The win is the absence of a call. No 3am hospital ride, no transplant-year setback, no toddler in a dialysis chair, no late-pregnancy emergency that started as a sandwich. The cost is near-zero — pasteurized swaps cost what raw costs, a fridge thermometer is ten dollars once. The effort is real but bounded: a learned shopping list, a reflex at restaurant menus, a habit of reheating deli meat until it steams. The catch is that the win is invisible by design — you only ever see the version of your year where nothing went wrong.
The food system is calibrated for an average healthy adult — a stomach acidic enough to kill most of what slips through, a placenta you don't have, a gut barrier that closes by your second birthday, an immune system not running at half throttle. For most adults that's enough. The Listeria on a deli slicer, the Salmonella in a soft-boiled egg, the Vibrio vulnificus in a Gulf oyster — the body usually clears them before you notice.
In four populations, that margin disappears. Pregnancy shifts the immune system to protect the fetus, and one of the costs is dropping the intracellular-pathogen defence the body uses against Listeria; the same bug that gives an average adult a brief fever can cross the placenta and end the pregnancy. After 65 the immune system slows and stomach acid drops — often further on a PPI for reflux — and the dose of a pathogen needed to make you sick falls. Children under five have a more permeable gut, no prior exposure, and a body small enough that fluid loss escalates fast. Anyone on chemotherapy, a TNF-alpha biologic, a transplant immunosuppressive, or a long course of steroids is running with the gate open. In all four, the same foods that produce a forgettable evening for someone else produce hospitalisation, organ failure, or death.
The pathogens that matter here have a particular character: they survive cold, hide in foods that look and smell normal, and can cause invasive disease — bloodstream, brain, placenta, kidney — not just gut symptoms. Listeria grows in the back of a refrigerator. Salmonella sits in a Caesar dressing for days. Shiga-toxin E. coli survives in a medium-rare burger and goes to a four-year-old's kidneys. Cronobacter sakazakii contaminates the powdered formula a parent is mixing one-handed at 2am. The thing they share isn't taste or smell or appearance. It's that for the average eater, almost nothing happens; for a vulnerable one, the worst-case is on the table.
Which group you're in changes which foods matter most
The closed list is mostly the same across the four groups — but the headline pathogen, and the items that would otherwise look like edge cases, shift.
Pregnant. The defining pathogen is Listeria, which crosses the placenta and produces stillbirth, preterm delivery, or neonatal meningitis with a combined fetal-or-neonatal-death rate around 20–30% across case series (Mateus et al. 2013; ACOG 2014). Deli meats and hot dogs not reheated to steaming, soft cheeses from unpasteurized milk, refrigerated smoked salmon, refrigerated pâté, and pre-cut melon held more than a few days are the canonical vehicles. The second concern is Toxoplasma gondii from undercooked pork, lamb, and venison, and from unwashed produce — a first-trimester infection that crosses the placenta is rarer but more devastating than a third-trimester one (Dubey and Jones 2008). The pregnancy list is the most extensive of the four.
Over 65. The risk profile broadens: Listeria still matters, plus Salmonella, Campylobacter, norovirus, and — if you live near the Gulf Coast, Atlantic, or any warm-water region — Vibrio from raw oysters. Reduced stomach acid is a quiet factor: PPIs and H2-blockers for reflux, age-related acid decline, and post-bariatric anatomy all lower the pathogen dose your stomach can neutralise, so the same plate of food carries more risk than it did at 40 (Lund and O'Brien 2011).
Under five (and especially under one). The headline danger is Shiga-toxin E. coli from undercooked ground beef, raw milk, and contaminated produce — the path to haemolytic uraemic syndrome. Salmonella and Campylobacter sit close behind. For infants under two months, the additional risk is Cronobacter from powdered formula — powdered formula is not sterile, and the bug grows fast in reconstituted formula at room temperature. Ready-to-feed liquid formula is sterile and is the safer default for the highest-risk infants (CDC 2024c; AAP 2021).
Immunocompromised. Transplant recipients carry roughly 50–100 times the listeriosis rate of the general population; biologic and steroid users sit lower but still well above baseline (Pouillot et al. 2012). Active chemotherapy, advanced HIV with low CD4, and end-stage renal disease pull a similar list. Chronic liver disease — cirrhosis, hepatitis B or C, hereditary haemochromatosis — deserves its own line: raw oysters are categorically off the table because Vibrio vulnificus uses iron-overload conditions to invade and kills about half the people it reaches in this group (Jones et al. 2014).
If you're caring for someone in one of these groups — feeding a toddler, cooking for an older parent, shopping for a partner mid-chemo — you're in the audience too. The decision happens in the grocery aisle and the restaurant menu, which is wherever the food is bought, not eaten.
What the dangerous version of the year looks like
The stakes here aren't cumulative. There's no slow trajectory, no biomarker that drifts wrong for years. A single Tuesday turkey sandwich. A single oyster on a Friday. A single pink burger at a birthday party. The next time the reader's calendar shows ordinary, the version where they ate the wrong thing has them in a hospital room instead.
For a third-trimester reader, the worst case starts as a low-grade fever and body aches — flu-like, easy to dismiss. By the time it's contractions, the placenta is already involved; by the time the obstetrician has the lab back, the baby has either been delivered emergently or won't be. People in this group don't forget what they ate that week; they catalogue it.
