1. Dietary Reference
Values for sodium:
challenges and
opportunities
Androniki Naska
National and Kapodistrian University of
Athens, School of Medicine
Chair of the EFSA Working Group on Dietary
Reference Values for minerals
16 October 2019
2. Sodium, a challenging nutrient
3
Sodium is an essential nutrient and is needed by the body
Sodium intake has been associated with increased blood pressure and risk of CVD
Is there an adequate and safe range of intake for sodium?
Universal salt reduction has been the subject of scientific controversy
Polarization of the scientific literature on sodium intakes and health outcomes
Conclusions of systematic reviews on the topic are diverging
PROPONENTS
“High sodium intakes increase blood pressure,
thus, the risk of cerebro-cardiovascular events”
“Salt reduction policies will produce major
public health benefits”
OPPONENTS
“Relationship between sodium intake and
clinical outcomes is U- or J-shaped”
“The harms associated with low sodium intakes
may mitigate any potential benefits of blood
pressure reduction”
3. EFSA Prometheus
4
EFSA Prometheus approach
(PROmoting METHods for
Evidence Use in Scientific
assessments)
http://www.efsa.europa.eu/interactive_pages/
prometheus/prometheus
4. Process
2016 –
2017
• Drafting of the background sections of the Opinion
• Protocol development
• Protocol published (ZENODO and PROSPERO)
2018
• Protocol implementation
2019
• Opinion finalisation, adoption
• Publication
• EFSA Journal
• Amended protocol published (ZENODO and PROSPERO)
Public
consultation
Public
consultation
+ technical
meeting
Open plenary
Dietary reference values for sodium,
Published in EFSA Journal on September 4, 2019
https://www.efsa.europa.eu/en/efsajournal/pub/5778
5. Methodological steps (adults)
6DRVs for sodium
Data integration
Expert Knowledge Elicitation (EKE)
BIOMARKERS AS
INDICATORS OF
REQUIREMENTS
No useful marker
SODIUM AND HEALTH CONSEQUENCES
Literature search
Study selection
(title/abstract and full text screening)
Risk of bias
appraisal
Data extraction
Data analysis
(Meta-analyses, meta-regression where possible)
Problem formulation
SODIUM BALANCE
Literature search
Study selection
Assessment of
the evidence
Problem formulation
Uncertaintyanalysis
BIOMARKERS OF
INTAKE
24-hour urinary
sodium excretions
6. Evidence
Metabolic studies that inform about systemic mechanisms to maintain a Na balance.
Several studies excluded because of methodological limitations
3 studies in adults and 1 study in adolescents were thoroughly reviewed
Findings
Balance maintained over a wide range of Na intake
Mean Na intake assessed in eligible balance studies ranged between 1.5 g and 4.9
g/day in adults and between 1.31 and 3.95 g/day in adolescents.
Rhythmical variations in the Na body pool independent of Na intake
Response of sympathetic nervous system and the renin–angiotensin–aldosterone
system to conserve Na evident at excretion below 100mmol/24 hours
Conclusion
Balance studies cannot be used to determine Na requirements
Can be used to inform about the levels of Na intake adequate to maintain a null
balance.
