4

Discussion and Future Directions

Objectively assessing the totality and quality of the evidence using a comprehensive systematic review is a required step in updating a Dietary Reference Intake (DRI) nutrient review. The process described in this report does not replace such a review. As this report establishes, an evidence scan supports an economical and objective method of assessing whether new evidence is sufficient to support a formal reexamination of a nutrient or nutrients for a DRI review.

This evidence scan was carried out to obtain new information on riboflavin in association with health outcomes, including adequacy, chronic disease risk, and toxicity. An iterative process of evaluation and adjustment was needed in order to obtain the most relevant evidence to inform identifying and selecting relevant studies. The committee was not asked to make recommendations; however, it comments on its methodological approach and discusses its findings and interpretation of the process to provide the study sponsors with greater context to support their interpretation and application of the reported results.

APPROACH TO THE EVIDENCE SCANNING PROCESS

The committee used its experience with evidence synthesis methodology and considered lessons learned from Brannon et al. (2016) and evidence from a previous evidence scan (NASEM, 2020). First, the committee recognized the need for an analytical framework to facilitate framing terms to identify relationships between dietary exposure and clinical outcomes and enable evaluating the relevance of the evidence specific to those relationships.

One new facet of this analytic framework relates to expanding DRIs to include chronic disease endpoints (NASEM, 2017) and applying and specifying the chronic disease risk reduction for sodium (NASEM, 2019). The original generic DRI framework (Russell et al., 2009) did not distinguish categories of outcomes (adequacy, chronic disease or toxicity); for this evidence review, these categories were needed because the evidence differs for each category. When chronic disease outcomes are considered for a specific nutrient DRI, an expanded analytic framework that distinguishes among these outcomes is needed. Specifically, it was used to guide developing search criteria using a PI[E]COD framework (see Chapter 2). This approach enabled the committee to reach consensus in its assessment of abstracts and full-text articles.

In addition, as noted in Brannon et al. (2016), the evidence scan must be tailored based on the nature of the evidence in the literature for health and clinical outcomes related to adequacy or chronic disease risk. It also needs to have sufficient breadth to identify new clinical outcomes but avoid identifying irrelevant citations. As noted in the methods, the committee adjusted the search to address these items.

The committee identified as an important lesson learned an understanding and appreciation of the critical nature of the iterative process. This became apparent at all levels of the screening. Without the ability to refine the screening criteria and data interpretation, the committee would likely have missed relevant evidence.

Design of the Search Process

The committee purposefully chose to be inclusive in the search design, such as going beyond the United States and Canada to include data from countries that the World Bank ranked as high or upper middle income at the time of the study. The committee used these two income groups because study results from them would likely be broadly similar to general U.S. and Canadian population groups. However, the committee also fully recognizes that such country-level assessments of applicability risks missing nuances, such as within-country differences and disparities in health, nutritional status, socioeconomic factors, and access to care. The committee realizes that contemporary populations frequently add supplements to their nutrient intake and that the search required articles to be indexed using terms for “diet” and other potentially restrictive concepts. Thus, potentially relevant articles not indexed with these terms may have been omitted. Additionally, the search was limited to English-language articles, so it may have missed studies from qualifying countries that were not published in English.

To illustrate, a challenge in evaluating the available literature for riboflavin status is the frequent inability to assess its effects independent of other B vitamins and separate the effects contributed by endogenous B vitamins naturally present in food from mandatory fortification and nutrient supplementation. Mandatory fortification is a critical consideration due to inconsistencies in policies among high- or upper-middle-income countries. Nutrient supplementation is of further concern given the widespread use of multivitamins. These concerns will likely apply to evaluation of other B vitamins, particularly niacin and thiamin. Thus, the results reported here must be interpreted in light of these possibilities.

The committee chose to expand the date range to capture articles that may not have been identified in the first DRI report on riboflavin (IOM, 1998). The committee extended its search to multiple databases and all potentially relevant peer-reviewed articles published from 1969 on. Although this evidence scan is not intended to supplant a comprehensive systematic review, the committee recognizes that its design, which is grounded in an analytic framework and informed by prespecified criteria, supports a search that is more wide ranging and inclusive than would be possible in a narrative review. Thus, while a number of studies retrieved duplicated those identified in the 1998 review, several additional studies of potential relevance were found.

Selection of Inclusion and Exclusion Criteria

The committee selected its inclusion and exclusion criteria to identify studies (or a subset of data within them) that would provide the most consistent estimate of riboflavin status and outcome relationships, including chronic disease and toxicity. The criteria reflect the expertise of the committee and subject-matter experts (see Appendix B for the agenda). The committee realized the importance of consulting with experts for its understanding of measuring and reporting riboflavin status and new and emerging evidence on health outcomes, including potential toxicities.

Discussion with the subject-matter experts helped the committee to elucidate associations between riboflavin exposures, markers of status, and outcomes. The committee used this

information in designing its analytic framework to determine if the evidence scan was appropriately comprehensive in identifying new approaches and outcomes that should be considered in developing the prespecified criteria. This approach helped to eliminate studies that, although corroborated in the literature, were not relevant because of small sample sizes or less rigorous methods than are currently considered acceptable. The criteria further allowed the committee to discriminate among studies that included data for irrelevant nutrients, including those from multinutrient supplements. In addition, discussing and refining the criteria during the screening process enabled the committee to consistently apply them based on a stronger shared understanding of their interpretation in the context of the studies identified. Future evidence scan committees may also find it helpful to review and discuss how to apply their criteria in the screening to ensure consistency in the process.

