Keto & Very-Low-Carb: The Honest Audit

Few diets generate more argument than keto — and much of it conflates three different questions: what ketosis does to blood sugar inside supervised medical programs, what it does to LDL cholesterol in individual bodies, and what it might do to lifespan, which no human trial has tested. This hub audits the claims one at a time.

What "Keto" Actually Means

Nutritional ketosis is a measurable metabolic state, not a brand: restrict digestible carbohydrate far enough — around 20–50 grams of net carbs a day — and the liver converts fat into ketone bodies, with blood β-hydroxybutyrate typically landing near 0.5–3 mmol/L. That range is what "in ketosis" means clinically.

Ketoacidosis is a different event wearing a similar name — a medical emergency in which ketones climb many-fold higher and the blood turns acidic, most classically when insulin is absent or insufficient. A blood ketone reading of 1.5 mmol/L after a week of low-carb eating is simply nutrition; the same reading alongside nausea and confusion is an emergency-room matter. The ketosis subtopic draws that line in full.

The Strongest Human Evidence

The most-quoted human test of keto is not a longevity study. Hallberg and colleagues (Diabetes Therapy, 2018) enrolled 349 adults with type 2 diabetes: 262 entered a continuous care intervention — a very-low-carb ketogenic diet plus remote coaching and clinician-managed medication adjustment — while 87 stayed in usual care.

At one year, intention-to-treat: mean HbA1c fell from 7.6% to 6.3%, weight fell about 13.8 kg, non-metformin prescriptions dropped from 56.9% to 29.7%, and 94% of insulin users reduced or eliminated insulin. Among the 204 participants with HbA1c reported at 1 year, 60.3% held HbA1c below 6.5% while on metformin or no glucose-lowering medication; 25% were off all such medication. Triglycerides fell 24%, HDL cholesterol rose 18%, and average LDL cholesterol rose about 10%.

Read it as a program result: a diet plus continuous medical supervision from a team titrating medications against biometrics — machinery a diet handout does not produce. The type 2 diabetes reversal subtopic carries the full evidence map.

Reading Those Numbers Honestly

The results are real and the design has soft spots. Four caveats belong next to these numbers:

• 🎯 Non-randomized: participants chose their group, and enrollees differ in motivation and support — adjustment narrows that gap but never closes it.
• 🩺 The medication layer is part of the therapy: clinicians made the reductions by titrating prescriptions against biometric feedback; copying the diet without that layer is a different experiment.
• 📉 Attrition was real but modest: 83% were still enrolled at one year and 74% at two — good retention, but the 2-year numbers rest on those who stayed.
• 💰 Industry funding: the largest datasets come from the program's operator; independent replication is the missing piece.

The 2-year headline: diabetes reversal in 53.5% of the treated group, remission in a further 17.6% (Athinarayanan et al., 2019) — striking numbers, carrying the caveats above.

The LDL Question Nobody Settles

Average LDL cholesterol rose about 10% in the supervised trial — unremarkable, except that the average hides a wide distribution, and the extremes have a name now. Norwitz and colleagues (2022) surveyed 548 adults on carbohydrate-restricted diets: the leaner respondents with high HDL and low triglycerides saw the largest LDL increases — a subgroup of 100 they labeled lean mass hyper-responders met a striking definition, LDL of 200 mg/dL or more alongside that favorable profile. One case report: LDL climbed from 95 to a peak of 545 mg/dL, then fell when carbohydrates returned.

A 2024 meta-analysis in the American Journal of Clinical Nutrition found LDL rose about 41 mg/dL on average in low-carb trials of normal-weight participants — and did not rise in heavier groups (Soto-Mota et al.). Put plainly: the leaner and metabolically healthier you start, the more likely your LDL drifts up.

What nobody has is outcome data. A 2022 commentary in the Journal of Clinical Lipidology, co-authored by senior researchers including Ronald Krauss, argued these cases "deserve urgent clinical attention and further research." The same group's phenotype paper argues the rises occur in otherwise low-risk profiles. Both are in print; no trial settles it — a genuine "we don't know yet." The LDL subtopic keeps the running ledger of that argument.

Where the Longevity Claims Come From

The lifespan claims rest on rodents: ketogenic diets in mice produce mixed results — some report better midlife health markers, others find no lifespan extension — and mouse metabolism diverges from ours. No human trial has tested a ketogenic diet against mortality or cardiovascular events. That evidence tier belongs to patterns people can maintain for decades, like the Mediterranean-style pattern. The longevity-claims subtopic catalogs the mouse studies in detail.

Two quieter problems: adherence (weeks are easy, years are hard) and the swap problem — replacing bread with butter is not automatically an upgrade (the Fats & Carbs topic unpacks the trade).

Who Should Not Do Keto

Keto is not neutral for everyone — these are the situations a supervised study would screen out:

• 🚫 Type 1 diabetes: insulin deficiency plus ketone production is the recipe for ketoacidosis; only specialist-supervised protocols apply.
• 💊 SGLT2 inhibitors: this common drug class carries label warnings about ketoacidosis — including cases with normal-looking blood sugar (callout below).
• 💉 Insulin and sulfonylureas: cutting carbohydrate while these drugs are active can push blood sugar dangerously low; adjust doses with a clinician first.
• 🤰 Pregnancy and breastfeeding: restricted diets are not recommended outside specialist supervision — nutrient needs are high and the margin small.
• 🫘 Kidney disease: protein and electrolyte shifts land differently in impaired kidneys — clinical nutrition territory.
• 🍽️ Eating-disorder history: highly restrictive rules can re-engage disordered patterns; involve a clinician.
• 🧒 Children and adolescents: ketogenic diets are supervised pediatric therapy for specific conditions — not a household decision.

How to Decide With a Clinician

The supervised studies got their results with scaffolding — measurements, medication management, follow-up. Borrow the scaffolding:

• 🧾 Medication list first: for insulin, sulfonylureas, SGLT2 drugs, and blood-pressure medications, the answer changes the plan.
• 📊 Baseline labs: a lipid panel, HbA1c, and kidney function — the numbers you will track against.
• 📅 A follow-up plan: recheck lipids and glycemic markers at three months, with an agreed LDL level that would change the approach.
• 🚪 An exit criterion: write down what result would make you stop, while you still feel fine. The who-should-not subtopic carries the fuller screening list.

Keto Questions, Answered Briefly

• ❓ Is keto better than other diets for weight? At 12 months, head-to-head, no — early losses are partly water, and by a year the field converges (Gardner et al., 2018).
• ❓ Is "keto flu" real? Transition symptoms in the first week or two are commonly reported; electrolytes and hydration are the usual levers.
• ❓ Can it be done vegetarian? Eggs, cheese, tofu, nuts, olive oil, non-starchy vegetables — the meat-heavy version is one shape, not the definition.

Go Deeper: Subtopics

• ⚗️ Ketosis: What It Actually Is — the metabolic state, drawn clearly. Read it →
• 🩸 The T2D Reversal Evidence — the supervised trials dissected. Read it →
• 🧪 The LDL Problem — the contested LDL evidence, in full. Read it →
• 🐭 The Longevity Claims vs the Evidence — the rodent studies vs. human reality. Read it →
• ⚠️ Who Should Not Do Keto — conditions and medications that change the calculus. Read it →

Related Topics