GiveBetter x USA

Center for Science in the Public Interest

Nutrition and food-safety research, consumer education, and policy advocacy

Research time: 30 min on GPT-6 Astra Light + GPT-6 Astra Medium
  • Research — reviewed programs, finances and impact evidence.

Updated: 2026-09-14

Donate

Summary

What do they do? CSPI researches food and health risks, informs consumers, and advocates for stronger public protections. An ordinary charitable gift supports its full portfolio, including scientific analysis, education, litigation and community partnerships—not a sodium-only purchase. Our sodium-pathway best guess remains about $480,000 per 10 USA QALYs, but it is a judgmental partial-health estimate rather than a measured whole-portfolio return.

Why we’re interested in this organization:

  • The sodium-to-blood-pressure mechanism is supported by randomized evidence, with a US disease model providing a finite health scale.

  • Specific technical work, a 2023 petition and the still-draft Phase II process identify a plausible prospective contribution beyond generic advocacy counts.

  • Three original financial years and separate charity–affiliate accounts support a consistent full-recipient cost denominator.

Our main reservations:

  • The leading uncertainty is whether CSPI causes additional sustained reform or merely brings forward reform that would happen anyway; a timing-only case is about $5.7 million per 10 QALYs.

  • No current unfunded sodium work package was identified; substantial marketable investments and possible replacement funding make immediate donation additionality uncertain.

  • The health scale is transported from an older simulation, while other portfolio effects, food substitutions and additional public or industry costs remain uncertain.

What do you get for your dollar? $484K per better life: ten additional quality-adjusted life years in USA. Sodium partial-health estimate; full charitable-recipient cost; judgmental policy contribution.

Prospective policy response -> otherwise-unavailable sustained sodium reformulation -> consumed sodium -> blood pressure -> finite cardiovascular health. Success-case consumed change=250*.5=125mg/day; probability-weighted full-scale national change=125*.2*.1*.5=1.25mg/day before clinical transport. These are model expectations, not monitored outputs. No comments-to-QALYs shortcut.

1. What do they do?

CSPI, EIN 23-7122879, is the charitable recipient behind the official giving page. Its retrieved page names Center for Science in the Public Interest and links to the same EIN; the official homepage links directly to it. No gift, form submission or outreach was made. The 501(c)(4) CSPI Action Fund is a separate organization. Shared personnel and cost reimbursements do not allow a charitable gift to take full credit for affiliate campaigns. Original return

The 2025–2030 strategy broadens the organization beyond nutrition and food safety into other health policy. The current impact report describes food labeling, school meals, food banking, chemicals, dietary guidance and science communication. The model charges all reported charitable-recipient expenses and quantifies only prospective sodium-related health; unrecognized volunteer and other external resource costs remain unresolved. Other work is assigned zero net health in the central calculation as an explicit simplifying judgment, not because it is known to have no value or harm.

The specific forward opportunity is better sodium guidance, transparent monitoring and industry accountability. CSPI's 2023 petition requested long-term targets, intermediate targets, compliance evaluation and a public product database. Its January 2025 comment sought stronger Phase II targets and more effective implementation. These actions establish relevant work; they do not establish the probability change caused by a new donation. Historical trans-fat and sodium-policy outcomes receive no retrospective credit.

2. Monitoring and information sharing

As reviewed in September 2026, FDA still describes Phase II as draft: voluntary three-year targets across 163 food categories, aiming at approximately 2,750 mg/day if achieved. A promised 2026 Phase I evaluation is an agency deliverable, not a measured outcome of CSPI support. We found no final Phase II implementation date or released complete Phase I evaluation in the targeted official-source check. This is a dated search conclusion, not proof that no further agency work exists.

The available FDA comparison covers 2010–2022, not a clean before–after test of the targets finalized in 2021. CSPI's reanalysis found small median changes and worse restaurant results, but its cross-category summaries are not weighted by categories' contributions to total sodium intake. Neither the percentage of categories improving nor the median percentage change proves a particular decline in consumed sodium. The FDA agenda's statement of an approximately 10% intake decline should not be read as a causal post-2021 estimate established by this comparison.