For a transplant recipient or someone in active chemotherapy, the worst case is a sudden fever and confusion that turns into a meningitis admission within 12 hours. Listeria meningitis kills about a quarter of the people it reaches. The rest get out of the hospital on a long course of IV antibiotics, often with a permanent line, and a transplant team that now files them as higher-risk for the rest of treatment.
For the parent of a four-year-old, the worst case starts as bloody diarrhoea on a Sunday afternoon. By Wednesday the child is pale and barely peeing; the platelets are dropping. Haemolytic uraemic syndrome is the kind of thing the family will talk about for the rest of the child's life — the dialysis catheter, the weeks in the unit, the kidney function that didn't quite come back (Tarr et al. 2005). About 15% of confirmed Shiga-toxin E. coli infections in young children take this path.
For an older adult with cirrhosis, the worst case is septic shock starting within 24 hours of eating a raw oyster. Half of the people in this group who get Vibrio vulnificus die. The ones who survive often lose tissue — the bug produces necrotising skin and soft-tissue infections that look like a flesh-eating bacteria headline (Jones et al. 2014).
None of this happens often. The U.S. has roughly 1,600 listeriosis cases a year, ~100 vibriosis deaths, ~30 STEC deaths, around 1.35 million Salmonella illnesses (Scallan et al. 2011). At the population level the per-meal odds look small. The asymmetry is what makes the entry worth the chore: the cost of skipping a Caesar salad is zero, the cost of skipping the wrong one is your year.
The closed list — what to skip, what to heat
The list below is the convergent guidance from the CDC, FDA, USDA-FSIS, ACOG, AAP, and WHO (CDC 2024; FDA 2024a; ACOG 2014). It's short. Memorise it once and the cost of compliance drops to almost nothing.
Then the handling adaptations. Two of them — the thermometer and the fridge temperature — change the protocol's failure rate by more than any individual food on the list.
That's the entire protocol. Outside this list, eat normally. The point of memorising it is so that the rest of the year stops being a series of decisions and goes back to being meals.
The myths that put people in the hospital
- "Raw milk has more nutrients." Pasteurization holds milk at 72 °C for 15 seconds. That doesn't measurably degrade calcium, protein, fat, or most vitamins. What it does kill is Salmonella, Campylobacter, E. coli, and Listeria. A 14-year U.S. review found roughly 150 times the outbreak rate per unit consumed for raw versus pasteurized dairy (Langer et al. 2012). There is no nutritional case strong enough to clear that gap for anyone in a vulnerable group.
- "If it smells fine, it's fine." Listeria, Salmonella, and Shiga-toxin E. coli don't produce off-odours or visible spoilage at the populations that make a vulnerable person sick. Smell is a check for spoilage organisms, not pathogens. The deli sandwich that ends the pregnancy will smell normal.
- "Freezing kills bacteria." Freezing mostly pauses bacteria — they wake up unchanged when the food thaws. The one common exception is Toxoplasma, whose tissue cysts in meat die after at least three days below −12 °C; that's a useful trick for game meat in pregnancy, not a general rule.
- "Lemon juice (or vinegar, or alcohol) cooks fish." Acid changes the texture but does not reliably kill Vibrio, norovirus, or parasites. Ceviche is raw fish for risk purposes.
- "Hard cheeses from raw milk are safe in pregnancy." Long-aged hard cheeses — true Parmigiano-Reggiano, aged cheddar, aged gouda — carry low Listeria risk because the cheese is dry and acidic enough that the bug struggles to grow. The guidelines from the CDC and ACOG still say pasteurized-only in pregnancy, because the downside of a single infection is large and the upside of an aged cheese is small. The conservative call is the right call here.
- "My fridge is cold enough." Roughly a quarter to a third of home refrigerators run above 4 °C. A $10 thermometer is the cheapest food-safety intervention in this entry.
- "It was only a tiny taste." Listeria's infectious dose in a vulnerable host is low — a few thousand organisms, easy to reach in a single bite of contaminated cheese. The "I just had a little" defence doesn't translate into "I'm safe."
Where the protocol breaks in real households
The list is short and the science is settled, but the failures are predictable. Knowing where the protocol breaks is half the work of keeping it.
- Restaurants are the blind spot. Caesar dressing, hollandaise, tiramisu, raw-yolk pasta sauces, and tartares are routine on menus and never flagged. Soft cheeses on charcuterie boards are usually unlabelled. Most pregnancy and immunocompromised exposures happen in kitchens the eater can't inspect. The reflex worth building: read the menu like a list of pathogens, ask if anything is uncertain, accept that a few favourites are off the table for a stretch.
- Deli-counter cross-contamination. Even pasteurized sliced meats pick up Listeria from the slicer biofilm; cleaning frequency varies wildly between stores. The reheat-to-steaming rule is the safer default than relying on any individual slice being clean.
- "Made with pasteurized milk" reading. European-style cheeses at farmers' markets and small delis often lack a clear label. Queso fresco from informal markets and unlabelled imports has caused multiple outbreaks. If the label doesn't say it, the cheese is off the list for the window.
- Powdered formula at room temperature. Cronobacter grows fast in reconstituted formula left out for a feeding round. The two-hour cumulative limit is real, and for an infant under two months the safer default is to switch to ready-to-feed liquid formula for that window (CDC 2024c).
