Oral single-ingredient nicotinic acid and gastrointestinal adverse events — formulation-specific risks and evidence limits

Useful facts when choosing a product

• The planned population is people exposed to oral single-ingredient nicotinic acid. Actual studies enrolled healthy adults, patients with dyslipidemia/diabetes/vascular disease, or hemodialysis patients. The planned population was not copied into verified recruitment facts. A single page-wide dose, duration and comparator is **not applicable**; study-specific values are in `study_contexts.json`. Nicotinic acid is not treated as interchangeable with nicotinamide/niacinamide, NR, NMN or esters. Topical and intravenous results were excluded. A single-ingredient oral product may still be used with other active drugs, so ingredient count and background therapy are separate. **IR, SR, ER and unknown release**, salt information, fasting/food, titration and actual exposure duration remain separated. Direct comparative GI values for dietary nutritional intake or confirmed deficiency treatment were not verified in scope. Recruitment as healthy or dyslipidemic does not establish biochemical niacin repletion. Sub-outcomes include **abdominal pain/cramps, dyspepsia, nausea, vomiting, diarrhea, serious ulcer/bleeding events, GI-related stopping, any-AE stopping and all-cause stopping**. Serious GI events differ from mild symptoms. Overlapping symptoms, recurrent events and stopping reasons are not added into an invented person-level any-GI rate. “Pass stool” is not renamed diarrhea without a definition.
• Baseline biochemical niacin-deficiency criteria/concentrations, total dietary intake and quantities of co-vitamins are commonly unreported in the adopted GI data. Healthy recruitment does not establish a passed deficiency test. Individual GI absolute risks in pregnancy, lactation, children, active ulcer/bleeding and severe hepatic/renal disease were not calculated from these studies. Small renal-population data and the prescribing label's renal-evidence limitations were not expanded into a claim that no human research exists. [S01, S02, S06, S07, S09] Study doses, titration and rescue medication are **not instructions to start, stop or modify individual treatment**. No equal-milligram release-form substitution or background-drug change is recommended. Earlier liver, urate, skin and cardiovascular pages remain unchanged linked material, not reassessed endpoints.
• Eligible direct GI rates for dietary nutritional exposure, confirmed deficiency treatment, pregnancy/lactation/children, baseline ulcer/bleeding and severe hepatic/renal disease remain unconfirmed in scope; not filled from other molecules/routes/combinations.

What the research actually shows

# Oral single-ingredient nicotinic acid and gastrointestinal adverse events — formulation-specific risks and evidence limits

TASK-1018 / R01-058 · NUT / R01 · GIHARM · Chat item 5/5

Evidence cutoff **2026-09-16**; supplied snapshot **2026-09-16T21:50:27+09:00**. This is completed bilingual clinical content, not a certificate of site-ID assignment or deployment.

## Thirty-second answer

**Safety: Warning.** Oral nicotinic acid can cause gastrointestinal symptoms such as nausea and vomiting and can lead to stopping during pharmacologic use; the safety label is Warning. Formulation, fasting/titration, population, comparator and event definitions differ, so no single GI incidence is established for all users. Serious GI excess with the laropiprant combination is not converted into a single-nicotinic-acid ulcer or bleeding rate. [S01, S02, S04, S05]

In a single-dose fasting IR 500 mg trial, nausea was **10/33 versus 1/35** and vomiting **4/33 versus 0/35**. These are observations in a small acute trial, not long-term user rates. The ER label separates GI symptoms from stopping and lists **active peptic ulcer and arterial bleeding as contraindications**. A contraindication is not itself an incidence estimate for new ulcers/bleeding or proof of individual causality in every patient. [S01, S02]

**Efficacy grade, score, axes and anchor: not applicable (null).** Human studies exist, hence `no_human_study=false`. Document quality **A** is the same author's scoped self-assessment, not independent expert verification, journal peer review, certification or a guarantee of no errors.

