The condition itself is silent — most people who have it never know. The risk is the trigger. The seven drugs with rock-solid evidence (dapsone, primaquine, tafenoquine, rasburicase, methylene blue, nitrofurantoin, phenazopyridine) can wreck red blood cells within days in someone deficient. Raw fava beans can do the same. Newborns are the other vulnerable window — untreated jaundice from undetected deficiency can cause permanent brain damage. The whole intervention is one test and a problem-list entry. Cheap, lifelong, and the difference between a silent gene and a foreseeable crisis you didn't see coming.
Red blood cells live for about four months and they live without a backup. They don't have mitochondria; they can't make new proteins; they can't repair much. The one enzyme they really need to keep working is glucose-6-phosphate dehydrogenase — G6PD. Its job is to keep a chemical called glutathione in its working form, and glutathione is what mops up the constant low-level oxidative damage that haemoglobin generates just by carrying oxygen around. Lose enough G6PD and the cleanup stops Luzzatto 2020.
With the cleanup stopped, haemoglobin starts to oxidise. Damaged haemoglobin clumps inside the cell into something called Heinz bodies; the cell membrane stiffens; the spleen recognises the damage and clears the cell, often by breaking it open right in the bloodstream. Multiply this across a meaningful fraction of your red cells over a day or two and you have a haemolytic crisis — jaundice from the released haemoglobin, dark urine from haemoglobin passing through the kidneys, fatigue from the missing red cells.
The trigger that started it all back in 1956 was the antimalarial drug primaquine. Some soldiers on it developed sudden anaemia; others didn't. Tracking down the difference led directly to the enzyme and the gene Cappellini and Fiorelli 2008. The list of triggers has grown since, but the mechanism is the same one every time: anything that pushes more oxidative work onto red blood cells than their crippled G6PD can keep up with.
How common, and why the map looks the way it does
About 400 to 500 million people carry a G6PD variant that produces meaningful deficiency — making this the most common enzyme defect in humans Nkhoma et al. 2009Cappellini and Fiorelli 2008. The map of who has it overlaps almost exactly with the historical map of malaria: sub-Saharan Africa, the Mediterranean rim, the Arabian Peninsula, the Indian subcontinent, Southeast Asia, parts of the Pacific. About one in ten Black American men is deficient. Prevalence runs from 5% to 30% across Mediterranean and Middle Eastern populations Howes et al. 2012.
The reason is malaria. Female carriers of the African A− variant have roughly half the risk of severe malaria of women without it — a large enough survival advantage to drive the variant up in frequency over the centuries it took malaria to shape African genetics Uyoga et al. 2015Ruwende et al. 1995. Men with the same variant don't get the malaria benefit (the genetics work differently for them) but inherited it along with their sisters. The trade — protection from a historical killer in exchange for vulnerability to a class of modern drugs and one particular bean — is the kind of compromise evolution makes when one side of the trade is the difference between living to reproduce.
What the variants actually do
There are hundreds of different mutations. Two matter for most readers. G6PD A−, the African variant, leaves about 12% of normal enzyme activity — enough to handle ordinary life, not enough to handle a strong trigger. Crises tend to be self-limited; the haemolysis burns through the oldest red blood cells and the newer ones are robust enough to hold the line Frank 2005. G6PD Mediterranean, found across southern Europe, the Middle East, and parts of South Asia, leaves about 4% activity. The same trigger produces a worse, longer crisis, and this is the variant most associated with fava-bean catastrophes Luzzatto et al. 2023.
Get tested. Tell people. Done.
The whole intervention is small. A blood test measures the enzyme activity in your red blood cells — either the older fluorescent spot test (yes/no answer) or a quantitative assay that gives a percentage of normal Frank 2005. Point-of-care biosensors that produce a number in a few minutes are now available in some clinics. Ask your doctor; in the US the test usually costs in the tens of dollars and is frequently covered when ordered for a clinical reason.
One catch on timing. If you've had a haemolytic crisis recently, testing within a few weeks can give a falsely normal answer — the deficient red blood cells have already been destroyed, leaving a young population that still has enough enzyme to look fine. Retest two to three months after the crisis if your doctor missed this and the result didn't fit the story Cappellini and Fiorelli 2008.
The trigger list
Most published lists are too long. A careful evidence review separates the drugs that clearly cause haemolysis in deficient people from drugs labelled by tradition with weak or in-vitro evidence behind the label Youngster et al. 2010. The short, strong list is the one to memorise.