The Screening Process

The abstract screening process, as usually occurs during evidence synthesis projects, revealed a considerable number of records that did not meet the inclusion criteria. Most studies conducted outside of the United States and Canada were excluded in the full-text screening, when the committee ascertained that the populations studied in lower-middle- and low-income countries were not comparable with those in both countries. Of the remaining studies, one was deemed relevant only when it described participants who met the inclusion criteria (see Chapter 2, Table 2-1). Frequently, the population studied, methods used, or data obtained were not sufficiently rigorous for the findings to be useful. A majority of studies (60 percent) were excluded due to an inability to assess the effects of riboflavin independently. At the full-text screen, 26 percent of articles were excluded because they reported insufficient data to assess dietary intake response and 24 percent because they did not report any of the outcomes of interest. Thus, at the final, full-text rescreening, 35 articles were determined eligible for data extraction (see Chapter 3, Figure 3-1).

FINDINGS

Assessment of Findings Relative to Methodology and Data Reporting

Study Methodology

The committee noted considerable heterogeneity in study methodologies, including to estimate riboflavin intake in blood and urine. In the first full-text screening, riboflavin levels were reported in blood and/or plasma. Because plasma flavin concentrations are less reliable indicators of riboflavin status than the erythrocyte glutathione reductase activation coefficient (EGRAC), the committee did not add plasma riboflavin to its inclusion criteria. However, studies reported either or both erythrocyte flavin/EGRAC and plasma riboflavin. The committee then decided to exclude those studies reporting only plasma levels of riboflavin (see Chapter 2, Table 2-1).

Another methodological variant was the assessment of riboflavin intake in urine. The methods reported included spot-, timed-, and 24-hour urine collection. Although all the methodologies have the potential for error, spot-urine collection errors in particular were noted as susceptible to variability in reported results, especially if the collection time was not

consistent. The committee thus noted this potential variability in data reporting for riboflavin status from urine collection methodology in its comments in Tables 3-2 and 3-3.

The committee’s assessment of studies with implications for potential toxicity related to riboflavin intake presented a unique challenge. Based on its discussion with subject-matter experts, the committee anticipated limited evidence, if any, would be identified. No relevant studies addressing toxicity were found in the screening steps. Structuring the criteria was difficult without clear indicators of toxicity reported in the current research. The committee therefore determined that capturing all possible articles relevant to toxicity would require broadening the search in any future scan or scoping review.

Data Reporting

The committee found inconsistencies in reporting riboflavin status in studies that combined food and supplement intakes. Additionally, the use of multinutrient supplements and foods with added nutrients in the study design or data reporting confounded the committee’s ability to interpret either adequacy or clinical outcome relationships. The committee also identified potential new outcomes for consideration. First, a number of studies identified associations between riboflavin status and blood pressure. However, in the final full-text screen, the committee identified only 6 of 11 articles related to blood pressure associations with riboflavin status as relevant (see Chapter 3, Table 3-3). Although the number of relevant articles was small, the committee nevertheless interpreted these results as an indicator for considering this outcome in future evidence reviews. The committee also reviewed studies for outcomes related to cognitive function but did not identify any relevant studies in its final full-text review. Further research may provide more support for this outcome in the future.

FUTURE DIRECTIONS

Ninety-one studies were identified as potentially relevant. From these, the committee identified only 35 that included relevant data. In developing its criteria for assessing methodological rigor, the committee realized that additional research would be useful in future evidence scans. Three priority topic areas emerged:

1. In the absence of a validated biomarker or surrogate marker and/or uncertainty about dietary intake estimates, the committee notes a need to identify a means to consider new and emerging intake estimate methodologies. This could include nontraditional methods to obtain estimates of nutrient status, such as -omics technology to detect riboflavin status.
2. For nutrients that can reduce the risk of chronic disease, validated surrogate biomarkers of risk will be essential for identifying which nutrients are of greatest interest to further investigate their impact on physiologic outcomes related to adverse health outcomes.
3. While the objective of the evidence scan is to examine the state of the evidence for a given nutrient of interest, the committee notes that because nutrients are not consumed in isolation and interact metabolically, developing a strategy to scan for nutrients in related clusters may be worth considering.

Concluding Statement

This report represents the committee’s interpretation of its charge. Given the purpose of the evidence scan to identify new evidence on adequacy, risk of chronic disease, and potential toxicities, the committee saw as its challenge the need to design the search to be broad enough to achieve its purpose but still specific enough to accurately reflect the current status of the evidence on riboflavin needed to support a DRI review. The committee was not charged to determine whether the evidence scan results are sufficient to support a formal DRI review. However, this report will inform the federal agencies about the type and quality of new evidence and assist them in assigning a priority to a comprehensive systematic review for riboflavin.

REFERENCES

This page intentionally left blank.