## 1. Planned question versus actual evidence

The planned population is people exposed to oral single-ingredient nicotinic acid. Actual studies enrolled healthy adults, patients with dyslipidemia/diabetes/vascular disease, or hemodialysis patients. The planned population was not copied into verified recruitment facts. A single page-wide dose, duration and comparator is **not applicable**; study-specific values are in `study_contexts.json`.

Nicotinic acid is not treated as interchangeable with nicotinamide/niacinamide, NR, NMN or esters. Topical and intravenous results were excluded. A single-ingredient oral product may still be used with other active drugs, so ingredient count and background therapy are separate. **IR, SR, ER and unknown release**, salt information, fasting/food, titration and actual exposure duration remain separated. Direct comparative GI values for dietary nutritional intake or confirmed deficiency treatment were not verified in scope. Recruitment as healthy or dyslipidemic does not establish biochemical niacin repletion.

Sub-outcomes include **abdominal pain/cramps, dyspepsia, nausea, vomiting, diarrhea, serious ulcer/bleeding events, GI-related stopping, any-AE stopping and all-cause stopping**. Serious GI events differ from mild symptoms. Overlapping symptoms, recurrent events and stopping reasons are not added into an invented person-level any-GI rate. “Pass stool” is not renamed diarrhea without a definition.

## 2. Direct single-ingredient IR and IR–SR comparisons

### 2.1 Mills 2003: one fasting IR 500 mg dose versus placebo

Of 68 healthy student volunteers, 33 received IR and 35 inert placebo. People with prior GI/liver disease and niacin intolerance were excluded. **Flushing** was primary; GI symptoms were additional self-reports. The n/N below were visually checked in the original PDF tables. Fixed GI observation hours, standardized symptom definitions and the authors' GI-specific statistical test/pair covariance were not reported. [S02, S12]

| Outcome / stopping distinction | Active | Comparator | Precision / difference | Source | |---|---|---|---|---| | Nausea | 10/33 (30%) | 1/35 (3%) | Recalculated crude difference +27.45 %p; original CI unreported | [S02] · M01 | | Vomiting | 4/33 (12%) | 0/35 (0%) | Recalculated crude difference +12.12 %p; original CI unreported | [S02] · M02 | | Stomach cramps | 2/33 (6%) | 0/35 (0%) | Recalculated crude difference +6.06 %p; original CI unreported | [S02] · M03 | | Pass stool, not relabelled as diarrhea | 3/33 (9%) | 0/35 (0%) | Recalculated crude difference +9.09 %p; original CI unreported | [S02] · M04 | | Withdrawal/medical attention for severe reactions with GI symptoms described | 3/33 (9%) | Unconfirmed | No comparative effect/CI generated | [S02] · M05 | | Intolerable rating, all causes | 6/33 (18.2%) | 0/35 (0%) | Recalculated crude difference +18.18 %p; original CI unreported | [S02] · M06 |

Nausea at 30% versus 3% was not converted into an **invented any-GI composite**. Among three withdrawals requiring medical attention, two had vomiting and one had severe cramps/nausea. Adding these to all four vomiting cases double-counts people. Six intolerable ratings are not six GI discontinuations. The authors' serious-event definition involved medication for relief and was not automatically relabelled hospitalization or life-threatening SAE. [S02]

**Statistical check:** the table reports p=.005 for both nausea and vomiting. Independent-arm marginal-table two-sided Fisher diagnostics were **0.00251158** for nausea and **0.05024651** for vomiting. For vomiting, uncorrected Pearson **0.03374504** and Yates-corrected **0.10794606** also did not reproduce .005. These diagnostics do not reconstruct the original pairwise analysis and do not silently replace the reported p. Strong vomiting significance is not the sole basis of the conclusion. Original CIs were unreported; no new CI or NNH was generated. [S02; calculations.json]

Limitations include the small acute fasting design, self-reported GI symptoms, possible functional unblinding from flushing, unblinded analysts, multiplicity and unknown symptom overlap. Possible unblinding is an editorial interpretation, not a measured failure rate. The result is not directly transferred to long-term, fed or gradually titrated exposure.

