Nitrofurantoin

Limits on what this shows

• Null findings: Metronidazole, a nitroimidazole and p-nitrobenzoic acid had no significant effect in the same system [15]. Whether 50 micromolar in vitro exposure is reached in mitochondria of treated people was not established in the sources read (the label gives peak plasma levels below 1 mcg/mL after 100 mg [1]); no unit conversion is attempted here.

Direction

Mitochondrial functional inhibition by the drug (toxicity mechanism); not a nutrient deficiency.

Regulatory status

The label states that in bacteria the reduced intermediates inhibit protein synthesis, aerobic energy metabolism, DNA, RNA and cell-wall synthesis [3]; it says nothing about human mitochondria.

Population evidence

None in humans. Lim 1986 [15] (abstract via a summarizing fetch tool): nitrofurantoin and nitrofurazone inhibited respiration of isolated C57B/6J mouse liver mitochondria; only complex I substrates (glutamate, beta-hydroxybutyrate, alpha-ketoglutarate) were affected, succinate (complex II) was not; at 50 micromolar, state 3 and dinitrophenol-uncoupled respiration with glutamate fell to 28 +/- 1% and 25 +/- 5% of control; both state 3 and state 4 were inhibited in mitoplasts; superoxide dismutase (240 micrograms/mL) plus catalase (200 micrograms/mL) prevented the state 3 inhibition. Rossi 1988 [16]: in rat hepatocytes oxygen uptake rose even when mitochondrial respiration was blocked, which indicates redox-cycling oxygen consumption rather than mitochondrial respiration.

Mechanism

Named in the source: redox cycling and reactive oxygen intermediates, with complex I identified as a target [15].

Red/orange hunt

One animal in vitro study; Tier B only, so yellow is the ceiling. Missing: any human or whole-organism measurement of mitochondrial function, or a second independent mitochondrial study. Tried: PubMed "nitrofurantoin AND mitochondria OR mitochondrial OR complex I OR respiration OR ATP" (429), "nitrofurantoin mitochondrial respiration" (3 records, read), "nitrofurantoin AND adenosine triphosphate" (10 records, read). Full text of Lim 1986 was not opened.

Counter-evidence held on record

An isolated-organelle result at a high concentration does not show mitochondrial damage in people taking the drug.

Limits on what this shows

• Reverse direction and null findings: Wijaya 2022 [20] (abstract via an OpenAlex record summary): in HepG2 cells and primary human hepatocytes nitrofurantoin radical anions activated Nrf2 and induced glutamate cysteine ligase, raising glutathione to a level like diethyl maleate; the rise lasted up to 48 hours after removal and protected against rotenone, paraquat and proteasome inhibitors. Lung glutathione rose in the mouse study [27]. Wang 2008 [19]: reactive oxygen species output tracked reductase activity but not cytotoxicity, so ROS generation alone did not explain toxicity.

Direction

Increased oxidant burden (a mechanism and toxicity entry, not a nutrient deficiency). Reverse-direction finding recorded below.

Regulatory status

The label does not describe oxidative stress as such; it states reduction by bacterial flavoproteins to reactive intermediates and, for term pregnancy and neonates, glutathione instability of red cells [1][3]. Acute, subacute and chronic pulmonary reactions are label-warned (chronic reactions generally with six months or longer of therapy) [1].