Balance studies
7
7. Prisma chart – BP and CVD
8
Records screened in the basis
of title and abstract (n=7,141)
Full–text articles assessed for
eligibility (n=402)
Studies included in the
qualitative synthesis (n=45)
36 experimental
9 observational
Studies included in
quantitive synthesis (meta–
analysis) (n=32)
Records identified
through database
searching, after
duplicates removed
(n=6,264)
Additional records
identified through
snowballing, after
duplicates removed
(n=877)
Records excluded
(n=6,731)
Systematic reviews (n=8)
Articles excluded, with
reasons (n=357)
IdentificationScreeningEligibilityIncluded
8. Findings
Significant effects of Na reduction on SBP by –3.9 mmHg (95%CI: –
5.1, –2.8 mmHg)
Significant effects of Na reduction on DBP by –2.0 mmHg (95%CI: –
2.8, –1.2 mmHg)
Linear dose–response over the range of mean UNa observed (49 –
209 mmol/24 h (1.3 – 4.8 g/day))
Mean SBP increased by 5.3 mmHg (95% CI: 3.6, 6.9 mmHg) for each
100 mmol (2.3 g)/24–h increase in mean UNa
Mean DBP increased by 2.6 mmHg (95% 1674 CI: 1.6, 3.7 mmHg) for
each 100 mmol (2.3 g)/24–h increase in mean UNa
Stronger association among hypertensive vs normotensive individuals
and among subjects aged 50 years vs subjects <50 years
Blood pressure in adults
9
9. Evidence
1 prospective cohort study (moderate RoB) on the long–term
relationship between UNa and blood pressure levels
2 RCTs (low RoB) and 2 prospective observational studies (low and
moderate RoB) on the relationship between UNa and risk of
hypertension
Findings
Support the positive relationship between UNa and blood pressure
levels
Blood pressure in adults
10
10. Evidence
No RCT eligible
Small number of prospective cohort studies
3 cohorts on the risk of stroke or on the risk of coronary heart disease
PREVEND (low RoB); EPOGH/FLEMENGHO and the Finnish cohort (moderate RoB)
3 cohorts on the risk of cardiovascular disease
TOPHI/II (low RoB); EPOGH/FLEMENGHO and InCHIANTI (moderate RoB)
No quantitative analysis
Findings
Limited conclusions
Risk of coronary heart disease: some evidence for a positive association
Risk of stroke: some evidence for an inverse association
Small number of studies and mechanisms unclear
Risk of cardiovascular disease: some evidence for a positive association
Cardiovascular diseases
12
11. An average requirement (AR) and a population reference intake
(PRI) can NOT be established for Na, because the distribution of the
requirement cannot be determined
Data relevant to the setting of DRVs for sodium
Balance studies: levels of Na intake adequate to maintain a null
balance
Relationship between Na and blood pressure or CVD risk: levels of
Na associated with a reduced risk of chronic diseases
Expert judgement, taking account of the associated uncertainties
Use of a formal Expert Knowledge Elicitation (EKE) process
Guidance of EFSA published in 2014
Relevant data and integration of the
evidence
13
12. Formal Expert Knowledge Elicitation (EKE)
Evidence–based judgements about a quantity of interest
Judgements expressed about the range of possible values for the quantity of interest
and their relative likelihoods
Limits bias;
Structured process improves rigour of reasoning;
Clear and unambiguous expression of uncertainty;
Rationale documented.
EKE conducted with the experts of the EFSA WG on DRVs for minerals
‘Sheffield’ protocol
Method designed to elicit the knowledge of a group of experts in a face–to–face
elicitation meeting
Presence of an elicitor essential.
EKE method
14
13. Data on Na and blood pressure or CVD risks could inform about
the levels of sodium intake associated to a reduced risk of chronic
diseases
Balance studies could inform about the levels of sodium intake which
are adequate to maintain a null sodium balance
EKE questions
Question 2 What is the lowest level of sodium intake which is adequate (i.e.
amount which allows to maintain sodium balance) for the majority (≥ 97.5%)
of the general population of adults?
Question 1 What is the lowest level of sodium intake at which the risk of
chronic disease (i.e. stroke, CHD) is minimised in the majority (≥ 97.5%) of
the general population of adults?
14. What is the lowest level of sodium intake at which the risk of
chronic disease (i.e. stroke, CHD) is minimised in the majority (≥
97.5%) of the general population of adults?
Group consensus uncertainty probability
distribution I
15. What is the lowest level of sodium intake which is adequate (i.e.
amount which allows to maintain sodium balance) for the
majority (≥ 97.5%) of the general population of adults?
Group consensus uncertainty probability
distribution II
16. A sodium intake of 2.0 g/day
represents a level of sodium
for which there is sufficient
confidence in a reduced risk of
CVD in the general adult
population.