### 2.2 McKenney 1994: SR versus IR

Groups of 23 received SR or IR, escalating 500→1000→1500→2000→3000 mg/day, six weeks per step. **All-cause withdrawal 18/23 versus 9/23** included GI, fatigue, laboratory abnormalities and other reasons. The accessible abstract mentions GI reasons in the SR arm, but **GI-specific n/N and CI remain unavailable without the full paper**. No ranking that SR is worse than ER for GI harm or that IR is safe was created. The prior 3131 liver assessment was not changed. [S06]

## 3. Official ER label: symptoms, stopping and spontaneous reports separated

The pooled database includes 245 unique ER participants and 157 placebo participants, median treatment 16 weeks. Dose-column denominators below are **non-additive and may overlap**; reactions are assigned to their initial dose of occurrence. Therefore 87+110+136+95 is not a new independent population size. Missing dose-specific durations, titration/co-medication details and exact event counts prevent treating these as independent randomized dose arms or a precise dose-response curve. [S01 §6.1/Table 2]

| Symptom (%) | Placebo N=157 | ER500 N=87 | ER1000 N=110 | ER1500 N=136 | ER2000 N=95 | |---|---:|---:|---:|---:|---:| | Diarrhea | 13 | 7 | 10 | 10 | 14 | | Nausea | 7 | 5 | 6 | 4 | 11 | | Vomiting | 4 | 0 | 2 | 4 | 9 |

These are **treatment-emergent percentages regardless of causal assessment** in the label. Some ER columns are below or equal to placebo; the pattern is not uniformly monotonic. Exact n was not reverse-engineered from percentages and a reported 0% was not declared universal zero risk. Independent quantitative abdominal-pain/dyspepsia values were not verified in this table.

| Outcome / stopping distinction | Active | Comparator | Precision / difference | Source | |---|---|---|---|---| | Stopping due to diarrhea | 2% (n unreported; N=245) | 0% (n unreported; N=157) | No comparative effect/CI generated | [S01] · LS-D | | Stopping due to nausea | 1% (n unreported; N=245) | 0% (n unreported; N=157) | No comparative effect/CI generated | [S01] · LS-N | | Stopping due to vomiting | 1% (n unreported; N=245) | 0% (n unreported; N=157) | No comparative effect/CI generated | [S01] · LS-V | | Stopping due to any adverse reaction | 16% (n unreported; N=245) | 4% (n unreported; N=157) | No comparative effect/CI generated | [S01] · LS-ALL |

The GI-reason figures of 2%, 1% and 1% are not added into a 4% any-GI stopping rate. **16% versus 4% is stopping for any adverse reaction**, not any-GI incidence. Original CIs and person overlap across GI reasons are unreported. [S01]

Postmarketing peptic ulcers, eructation and flatulence are **spontaneous-report signals**. Unknown exposed population, duration, frequency and individual causal attribution prevent mixing them with RCT incidence. Active-peptic-ulcer/arterial-bleeding contraindications and hepatic/renal cautions remain explicit. This is not advice to substitute equal milligrams across ER, SR and IR. [S01 §§2.2,4,6.2,8.6/8.7]

## 4. Add-on and combination evidence: attribution limited

### 4.1 AIM-HIGH: single-ingredient ER added to statin-based therapy

Prior originals supplied NIASPAN ER 1500–2000 mg/day, IR nicotinic acid 100–150 mg/day in control tablets, and LDL-targeted simvastatin with optional ezetimibe in both arms. Control is not pure non-exposure; a common treatment policy does not mean identical actual statin doses. Active tolerability run-in and selection further constrain application. [S01, S03; prior 3082/3130/3131]

The newly relevant GI primary table was inaccessible with the current tools. Indexed candidate numbers were not presented as fully verified original-table values: the **adopted new GI effect remains null**. This is not `no_human_study=true` or “no GI effect.” Obtaining the original table is a revision trigger, not clinical research delegated to Codex.