For marginal impact, useful measurements would be a dated sodium work plan, staffing or analysis that new money changes, whether comparable work is already funded, a policy response traceable to that work, and sales-weighted consumed sodium sustained after reformulation. Population blood pressure and cardiovascular outcomes would test downstream benefits. Comments submitted, media coverage, membership and coalition victories are intermediate activities, not QALYs.

3. Qualitative assessment

The randomized-trial meta-analysis includes 133 studies and supports sodium reduction lowering blood pressure. The average sodium reduction in those studies was much larger than this model's 125 mg/day realized opportunity, so the pooled blood-pressure drop is not applied directly to food reformulation. Potassium-enriched salt trials support a related mechanism, but added potassium and their populations prevent treating those effects as a US packaged-food trial.

The US IMPACT simulation follows adults aged 30–84, with competing mortality and finite health utility. Its discounted 20-year sodium scenarios provide a health scale, not charity attribution. The 2.1 million minus 0.7 million QALY difference is interpolated using the 565 mg/day endpoint difference. Endpoint interpolation is approximate because the original exposure ramps differ. The 0.7 transport factor is a judgment about this and future population, treatment and dietary changes. The model does not add life-years to QALYs or extrapolate benefits indefinitely; its original clinical lag is not imposed a second time.

Organizational influence is less secure. Counsel’s account of the 2015 lawsuit confirms that FDA responded after a delay challenge, but denied the requested mandatory action while releasing voluntary targets. The 2023 petition and subsequent intermediate-target proposal show alignment with an identifiable request, while government commitments and other advocates provide strong counterfactual explanations. We retain a two-percentage-point effective charitable contribution over five years as a plausible but very uncertain judgment: a 20-point coalition increment times a 10% charitable-CSPI share. This is not a measured frequency or a claim that CSPI deserves 10% of all FDA outcomes.

The most consequential alternative is acceleration rather than durable additional reform. If the same finite health stream would occur three years later without the credited effort, the present-value gain is only about 8.5% of the durable-change calculation under a simple shift approximation. That sensitivity is approximately $5.7 million per 10 QALYs; a half-durable, half-three-year-acceleration mixture is about $890,000. Neither split is empirically identified. These comparisons make the counterfactual concrete without forcing the best guess to zero simply because attribution is uncertain.

Potential harms include nutritionally poor replacements, food costs or access effects, microbiological tradeoffs and adverse health-policy effects elsewhere. Unrestricted support may also create substantial omitted benefits. Zero and net-negative cases are retained. A positive sodium figure is not a proven lower bound on the full portfolio.

4. What do you get for your dollar?

Our retained best guess is $484,088 per 10 USA QALYs, appropriately communicated as about $480,000. It charges five years of the full charitable recipient's FY2025 expense, $89.63 million, against about 1,852 expected USA QALYs. A $10,000 gift-equivalent corresponds to about 0.207 expected QALY under the proportional funding model. This is an expected value across uncertain policy outcomes, not a promise of a detectable individual result or a restricted project quote.

The native sequence is an assumed 250 mg/day opportunity beyond the Phase I counterfactual, 50% implementation giving 125 mg/day in the success case, then a 2% effective charitable policy contribution and 50% funding response. At the full five-year cost scale that corresponds to a probability-weighted 1.25 mg/day national average improvement before clinical transport. This is an expectation across scenarios, not a measured intake change or uniform benefit to every person. The $10,000 normalization credits about 0.000112 percentage points of the defined reform opportunity; exact proportionality for tiny gifts is a modeling convention.