Population evidence

No human in vivo oxidative-stress marker (glutathione, malondialdehyde, antioxidant enzymes) in nitrofurantoin users was found. Human-cell in vitro: erythrocytes at 840 micromolar for about 21 hours lost 65% of glutathione (control decline reported as 43% in the same abstract summary) with a 5-fold rise in hydrogen peroxide; EC50 for glutathione and ATP depletion ranged from about 10 to 100 micromolar across donors [13]. A review of lung toxicity notes it has no human redox or glutathione measurements and relies on a single rat study [22]. Rats, 200 mg/kg subcutaneous: transient fall in lung glutathione with a rise in oxidized glutathione, increased lipid peroxidation (diene conjugates, TBARS), superoxide dismutase and glutathione peroxidase unchanged, catalase and glutathione reductase transiently lower; numbers not in the abstract [17]. Isolated rat hepatocytes: oxygen uptake rose even with KCN or antimycin A present, glutathione depletion with near-stoichiometric GSSG formation preceded cell death, nitrofurantoin reversibly inactivated glutathione reductase, and the glutathione depletion and GSSG formation did not occur in hepatocytes from low-selenium-diet rats lacking glutathione peroxidase activity [16]. Mice 40 mg/kg i.p.: liver glutathione and catalase fell at 4 hours, lung glutathione rose, lung glutathione peroxidase fell, and liver MDA rose only in vitamin E-deficient mice [27]. Rats: activation of AP-1, HIF-1 and STAT3 with heme oxygenase-1 induction in liver [21].

Mechanism

Named. NADPH-dependent cytochrome P450 reductase and NADH-preferring cytochrome b5 reductase redox-cycle nitrofurantoin to superoxide, hydrogen peroxide and, with redox-active iron, hydroxyl radicals (recombinant enzymes and cell lines) [18][19].

Red/orange hunt

No Tier A human study measures an oxidative-stress marker with nitrofurantoin, so orange is not reachable. Missing: any human in vivo marker. Tried: PubMed "nitrofurantoin AND (oxidative stress OR redox cycling OR superoxide OR glutathione)" (189 results, 10 shown) and "nitrofurantoin AND (lung OR pulmonary) AND (glutathione OR lipid peroxidation OR antioxidant)" (429, not read); a targeted "Suntres nitrofurantoin" lookup returned the rat paper [17].

Counter-evidence held on record

A laboratory mechanism is not a finding that a person's antioxidant status is low, and antioxidant supplements are not supported by any source here.

Limits on what this shows

• Null findings: No human study measuring iron or ferritin in nitrofurantoin users was found. Hemolytic anemia is an outcome of the oxygen-delivery entry and is not an iron-status finding.

Direction

Redistribution and possible excess of reactive (labile) iron in a redox-stress model; not iron depletion. Any iron supplementation would not be supported by this entry.

Regulatory status

The label text read has no iron interaction or iron statement [1][2][3].

Population evidence

None in humans. Sturm 2006 [23] (HepG2 cells; abstract via a summarizing fetch tool): nitrofurantoin increased the labile iron pool but, unlike antimycin A, did not stimulate non-transferrin-bound iron uptake; ebselen reversed the nitrofurantoin effect; no numbers in the abstract. Monteiro 1989 [24] (cell-free, xanthine oxidase): nitroaromatic radicals including nitrofurantoin released ferritin iron more efficiently than an equivalent superoxide flux; superoxide dismutase inhibited release in air. Wang 2008 [19]: hydroxyl radical formation by nitrofurantoin redox cycling requires redox-active iron.

Mechanism

Named: one-electron reduction products release or use iron, feeding Fenton-type chemistry.

Red/orange hunt

In vitro and cell-free Tier B only. Missing: iron indices in treated humans or animals. Tried: PubMed "nitrofurantoin AND iron" (42 results, 10 read), "nitrofurantoin iron copper zinc" (no results, 404), label text of three products.

Counter-evidence held on record

A laboratory iron signal under oxidant stress says nothing about whether a person has too little or too much iron.

Limits on what this shows

• Why no grade: Dual condition met: no Tier B or C source supports even yellow, and no circulating claim on copper and nitrofurantoin was found to rebut as white. This is a disclosed gap, not a finding of safety. Re-check with open-web search restored.