### 4.2 HPS2-THRIVE: ER nicotinic acid plus laropiprant

The intervention is the **combination of ER nicotinic acid 2 g plus laropiprant 40 mg/day** versus matched placebo, with simvastatin 40 mg/day and optional ezetimibe 10 mg/day in both arms. There were 12,838 versus 12,835 participants and median follow-up 3.9 years. Whole-cohort aspirin use of 86.3% is not verification of each randomized arm's rate. This design does not isolate the contribution of nicotinic acid alone. [S04, S05]

| Outcome / stopping distinction | Active | Comparator | Precision / difference | Source | |---|---|---|---|---| | Serious gastrointestinal event | 620/12,838 (4.8%) | 491/12,835 (3.8%) | Reported rate ratio 1.28; 95% CI 1.13–1.44; +1.0±0.3 %p (SE) | [S04] · H-SGI | | Serious bleeding at any site, not a GI-specific count | 326/12,838 (2.5%) | 238/12,835 (1.9%) | Reported rate ratio 1.38; 95% CI 1.17–1.62 | [S04] · H-BLEED | | Stopping for upper GI reason | 227/12,838 | 104/12,835 | Recalculated crude difference +0.96 %p; original CI unreported | [S05] · H-STOP-U | | Stopping for lower GI reason | 205/12,838 | 73/12,835 | Recalculated crude difference +1.03 %p; original CI unreported | [S05] · H-STOP-L | | Stopping for other GI reason | 63/12,838 | 42/12,835 | Recalculated crude difference +0.16 %p; original CI unreported | [S05] · H-STOP-O | | Author-reported any-GI reason stopping | 495/12,838 (3.9%) | 219/12,835 (1.7%) | +2.1 %p (SE 0.2) | [S05] · H-STOP-GI | | Stopping for any medical reason | 2,107/12,838 (16.4%) | 1,020/12,835 (7.9%) | Recalculated crude difference +8.47 %p; original CI unreported | [S05] · H-STOP-MED | | All-cause stopping | 3,256/12,838 (25.4%) | 2,136/12,835 (16.6%) | Recalculated crude difference +8.72 %p; original CI unreported | [S05] · H-STOP-ALL |

The reported serious-GI **rate ratio 1.28 (95% CI 1.13–1.44)** is log-rank based. The crude count-based risk ratio **1.2624** is a different statistic and does not replace it. The crude GI-stopping absolute difference **+2.1495 percentage points** aligns with the reported rounded **+2.1 percentage points (SE 0.2)**. All-site bleeding at 326 versus 238 is not a GI-bleeding count. Precise cause-specific n/N and CI for ulcers, GI bleeding, dyspepsia and diarrhea were not verified from the accessible main/supplementary material. [S04, S05]

The original PDF Table 2 was **read as parsed text and the official abstract was read, but visual PDF table rendering failed**. No claim is made that the HPS table image was visually verified. The stopping table was checked in the 2013 primary HTML. The 2013/2014 publications are one trial, not two independent replications.

During active combination run-in, GI-related withdrawal was **2,117/38,369 (5.5%)** and all-cause withdrawal **12,696/38,369 (33.1%)**, over a mean phase of 7.4 weeks. The preceding statin-stabilization phase differs in population and duration and is not a parallel randomized control. Selecting out people unable to tolerate treatment limits applying subsequent randomized-period stopping rates to all new users. Stopping is not loss to follow-up; exact GI-specific attrition denominators were not confirmed. [S05]

## 5. Contrary, unreported and different-population evidence

**ADVENT:** safety denominators in this type 2 diabetes trial were placebo 49, ER1000 45 and ER1500 52. Two of the 47 randomized to ER1000 never received drug. Some participants also used statins and glucose-lowering drugs. The all-AE/stopping numbers below were not substituted for GI-specific incidence. [S07]

| Outcome / stopping distinction | Active | Comparator | Precision / difference | Source | |---|---|---|---|---| | Any treatment-emergent AE | 31/45 | 36/49 | Recalculated crude difference -4.58 %p; original CI unreported | [S07] · AD-AE-1000 | | Any treatment-emergent AE | 40/52 | 36/49 | Recalculated crude difference +3.45 %p; original CI unreported | [S07] · AD-AE-1500 | | Stopping due to any AE | 3/45 | 5/49 | Recalculated crude difference -3.54 %p; original CI unreported | [S07] · AD-STOP-1000 | | Stopping due to any AE | 7/52 | 5/49 | Recalculated crude difference +3.26 %p; original CI unreported | [S07] · AD-STOP-1500 |