Sulfa drugs as a class — sulfamethoxazole-trimethoprim (Bactrim), sulfasalazine, sulfanilamide — sit in a fuzzier category. Real-world data suggests Bactrim and sulfasalazine carry less haemolytic risk than the historical label implies, while older sulfa agents like sulfanilamide and sulfacetamide are clearer triggers Youngster et al. 2010. The conservative move is to flag any sulfa drug for your pharmacist and let them weigh the alternative.
Other commonly listed agents — high-dose aspirin, intravenous gram-doses of vitamin C, chloramphenicol, some quinolone antibiotics — have weaker evidence behind their inclusion. They are not bright-line forbidden the way the list above is; risk depends on dose, the specific variant, and what else is going on. A good prescriber, told about your status, can navigate this.
Non-drug triggers
One specific situation worth flagging for any G6PD-deficient adult and the people around them: if methaemoglobinaemia is the emergency (lips and fingers turning blue, blood sample looking chocolate-brown), the standard antidote is methylene blue — which is contraindicated here. The alternatives are slower (intravenous vitamin C, exchange transfusion, hyperbaric oxygen) but real; an emergency team that knows your status will pick correctly.
What people get wrong
"It's a sulfa allergy." No. An allergy is the immune system overreacting; this is a chemistry problem inside the red blood cell. The reactions look different — allergy gives you hives or anaphylaxis within minutes, G6PD-driven haemolysis gives you jaundice and dark urine a day or two after the dose. The fix is also different: an allergy never goes away with avoidance of one drug; G6PD deficiency stays exactly the same drug-list-wide for life.
"Only men get it." Wrong, and clinically dangerous. The gene is on the X chromosome, so men with one bad copy are uniformly affected. But women have two X chromosomes, and one of them is randomly switched off in every cell early in development. A woman who inherits one bad copy ends up with a mosaic — some of her red blood cells are fine, some are deficient. Depending on how the random switching went, she can be functionally normal, or just as vulnerable as her brother Luzzatto 2020. The older yes/no spot test averages across her red cells and often misses these women. If you're a woman in a high-prevalence family, ask for a quantitative test, not the spot test.
"My ancestry isn't Mediterranean or African, so I'm fine." The map is statistical, not absolute. New York State's universal newborn screening data found that risk-factor-based testing would have missed about 44% of affected infants, including a meaningful fraction of white and Ashkenazi Jewish children. Ancestry is a strong hint, not a screening test.
"I ate fava beans once and was fine, so I'm fine." Tolerance is variable in ways that aren't fully understood. The same person can eat fava beans on Monday with no effect and crash on Wednesday. Cooking matters, dose matters, what else is happening in your body matters. Treat a confirmed deficiency as a real constraint, not a personally-observed risk.
Who specifically should test
The risk is concentrated in four broad ancestries: sub-Saharan African (roughly 10–20% of men deficient in many populations), Mediterranean (especially Sardinian, Greek, southern Italian, Cypriot), Middle Eastern Arab, and Southeast Asian (especially Thai, Lao, Cambodian, southern Chinese, Filipino). Ashkenazi Jewish populations also carry the Mediterranean variant at appreciable frequency Howes et al. 2012Nkhoma et al. 2009.
Two specific scenarios where the test is worth pushing for regardless of ancestry: an unexplained jaundice episode (yours or a close relative's) after a specific antibiotic or after fava beans, or a newborn with severe jaundice that didn't respond well to phototherapy.
If you're starting a family
Two reasons to know your status before pregnancy. First, primaquine and tafenoquine — the relapse-preventing antimalarials — are contraindicated in pregnancy regardless of your G6PD status, because the fetus might be deficient even if you're not. Second, if your baby inherits the deficiency, neonatal jaundice is the immediate window of risk; flagging this before delivery means the hospital is watching for it from day one rather than reacting to a bilirubin number on day four Kemper et al. 2022.
How this goes wrong in the real world
Almost every preventable G6PD crisis comes down to one of four documentation or recognition failures.
The drug is prescribed by someone who didn't know. You walk into urgent care with a UTI; you get nitrofurantoin. You have a cellulitis; you get Bactrim. The prescriber didn't ask, you didn't think to mention something you've never had a crisis from before. Two days later you're jaundiced. This is the single most common scenario, and it's the one a problem-list entry in your medical record prevents.
The test was done during a crisis and read as normal. Acute haemolysis destroys exactly the cells the test is supposed to measure, leaving young red cells with enough residual enzyme to look fine. The diagnosis gets missed at the moment it matters most. If your story doesn't fit the result, retest two to three months later Cappellini and Fiorelli 2008.
A female carrier was told she's normal by the old spot test. The fluorescent spot test averages across her red cells and frequently misses women whose mosaic happens to favour deficient cells. She gets the same drug her brother would have been warned away from, and crashes. The quantitative test catches her; the spot test doesn't Luzzatto et al. 2023.
The newborn was discharged before the bilirubin peak. Hospitals send healthy-looking babies home on day two; the bilirubin peaks on day four or five. A G6PD-deficient baby can climb from cheerfully jaundiced to dangerously jaundiced in that window without much warning, and the warning signs are subtle until they aren't. This is what universal newborn screening in places like Singapore, Sardinia, and Greece is designed to catch Kemper et al. 2022Kaplan and Hammerman 2010.
What an unrecognised case actually looks like
Day one, you take the drug — say, nitrofurantoin for a urinary infection. You feel fine. Day two, still fine, mildly tired but you wrote off the UTI being annoying. Day three, the tired hits differently — climbing stairs is suddenly hard, and someone at work says you look a little yellow. You go to the bathroom and the toilet bowl has tea in it. By the time you're at the emergency room your eyes are clearly yellow, your heart is racing, and a blood draw shows your haemoglobin has dropped by a third in seventy-two hours Frank 2005.
The crisis is usually survivable, especially if you get to a hospital. Transfusion stabilises you. The drug gets discontinued and you turn the corner over the next week or two. But the kidneys take the hit from haemoglobin passing through them — acute kidney injury is a real complication of a severe crisis, and a small fraction of patients end up needing dialysis. The Mediterranean variant in particular has historically killed people from bad favism episodes.
For a newborn, the stakes compress. Untreated severe jaundice in the first week of life can cross into the brain — kernicterus — with a roughly 10% chance of death and around a 70% chance of permanent damage in survivors: deafness, cerebral palsy, intellectual impairment Kemper et al. 2022. This is rare in modern hospitals with active screening, but every case is a child whose life is permanently constrained by a condition that could have been flagged on day one of life with a heel prick.
The Sardinian screening programme, running across decades, essentially eradicated kernicterus from G6PD deficiency in a population where the Mediterranean variant is common and favism used to be a recurring public-health problem. The cost was a heel prick on every newborn and a counselling session for every positive. The benefit was a generation of children who never inherited their grandparents' fear of fava beans because the danger had been disarmed for them on day one.
What changes once you know
Week one, nothing visibly changes. You added a line to a medical record and stuck a card in your wallet. The intervention is administrative; the body it acts on hasn't moved.
What changes is foreseeable encounters years from now. The next time you have a urinary infection — usually some random Tuesday — the pharmacist looks at the screen and says, "let's pick something else for you." The dentist running through a pre-procedure med list catches a flag. The travel-medicine clinic prescribing primaquine for an upcoming trip switches to a different regimen. Each of these is a non-event that would otherwise have been three days of haemolysis you didn't see coming.
Multiply that across forty or fifty years of medical encounters. In aggregate, the people in your life — the partner who would have driven you to the emergency room, the colleagues who would have visited you in the hospital, the kids who would have heard you were sick — never see the crisis happen. The flag did its work invisibly. This is the boring kind of payoff that prevention earns: an absence of stories.
For parents of a G6PD-deficient newborn, the payoff is more compressed. The hospital watches the bilirubin curve closely in the first week. Phototherapy starts earlier if it's needed. The fava bean talk happens at the first feeding milestone instead of after a crisis. By the time the child is school-age, they know the rules the way other kids know about a peanut allergy — a constraint that's been part of their life as long as they've been around, not a discovery made the hard way.
Adjacent topics you may want to look into: sickle cell trait and other inherited red-blood-cell conditions that overlap geographically with G6PD deficiency; pharmacogenetic testing more broadly (a single blood test now covers dozens of drug-response variants alongside G6PD); newborn screening panels and what your country's actually includes; the malaria treatment landscape, where point-of-care G6PD testing is changing access to relapse-preventing drugs in endemic countries.
- — The whole intervention is one chart entry — G6PD on your record so every doctor and pharmacist sees it.
- — G6PD status is a pharmacogenetic fact: it tells doctors which drugs and foods can trigger a crisis.
- — Nitrofurantoin, a go-to UTI antibiotic, can rupture red cells if you're G6PD-deficient — flag the gene before you fill that script.
- — G6PD deficiency is inherited; one test settles it for life and goes on your chart.
- — Like HLA-B*57:01, G6PD status is a one-time genetic check that flags drugs to avoid.
- — When the wrong drug or fava beans hit, G6PD breaks red cells fast — it's one of the hemolytic causes behind an unexplained anemia.
Substance and claimed effects
Glucose-6-phosphate dehydrogenase (G6PD) deficiency is the most common human enzyme defect, affecting an estimated 400–500 million people worldwide Cappellini and Fiorelli 2008Nkhoma et al. 2009. It is an X-linked recessive disorder of the G6PD gene on Xq28, with more than 200 known mutations producing functional variants of varying severity Luzzatto et al. 2023. G6PD catalyses the first step of the pentose phosphate (hexose monophosphate) pathway, generating NADPH; in mature red blood cells, which lack mitochondria and protein-synthesis machinery, this is the sole source of NADPH and therefore the sole engine for regenerating reduced glutathione that quenches reactive oxygen species. Without it, red cells cannot survive oxidative challenge. The entry covers the condition itself and the consequences that follow from it: vulnerability to acute haemolytic anaemia after exposure to oxidant drugs, fava beans, and certain infections; neonatal jaundice with kernicterus risk; chronic non-spherocytic haemolytic anaemia in severe variants; and the heterozygote advantage against falciparum malaria that explains the geographic distribution.
Evidence by addressing question
Mechanism
G6PD oxidises glucose-6-phosphate to 6-phosphogluconolactone, reducing NADP+ to NADPH. NADPH reduces oxidised glutathione (GSSG) to reduced glutathione (GSH), which in turn detoxifies hydrogen peroxide via glutathione peroxidase and reduces protein and membrane thiols oxidised by reactive oxygen species Luzzatto et al. 2020. Other cell types compensate for G6PD loss through NADPH from the malic enzyme and isocitrate dehydrogenase pathways; the mature erythrocyte cannot, because it lacks the nuclei and mitochondria to express those alternatives. When NADPH is depleted, haemoglobin is oxidised to methaemoglobin; thiol groups on globin chains crosslink and precipitate as Heinz bodies; the membrane skeleton becomes rigid; and the affected cells are cleared by the spleen, often haemolysing intravascularly when the oxidant load is high Cappellini and Fiorelli 2008.
The favism pathway is mechanistically distinct from the drug pathway only in the upstream chemistry. Fava beans contain the β-glucosides vicine and convicine at up to 2% of dry weight; intestinal hydrolysis releases divicine and isouramil, which redox-cycle in the bloodstream and generate superoxide and hydrogen peroxide that oxidise NADPH and GSH faster than the deficient cell can regenerate them Luzzatto et al. 2020. Cooking partially inactivates the glucosidases, which is part of why raw or undercooked beans carry the highest risk. Heinz-body anaemia after primaquine in 1956 was the inaugural pharmacogenetic syndrome and the discovery context for the enzyme itself.
Older red cells have the lowest residual G6PD activity (the enzyme decays through the cell's 120-day lifespan); acute haemolysis preferentially destroys them, leaving a younger population of cells with more residual enzyme. This explains the self-limited course of the A− variant: haemolysis is essentially over within 7–14 days even with the trigger still present, because the surviving cells have enough enzyme to handle the load Frank 2005. The Mediterranean variant has so little residual activity (~4%) that this rebound mechanism fails, producing more severe and prolonged crises Luzzatto et al. 2023.
Evidence
Epidemiology. A systematic review of 280 prevalence estimates from 88 countries reported a global allele frequency of approximately 4.9% with a wide range across regions, peaking above 20% in parts of sub-Saharan Africa, the Arabian Peninsula, southern Italy and Greece, the Thai–Lao border, and the Solomon Islands Nkhoma et al. 2009. A model-based geostatistical map estimated 330–400 million affected individuals in malaria-endemic countries alone Howes et al. 2012. In the United States, prevalence among Black males is approximately 10%; among males of Mediterranean, Middle Eastern, and Southeast Asian ancestry it ranges from 5–25% depending on subpopulation Frank 2005.
Drug-induced haemolysis. The discovery of G6PD deficiency followed observations in 1956 that primaquine produced haemolysis in a subset of African-American soldiers. A 2010 evidence-based review classified medications by quality of evidence and concluded that only a short list — dapsone, methylene blue, nitrofurantoin, phenazopyridine, primaquine, rasburicase, and tolonium chloride — carried unambiguous evidence of haemolytic risk; many drugs historically labelled as contraindicated had weak or absent supporting evidence, often deriving from in vitro assays or single case reports Youngster et al. 2010. Tafenoquine, a long-acting 8-aminoquinoline, is contraindicated below 70% G6PD activity per FDA labelling CDC/Haston et al. 2019. Rasburicase, used for tumour lysis syndrome, carries an FDA boxed warning for haemolysis and methaemoglobinaemia in G6PD-deficient patients FDA 2020.
Neonatal hyperbilirubinaemia and kernicterus. G6PD deficiency is a major risk factor for severe neonatal hyperbilirubinaemia, with risk arising from both increased haemoglobin breakdown and impaired bilirubin conjugation. A Sardinian screening programme that tested ~34,000 newborns found 18.4% G6PD-deficient (24.5% boys, 11.8% girls); zero developed kernicterus over the programme's follow-up, against a baseline rate of preventable cases in the years prior. The 2022 AAP guideline for hyperbilirubinaemia management names G6PD deficiency as one of the principal hyperbilirubinaemia neurotoxicity risk factors and recommends testing in any infant with hyperbilirubinaemia poorly responsive to phototherapy or with consistent family ancestry Kemper et al. 2022Kaplan and Hammerman 2010.
Malaria selection. G6PD deficiency confers heterozygous-female protection against severe Plasmodium falciparum malaria. In a Kenyan case-control and cohort study of >6,000 children, heterozygous G6PD A− females had reduced severe-malaria risk and reduced mortality, while hemizygous males showed no protection against severe malaria and possibly elevated risk of severe malarial anaemia Uyoga et al. 2015. An earlier landmark Nature paper documented natural selection acting on both hemi- and heterozygotes in Nigeria Ruwende et al. 1995; Luzzatto's 2015 commentary integrates the human-genetic evidence with a balanced-polymorphism model and concludes that female heterozygote advantage is the dominant selection mechanism, with hemizygous males paying the cost Luzzatto et al. 2020.
Protocol
The clinical action is single-step: test once, document the result, then apply a lifelong avoidance list. Quantitative spectrophotometric assay of erythrocyte G6PD activity (the reference standard) or fluorescent spot testing (a qualitative screen) makes the diagnosis Frank 2005. Both tests are unreliable during or immediately after an acute haemolytic episode, because the deficient older red cells have been destroyed and the residual younger population has near-normal activity; retesting 2–3 months after the crisis is the conventional workaround Cappellini and Fiorelli 2008. Point-of-care quantitative biosensors (e.g. SD Biosensor STANDARD G6PD; CareStart) now provide enzyme activity at the bedside within minutes and have WHO prequalification for use prior to primaquine and tafenoquine prescribing. Documentation matters as much as the test itself: medic-alert tags, electronic health record problem-list entries, and pharmacist awareness prevent the foreseeable failure mode of an emergency-room provider prescribing a sulfonamide or nitrofurantoin to an unidentified patient.
Contraindications (triggers to avoid)
The high-evidence drug list, drawn from Youngster et al. 2010 and contemporary FDA labelling: primaquine, tafenoquine (8-aminoquinoline antimalarials), dapsone (leprosy, dermatitis herpetiformis, Pneumocystis prophylaxis), rasburicase (tumour lysis prophylaxis — FDA boxed warning), methylene blue / methylthioninium chloride (methaemoglobinaemia, sentinel-node mapping — paradoxically the standard methaemoglobinaemia antidote that becomes contraindicated here), nitrofurantoin (urinary antiseptic), phenazopyridine (urinary analgesic), tolonium chloride / toluidine blue. Sulfonamides as a class are conventionally listed; real-world risk varies considerably across agents, with sulfamethoxazole/trimethoprim and sulfasalazine showing lower haemolytic risk than the historical label implies, while sulfanilamide and sulfacetamide retain clear association Youngster et al. 2010. Other agents commonly listed but with weaker evidence include high-dose aspirin (>1 g/day), high-dose vitamin C (intravenous gram-doses), chloramphenicol, and some quinolones; clinical risk depends on dose, variant severity, and concurrent oxidative load.
Non-drug triggers: raw or undercooked fava beans (mainly Mediterranean variant; Vicia faba exposure includes ingestion and inhalation of bloom pollen in classical case reports); naphthalene (mothballs); henna; infections, particularly viral hepatitis, typhoid, pneumonia, and severe sepsis — infection-triggered haemolysis is in fact the most common cause of crisis in many series Cappellini and Fiorelli 2008. Diabetic ketoacidosis can also precipitate haemolysis through endogenous oxidative stress.
Methaemoglobinaemia in a G6PD-deficient patient is a therapeutic conundrum: methylene blue, the standard antidote, is itself contraindicated. Ascorbic acid (intravenous), high-dose vitamin C combined with cobalamin, exchange transfusion, and hyperbaric oxygen are second-line options; onset is slower than methylene blue.
Misconceptions
Three common errors. First, G6PD deficiency is not a sulfa allergy — the mechanism is direct pharmacological oxidation of erythrocyte glutathione, not IgE-mediated hypersensitivity, and the clinical pattern (delayed haemolysis 24–72 hours after exposure, dark urine, jaundice) differs from urticaria/anaphylaxis Frank 2005. Second, the condition is not strictly "male-only." Female heterozygotes have a mosaic red-cell population due to X-inactivation; depending on the lyonisation pattern, they can have anywhere from 5% to 95% deficient red cells and can experience haemolytic crises indistinguishable from hemizygous males Luzzatto et al. 2020. The fluorescent spot test, which measures averaged enzyme activity, frequently misclassifies these women as normal. Third, fava beans are not always dangerous to anyone with G6PD deficiency: many people with the African A− variant eat them without incident, while the Mediterranean variant carries a much higher per-exposure risk. The conservative clinical position is total avoidance for any confirmed deficient individual, but the population-level reality is variable.
Audience
Highest carrier and hemizygote frequencies in: sub-Saharan African ancestry (African A−; ~10–20% male prevalence in many populations), Mediterranean ancestry (G6PD Mediterranean / B−; especially Sardinia, Greece, southern Italy, Cyprus), Middle Eastern Arab ancestry (~2–31% across countries with consanguinity-driven enrichment), and Southeast Asian populations (G6PD Mahidol, Viangchan, Kaiping; especially Thailand, Laos, Cambodia, southern China, the Philippines) Howes et al. 2012Nkhoma et al. 2009. Ashkenazi Jewish populations carry the Mediterranean variant at appreciable frequency. Affected ancestry alone is not a clinical diagnosis: New York State's universal newborn screening implementation found that 13% of identified deficient infants had parents identifying as white or Ashkenazi Jewish and 22% as Puerto Rican or Dominican — risk-factor-based screening would have missed roughly 44% of cases.
Failure modes
Most preventable harm in G6PD deficiency comes from documentation and recognition failures rather than from any inherent unmanageability of the condition. (1) The qualitative fluorescent spot test misclassifies female heterozygotes whose lyonisation pattern produces an averaged red-cell activity above the deficiency cutoff but whose deficient subpopulation can still haemolyse; quantitative testing is preferable when female status matters clinically Luzzatto et al. 2023. (2) Testing during acute haemolysis returns a false-normal result because the deficient cells have already been destroyed; the diagnosis is missed exactly when it is most clinically important, and patients are sent home thinking they are normal until the next crisis. (3) Emergency-department and urgent-care prescribing of nitrofurantoin (for cystitis), sulfamethoxazole-trimethoprim (for cellulitis or sinusitis), or dapsone (for various dermatologic indications) without history-taking is a routine trigger when status is undocumented or unknown to the patient. (4) Neonatal discharge before peak bilirubin (typically days 3–5) can miss kernicterus risk, particularly in infants with no obvious haemolysis but baseline impaired bilirubin clearance from deficient hepatic G6PD Kemper et al. 2022.
Stakes
Most G6PD-deficient people pass through life with no episodes, because they happen not to encounter a trigger and not to be prescribed a contraindicated drug. The minority who do encounter a trigger experience acute intravascular haemolysis 24–72 hours after exposure: jaundice, dark cola-coloured urine (haemoglobinuria), tachycardia, abdominal or back pain, and progressive weakness over days; severe crises require hospitalisation, transfusion, and occasionally cause acute kidney injury from haemoglobin nephropathy Frank 2005. Neonatal stakes are higher per event: untreated severe hyperbilirubinaemia produces kernicterus with mortality ~10% and long-term morbidity (deafness, cerebral palsy, intellectual impairment) in ~70% of survivors Kemper et al. 2022. The Mediterranean and severe Asian variants account for the bulk of fatal favism cases, historically a recognised public health problem in Sardinia and southern Italy before screening programmes.
Payoff
Documented status confers near-complete avoidance of foreseeable crises. The Sardinian screening programme essentially eradicated kernicterus from G6PD deficiency over two decades; Singapore similarly reports zero G6PD-attributable newborn deaths since universal screening implementation. For adults, the payoff is a stable problem-list entry that prevents the foreseeable prescribing mistake — a small ongoing administrative cost (telling new clinicians, pharmacists, dentists) in exchange for a low-probability, high-severity averted event.
The credibility range
Optimist case (the maximalist position on G6PD relevance). G6PD deficiency is one of the most prevalent inherited conditions on Earth, and the affected population is concentrated in precisely the regions where antimalarials are routinely prescribed. Universal newborn screening is cheap, fast, and demonstrably eliminates kernicterus where implemented; the failure of high-income countries to adopt it is regulatory inertia rather than scientific debate. The drug list is firmly established for the high-evidence agents, and every emergency department should treat G6PD status as a routine vital piece of history. The condition is benign only conditionally — it becomes lethal in the presence of a trigger, and our pharmacopoeia continues to add candidate triggers (rasburicase, tafenoquine, immunotherapy combinations).
Skeptic case (the deflationary position). Most G6PD-deficient people are asymptomatic their entire lives. Many drugs historically labelled as contraindicated have weak supporting evidence — single case reports, in vitro assays, mechanistic plausibility rather than clinical events Youngster et al. 2010. The drug-induced haemolysis literature predates modern pharmacovigilance and conflates real triggers with co-administered drugs. Universal newborn screening produces high false-positive rates, generates familial anxiety, and most identified infants would never have had a crisis. Sulfa drugs in real-world data carry less haemolytic risk than their label implies, and methylene blue's contraindication is routinely overridden in emergencies without obvious harm. The condition is genuinely benign for most people; the clinical-attention budget is better spent elsewhere.
Author's call. The deflationary view captures something real — overcaution on weak-evidence drugs and on universal screening in low-prevalence populations is overcaution. The high-evidence drug list (the seven absolute contraindications) is the load-bearing piece and should be treated with bright-line discipline. Beyond that list, prescribing decisions should weigh dose, variant severity, and clinical alternative. Documentation of status remains the highest-leverage action: it costs essentially nothing once known, and it converts a high-severity unpredictable event into a managed problem. Newborn screening in high-prevalence populations passes any sensible cost-benefit threshold; in low-prevalence populations the case is debated but trending toward universal as point-of-care quantitative testing becomes cheaper. Evidence quality is high overall; controversy is low and concentrated at the margins of the drug list and screening policy.
Stakeholder and incentive map
- Antimalarial programmes — the dominant institutional stakeholder. WHO, national malaria control programmes, MMV (Medicines for Malaria Venture), and PATH have driven point-of-care G6PD testing adoption to enable safe primaquine and tafenoquine use for P. vivax radical cure. Incentive: end malaria transmission; cannot use the only relapse-prevention drugs without knowing G6PD status.
- Newborn screening programmes — Greece, Israel, Singapore, Sardinia, Philippines, and increasingly U.S. states have G6PD on the panel; the AAP and AAFP advocate selective testing in high-risk infants. Incentive: prevent kernicterus.
- Pharmacovigilance and drug labelling — FDA, EMA, NICE maintain contraindication labels; the rasburicase boxed warning is a notable post-marketing addition FDA 2020. Incentive: liability and regulatory safety.
- Patient advocacy — small but established (G6PD Deficiency Foundation, regional favism societies). Influence is modest given the condition's low public profile in the U.S.
- Counter-pressure — there is no significant commercial counter-incentive. Drug manufacturers occasionally push back on listing of marginal agents but generally accept G6PD labelling. The deflationary pressure is mostly academic — calls for restricting the drug list to high-evidence agents to avoid undertreatment of common infections.
Population variability
- Variant severity dominates clinical presentation. WHO Class A (chronic non-spherocytic haemolytic anaemia, no triggers required) is rare and severe; Class B (the common polymorphic variants, including African A− and Mediterranean) is overwhelmingly the clinical reality and produces episodic trigger-dependent haemolysis Luzzatto et al. 2023. Within Class B, the Mediterranean variant has ~4% residual activity and is associated with favism; African A− has ~12–20% residual activity and a milder, more self-limited pattern.
- Sex. Hemizygous males express the variant uniformly. Heterozygous females have mosaic red-cell populations from X-inactivation, producing a continuous distribution of effective enzyme activity. The conventional dichotomy "carrier vs. affected" understates female risk; women with skewed X-inactivation can be clinically affected and can be missed by qualitative screening Luzzatto et al. 2023.
- Age. Neonates are uniquely vulnerable to hyperbilirubinaemia regardless of haemolysis, because hepatic G6PD is also reduced and bilirubin conjugation is impaired Kemper et al. 2022. Older adults can present with first crisis in late life, often triggered by a newly prescribed urinary antibiotic.
- Concurrent oxidative load. Infection, diabetic ketoacidosis, and combinations of mild oxidant exposures can summate beyond what any single trigger would produce. The same drug at the same dose can be tolerated once and trigger haemolysis another time depending on background.
Knowledge gaps
- Many drugs on conventional avoidance lists have weak underlying evidence; head-to-head pharmacovigilance data are sparse, and the field has not systematically re-evaluated the historical list with modern methods Youngster et al. 2010.
- The real-world incidence of clinically significant haemolysis from sulfamethoxazole-trimethoprim and similar agents is uncertain; reported case rates are far below what the contraindication label suggests.
- Whether universal newborn screening in low-prevalence populations passes a cost-benefit threshold is contested; the relevant trial would be expensive and is unlikely to be run.
- The Mediterranean variant's selective advantage against P. vivax is suggestive but not as firmly established as the African variant's advantage against P. falciparum.
- Mechanisms of variation in trigger sensitivity — why the same person can eat fava beans without incident in one exposure and crash in another — remain unexplained.
Scope coverage relative to the brief. The input description named four threads — red-cell vulnerability to oxidative triggers (drugs and foods), hemolysis risk, and prescribing precautions. All four are covered end to end (mechanism, evidence, contraindications, failure-modes). No narrowing.
Category placement. Filed under medical rather than screening. The blood test is single-event and the lifelong work is medication-aware living — a clinical-condition pattern more than a recurring screening regimen. Re-filing to screening would be defensible; flagging in case a reviewer prefers it.
Drug list framing. Led with the high-evidence short list (seven agents per Youngster et al. 2010) rather than the traditional long list. This is mildly contrarian — many clinical references still publish 30+ drug lists. The risk of the conservative long list is that it leads to under-treatment of common infections; the risk of the short list is that an idiosyncratic case from a weak-evidence agent gets missed. Chose to anchor on the evidence-based review and explicitly note that sulfa drugs and weaker-evidence agents need pharmacist input. Reviewer call: would also accept a longer warning list if the editorial preference is maximally conservative.
Audience scoping decisions. Did not narrow audience.gender to male despite X-linked inheritance. Female heterozygotes can be clinically affected through skewed X-inactivation; the fluorescent spot test misses many of them. Narrowing to male would mis-signal that women in high-prevalence families are safe. The audience addressing section names the four high-prevalence ancestries explicitly within the body.
Rating difficulties. Most benefit dimensions are 0 because the condition itself confers no benefit. The non-zero scores (health_short_term 3, longevity 2) reflect the avoided crisis when status is known and acted on, not a positive effect of having the gene. Considered scoring 0 across the board with a high evidence rating, but that would understate the practical health stakes; landed on the avoided-harm framing.
Future links to wire in once they exist.
- Sickle cell trait — geographic and ancestral overlap, same one-time-knowledge action pattern
- Pharmacogenetic testing more broadly (CYP2D6, TPMT, DPYD) — same actionable-genetics frame
- Lp(a) testing — different topic, same one-test-once-life pattern that this entry exemplifies
- Newborn screening — if the catalogue grows a parent-facing entry on what the heel-prick panel covers
Separate-entry candidates surfaced during the write.
- Methylene blue and methaemoglobinaemia management — covered briefly here, but the broader topic (toxic exposures, blue-baby syndrome, anaesthetic complications) could warrant its own entry under medical
- Fava beans / favism — interesting as a folkloric-to-genetic story; could be its own short entry under food, cross-linked from here
- Point-of-care diagnostics — the G6PD biosensor story is one example of a broader shift; could anchor a productivity- or medical-category entry
Action and cadence. know + once. Considered test + once since the test is the gating step, but the substantive lifelong work is the knowing and the documentation, not the testing itself. know better captures that the reader's job is awareness rather than recurring measurement.
G6PD Deficiency
One quantitative G6PD assay (~$30–100, often covered by insurance when ordered for clinical reason); no recurring cost. Point-of-care biosensors are reducing this further.
One blood draw, then a small ongoing administrative load: telling new clinicians, pharmacists, and dentists; reading drug labels for the high-evidence trigger list; avoiding raw fava beans. Not daily effort.
Most common human enzyme defect with decades of literature (Cappellini and Fiorelli 2008; Luzzatto et al. 2020), WHO classification updated 2022 (Luzzatto et al. 2023), FDA boxed warnings on rasburicase (FDA 2020) and tafenoquine (CDC/Haston et al. 2019), AAP 2022 hyperbilirubinaemia guideline naming G6PD as a principal neurotoxicity risk factor (Kemper et al. 2022), 280-study global prevalence meta-analysis (Nkhoma et al. 2009).
Untriggered, the condition is silent. When a contraindicated drug or fava beans triggers crisis, intravascular haemolysis produces jaundice, haemoglobinuria, weakness, and hospitalisation within 24–72 hours (Frank 2005); documenting status converts a high-severity unpredictable event into a managed one. Score reflects the avoided crisis, not the baseline state.
Neonatal kernicterus carries ~10% mortality and ~70% long-term morbidity in survivors; documented G6PD status with phototherapy and transfusion protocols essentially eliminates this outcome where screening is in place (Kemper et al. 2022). Adult haemolytic crises are rarely fatal but acute kidney injury from haemoglobin nephropathy is a real mortality contributor in severe variants.