What was searched

PubMed "nitrofurantoin AND copper" (10 records read): none human; most are analytical sensors using copper nanoclusters or frameworks to detect the drug; Iscan 1993 (sheep lung and liver nitrofurantoin reductase inhibited by copper in vitro; listing level), which is copper acting on an enzyme, not the drug acting on copper; one bacterial gene-transfer study using copper as a stressor. PubMed "nitrofurantoin iron copper zinc": no results. Label text of three products: no copper statement [1][2][3]. Consumer pages fetched (Drugs.com food interactions, Medical News Today): no copper or mineral depletion statement [40][41].

Channel outcomes

The PubMed channel worked, so the absence of human or biological copper studies is a genuine gap on that channel. The consumer-claim audit was tooling-limited: the shared web-search budget was exhausted, so open-web searches for copper claims could not be run. Szilagyi 2018 [18] and Wang 2008 [19] mention transition metals or redox-active iron in hydroxyl radical formation, not copper specifically.

Repair pass 2026-10-08

Added PubMed "nitrofurantoin AND ceruloplasmin" (fetch returned an error page, which in this session's earlier zero-hit queries meant no matching records) and a combined query (nitrofurantoin with ceruloplasmin, serum copper, copper deficiency, serum zinc, zinc deficiency) that PubMed did not parse into a results page, so it counts as no result, not as a negative finding. The open-web search tool reported its shared budget exhausted, so consumer-claim searches for copper still could not be run. No Tier B or C source for copper and nitrofurantoin was found; the entry stays ungraded as a genuine gap on PubMed and tooling-limited for the consumer audit.

Limits on what this shows

• Null findings: Stewardson found no global change for nitrofurantoin [30]; Elvers and Baldanzi describe little or few effects [31][32]. The Elvers review may include the Vervoort and Stewardson studies (not verified), and Vervoort and Stewardson share an author (A. Stewardson is a co-author of Vervoort), so these are not fully independent.

Direction

Small compositional disturbance, directionally benign in the measured taxa (a risk entry, not a nutrient deficiency; no supplementation implication).

Regulatory status

Label: pseudomembranous colitis reported sporadically; C. difficile-associated diarrhea "has been reported with use of nearly all antibacterial agents, including nitrofurantoin", including over two months after therapy [1][2]. Diarrhea is a minor trial adverse event (below 1%) [1].

Population evidence

Vervoort 2015 [29] (UTI patients on nitrofurantoin versus untreated healthy controls, 16S pyrosequencing; before, within 48 hours after, and 28 days after treatment; abstract via an OpenAlex record summary and a PubMed summary, per-group n not retrieved): Actinobacteria rose by 19.6% at the end of treatment (P = 0.026) and returned to within 1.0% of baseline by day 28; the rise was driven by Bifidobacteriaceae (81.0% of the phylum, 95% CI +/- 7.4%); no other significant impact. Stewardson 2015 [30] (outpatients with UTI given ciprofloxacin or nitrofurantoin, untreated UTI patients and household contacts, 16S at three time points; abstract via a summarizing fetch tool, group sizes and p values not retrieved): ciprofloxacin had a significant global effect, nitrofurantoin did not; nitrofurantoin was associated with a decrease in Clostridium and an increase in Faecalibacterium. Baldanzi 2026 [32] (Swedish prescribing register linked to fecal metagenomes of 14,979 adults, 8-year look-back; article page via a summarizing fetch tool): nitrofurantoin was associated with only a few species, far fewer than clindamycin, fluoroquinolones and flucloxacillin; use within the previous year was associated with higher species diversity (q = 0.043), which the authors attribute to chance or collider bias and which was not seen in the single-course analysis; the count of nitrofurantoin users was not reported in the text read. Elvers 2020 [31] (systematic review, 31 studies, up to May 2020; abstract via an OpenAlex record summary, and in the repair pass the open-access article page via a summarizing fetch tool): phenoxymethylpenicillin, nitrofurantoin and amoxicillin had very little effect on the gut microbiome. Group sizes retrieved in the repair pass : the review's text describes three nitrofurantoin studies with 25 participants in total. The Stewardson study (ref 53 in the review) had 10 UTI patients on nitrofurantoin macrocrystals, with no significant global effect, a higher Faecalibacterium proportion and a lower Clostridium (Clostridiaceae) proportion at the end of treatment. A second study of 8 people with uncomplicated UTI (ref 54, author not named in the text read) found no significant impact apart from a temporary rise in Actinobacteria, specifically Bifidobacterium, returning to pre-antibiotic levels 31 to 43 days after stopping; this matches the Vervoort abstract, but the review text read does not name Vervoort, so that identification is an inference and Vervoort's own n is unconfirmed. A third study of 7 women with recurrent UTI (ref 55, author not named) found enterobacteria, enterococci and yeasts unchanged during treatment with no resistant Gram-negative strains detected. The review's reference list and Table 2 were not visible in the text read. These are very small samples (10, 8 and 7).

Mechanism

Named but weak: antibacterial action on gut organisms is limited by low systemic and tissue exposure (a reasoned argument in a 2026 review, not a measured result; listing level) [33]; the label records the C. difficile risk [1].

Red/orange hunt

Three human sources with this drug in the cohort exist, but they agree that the effect is small, transient and confined to a few taxa, so the evidence supports yellow, not a harm grade. Missing for a possible orange: Stewardson p values and the other-arm sizes (the nitrofurantoin arm is 10 per the Elvers review), Vervoort per-group n (8 is inferred only), taxon tables of Baldanzi. Repair pass: the three human studies total 25 nitrofurantoin users, none showing a harmful shift, so even full text would not move this above yellow. Tried: PubMed "nitrofurantoin gut microbiota" (18 results, 10 read), "nitrofurantoin resistome gut" (42, 10 read), several variants that returned 429; Stewardson full text (clinicalmicrobiologyandinfection.com 403, europepmc.org blocked by robots.txt, Europe PMC REST 429). Repair pass: OpenAlex works/doi lookup (no abstract index, bronze OA, repository copy listed), University of Geneva archive page unige:74165 (full text restricted to UNIGE login, abstract only, no counts), ScienceDirect page (blocked by robots.txt), Oxford Academic page for Vervoort (abstract only, full text paywalled, no sample sizes), BMJ Open page for Elvers (opened, gave the 10, 8 and 7 figures above).

Counter-evidence held on record

Most people return to baseline within weeks, and the measured shifts are small; nothing here supports a probiotic claim or implies vitamin loss.

Limits on what this shows

• Null findings and counter-evidence: In the Clalit cohort nitrofurantoin was prescribed to 1,366 G6PD-deficient patients without a hemolysis admission [8]; the real-world signal is much smaller than the label suggests. The pharmacovigilance rates (about 1 per 100,000 courses at worst, 42 G6PD cases against at least 245 million exposures) are low, the deaths were not shown to be caused by the drug [7][9], and the early red-cell survival studies suggest less oxidant hemolysis than primaquine [7]. The 2022 review's authors suggest that a 3 to 5 day course at 200 mg total daily dose may be used without G6PD screening when accompanied by appropriate advice [7]. Wikipedia (Tier D) states nitrofurantoin is contraindicated in G6PD deficiency [42]; the US label text read lists it under Warnings [1][2].
• At-risk subgroups: G6PD deficiency (variant and dose matter); neonates under one month; women at term or in labor; anyone with unrecognized deficiency on a longer or higher-dose course. The case reports above include pregnant and older patients.

Direction

Drug toxicity outcome (loss of red-cell mass and oxygen-carrying capacity; a fraction of hemoglobin converted to methemoglobin). It is not an iron, folate or other nutrient deficiency (Hard Rule 2) and implies no supplementation. Iron is a separate entry.

Regulatory status

Label (Macrobid, read through a summarizing fetch tool) [1]: hemolytic anemia of the primaquine-sensitivity type has been induced; hemolysis "appears to be linked to a glucose-6-phosphate dehydrogenase deficiency in the red blood cells"; the label gives the prevalence of deficiency as about 10 percent in Black patients and in some Mediterranean and Near-Eastern groups; hemolysis ceases when the drug is withdrawn. Contraindicated in pregnant patients at term (38 to 42 weeks), in labor and delivery, and in neonates under one month because of immature erythrocyte enzyme systems (glutathione instability) [1][3]. Trial laboratory events of 1% to 5% include decreased hemoglobin. "Cyanosis secondary to methemoglobinemia has been reported rarely." Hemolytic anemia and megaloblastic anemia are listed among laboratory abnormalities [1][2]. G6PD deficiency appears in the Warnings text read, not in the Contraindications text read [1][2].

Population evidence

Recht 2022 [7] (literature review plus FAERS searched July 2020 and a VigiBase file; full text read through a summarizing fetch tool, first 100,000 of 101,189 characters): 318 hemolytic anemia episodes, 10 deaths (all from pharmacovigilance databases, causal link not established), 42 episodes in confirmed or highly probable G6PD deficiency, at least 245 million exposures. Mostly case reports; the authors did not grade quality and a meta-analysis was not feasible. Case regimens in G6PD-deficient patients included 100 mg three times daily for 4 days (Lavelle), 300 mg/day for 2 days (pregnant) and 800 mg daily for 14 days. A 1990 estimate cited there put hemolytic events at about 1 per million exposures. Gait 1990 [9]: Norwich Eaton adverse-reaction database, about 130 million US courses since 1953, 127 hemolytic reports, worst-case incidence 1 in 100,000 courses, 87% of patients with recorded outcomes fully recovered, no well-documented neonatal hemolysis cases. Gronich 2024 [8] (Clalit database, abstract via an OpenAlex record summary): 31,962 patients with G6PD activity below 6 U/g Hb followed 2010 to 2022; 71 hemolysis hospitalizations (0.2%), of which 51 followed fava beans, 6 infection and 3 were suggested to be medication-caused (nitrofurantoin, phenazopyridine, an analgesic). Repair pass: a second, independent record of the same abstract (PubMed single-record lookup) confirmed these figures and added that 1,366 G6PD-deficient patients were safely prescribed nitrofurantoin (among 31,875 patients without major hemolysis), alongside hundreds to thousands prescribed ciprofloxacin, sulfamethoxazole/cotrimoxazole, hydroxychloroquine and others; the abstract does not say how many of the 3 medication-suggested cases involved nitrofurantoin, so no nitrofurantoin-specific hemolysis rate or confidence interval can be calculated from it. Recht 2022 abstract (second record, PubMed lookup): 318 episodes, 10 deaths, 42 (13%) in confirmed or highly probable G6PD deficiency, at least 245 million exposures; the authors state no causal link to death was reported and that a precise risk estimate in G6PD-deficient people was not possible because few reports come from high-prevalence regions. Red-cell survival (cited inside [7], primary papers not opened): in G6PD-deficient subjects (Canton and B-Chinese variants) nitrofurantoin shortened red-cell half-life from 19 to 28 days to 4 to 14 days (Chan 1976), versus 3 to 5 days with primaquine 15 to 45 mg/day; in two primaquine-sensitive men given 800 mg daily for 14 days, glutathione fell from 45 to 30 mg/dL and the hematocrit fell by only 20% to 25% of the fall seen with primaquine 30 mg daily (Powell 1963). Case reports : Bhatt 2023 [10] (woman in her 60s, day 5 of a 7-day course, hemoglobin 121 to 89 g/L, hemolysis markers, G6PD deficiency, full recovery after stopping); Lavelle 1976 [11] (69-year-old man, 4 days of therapy, falling hemoglobin, reticulocytosis, lactic acidosis, resolved within 3 days of withdrawal); Padhiyar 2026 [12] (G6PD deficiency, hemolysis and methemoglobinemia, listing level only, methemoglobin percentage not stated).