FY2023, FY2024 and FY2025 recipient expenses are $18,005,982, $17,289,504 and $17,926,322, averaging $17,740,603. Original returns show no event, rental, gaming or inventory expense addback. The latest tax return and consolidating audit agree on recipient identity but differ by $9,781 in expenses: tax $17,926,322 versus charity-only audit $17,916,541. We retain the higher tax figure and disclose the unresolved 0.055% difference. Neither the $19.52 million consolidated affiliate expense nor a sodium-only budget substitutes for recipient cost. Blank Schedule D reconciliation does not establish that all unrecognized volunteer resources are zero.

The $9,781 is a paired difference: after removing unrealized investment gains, tax revenue is also $9,781 above audit revenue. Both statements therefore imply the same $2,805,236 decrease in net assets. The accounting classification is unresolved, but this is not an unexplained loss of net assets.

The audit allocates charity-only expense to public education ($3.188 million), Nutrition Action ($3.582 million), special projects ($8.361 million), membership development ($0.255 million), management ($0.314 million) and fundraising ($2.216 million). Joint education/fundraising costs are already included. The FY26 numbers in the impact report are not substituted for a completed original fiscal year: they are unaudited, their reporting basis is unclear, and the document contains editorial remnants. Using its $20.124 million expense number only as a scope-uncertain forward-cost sensitivity raises the retained price to about $540,000.

The full-recipient donor denominator includes staff, management, fundraising, occupancy and reported grants. Government monitoring, food-industry reformulation and participant resources outside the charity are additional social costs; historical modeled savings are not donor cash returns. A current social-cost estimate needs new implementation data. Favorable and poor scenarios remain about $15,000 and $1.6 billion per 10 QALYs, not confidence bounds. Zero or negative health has no positive cost-per-QALY price.

Model, assumptions and sensitivity

Whole ordinary charitable-recipient gift. Five years of full FY2025 recipient expense at constant real scale; sodium is the only quantified health pathway. No restricted-project price, no affiliate double count, no historical victory credit. Additional external social costs remain unquantified.

USA 50 states plus DC. g=.99 remains a judgmental national-coverage allowance, not a measured territorial allocation. The underlying clinical model covers ages30–84, so the discounted scale is not all-age measured health; no population multiplier fills missing age groups.

H=(2100000-700000)*(dm/565)*e*t/(1.03^L); S=fDurable+(1-fDurable)*(1-1.03^(-accelerationYears)); Qall(G)=G/(C*Y)*b*(p*a*H*S-h); Qusa=Qall*g; price10=10*G/Qusa only when Qusa>0. Central C=17926322,Y=5,dm=250,e=.5,t=.7,L=5,p=.2,a=.1,b=.5,g=.99,h=0,fDurable=1. Existing alpha values reproduced; additional timing scenarios vary S.