Reported nonsignificance for other individual AEs is **not proof of GI equivalence or harmlessness**. GI-specific n/N and CI are unconfirmed. Two dose comparisons sharing a placebo group are not added as independent studies.

**Galal 2025, hemodialysis:** the three-month ER500 mg/day-plus-usual-care versus usual-care report mentions GI monitoring but does not provide verified symptom-specific event counts/denominators or definitions. The 53 randomized and 50 completers (25/25) were not substituted as an invented GI denominator. The death-arm discrepancy already recorded in 3132 and the “no serious events” narrative remain separate; this wording is not converted to “zero GI events.” Current lack of access to the Kang CKD full text is not evidence of harmlessness either. [S09, S10]

**CDC original call review:** nausea/vomiting signals after nonmedical/high-dose exposures have incomplete formulation, exposed-population denominator, co-ingestion and individual-duration information. They are not used as trial incidence or nutritional-intake risk. Other case reports cited by that report were not counted as directly verified primary case data. [S11]

## 6. Nutritional status, baseline illness and current safety limits

Baseline biochemical niacin-deficiency criteria/concentrations, total dietary intake and quantities of co-vitamins are commonly unreported in the adopted GI data. Healthy recruitment does not establish a passed deficiency test. Individual GI absolute risks in pregnancy, lactation, children, active ulcer/bleeding and severe hepatic/renal disease were not calculated from these studies. Small renal-population data and the prescribing label's renal-evidence limitations were not expanded into a claim that no human research exists. [S01, S02, S06, S07, S09]

Study doses, titration and rescue medication are **not instructions to start, stop or modify individual treatment**. No equal-milligram release-form substitution or background-drug change is recommended. Earlier liver, urate, skin and cardiovascular pages remain unchanged linked material, not reassessed endpoints.

## 7. Final content, classification and revision conditions

**M / gut:** actual pharmacologic high-dose medicinal formulations dominate the adopted conclusion, so provisional S is changed to M. The OTC supplement IR trial's commercial setting remains described; this does not classify all foods/supplements as M. No new semantic code, site ID or URL was issued.

The latest supplied prompt §5 routes this as a completed safety report with **`grading_status=not_applicable_or_policy_missing`** and **`publication_status=needs_format_mapping`**. The older grading cases 32/39 and their direction discussion are recorded in `grading_audit.json`; no inverted efficacy A–F label or invented anchor is used. `grade/score/axes/suggested_grade/score_anchor=null`, `no_human_study=false`. Warning is a scoped qualitative content decision among the four supplied labels, not a new quantitative threshold.

**`completed_with_uncertainty` / `completed_with_declared_scope`**. Accessible original tables, symptom denominators, registries/supplements, formal corrections or directly relevant populations can trigger an ID-preserving revision. Current gaps and revision conditions are in `unresolved.json` U01–U10. No new clinical judgment, value or grade is left as a Codex task.

The complete 3,133-record index and 11 candidate originals were compared; no exact independent completed GI duplicate was identified. IDs 128,2346,3082,3083,3091,3092,3129,3130,3131,3132,3133 were reused for boundaries/sources, with every original unchanged. Actual publication 54→3130,55→3131,56→3132,57→3133 is distinguished through supplied native snapshots and the user's technical receipt; no live server recheck occurred. This fifth **research handoff** does not rotate the chat before actual deployment/ledgers/MD are completed.

Why the safety label is Warning

Older grading cases 32/39 discuss support for harm claims, whereas the supplied v2 §5 explicitly routes safety-only questions to a separate report without the efficacy A–F formula. No inversion of Warning into F or invented E+ efficacy score.