A sodium intake of 2.0 g/day
is likely to allow most of the
general population to maintain
sodium balance
2.0 g of sodium per day is a
safe and adequate intake for
the general EU population of
adults
Conclusion – DRVs for adults
18
17. Safe:
the concept of a safe intake has been used in previous assessments regarding a daily
intake of a nutrient which does not give rise to concerns about adverse health
effects, in instances when a tolerable upper intake level (UL) could not be established.
Adequate:
An adequate intake (AI) is the value estimated when a population reference intake (PRI)
cannot be established because an average requirement (AR) cannot be determined (EFSA
NDA Panel, 2010).
The AI is the level of intake that is assumed to be sufficient based on
observations from groups of apparently healthy people.
Terminology
19
18. Mean/median intake of Na in European populations exceeds the safe and
adequate intakes proposed
Low risk of inadequate (insufficient) intake in European populations
Concerns relate to ‘excess’ intake of Na
For populations: value proposed can be used to inform the setting of population
goals for the reduction of Na intake
For individuals: if the usual intake of Na exceeds this value, it could be associated
with an increased risk of CVD, including concurring risk factors such as primary
hypertension.
In practice
20
19. Requirement for the daily accretion rate of sodium in fetal and
maternal tissues can be met by adaptive changes that maintain Na
homeostasis during pregnancy
No evidence that Na requirement of lactating women differs from the
requirement of non–lactating women
Lack of data from which an AR could be derived for infants
Upwards extrapolation from the estimated Na intake of fully breast–fed
infants during the first 6 months of life (120 mg/day)
Lack of data from which an AR could be derived for children
Downwards extrapolation from the reference value for adults, based on
the AR for energy and including a growth factor
Conclusions – other population groups
20. Safe and Adequate Intake
(g/day)
1–3 years 1.1
4–6 years 1.3
7–10 years 1.7
11–17 years 2.0
≥ 18 years 2.0
Conclusions
22
Adequate Intake
(g/day)
7–11 months 0.2
21. Contributors
25
NDA Panel Members
Dominique Turck (chair)
Jacqueline Castenmiller
Stefaan de Henauw
Karen-Ildico Hirsch-Ernst
John Kearney
Helle Katrine Knutsen
Alexandre Maciuk
Inge Mangelsdorf
Harry J McArdle (vice-chair)
Androniki Naska
Carmen Pelaez
Kristina Pentieva
Alfonso Siani
Frank Thies
Sophia Tsabouri
Marco Vinceti
EFSA WG on DRVs for minerals
Androniki Naska (chair)
Peter Aggett
Susan Fairweather-Tait
Ambroise Martin
Hildegard Przyrembel
Alfonso Siani
Marco Vinceti (vice-chair)
EFSA NUTRI staff
Agnès de Sesmaisons Lecarré
Silvia Valtueña Martinez
Anja Brönstrup (SO)
Jelena Gudelj Rakic (SNE)
Olga Vidal Pariente, Qingqing Sun,
Ester Artau Cortacans (trainees)
Pauline Lachevre (guest scientist)
EFSA AMU staff
Laura Ciccolallo
Elisa Aiassa
Andrea Bau
Daniela Tomcikova
Irene Munoz Guajardo
EFSA WG on uncertainty analysis
Andy Hart
Laura Martino (AMU staff)
Caroline Merten (SCER staff)
22. 34 opinions published
Cover protein, carbohydrates,
fats, energy, 14 vitamins and 15
minerals
Useful resources
Topic on DRVs on EFSA’s
website
Special issue on DRVs in EFSA
Journal
DRVs Finder
DRVs at EFSA
26
Poster presentation 8:30-9:15, 17th October Liffey B (Level 1)
Poster Session 2, Station 23, Presentation 2.