- Time stacking. The grocery checkout, the car ride home, the counter while you set the table, the buffet warmer that didn't get warm enough — each segment counts toward the four-hour cumulative limit. The rule is cumulative across the food's life, not per-stop.
- Kitchen drift in older-adult households. Cohort studies find refrigerators unmeasured, thermometers absent, leftover-age tracking gone — the protocol fails before it gets tested (Lund and O'Brien 2011). If the reader is over 65 living alone or shopping for a parent who is, the highest-leverage intervention is a $10 fridge thermometer and a sticky note on the fridge door with the four big rules.
- The post-trip slip. A holiday, a wedding, a family gathering — somebody's well-meaning soft cheese, somebody's homemade tiramisu. The vulnerable window is the one to be most boring at; an aunt's "but I made it myself" is exactly the kind of source that has no idea what was in the unpasteurized milk.
What the boring version of the year looks like
The payoff is the absence of a story. Nothing happens. The third trimester runs to delivery on its own timeline. The first transplant year ticks past its anniversary without an unplanned admission. The toddler gets the usual run of daycare stomach bugs that resolve on the couch in two days. The 78-year-old grandparent comes back from the cruise with photos, not a hospital bracelet.
None of this looks like anything. The reader who skipped the deli sandwich never finds out which sandwich would have been the wrong one — that's not how counterfactuals work. What they get is a quiet year that they otherwise might not have had, and after a few weeks the protocol stops being a series of conscious decisions and becomes how their kitchen runs.
The cognitive payoff arrives sooner than the medical one. The first time a reader hits a restaurant menu after they've memorised the list, the decision happens in under a second: this side is fine, the dressing on it isn't, ask for the vinaigrette. The first time they reach the deli counter, they're already thinking about whether the sandwich is going to be reheated or not. The mental load of a high-stakes window — pregnancy, post-transplant, a child in the house — is heavy enough without re-deciding food at every meal. Decision-loading the list once buys back the bandwidth.
When the window ends — the baby is born, the transplant year clears, the child turns five, the chemo cycle finishes — most readers just go back to eating normally. A few keep the thermometer habit and the fridge dial set right; those are the cheapest possible carry-overs and worth keeping. The rest of the protocol relaxes back to the average-adult bar.
Adjacent topics the reader may want to look at separately: methylmercury and fish choices in pregnancy; travellers' diarrhoea and water-source contamination abroad; foodborne botulism from home canning; chemical and allergen contamination of foods; general food poisoning in healthy adults (lower-stakes, different protocol); fridge organisation and leftover storage as a household practice. Each carries its own list and its own logic; this entry covers only the pathogen risk specific to vulnerable groups.
Substance and claimed effects
The substance of this entry is the class of high-risk foods for vulnerable groups — foods that carry an elevated probability of transmitting an enteric or invasive pathogen to readers whose host defences are diminished. Four populations bear almost all of the excess hospitalization and mortality from foodborne disease in the United States: adults aged 65 or older, children under five (and especially infants under one), pregnant people, and the immunocompromised (organ-transplant recipients, active cancer chemotherapy, advanced HIV, biologic and steroid users, chronic liver disease, severe diabetes, end-stage renal disease) Scallan et al. 2011, CDC 2024. The claimed effect is large and population-wide: avoiding a closed list of specific foods (raw or unpasteurized dairy, deli meats and refrigerated pâté not reheated, soft cheeses from unpasteurized milk, raw sprouts, raw oysters and other raw molluscs, undercooked eggs, undercooked poultry and ground meat, refrigerated smoked seafood, unpasteurized juice, raw cookie dough, melon held more than seven days after cutting, powdered infant formula in infants under two months) plus a small handful of handling adaptations (reheat-to-steaming, internal-temperature thermometer use, the four-hour rule, refrigerator at 4 °C, separate-raw-and-cooked) collapses the listeriosis, salmonellosis, STEC-haemolytic uraemic syndrome, vibriosis, toxoplasmosis, and cronobacter rates these readers face in proportion to baseline incidence in their group Silk et al. 2012, Pouillot et al. 2012, Lund and O'Brien 2011. The claim is not preventive in the cardiometabolic sense (no slow trajectory bent); it is risk-reduction against acute, often catastrophic events — stillbirth, meningitis, dialysis, sepsis, death — whose individual-incident rate is low but whose conditional severity in the vulnerable groups is extreme.
Evidence by addressing question
mechanism
Five mechanism stories carry the entry, one per major pathogen family.
Listeria monocytogenes is the headline pathogen because it grows at refrigerator temperatures (down to roughly −0.4 °C), survives in moist deli-counter biofilms, and crosses the placenta and the blood-brain barrier. Pregnant people are infected at about 13 times the general-population rate and account for ~17% of all listeriosis cases despite being ~5% of the population; in the infected pregnancy the bacterium colonises the placenta and reaches the fetus, producing fetal loss, preterm delivery, or neonatal meningitis with a combined fetal-or-neonatal-death rate of roughly 20–30% across case series Silk et al. 2012, Mateus et al. 2013, ACOG 2014. In non-pregnant adults Listeria targets the central nervous system, producing meningitis or bacteraemia with a case-fatality of ~20–30%; risk rises steeply with age and immunosuppression, with adults over 65 carrying ~4× the listeriosis rate and transplant recipients ~50–100× Pouillot et al. 2012.
