this report the Chronic Disease Risk Reduc- tion Intake (CDRR), which is distinct from the AI and UL. In moving from the previous DRI model to an expanded model, the committee needed to consider conceptual interrelationships among the DRI categories. REFERENCES Balshem, H., M. Helfand, H. J. Schunemann, A. D. Oxman, R. Kunz, J. Brozek, G. E. Vist, Y. Falck-Ytter, J. Meerpohl, S. Norris, and G. H. Guyatt. 2011. GRADE guidelines: 3. Rating the quality of evidence. Journal of Clinical Epidemiology 64(4):401-406. Berkman, N. D., K. N. Lohr, M. Ansari, M. McDonagh, E. Balk, E. Whitlock, J. Reston, E. Bass, M. Butler, G. Gartlehner, L. Hartling, R. Kane, M. McPheeters, L. Morgan, S. C. Morton, M. Viswanathan, P. Sista, and S. Chang. 2013. Grading the strength of a body of evidence when assessing health care interventions for the Effective Health Care Pro- gram of the Agency for Healthcare Research and Quality: An update. In Methods Guide for Effectiveness and Comparative Effectiveness Reviews. Rockville, MD: Agency for Healthcare Research and Quality. Biomarkers Definitions Working Group. 2001. Biomarkers and surrogate endpoints: Pre- ferred definitions and conceptual framework. Clinical Pharmacology and Therapeutics 69(3):89-95. Guyatt, G., A. D. Oxman, E. A. Akl, R. Kunz, G. Vist, J. Brozek, S. Norris, Y. Falck-Ytter, P. Glasziou, H. DeBeer, R. Jaeschke, D. Rind, J. Meerpohl, P. Dahm, and H. J. Schunemann. 2011a. GRADE guidelines: 1. Introduction-GRADE evidence profiles and summary of findings tables. Journal of Clinical Epidemiology 64(4):383-394. Guyatt, G., A. D. Oxman, S. Sultan, P. Glasziou, E. A. Akl, P. Alonso-Coello, D. Atkins, R. Kunz, J. Brozek, V. Montori, R. Jaeschke, D. Rind, P. Dahm, J. Meerpohl, G. Vist, E. Berliner, S. Norris, Y. Falck-Ytter, M. H. Muran, and H. J. Schunemann. 2011b. GRADE guidelines: 9. Rating up the quality of evidence. Journal of Clinical Epidemiol- ogy 64(12):1311-1316. Higgins, J. P. T., and S. Green. 2011. Cochrane handbook for systematic reviews of interven- tions: Version 5.1.0. http://handbook.cochrane.org (accessed December 16, 2018). Higgins, J. P. T., S. G. Thompson, J. J. Deeks, and D. G. Altman. 2003. Measuring inconsis- tency in meta-analyses. BMJ 327(7414):557-560. IOM (Institute of Medicine). 1994. How should the Recommended Dietary Allowances be revised? Washington, DC: National Academy Press. IOM. 2002/2005. Dietary Reference Intakes for energy, carbohydrate, fiber, fat, fatty acids, cholesterol, protein, and amino acids. Washington, DC: The National Academies Press.
60 DIETARY REFERENCE INTAKES FOR SODIUM AND POTASSIUM IOM. 2006. Dietary Reference Intakes: The essential guide to nutrient requirements. W Â ashington, DC: The National Academies Press. IOM. 2010. Evaluation of biomarkers and surrogate endpoints in chronic disease. ÂWashington, DC: The National Academies Press. NASEM (National Academies of Sciences, Engineering, and Medicine). 2017a. Guiding prin- ciples for developing Dietary Reference Intakes based on chronic disease. Washington, DC: The National Academies Press. NASEM. 2017b. Guiding principles for developing Dietary Reference Intakes based on chronic diseaseâHighlights from the consensus report. https://www.nap.edu/resource/24828/ GuidingPrinciplesforDRIs-ReleaseSlides.pdf (accessed January 28, 2019). Newberry, S. J., M. Chung, C. A. M. Anderson, C. Chen, Z. Fu, A. Tang, N. Zhao, M. Booth, J. Marks, S. Hollands, A. Motala, J. K. Larkin, R. Shanman, and S. Hempel. 2018. Sodium and potassium intake: Effects on chronic disease outcomes and risks. Rockville, MD: Agency for Healthcare Research and Quality. Schunemann, H., J. Brozek, G. Guyatt, and A. D. Oxman. 2013. Introduction to GRADE handbook. https://gdt.gradepro.org/app/handbook/handbook.html (accessed January 16, 2019). Sterne, J. A., M. A. Hernan, B. C. Reeves, J. Savovic, N. D. Berkman, M. Viswanathan, D. Henry, D. G. Altman, M. T. Ansari, I. Boutron, J. R. Carpenter, A. W. Chan, R. Churchill, J. J. Deeks, A. Hrobjartsson, J. Kirkham, P. Juni, Y. K. Loke, T. D. Pigott, C. R. Ramsay, D. Regidor, H. R. Rothstein, L. Sandhu, P. L. Santaguida, H. J. Schunemann, B. Shea, I. Shrier, P. Tugwell, L. Turner, J. C. Valentine, H. Waddington, E. Waters, G. A. Wells, P. F. Whiting, and J. P. Higgins. 2016. ROBINS-I: A tool for assessing risk of bias in non-randomised studies of interventions. BMJ 355:i4919. Taylor, C. L. 2008. Framework for DRI development: Components âknownâ and components âto be explored.â https://www.nal.usda.gov/sites/default/files/fnic_uploads/Framework_ DRI_Development.pdf (accessed April 9, 2019). Yetley, E. A., A. J. MacFarlane, L. S. Greene-Finestone, C. Garza, J. D. Ard, S. A. Atkinson, Â D. M. Bier, A. L. Carriquiry, W. R. Harlan, D. Hattis, J. C. King, D. Krewski, D. L. OâConnor, R. L. Prentice, J. V. Rodricks, and G. A. Wells. 2017. Options for basing Â Dietary Reference Intakes (DRIs) on chronic disease endpoints: Report from a joint US-/Canadian-sponsored working group. American Journal of Clinical Nutrition 105(1):249S-285S.