Counterevidence and limits. **ADVENT:** safety denominators in this type 2 diabetes trial were placebo 49, ER1000 45 and ER1500 52. Two of the 47 randomized to ER1000 never received drug. Some participants also used statins and glucose-lowering drugs. The all-AE/stopping numbers below were not substituted for GI-specific incidence. [S07] | Outcome / stopping distinction | Active | Comparator | Precision / difference | Source | |---|---|---|---|---| | Any treatment-emergent AE | 31/45 | 36/49 | Recalculated crude difference -4.58 %p; original CI unreported | [S07] · AD-AE-1000 | | Any treatment-emergent AE | 40/52 | 36/49 | Recalculated crude difference +3.45 %p; original CI unreported | [S07] · AD-AE-1500 | | Stopping due to any AE | 3/45 | 5/49 | Recalculated crude difference -3.54 %p; original CI unreported | [S07] · AD-STOP-1000 | | Stopping due to any AE | 7/52 | 5/49 | Recalculated crude difference +3.26 %p; original CI unreported | [S07] · AD-STOP-1500 | Reported nonsignificance for other individual AEs is **not proof of GI equivalence or harmlessness**. GI-specific n/N and CI are unconfirmed. Two dose comparisons sharing a placebo group are not added as independent studies. **Galal 2025, hemodialysis:** the three-month ER500 mg/day-plus-usual-care versus usual-care report mentions GI monitoring but does not provide verified symptom-specific event counts/denominators or definitions. The 53 randomized and 50 completers (25/25) were not substituted as an invented GI denominator. The death-arm discrepancy already recorded in 3132 and the “no serious events” narrative remain separate; this wording is not converted to “zero GI events.” Current lack of access to the Kang CKD full text is not evidence of harmlessness either. [S09, S10] **CDC original call review:** nausea/vomiting signals after nonmedical/high-dose exposures have incomplete formulation, exposed-population denominator, co-ingestion and individual-duration information. They are not used as trial incidence or nutritional-intake risk. Other case reports cited by that report were not counted as directly verified primary case data. [S11]

Review performed and remaining limitations

Direct GI rates for nutritional/dietary/confirmed-deficiency use and pregnancy/lactation/children — Obtain direct data with matched population, molecule, dose/titration, duration, definition and arm n/N Exact ER-label symptom/stopping counts, dose-column overlap, individual-trial co-medication/windows — Obtain original trial CSR/GI tables and person-overlap information Mills vomiting 4/33 versus 0/35: reported p=.005 not matched by standard independent 2×2 diagnostics — Obtain original GI test specification/pair-level data or formal correction AIM-HIGH GI primary table/cause denominators; indexed candidates not adopted — Verify definitions, arm events, period and precision in an accessible original table HPS serious-GI cause-specific tables and visual PDF table check — Obtain original supplement/images; ingredient attribution additionally requires a separating design SR versus IR GI-specific stopping and detailed covariates — Obtain McKenney full GI tables ADVENT/hemodialysis symptom-specific GI numerators/denominators and missingness — Obtain original GI safety tables/verified reports Registry-paper GI prespecification/analysis concordance — Access registry body/historical versions Previously documented Galal death-allocation/no-serious-event narrative boundary — Future author correction/data and a separate revision trail; no rewriting of 3132 now Safety-only A–F/score algorithm and new site ID/format — Technical safety mapping/ID without new clinical values; efficacy score remains inapplicable

Evidence Table

Receipt — 13 References

Evidence access cutoff: 2026-09-16. Each citation states its actual access level (full-text transcription, abstract, index, or other) and remaining limitations. Bibliographic checking does not certify the study results or treatment efficacy.

What this document does and does not do

Chamgap is an information source. It reports what research has and has not confirmed; it does not tell readers what to take or buy. That decision belongs to readers and, when needed, medical or legal professionals. This verdict reflects literature available up to the search date and may change as new research appears. Nothing here is medical advice.