Salmonella, Campylobacter, and Shiga-toxin-producing E. coli (STEC) are the high-volume causes. Salmonella alone produces ~1.35 million U.S. illnesses and ~26,000 hospitalizations a year, concentrated in children under five and adults over 65 Scallan et al. 2011. STEC O157:H7 is the front-of-mind paediatric danger: in children under five, about 15% of confirmed O157:H7 infections progress to haemolytic uraemic syndrome (HUS) — microangiopathic haemolytic anaemia, thrombocytopaenia, acute kidney injury — with substantial dialysis and lifelong renal-function risk Tarr et al. 2005. The pathophysiology is Shiga-toxin endothelial damage to glomerular capillaries; antibiotics are contraindicated because they increase toxin release.
Vibrio vulnificus in raw oysters seeds an unusually concentrated mechanism: the bacterium requires a host with low gastric acidity or iron-overload conditions to invade beyond the gut. In adults with chronic liver disease (cirrhosis, hepatitis B/C, haemochromatosis) the organism enters the bloodstream from oyster ingestion and produces necrotising septicaemia with a case-fatality near 50% — among the highest of any foodborne pathogen Jones et al. 2014.
Toxoplasma gondii reaches humans through tissue cysts in undercooked meat (especially pork, lamb, venison) and through oocysts on unwashed produce and from cat litter. In pregnancy, primary maternal infection crosses the placenta in ~30% of cases, with first-trimester infection producing the most severe congenital sequelae (chorioretinitis, hydrocephalus, intracranial calcifications); estimated U.S. seroprevalence in reproductive-aged women is ~9–11% Dubey and Jones 2008.
Cronobacter sakazakii contaminates powdered infant formula (a non-sterile product) and produces neonatal meningitis and necrotising enterocolitis with case-fatality of ~40% in infants under two months; the organism cannot be excluded from powdered formula at manufacture, so the prevention path is hot-water reconstitution (≥70 °C kills the organism) and prompt feeding CDC 2024c, Payne and Paglione 2018, AAP 2021.
The common-thread mechanism across vulnerable groups is host-defence shortfall: physiologic immune downregulation in pregnancy (Th1→Th2 shift protects the fetus but degrades intracellular-pathogen clearance), immunosenescence and reduced gastric acid in older adults, immature gut barrier and underdeveloped adaptive immunity in young children, and direct iatrogenic immunosuppression in transplant and biologic patients Lund and O'Brien 2011.
evidence
The epidemiology is unusually clean. Scallan et al.'s burden estimate — ~48 million U.S. foodborne illnesses, 128,000 hospitalizations, 3,000 deaths annually — is the canonical denominator and breaks down by pathogen, food vehicle, and severity Scallan et al. 2011. Pouillot et al.'s FoodNet age-stratified analysis quantifies the listeriosis incidence-rate ratio at ~13 for pregnant women, ~4 for adults 65+, and several-fold higher again within the 65+ band for those with cancer, transplant, or end-stage renal disease Pouillot et al. 2012. Painter et al.'s outbreak-attribution analysis of 1998–2008 U.S. data assigns the largest share of foodborne deaths to dairy (Listeria), poultry (Salmonella, Campylobacter), and produce (STEC, Salmonella), with seafood third for Vibrio mortality Painter et al. 2013. Langer et al.'s 1993–2006 unpasteurized-dairy review found nonpasteurized products responsible for ~150× the outbreak rate per consumption-adjusted unit compared with pasteurized — the foundational evidence behind the unpasteurized-milk-and-soft-cheese rule Langer et al. 2012. The cantaloupe-Listeria mechanism is well-documented from outbreak investigations including the 2011 Jensen Farms outbreak (33 deaths, one fetal loss) and 2023–2024 cantaloupe Salmonella outbreaks Pouget et al. 2019.
The handling-side evidence is consistent. Time-temperature kinetics for poultry are well-characterised: holding internal temperature at 74 °C (165 °F) for 15 seconds achieves a 7-log Salmonella reduction; ground meat requires 71 °C (160 °F) end-point to compensate for surface-pathogen distribution; whole-muscle beef tolerates a lower 63 °C with rest because pathogens are surface-only USDA 2020, FDA 2024b. Reheating ready-to-eat meats to ≥74 °C reliably kills Listeria, the basis for the deli-meat reheat rule FDA 2024a. The "danger zone" 4 °C–60 °C and two-hour cumulative rule come from generation-time data on mesophilic foodborne organisms; the 70 °C reconstitution temperature for powdered formula is calibrated to Cronobacter thermal-death kinetics CDC 2024c.
Behaviour-change evidence is weaker. RCTs of food-safety education show modest knowledge gains and smaller behaviour gains (5–15 percentage points on observed-behaviour endpoints), and self-reported handling in older-adult cohorts shows persistent gaps — refrigerator temperatures often above 4 °C, no thermometer use, leftovers held beyond safe windows Powell et al. 2011, Lund and O'Brien 2011. The mechanism-side evidence is much stronger than the behaviour-change-side evidence — we know exactly which foods are dangerous and why; we know less about how to make households reliably avoid them.
protocol
The closed list of foods that the FDA, CDC, ACOG, AAP, and WHO position as "avoid entirely" or "heat first" for vulnerable groups converges across all five bodies. The avoidance list:
- Unpasteurized (raw) milk and any cheese, yogurt, or ice cream made from it. Soft cheeses (brie, camembert, feta, blue, queso fresco / blanco / panela) unless the label explicitly says "made with pasteurized milk." Hard, aged cheeses from raw milk are lower-risk but still excluded for pregnancy and severe immunosuppression Langer et al. 2012, FDA 2024a.