C
17926322 USD per recipient fiscal year (observed). FY2025 Form 990 Part I/IX, all functional expenses; Part VIII direct-cost lines empty. [cspi-99025]
Y
5 years of full recipient costs (judgment). Sustained advocacy horizon; not one annual budget assigned a full national reform.
Qopt
2100000 discounted QALYs over original 20-year simulation (observed). Published simulated optimal scenario, not measured outcomes. [sodium-us-model]
Qshort
700000 discounted QALYs over original 20-year simulation (observed). Published short-term-only scenario. [sodium-us-model]
Sshort-minus-Sopt
565 mg sodium/day (observed). Reported 2036 medians 2789 and 2224; difference is an interpolation reference, not current intake. [sodium-us-model]
deltaS
250 mg sodium/day beyond assumed counterfactual (judgment). Phase I target3000 minus Phase II2750; achievable further reduction is uncertain and the current baseline is not assumed empirically known. [fda-sodium] [fda-progress]
e
0.5 fraction of incremental sodium opportunity implemented (judgment). Discounts voluntary noncompliance, purchasing substitutions, and incomplete persistence; not fraction of food categories with any decrease. [cspi-sodium-comment] [fda-progress]
t
0.7 clinical and future-horizon transport factor (judgment). Old population, treatment trends, nonlinearity, different reformulation ramp, and rebasing to future benefits justify a broad discount. [sodium-us-model] [sodium-bp] [ssass]
L
5 additional years of benefit delay (judgment). Five-year delay before the rebased already-discounted health stream; no claim of a final Phase II date. Original modeled clinical lag is not reapplied. [fda-priorities26]
p
0.2 incremental campaign probability of achieving modeled opportunity (judgment). Subjective 20-percentage-point coalition contribution over five years relative to government/industry background action; broad 0–.5 scenarios. [cspi-sodium-comment] [fda-sodium]
a
0.1 charitable CSPI share of coalition increment (judgment). A tenth is a cautious opinion given sustained technical/legal work and many other actors; not a causal audit. Tested .01–.2. [cspi-sodium-comment] [cspi-99025] [cspi-litigation-history]
b
0.5 additional effort enabled per dollar (judgment). Allows half to replace reserves, donors, or otherwise-funded work over five years; reserves and loss are observed, this response is not. [cspi-audit25]
h
0 net other-pathway QALYs lost per five-year effort (judgment). Central assigns zero net non-sodium benefit or harm. Negative-outcome stress tests retain loss even if policy influence is zero.
g
0.99 USA share of credited sodium health (judgment). Near-national US pathway with small geographic/coverage allowance; not an audited outcome-weighted share.
marginal-resource-cost
null incremental public and industry USD (unknown). Required for a current social-cost model; historical external implementation/savings estimates are not added as donor savings. [sodium-us-model]
fDurable
1 fraction of credited influence producing additional finite reform rather than timing only (judgment). Central preserves alpha durable-change interpretation; timing-only and50:50 cases reveal dependence on that judgment. [fda-sodium] [cspi-petition23]
accelerationYears
3 years earlier identical finite health stream in timing sensitivity (judgment). Simple present-value shift approximation, not a rerun of population survival. Both streams have finite20-year duration; later counterfactual ends3years later.

Sodium partial-health best guess; durable-change interpretation: Cost: $10K; USA QALYs: 0.2065739312284309; all-population QALYs: 0.20866053659437464. Judgment scenario, not a confidence bound or restricted purchase. Inputs {"C":17926322,"dm":250,"e":0.5,"t":0.7,"L":5,"p":0.2,"a":0.1,"b":0.5,"h":0,"g":0.99,"f":1,"accel":3}; H=187025.79838417712; timing multiplier=1. G=$10,000 and Y=5. Price=484088.1877269368. h is not multiplied by policy-success probability.

Favorable joint judgments: Cost: $10K; USA QALYs: 6.500992669904147; all-population QALYs: 6.566659262529442. Judgment scenario, not a confidence bound or restricted purchase. Inputs {"C":17926322,"dm":350,"e":0.9,"t":1,"L":2,"p":0.5,"a":0.2,"b":0.8,"h":0,"g":0.99,"f":1,"accel":3}; H=735725.3025274082; timing multiplier=1. G=$10,000 and Y=5. Price=15382.266228808781. h is not multiplied by policy-success probability.

Weak policy influence and implementation: Cost: $10K; USA QALYs: 0.00006109456721225645; all-population QALYs: 0.00006171168405278428. Judgment scenario, not a confidence bound or restricted purchase. Inputs {"C":17926322,"dm":100,"e":0.1,"t":0.3,"L":10,"p":0.05,"a":0.02,"b":0.1,"h":0,"g":0.99,"f":1,"accel":3}; H=5531.31759746238; timing multiplier=1. G=$10,000 and Y=5. Price=1636806749.9779024. h is not multiplied by policy-success probability.

One-percent charitable share of coalition increment: Cost: $10K; USA QALYs: 0.020657393122843086; all-population QALYs: 0.020866053659437462. Judgment scenario, not a confidence bound or restricted purchase. Inputs {"C":17926322,"dm":250,"e":0.5,"t":0.7,"L":5,"p":0.2,"a":0.01,"b":0.5,"h":0,"g":0.99,"f":1,"accel":3}; H=187025.79838417712; timing multiplier=1. G=$10,000 and Y=5. Price=4840881.87726937. h is not multiplied by policy-success probability.