Principles:
Impartiality: free from preconceptions due to prior knowledge
Excellence in scientific assessment: outstanding credentials of the experts involved + use of sound scientific methodological approaches
Responsiveness: timeless and effectiveness of the assessment
Openness and transparency: engaging with relevant parties; increasing clarity and reproducibility of the assessment
4-step process designed to fulfil these principles: plan, conduct, verify report
Result: fit-for-purpose,evidence-based scientificadvice for decision-making
Biomarkers:
Findings
Homeostatic mechanisms maintain systemic distribution, acquisition and excretion of Na, including plasma Na concentration/activity, as a means of maintaining water homeostasis
Hyponatraemia and hypernatraemia related to disorders affecting water and electrolyte balance; seldom due to inappropriate Na intake
Plasma Na concentration does not accurately reflect Na body content
Conclusion
No appropriate biomarkers of Na status that can be used for deriving DRVs for sodium
Presentation will focus on the two other lines of evidence
Do we need to define what balance studies are or will the audience be familiar ?
[ADA] We can address it in Q&A section if it arises
Based on the review of the available evidence and the findings which have just been presented, the Panel concluded that the sodium requirement or its distribution in the population could not be determine; so, an AR and population reference intake (PRI) for sodium could not be established.
Still, studies informing the relationship between sodium intake and blood pressure levels or CVD risks and balance studies on sodium were considered to provide relevant data to inform the setting of DRVs for sodium.
In effect, data on the relationship between sodium intake and levels of blood pressure or CVD risks could inform about the levels of sodium intake associated to a reduced risk of chronic diseases AND balance studies could inform about the levels of sodium intake that are adequate to maintain a null sodium balance.
Because of the limited evidence available and of the associated uncertainties, it was not possible to identify such levels of sodium intake with certainty.
Expert judgement was required to weigh the available evidence and take account of the associated uncertainties. Thus, experts decided to use a formal Expert Knowledge Elicitation (EKE) to weigh and integrate the evidence AND I am going to illustrate the method in the following slides. EFSA published a guidance on EKE in 2014. I invite you to download it from EFSA’s website in case you would like to know more about the approach.
EFSA published a guidance on EKE in 2014. I invite you to download it from EFSA’s website in case you would like to know more about the approach. Let me summarize here the main elements.
The EKE is a formal method which is used to support experts in making evidence-based judgements about a quantity of interest (typically a parameter) in a structured way. What is elicited is a distribution which express judgements about the range of possible values of the parameter of interest, and their relative likelihoods.
It has several advantages:
Designed to mitigate psychological biases affecting expert judgements
Structured process improves rigour of reasoning
Allows a clear and unambiguous expression of uncertainty about the true value of the parameter of interest
Makes the rationale for the judgements transparent and documented
For sodium, the EKE was conducted with the experts of the EFSA WG on DRVs for minerals, who had been in charge of reviewing the available evidence and were well aware of the body of evidence.
Judgements were elicited following the ‘Sheffield’ protocol which is designed to elicit the knowledge of a group of experts in a face–to–face elicitation meeting.
The result is an uncertainty probability distribution that represents the experts aggregated judgements achieved via discussion.
It is be noted that, in this process, the presence of an elicitor is essential. The elicitor is a professional who is typically not in the field of interest and acts as an objective facilitator of the discussions.
Two separate elicitations were conducted in relation to the two sources of evidence available.
In the first one, experts had to express their judgements about the levels of sodium intake associated to a reduced risk of cardiovascular diseases. The elicitation question used in the exercise is displayed on the screen.
In the second oelicitation, experts had to express their judgements about the levels of sodium intake which are adequate to maintain a null sodium balance. The elicitation question used in the exercise is displayed on the screen.
In the next two slides I am going to present the outcome of the two elicitations. My talk will focus on the principles of the exercise. I am not going to enter into the details of the rationale for each judgement, which are described in the Opinion. Still, feel free to send specific questions through the system.
Two separate elicitations were conducted in relation to the two sources of evidence available.
In the first one, experts had to express their judgements about the levels of sodium intake associated to a reduced risk of cardiovascular diseases. The elicitation question used in the exercise is displayed on the screen.
What you see on the screen is the group consensus distribution obtained in response to the first question. Thus it expressed the expert judgements about the lowest level of sodium intake at which the risk of chronic disease is minimised in the majority of the general population of adults.
You might skip the text in bold below?