- Deli meats, hot dogs, and refrigerated pâté or meat spreads — unless reheated to steaming (≥74 °C internal). Shelf-stable pâté in cans is fine ACOG 2014.
- Refrigerated smoked seafood (smoked salmon, trout, whitefish labelled "nova-style", "lox", "kippered", or "jerky") unless cooked into a dish reaching ≥74 °C. Shelf-stable canned or pouched is fine.
- Raw or undercooked seafood — raw oysters and other raw molluscs especially (Vibrio, norovirus, hepatitis A); sushi and ceviche; raw clams; partially cooked shrimp Jones et al. 2014.
- Raw or undercooked eggs — runny yolks, soft-boiled, hollandaise, Caesar dressing, homemade mayonnaise, mousse, tiramisu, raw cookie dough/batter, eggnog. Pasteurized in-shell eggs sidestep this.
- Raw or undercooked meat and poultry — including steak tartare, carpaccio, rare burgers, undercooked sausage, pink chicken.
- Raw sprouts (alfalfa, mung bean, clover, radish) — the warm-moist sprouting environment amplifies any seed-borne Salmonella or E. coli; sprouts have produced ≥50 documented outbreaks since 1990. Cook thoroughly or skip.
- Unpasteurized juice and cider — must be heat-treated; look for the warning label.
- Cut melon (especially cantaloupe) — Listeria and Salmonella colonise the rind and transfer to the flesh on cutting; cut melon held above 4 °C for more than 4 hours or refrigerated more than 7 days is unsafe Pouget et al. 2019.
- Powdered infant formula for infants under two months, premature infants, or immunocompromised infants — reconstitute with water at ≥70 °C and use within 2 hours; ready-to-feed liquid formula is sterile and preferred for the highest-risk infants AAP 2021, CDC 2024c.
Handling adaptations:
- Use a food thermometer. Whole cuts of beef, pork, lamb, veal: 63 °C (145 °F) with 3-minute rest. Ground meats: 71 °C (160 °F). Poultry (whole and ground): 74 °C (165 °F). Reheated leftovers and ready-to-eat meats: 74 °C (165 °F) until steaming USDA 2020.
- Refrigerator at ≤4 °C, freezer at ≤−18 °C. Verify with a fridge thermometer; many home fridges run warm.
- Two-hour rule (one hour above 32 °C ambient): perishable food out of refrigeration accumulates time toward the four-hour cumulative limit; past that, discard.
- Separate raw and cooked — separate boards, separate utensils, separate fridge zones; raw meat on the bottom shelf.
- Wash hands, surfaces, and produce. Twenty-second soap-and-water hand-wash before handling food and after handling raw meat, eggs, or unwashed produce; rinse all produce under running water before eating, including melons before cutting WHO 2006.
- Leftovers — refrigerate within 2 hours; consume within 3–4 days for most items; reheat to 74 °C until steaming.
contraindications
No contraindications to the substance itself (food safety is universally net-beneficial). The reverse is the relevant frame: the entry's recommendations are *more* binding under specific contraindications to the vulnerable-group definition — pregnancy throughout, chemotherapy, transplant immunosuppression, biologics including TNF-α inhibitors, chronic systemic steroid use, advanced HIV with CD4<200, chronic liver disease, end-stage renal disease, haemochromatosis (specifically raises Vibrio risk via iron overload). These are not contraindications to the entry — they are intensifiers Jones et al. 2014, Lund and O'Brien 2011.
audience
Four core audiences with different highest-risk pathogens:
- Pregnant people. Listeria (13× rate, ~20–30% fetal/neonatal death) and Toxoplasma (congenital sequelae from first-trimester infection) dominate. The avoidance list is most extensive here, and the cost of a slip is highest Silk et al. 2012, ACOG 2014, Dubey and Jones 2008.
- Adults 65+. Listeria (~4× rate, sharply higher with comorbidity), Salmonella, Vibrio in shellfish, and norovirus dominate. Reduced gastric acid (often pharmacologic — PPIs), immunosenescence, and higher comorbidity burden raise the dose-response curve Pouillot et al. 2012.
- Children under five (especially infants under one). STEC-HUS, Salmonella, and Campylobacter dominate; Cronobacter in infants under two months. The child's gut barrier is more permeable, the immune repertoire is naïve, and dehydration risk is steeper Tarr et al. 2005, CDC 2024c.
- Immunocompromised adults. Listeria, invasive Salmonella, and (in chronic liver disease) Vibrio dominate. Transplant recipients carry 50–100× the listeriosis rate of the general population Pouillot et al. 2012.
misconceptions
Several myths actively cause illness:
- "Raw milk has more nutrients." Pasteurization (72 °C for 15 seconds) does not measurably degrade calcium, protein, fat, or most vitamins; the 1993–2006 outbreak data show ~150× the outbreak rate per unit consumed for raw versus pasteurized dairy, dominated by Salmonella, Campylobacter, STEC, and Listeria Langer et al. 2012.