Only10% additional effort: Cost: $10K; USA QALYs: 0.04131478624568618; all-population QALYs: 0.04173210731887493. Judgment scenario, not a confidence bound or restricted purchase. Inputs {"C":17926322,"dm":250,"e":0.5,"t":0.7,"L":5,"p":0.2,"a":0.1,"b":0.1,"h":0,"g":0.99,"f":1,"accel":3}; H=187025.79838417712; timing multiplier=1. G=$10,000 and Y=5. Price=2420440.9386346843. h is not multiplied by policy-success probability.

Three-year acceleration, no durable extra reform: Cost: $10K; USA QALYs: 0.017529521024939196; all-population QALYs: 0.017706586893877975. Judgment scenario, not a confidence bound or restricted purchase. Inputs {"C":17926322,"dm":250,"e":0.5,"t":0.7,"L":5,"p":0.2,"a":0.1,"b":0.5,"h":0,"g":0.99,"f":0,"accel":3}; H=187025.79838417712; timing multiplier=0.0848583406468405. G=$10,000 and Y=5. Price=5704662.429608334. h is not multiplied by policy-success probability.

Half durable change, half three-year acceleration: Cost: $10K; USA QALYs: 0.11205172612668506; all-population QALYs: 0.11318356174412632. Judgment scenario, not a confidence bound or restricted purchase. Inputs {"C":17926322,"dm":250,"e":0.5,"t":0.7,"L":5,"p":0.2,"a":0.1,"b":0.5,"h":0,"g":0.99,"f":0.5,"accel":3}; H=187025.79838417712; timing multiplier=0.5424291703234203. G=$10,000 and Y=5. Price=892444.975697568. h is not multiplied by policy-success probability.

Unaudited scope-uncertain FY26 expense sensitivity: Cost: $10K; USA QALYs: 0.18401294879736788; all-population QALYs: 0.18587166545188674. Judgment scenario, not a confidence bound or restricted purchase. Inputs {"C":20124186,"dm":250,"e":0.5,"t":0.7,"L":5,"p":0.2,"a":0.1,"b":0.5,"h":0,"g":0.99,"f":1,"accel":3}; H=187025.79838417712; timing multiplier=1. G=$10,000 and Y=5. Price=543440.0168768471. h is not multiplied by policy-success probability.

No incremental policy influence: Cost: $10K; USA QALYs: 0; all-population QALYs: 0. Judgment scenario, not a confidence bound or restricted purchase. Inputs {"C":17926322,"dm":250,"e":0.5,"t":0.7,"L":5,"p":0,"a":0.1,"b":0.5,"h":0,"g":0.99,"f":1,"accel":3}; H=187025.79838417712; timing multiplier=1. G=$10,000 and Y=5. Price=not defined for nonpositive health. h is not multiplied by policy-success probability.

No sodium gain plus adverse portfolio effect: Cost: $10K; USA QALYs: -0.005522605250536055; all-population QALYs: -0.005578389141955612. Judgment scenario, not a confidence bound or restricted purchase. Inputs {"C":17926322,"dm":250,"e":0.5,"t":0.7,"L":5,"p":0,"a":0.1,"b":0.5,"h":100,"g":0.99,"f":1,"accel":3}; H=187025.79838417712; timing multiplier=1. G=$10,000 and Y=5. Price=not defined for nonpositive health. h is not multiplied by policy-success probability.

Central sodium gain offset by1000QALY other-portfolio loss: Cost: $10K; USA QALYs: 0.15134787872307034; all-population QALYs: 0.15287664517481853. Judgment scenario, not a confidence bound or restricted purchase. Inputs {"C":17926322,"dm":250,"e":0.5,"t":0.7,"L":5,"p":0.2,"a":0.1,"b":0.5,"h":1000,"g":0.99,"f":1,"accel":3}; H=187025.79838417712; timing multiplier=1. G=$10,000 and Y=5. Price=660729.4455905495. h is not multiplied by policy-success probability.

Counterfactual: Phase I-scale progress, government activity, private reformulation, and other advocates continue without this funding. Only a further 250 mg/day opportunity is considered; p is an incremental probability change attributable to the broader campaign rather than the unconditional chance FDA acts. b removes expenditure that would occur from reserves or replacement funding anyway.

Attribution: p=.2 is a subjective campaign-induced probability increment over five years; a=.1 allocates a tenth of that increment to charitable CSPI, excluding credit owed to its Action Fund, other advocates, researchers, officials, and manufacturers. The resulting p*a=.02 is not measured. Because all portfolio costs are charged, no additional sodium budget fraction is multiplied in; doing both would double-charge portfolio allocation.

Retained judgmental sodium partial-health best guess, not a measured portfolio return. Durable change, policy acceleration, clinical transport and funding replacement are separate uncertainties. Central other-portfolio net health=0 is a judgment, not a proven lower bound. Missing observed marginal inputs alone do not force null.

Sensitivity

  • Central484088.19; timing-only5704662.43; half-durable mixture892444.98 dollars per10USAQALYs. The timing contrast assumes identical finite health streams shifted3years; it is not a new disease simulation.
  • Five full annual recipient budgets require896.3161netUSAQALYs to meet1million dollars per10, or8963.161 to meet100000. Central netUSAhealth is1851.5554.
  • Central effective p*a=.02. At b=.5 and g=.99, about .00968 durable effective attribution is required for1million dollars per10. If all influence is only3year acceleration, the requirement is about11.4percentage points instead.
  • A net other-portfolio loss of3740.516QALYs per full5year effort would erase central sodium gain before funding/geographic scaling. Zero-other-health is not a lower-bound claim.
  • Changing costs to the unverified FY26 impact-report scope increases price by about12.3%; it does not establish a current funding gap. The retained original FY25 tax/audit cost difference is only0.055%.
  • The original clinical-model uncertainty does not contain CSPI attribution, donation additionality, policy timing or whole-portfolio effects. Do not display it as a charity confidence interval.

Unresolved inputs

  • A donor-contingent sodium work package with budget, timing, reserve policy and replacement-funder counterfactual.
  • Whether the work changes eventual strength/compliance or only advances otherwise-identical reform, and by how long.
  • Sales-weighted consumed sodium and a current clinical model with explicit implementation ramps and survival.
  • Charity-versus-Action-Fund allocation of prospective work, external implementation costs and non-overlapping other-portfolio health.
  • Explanation of the9781dollar tax/audit expense difference and unrecognized in-kind resources; this small residual does not dominate the health model.

5. Funding and previous grants

The FY2025 accounts show $31.413 million in charitable-recipient net assets, including $27.924 million without donor restrictions. That is not all cash: the charity held $1.871 million cash, $4.611 million current investments and $28.627 million long-term investments. Consolidated operating cash use was $6.671 million; it must not be mislabeled charity-only cash burn. The audit's consolidated one-year-available financial-assets measure was $2.587 million after restrictions and a $1 million board reserve, but it excludes long-term investments that the audit says can be converted to cash. Thus neither a funding cliff nor a conclusion of no funding need follows mechanically.

Restricted purposes and a $1.392 million donor-restricted endowment constrain some funds. Board reserves can be released by resolution; marketable long-term holdings weaken a claim that donations automatically determine immediate staffing. The Action Fund had negative ending net assets of about $119,000 and substantial cost-sharing with CSPI; this does not turn an unrestricted charitable gift into an Action Fund gift or justify consolidated attribution.

We retain 50% additional effort over a five-year horizon as an explicit judgment after reserve and donor replacement. The annual loss, current work and fundraising support positive future capacity, while marketable assets and uncertain program allocation justify very wide alternatives. The public record did not identify an uncovered sodium-specific staff position, analysis contract, monitoring project or litigation budget that this gift would fund. That is the next donor-readiness bottleneck, not a reason to erase every judgment-based estimate.

CSPI remains an eligible research candidate, but this review alone does not support a top-pick endorsement. A donor who values uncertain policy opportunities may find the partial-health estimate useful. A stronger recommendation would require a credible near-term work package, what changes at the margin, its actual charity–affiliate allocation, and whether it changes final policy strength or only timing. No contact was made to resolve those questions.

Annual expenses

Organization-level spending, including programs, administration and fundraising. The research list averages three comparable, consecutive full fiscal years when available.

  • FY 2023: $18.0M; Center for Science in the Public Interest, EIN23-7122879, 12-month period, Recipient Form990 accrual functional expenses plus any direct operating costs netted against revenue. Source
  • FY 2024: $17.3M; Center for Science in the Public Interest, EIN23-7122879, 12-month period, Recipient Form990 accrual functional expenses plus any direct operating costs netted against revenue. Source
  • FY 2025: $17.9M; Center for Science in the Public Interest, EIN23-7122879, 12-month period, Recipient Form990 accrual functional expenses plus any direct operating costs netted against revenue. Source

6. Sources

  1. FY2025 consolidated audit and recipient-level consolidating statements. CSPI / Calibre CPA Group. Published: not stated; retrieved: 2026-09-14.
  2. Form 990, fiscal year ended June 30, 2025. CSPI / Internal Revenue Service. Published: not stated; retrieved: 2026-09-14.
  3. Form 990, fiscal year ended June 30, 2024. CSPI / Internal Revenue Service via ProPublica. Published: not stated; retrieved: 2026-09-14.
  4. Form 990, fiscal year ended June 30, 2023. CSPI / Internal Revenue Service via ProPublica. Published: not stated; retrieved: 2026-09-14.
  5. 2026 Impact Report. CSPI. Published: not stated; retrieved: 2026-09-14.
  6. Comments on FDA Phase II voluntary sodium targets. CSPI. Published: 2025-01-13; retrieved: 2026-09-14.
  7. Sodium reduction in the food supply: Phase I and draft Phase II. FDA. Published: not stated; retrieved: 2026-09-14.
  8. 2010–2022 preliminary sodium assessment. FDA. Published: not stated; retrieved: 2026-09-14.
  9. Human Foods Program 2026 priority deliverables. FDA. Published: not stated; retrieved: 2026-09-14.
  10. US FDA voluntary sodium reformulation microsimulation. Pearson-Stuttard, Kypridemos et al. / PLOS Medicine. Published: 2018-04-10; retrieved: 2026-09-14.
  11. Effect of dose and duration of sodium reduction on blood pressure: randomized-trial meta-analysis. Huang et al. / BMJ. Published: 2020-02-25; retrieved: 2026-09-14.
  12. Effect of salt substitution on cardiovascular events and death. Neal et al. / New England Journal of Medicine. Published: 2021-08-29; retrieved: 2026-09-14.
  13. CSPI v. FDA: sodium-petition litigation. Public Citizen, counsel in the case. Published: not stated; retrieved: 2026-09-14.
  14. Official homepage donation link. CSPI. Published: not stated; retrieved: 2026-09-14.
  15. Citizen petition requesting long-term sodium targets, monitoring and public database. CSPI. Published: 2023-04-06; retrieved: 2026-09-14.
  16. 2025–2030 strategic plan. CSPI. Published: not stated; retrieved: 2026-09-14.
  17. Current original-accountability document index. CSPI. Published: not stated; retrieved: 2026-09-14.
  18. Randomized dietary sodium/BP meta-analysis, original PDF. Huang and colleagues / BMJ. Published: 2020-02-25; retrieved: 2026-09-14.
  19. Appraising the FDA’s Sodium Reduction Efforts. Musicus and colleagues / JAMA. Published: 2025-03-24; retrieved: 2026-09-14.