As eligible studies in relation to this source of evidence used UNa as marker of sodium intake, the scale used for the elicitation referred to UNa and was expressed in mmol/24 h.
In drawing this distribution the experts integrated the evidence about the relationship between Na and blood pressure and cardiovascular disease incidence. In particular, they took into consideration:
The strong evidence coming from experimental studies for a positive relationship between blood pressure and UNa in the range covered by the studies (ca. 50 to 200 mmol/day). Blood pressure is a well established risk factor of cardiovascular diseases and any increase in blood pressure level was considered adverse.
Experts also considered
The limited evidence from observational studies for a positive association between UNa and risk of hypertension, coronary heart disease and total cardiovascular diseases in the one hand
And the limited evidence from observational studies for an inverse association between UNa and risk of stroke in the other hand. It is important to stress that in drawing their ditrubutin the experts integrated the substantial uncertainty related to that relationship, due to the low number of studies eligible and the unclear mechanisms by which UNa could be inversely associated with the risk of stroke.
As can be seen from the figure, looking at the evidence, the experts agreed that the level of interest was unlikely to be above 120 mmol/day and very unlikely to be below 40 mmol/day. Rather, they agreed that it was very likely to be between 40 and 120 mmol/day. The probability that this value is in the ranges 40 and 80 mmol/day and 80 and 120 mmol/day was judged to be almost equal, and is ‘centered’ around 80 mmol/24 hour (equivalent to 2 g/day). It is to be noted that a slightly higher probability (50% vs 45%) was given to the range 80 to 120 mmol/day in the light of the evidence on stroke; the fact that the difference in the probability assigned to the two ranges is small (5%) actually reflects the substantial uncertainty associated with the evidence on stroke.
Importantly, the distribution captures the uncertainty of the experts about the lowest level of sodium intake at which the risk of chronic disease (stroke and coronary heart disease in particular) is minimised in the general population.
In the second elicitation, experts had to express their judgements about the levels of sodium intake which are adequate to maintain a null sodium balance. The elicitation question used in the exercise is displayed on the screen.
What you see on this screen is the group consensus distribution obtained in response to the second question. This time it expressed the expert judgements about the lowest level of sodium intake which is adequate to maintain a null sodium balance for the majority of the general population of adults.
Again, you might skip the parts in bold?
In relation to that question the scale used for the elicitation was expressed in g/day, keeping the unit used in the balance studies.
Again, the detailed rationale for the distribution is provided in the Opinion and I am just going to outline the main elements.
As can be seen from the figure, the experts considered that an adequate level of sodium intake was most likely in the range from 1.0 to 2.5 g per day and actually more likely to be between 1.5 and 2.0 g. The distribution of the probabilities in the range 1.0 to 2.5 integrates several elements:
First, evidence from one well designed balance study indicates that sodium balance could be maintain under heat stress with sodium intake of 1.5 g/day; however, this study involved a small number of young men, thus there is uncertainty regarding the general population
Second, there is a lack of data on the health effects of a sustained activation of the sympathetic nervous system and the RAAS system, which becomes apparent with sodium intake below 1 to 2 g.
- Overall, the probability distribution is shifted towards range 1.5–2.5 g/day to cover most of the population and considering the lack of data from balance studies below 1.5 g/day.
Similar to the previous example, the distribution captures the uncertainty of the experts about the lowest level of sodium intake which allows to maintain sodium balance for most of the general population.
In the next step parametric distributions were fitted to the group consensus distributions. At this stage, UNa values expressed in mmol/24 h were converted to g/day.
Based on the two fitted parametric distributions, cumulative uncertainty density functions were derived. They are represented here in the Figure. Again, it is important to note that the curves reflect the uncertainty of the experts about the levels elicited. They do NOT represent the probability (or risk) of a given outcome, be it cardiovascular disease or sodium ‘imbalance’, associated with a given level of sodium intake.
These cumulative uncertainty density functions were helpful for the Panel to integrate the two sources of evidence and derive reference values for sodium. And I am going to illustrate how.