- "If it smells fine, it's fine." Listeria, Salmonella, and STEC don't produce off-odours or visible spoilage at populations sufficient to make a vulnerable person sick.
- "Freezing kills bacteria." Freezing suspends most bacteria but does not kill them at populations that matter; thawed food is as contaminated as the food going in. Toxoplasma tissue cysts are killed by freezing meat below −12 °C for at least 3 days; this is the one common exception.
- "Lemon juice / vinegar / alcohol cooks fish." Ceviche acid denatures protein but does not reliably kill Vibrio, norovirus, or parasites.
- "Hard cheeses from raw milk are fine in pregnancy." Aged hard cheeses (Parmigiano-Reggiano, aged cheddar, aged gouda) carry low Listeria risk because of low water activity and acid; CDC and ACOG nonetheless recommend pasteurized-only in pregnancy out of cost-benefit caution given the catastrophic downside of a single infection.
- "My fridge is cold enough." Surveys consistently find ~20–30% of household refrigerators running above 4 °C; Listeria grows at 4 °C and accelerates above it.
failure-modes
Where the protocol fails in real households:
- Restaurant blind spot. Caesar dressing, hollandaise, tiramisu, raw-yolk pasta sauces, and tartares are routine on menus and not flagged. Most pregnancy and immunocompromised exposures happen at restaurants where the eater cannot inspect the kitchen.
- "Made with pasteurized milk" reading. European-style cheeses sold at delis and farmers' markets often lack a clear label; queso fresco from informal markets is a recurring outbreak source.
- Deli-counter cross-contamination. Even pasteurized sliced meats acquire Listeria from the slicer biofilm; the reheat-to-steaming rule is the safer default than relying on the meat's own status.
- Powdered formula at room temperature. Cronobacter grows rapidly in reconstituted formula; the 2-hour cumulative limit is not the 4-hour rule that adults use.
- Time stacking. A meal that sat out for 30 minutes at the grocery checkout, 45 in the car, and 90 on the counter before reheating has used the 2-hour rule three times over — each segment counts.
- Older-adult kitchen drift. Multiple cohort studies of older-adult households find refrigerator temperatures unmeasured, thermometers absent, leftover-age tracking gone — the protocol fails before it's tested Lund and O'Brien 2011.
stakes
The stakes are episodic and catastrophic, not cumulative. A single Listeria-positive turkey sandwich in the third trimester can produce stillbirth; a single Vibrio-positive oyster in a cirrhosis patient can produce septic shock in 24 hours; a single STEC-positive hamburger in a four-year-old can produce HUS and lifelong renal impairment. Annualised, U.S. listeriosis kills ~260 people, vibriosis ~100, and STEC ~30; STEC-HUS accounts for ~70% of paediatric acute kidney injury in industrialised countries Scallan et al. 2011, Tarr et al. 2005. Most affected households did not consider themselves high-risk before the event.
payoff
The payoff is the absence of a counterfactual catastrophe — a measurement problem common to high-leverage prevention. Quantitatively, sustained adherence in a high-risk pregnancy collapses the listeriosis-from-food rate by an order of magnitude (the bulk of cases trace to deli meats, soft cheeses, and refrigerated smoked seafood, all on the avoidance list); in adults 65+ with cirrhosis, avoiding raw oysters during May–October collapses the warm-water Vibrio season-attributable risk to near zero Jones et al. 2014, Mateus et al. 2013. The felt payoff is a quieter pregnancy or post-transplant year — no week-long viral-feeling fever turning into hospitalisation, no late-pregnancy emergency that started as a sandwich.
practicalities
Cost is near-zero — pasteurized dairy is the same price as raw, ready-to-feed liquid formula costs more than powdered but only for the highest-risk-infant period (typically 2 months), thermometers are $10–$20 one-time. The effort burden is real but bounded: a sustained shopping-list shift, a learned reflex around restaurant menus, a habit of reheating deli meat and using a thermometer. The bulk of the protocol is decision-loaded once (what to skip) and then runs as background.
out-of-scope
Excluded from this entry: general food poisoning in healthy adults (lower-stakes, different protocol); travellers' diarrhoea and water-source contamination (separate entry); chemical and allergen contamination (allergy entry); fish mercury and methylmercury in pregnancy (separate substance, different mechanism); raw-milk legalisation politics (regulatory/political, not health); foodborne botulism from home canning (low-volume, different protocol). Each is a legitimate adjacent topic but shares neither the pathogen list nor the vulnerable-group framing.
Credibility range
Optimist case
The mechanism-and-epidemiology evidence is among the strongest in the catalogue. Listeria's predilection for refrigerated ready-to-eat foods is a textbook microbiology fact; the FoodNet-derived incidence-rate ratios for pregnancy and 65+ are clean and replicated; the U.S. raw-milk dairy data show a clear ~150× outbreak-rate signal versus pasteurized; STEC-HUS in children is one of the best-characterised acute paediatric emergencies and tracks tightly to undercooked ground meat and raw-milk exposure; Vibrio-cirrhosis case-fatality at ~50% from raw oysters is documented in multiple state-level surveillance systems. The avoidance list and handling adaptations converge across the CDC, FDA, USDA, ACOG, AAP, and WHO with no meaningful disagreement. The intervention is cheap, the downside of acting is essentially zero, and the upside of skipping a single high-risk exposure during a high-risk window is large.
Skeptic case
Two skeptic threads carry weight. First, **base-rate framing**: U.S. listeriosis is ~1,600 cases and ~260 deaths a year against ~330 million people and trillions of food exposures, so the per-exposure probability for any one deli sandwich is very small; rigid adherence to the full avoidance list represents tens to hundreds of thousands of joyless decisions for an expected-value gain that, individually, looks tiny. Second, **behaviour-change RCT data are weak**: most food-safety interventions move knowledge more than behaviour, and the same older-adult cohorts that read the guidance run their refrigerators at 6 °C and skip thermometers anyway. There is no high-quality RCT showing that telling pregnant people to avoid soft cheese in a population-level intervention reduces listeriosis incidence (it would be unethical to run as a negative-control trial); the inference is from outbreak attribution. Some specific items (aged hard cheeses from raw milk, sushi from reputable establishments, low-Vibrio-season oysters) are arguably over-restricted relative to per-exposure risk.
Author's call
Strongly toward the optimist case for the four named vulnerable groups, with two craft notes. The base-rate skeptic is right that the per-exposure odds are small, but the asymmetry — catastrophic downside, near-zero cost of avoidance — is the textbook expected-value win, and outbreak data demonstrate that the listed foods are the actual vehicles, not folklore. Adherence to the closed list is high-evidence, high-leverage, low-cost. The two craft notes: (1) honesty about behaviour-change difficulty, especially in older-adult solo households; (2) avoiding the wellness-paranoia register that turns pregnancy into a year of fear — this entry is a closed list of foods, not a generalised threat about eating.
Stakeholder and incentive map
- Public-health agencies (CDC, FDA, USDA-FSIS, EFSA, WHO, NICE). Conservative by design — the cost-asymmetry of a missed warning is high. Their guidance is the operational basis for this entry.
- Obstetric and paediatric specialty bodies (ACOG, AAP, RCOG). Aligned with public-health agencies; provide the pregnancy- and infant-specific protocols.
- Food industry. Generally aligned with safety guidance (regulatory and liability incentives push the same way); some lobbying around raw-milk legalisation and produce traceability runs counter, but the consumer-facing labelling industry mostly converges.
- Raw-milk and "real food" advocates. Push against pasteurization rules on culinary, libertarian, and traditional-food grounds; their evidence base on nutrient and probiotic preservation is weak and routinely overstated. Outbreak data run squarely against the consumer-safety claim.
- Foodservice industry. Mixed incentives — restaurants want the menu flexibility, but liability and reputational cost of an outbreak is large. Pregnancy- and immunocompromise-aware menu labelling is uncommon in the U.S., common in parts of Europe.
Population variability
- Pregnancy raises listeriosis incidence ~13×; toxoplasma transmission risk rises with gestational age (~10% first trimester, ~60% third), severity inverts (first-trimester transmission is rarer but more devastating).
- Age 65+ raises listeriosis incidence ~4× at baseline and another several-fold with chemotherapy, transplant, dialysis, or cirrhosis. Reduced gastric acid (age, PPI use, post-bariatric anatomy) lowers the inoculum threshold for Salmonella, Vibrio, and Listeria.
- Children under five show distinctly elevated STEC-HUS, Salmonella, and Campylobacter rates; the under-two cohort additionally faces Cronobacter from powdered formula. Breastfed-only infants are at lower bacterial-gastroenteritis risk.
- Immunocompromised. Transplant recipients carry the highest listeriosis incidence-rate ratios (50–100× in some series); biologic and chronic-steroid users face moderately elevated rates with less precise quantification.
- Chronic liver disease and haemochromatosis specifically raise Vibrio vulnificus mortality from raw oysters to ~50%.
- Geographic and seasonal variability. Vibrio risk concentrates in Gulf Coast warm-water months (May–October) and is rising with sea-temperature warming; Listeria risk is seasonally flat; Salmonella peaks in summer poultry-handling.
Knowledge gaps
- Behaviour-change interventions that actually work. Education studies show small effects; the household-level intervention that reliably moves refrigerator temperatures, thermometer use, and restaurant-ordering behaviour in older adults is undefined.
- Per-exposure dose-response in the vulnerable groups. Population incidence is well-characterised; individual exposure-risk estimates are coarse — we know deli meat is a major Listeria vehicle but not the conditional probability per slice for a third-trimester eater.
- The aged-cheese question. Whether truly aged hard cheeses (low water activity, low pH) from raw milk are safe in pregnancy is genuinely uncertain; current guidance errs strongly conservative.
- Sushi and ceviche risk stratification. Per-exposure pathogen rates from commercial sushi (which uses parasite-killing freezing) versus home-prepared raw-fish dishes are inadequately separated in the literature.
- Climate-driven Vibrio expansion. Northward spread of V. vulnificus into formerly cool-water states is documented but the updated risk geography for cirrhotic readers is still settling.
- Cronobacter outside powdered formula. Recent outbreaks (2022 Abbott recall) raised manufacturing-side and home-handling questions whose evidence base is still maturing.
Scope notes:
- The brief named four vulnerable groups (older adults, young children, pregnant people, immunocompromised) and three food families (raw, unpasteurized, ready-to-eat) — all are covered. The pregnant audience gets slightly more body weight than the others because the canonical pathogen (Listeria) and the most extensive avoidance list both concentrate there; the four-groups audience section keeps the breakdown explicit so this doesn't read as a pregnancy-only entry.
- Audience meta field left blank intentionally — the entry's actual audience (the four vulnerable groups + their caregivers) doesn't decompose cleanly into the catalogue's gender/age-band scoping. Applicability handles the reach calculation via the avoidance/awareness rule (currently-affected ~25–30%, lifetime decision audience ~most adults). The article body carries the per-group differentiation.
- Contraindications field left empty — this is a protective entry; the closest concept (pregnancy / immunocompromise) is the qualifying audience, not a contraindication.
Hard calls:
- Aged hard cheeses in pregnancy. Genuine evidence ambiguity (low water activity / low pH lowers Listeria risk substantially) but CDC and ACOG guidance is pasteurized-only. Followed guidance; named the ambiguity in `misconceptions` so the reader sees the tradeoff rather than thinking the rule is hidden.
- Sushi from reputable establishments. Commercial sushi is required to use parasite-killing freezing, which materially reduces some risks (anisakis) but does not address Vibrio or norovirus on the raw fish surface. Kept on the skip list for the four groups; arguably over-restricted for the typical reputable-restaurant case, but the per-exposure-evidence isn't there to relax the rule for a vulnerable audience.
- Vibrio seasonal/geographic nuance. Compressed to "warm-water region" rather than enumerated; the geography is moving with sea-temperature warming and any specific list would date quickly.
Rating difficulty:
- longevity at 3, not 4. The catastrophic-event prevention is real (Listeria fetal mortality, Vibrio septic mortality, STEC-HUS), but the absolute mortality numbers per population are modest. Score reflects "meaningful disease prevention," not "dominant longevity effect."
- applicability at 4. The avoidance/awareness lift in `meta.md` §6 applies cleanly: lifetime decision audience (anyone who will ever be pregnant, parent a young child, age past 65, or face immunosuppression) is most adults, even though the point-prevalence audience is ~25–30%.
- pull at 1. The reader is being asked to skip foods they often like (soft cheese, sushi, deli sandwiches, runny yolks, raw oysters) for an invisible counterfactual win. The writing has to do work here — the relief lever and the asymmetry framing are how the entry earns the read.
Future-link candidates (entries that don't exist yet but should cross-link once they do):
- Methylmercury and fish choices in pregnancy
- Travellers' diarrhoea and water safety abroad
- Foodborne botulism and home canning
- Fridge organisation, leftover storage, and household kitchen hygiene
- Toxoplasmosis (broader entry — covers cat-litter exposure beyond food)
- Proton-pump inhibitors and infection risk in older adults
Separate-entry candidates surfaced during writing:
- Cronobacter and powdered infant formula — narrow enough to warrant its own entry once the catalogue covers infant feeding more comprehensively.
- Raw-milk politics and nutrition claims — the misconceptions paragraph handles it for now, but a focused myth-buster entry is plausible.
High-Risk Foods for Vulnerable Groups
Near-zero. Pasteurized dairy is the same price as raw; a fridge thermometer is a one-time ~$10–20 purchase; ready-to-feed liquid formula costs more than powdered but only for the highest-risk infants (under two months, immunocompromised, premature).
A sustained shopping-list shift, a learned restaurant-menu reflex (Caesar, hollandaise, tartare, sushi, soft cheese), a habit of reheating deli meat to steaming, and thermometer use. Decision-loaded once, then runs as background; a few minutes daily of attention but no major lifestyle reorganisation (Lund and O'Brien 2011).
Strong epidemiologic and outbreak-attribution evidence; consistent guidance across CDC, FDA, USDA-FSIS, ACOG, AAP, WHO, and NICE. RCT-impossible (would be unethical to randomise vulnerable readers to high-risk foods), but FoodNet incidence-rate ratios for pregnancy (~13×) and 65+ (~4×) are clean and replicated; the 1993–2006 raw-milk outbreak rate is ~150× pasteurized (Scallan et al. 2011; Pouillot et al. 2012; Langer et al. 2012).
Prevents catastrophic single-exposure events with high case-fatality in the vulnerable groups: listeriosis (~20–30% mortality in non-pregnant adults, ~20–30% combined fetal/neonatal death in pregnancy), Vibrio vulnificus in cirrhosis (~50% case-fatality from raw oysters), STEC-HUS in children (substantial dialysis and lifelong-renal-impairment risk). Adherence collapses the population-attributable risk meaningfully (Pouillot et al. 2012; Jones et al. 2014; Tarr et al. 2005).
For pregnant, immunocompromised, 65+, and under-five readers, avoiding the closed list prevents acute episodic illness — Listeria gastroenteritis/bacteraemia, Salmonella, STEC, vibriosis — whose individual-incident rate is low but conditional severity is high. Not a 'felt wellness improvement' in normal weeks; a real-but-small reduction in the rate of feel-terrible-for-a-week-or-worse episodes (Scallan et al. 2011).
The peace-of-mind dividend during a high-stakes window (third trimester, post-transplant year, infant under one) is real but small — fewer late-night second-guesses about a sandwich. Not a primary reason to do this; surfaces as a side-benefit of decision-loading the list once.