Mechanism

Named. Nitrofurantoin redox-cycles with NADPH-dependent flavoenzymes to superoxide and hydrogen peroxide, consuming reduced glutathione; red cells that cannot regenerate NADPH (G6PD deficiency, immature neonatal enzyme systems) are most exposed. In vitro with human erythrocytes at 840 micromolar over about 20 hours, hydrogen peroxide production rose 5-fold, methemoglobin exceeded 20% of total hemoglobin and glutathione fell 65% [13]; in a G6PD-deficient patient's red cells nitrofurantoin stimulated glucose use and lactate production and inhibited reduced glutathione generation [11]. Methemoglobin formation after nitrofurantoin is also shown in rats exposed to UV-A light (0.97 +/- 0.37% versus 0.5% control; listing level) [14].

Red hunt

Red needs two independent Tier A sources with consistent direction, clinically meaningful size and no substantial contradicting Tier A evidence. The label [1][2], the pharmacovigilance synthesis [7] and the red-cell survival data cited in it point one way, but the largest direct cohort [8] shows 1,366 exposed patients without hemolysis and the incidence is about 1 in 100,000 courses [9]; that is substantial contradicting evidence on frequency, and the entry is a toxicity outcome, so it is held at orange with the boundary in the title. Effect sizes were retrieved for the cohort and pharmacovigilance sources; the primary red-cell survival papers (Chan 1976, Powell 1963) were not opened. Repair pass: Gronich 2024 and Recht 2022 were each confirmed by a second PubMed record. Gronich is a second independent cohort with this drug, but it runs against a clinically large effect (1,366 exposed without a hemolysis admission, and no per-drug rate or CI in the abstract), so red remains out of reach; the entry stays orange.

Counter-evidence held on record

Hemolysis here is an adverse drug effect in a defined subgroup. It is not a sign of low iron, folate or B12, and nothing in this entry supports taking a supplement.

Limits on what this shows

• Null findings: No study found with an unchanged-ATP result for nitrofurantoin. The red-cell study used a concentration range that may exceed typical blood levels (label peak below 1 mcg/mL [1]); relevance in vivo is unestablished.

Direction

Reduced ATP in a toxicity model (energy-metabolism entry, kept separate from Mitochondrial burden and from Oxygen delivery/utilization); not a nutrient deficiency.

Regulatory status

The label makes no statement about human ATP; bacterial aerobic energy metabolism inhibition is described [3].

Population evidence

Dershwitz 1985 [13] (human erythrocytes in vitro; abstract via a summarizing fetch tool): significant differences from control after 12 hours, largest at 18 to 21 hours; at 21 hours 840 micromolar nitrofurantoin lowered ATP by 75%, against a control decline of 29% in the abstract summary (a separate summary of the same record gave 43% for controls, so the control figure is uncertain); EC50 for ATP depletion about 10 to 100 micromolar depending on donor; echinocyte formation. Valeri 1997 (search-listing record only, within [43]): ATP and energy charge were compromised in rat hepatocytes exposed to 400 micromolar nitrofurantoin, no numbers in the record. State 3 (ADP-stimulated) respiration with complex I substrates fell to about a quarter of control at 50 micromolar in mouse liver mitochondria [15]; this is an indirect surrogate and no ATP was measured there.

Mechanism

Plausible and partly named: redox-cycling oxidant stress and NADPH/glutathione depletion in erythrocytes [13]; complex I inhibition in mitochondria [15].

Red/orange hunt

Only Tier B in vitro data exist. Missing: ATP or adenine nucleotides in tissue from treated humans or animals. Tried: PubMed "nitrofurantoin AND adenosine triphosphate" (10 records read), "nitrofurantoin AND ATP" (429 once, then re-run with the longer phrase); the original ATP-specific record Dershwitz was retrieved by a single-record multi-uid lookup.

Counter-evidence held on record

A drop in ATP in red cells in a dish is not evidence of low energy or fatigue in people on a short course.