In short, the red curve corresponds to the elicitation of the lowest level of sodium intake at which the risk of chronic disease (i.e. stroke, CHD) is minimised. For the interpretation of the curve, it is important to bear in mind that it integrates evidence going into opposite directions. In the one hand some evidence for a positive association between Na and risk coronary heart diseases and, in the other hand, some evidence for an inverse association between Na and risk of stroke. For this reason, going towards the right of the curve, the uncertainty that the risk of CHD is minimised increases; In contrast as one moves towards the left, the uncertainty that the risk of stroke is minimised increases.
The blue curve is the cumulative density function corresponding to the elicitation of the lowest level of sodium intake that allows to maintain sodium balance for most of the general population. In this case, the interpretation of the curve is unidirectional: as one goes from zero towards the right, the confidence of including the level of sodium intake that is adequate to maintain balance increases.
In identifying a reference value for sodium, the experts first aimed at identifying a level of Na intake associated with a reduction of CVD risk. The identification of such value was constrained by the bidirectional interpretation of the curve. Therefore the median level of the distribution, was chosen. Indeed, at the median, there is equal certainty that the level of sodium that it is associated with a reduction of cardiovascular risk is neither overestimated nor underestimated. As can be seen from the curve the median level corresponds to 2g sodium/day (red dot),
Looking at the blue curve, a level of sodium intake of 2 g/day is likely (i.e. with a 75% probability) to be adequate to maintain sodium balance for most of the general adult population (see blue dot). A probability of 75% was considered to be sufficiently conservative.
Therefore, the panel concluded that 2.0 g of sodium per day is a safe and adequate intake for the general EU population of adults
And now I would like to clarify the terminology used here.
The types of dietary reference values were defined in 2010 by the EFSA NDA Panel in an opinion on the principles on deriving and applying DRVs. This opinion includes a definition for each type of DRV and provide guidance on how to use these values.
It is to be noted that, although the term ‘safe intake’ was not defined in that opinion, the concept of a safe intake has been used in previous assessments when providing advice on a daily intake of a nutrient which does not give rise to concerns about adverse health effects, in case a tolerable upper intake level (UL) could not be established. So it is not a new concept in relation to nutrient reference values. The reference value for sodium is called ‘safe’ as the value proposed takes account of an increased risk of CVD at higher levels of sodium intake when exposure is prolonged.
As defined in the principles opinion, an adequate intake (AI) is the value estimated when a population reference intake (PRI) cannot be established because an average requirement (AR) cannot be determined. The setting of an AI involves more expert judgement than is used for determining an AR or PRI. The practical implication of an AI is similar to that of a PRI in the sense that it describes the level of intake of a nutrient that is considered adequate for good health. The distinction in the terms relates to the different strength of the scientific basis on which they rest. The reference value for sodium is called ‘adequate’ in line with this definition.
Challenges:
Sodium is a “special” nutrient in that adequate and safe intakes lie in a narrow range
Polarized literature
Need to integrate various lines of evidence: balance data, evidence from intervention studies, evidence from observational studies : how to make them talk to each other?
Opportunities:
Apply the principles of EFSA Prometheus approach
Thorough analysis of uncertainties
Integration of the evidence through an EKE
Engagement with stakeholders
Open a discussion about the DRVs framework and its limitations (terminology, principles), in particular when it comes to integrate epidemiological evidence in setting reference values
There is a need for studies, using robust assessment methods for sodium intake and the outcome of interest, investigating:
Challenges:
Sodium is a “special” nutrient in that adequate and safe intakes lie in a narrow range
Polarized literature
Need to integrate various lines of evidence: balance data, evidence from intervention studies, evidence from observational studies : how to make them talk to each other?
Opportunities:
Apply the principles of EFSA Prometheus approach
Thorough analysis of uncertainties
Integration of the evidence through an EKE
Engagement with stakeholders
Open a discussion about the DRVs framework and its limitations (terminology, principles), in particular when it comes to integrate epidemiological evidence in setting reference values
There is a need for studies, using robust assessment methods for sodium intake and the outcome of interest, investigating: