Substance + claimed effects

Walking after meals is a short, low-to-moderate-intensity walk — typically 10 to 20 minutes — initiated within roughly half an hour of finishing a meal, ideally within the first 15 minutes. The substance is unusual in the catalogue in that it is not a new behaviour: most adults already walk; the lever here is the timing of a short bout relative to food. Claimed effects fall across several dimensions and are addressed holistically per entry.md §1a: a substantial blunting of the post-meal blood-glucose spike Reynolds et al. 2016 DiPietro et al. 2013 Engeroff et al. 2023; a parallel reduction in postprandial insulin Buffey et al. 2022 Dunstan et al. 2012; faster gastric emptying and reduced functional dyspepsia signs in healthy adults Franke et al. 2008; a modest reduction in postprandial triglyceridemia when bouts are accumulated across the day Miyashita et al. 2008; and a behaviour-architecture lift in daily activity accumulation that compounds over time because it attaches to an event (the meal) that already recurs reliably. The effect on reflux is direction-dependent and is treated honestly here: light walking is generally well tolerated and may reduce reflux events versus reclining after a meal Festi et al. 2009, but vigorous walking or bending while full of food can worsen symptoms in susceptible people. Longevity, day-to-day energy, focus, and downstream cardiovascular risk are downstream of those core effects; the dossier traces each.

Evidence by addressing question

Mechanism

Science / mechanism. The chain that explains the glucose effect is well characterised. After a mixed meal, gut-derived glucose appears in the systemic circulation within 15 to 30 minutes and peaks around 45 to 75 minutes; insulin rises in parallel, and skeletal muscle is the dominant disposal site for the glucose load, taking up roughly 70–80% of the postprandial bolus in healthy adults. When skeletal muscle contracts during this window, two routes for glucose entry operate in parallel: the insulin-dependent route through the PI3K–Akt pathway, and a contraction-stimulated, insulin-independent route in which AMPK and mechanical / calcium signalling traffic GLUT4 vesicles to the sarcolemma. The second route is the one that walking exploits: a person whose insulin signalling is impaired (insulin resistance, type 2 diabetes) can still clear glucose into muscle by contracting it. Even very light contraction is enough — the soleus, a slow-twitch postural muscle, can sustain oxidative glucose disposal for hours at low metabolic cost Hamilton et al. 2022, which is why a slow stroll and an upright posture both move the needle far more than their metabolic cost would predict.

Gastric and biliary mechanisms run alongside. Upright posture and rhythmic abdominal-wall activity during walking accelerate gastric emptying of liquids and semi-solid meals versus seated rest Franke et al. 2008. Faster emptying spreads the carbohydrate appearance curve, lowering its peak even before muscle uptake is factored in. For triglycerides, the mechanism is dual: muscle contraction raises lipoprotein lipase activity in capillary beds within hours, increasing chylomicron-TG clearance, and the bout itself draws on intramuscular and circulating lipids as substrate when carbohydrate availability is high. The effect is dose-dependent on accumulated walking time across the postprandial day rather than on any single bout's intensity Miyashita et al. 2008.

Evidence

Science. The strongest single trial is Reynolds et al. 2016, a randomised crossover in 41 adults with type 2 diabetes comparing 30 minutes of walking once daily versus three 10-minute walks each beginning within 5 minutes of finishing a meal. Postprandial glucose, measured by CGM, was 12% lower on average across the after-meal arm than under the equivalent total dose of walking taken any time, with the largest single-meal effect — about 22% — after dinner, the meal that carried the most carbohydrate and was followed by the longest period of inactivity Reynolds et al. 2016. DiPietro et al. 2013 ran a similar design in older adults with impaired glucose tolerance: three 15-minute moderate walks timed to start 30 minutes after each meal lowered 24-hour glycaemia equivalently to a single 45-minute morning or afternoon walk, and lowered post-dinner glycaemia more, including reducing nocturnal hyperglycaemia DiPietro et al. 2013. Manohar et al. 2012 quantified the dose-response in healthy adults and type 1 diabetics using accelerometers — every additional minute of walking within the first 90 minutes after a meal incrementally reduced glycaemic excursion, in a near-linear relationship up to roughly 30 minutes Manohar et al. 2012. Van Dijk et al. 2013 confirmed the effect with structured cycling in type 2 diabetics: moderate-intensity exercise within an hour of the meal reduced 24-hour mean glucose by approximately 0.7 mmol/L versus sedentary control van Dijk et al. 2013. Pahra et al. 2017 replicated the post-meal-split versus single-bout comparison in newly diagnosed type 2 diabetes and showed greater reductions in both fasting and postprandial glucose with the split protocol Pahra et al. 2017.

The two meta-analyses align. Engeroff et al. 2023 pooled 38 trials in healthy adults and type 2 diabetics: post-meal exercise produced larger postprandial glucose reductions than equivalent pre-meal exercise (standardised mean difference favouring post-meal of roughly 0.4), with the effect strongest when exercise started within 30 minutes of the meal and lasted at least 15 minutes Engeroff et al. 2023. Buffey et al. 2022 specifically isolated light-intensity walking as an interruption of prolonged sitting and pooled seven studies showing significant reductions in postprandial glucose (SMD ≈ -0.7) and insulin (SMD ≈ -0.4) versus continuous sitting, with light walking outperforming standing breaks Buffey et al. 2022. Aqeel et al. 2020 reviewed exercise-timing studies and found the post-meal window the most consistent direction of effect, with effect sizes comparable to a low-dose oral antidiabetic medication when the bout was timed to the glucose-appearance peak Aqeel et al. 2020. Bellini et al. 2022 added meal-composition data: post-meal walking blunted glucose response after high-carbohydrate and mixed meals, with the largest absolute reduction after the highest-glycaemic-load meal Bellini et al. 2022.

Practice / clinical consensus. The American Diabetes Association's 2016 position statement explicitly recommends interrupting prolonged sitting with light activity every 30 minutes and notes that postprandial walking is among the simplest tools for postprandial glucose control in type 2 diabetes Colberg et al. 2016. Diabetes educators routinely prescribe a 10–15 minute post-dinner walk as a first-line behavioural lever before medication titration. The cardiovascular practice community has adopted "movement snacks" (Dunstan, ADA, the European Society of Cardiology lifestyle group) as a structural recommendation; postprandial walking is the canonical instance.

Community / lay evidence. Continuous glucose monitor users on diabetes forums, biohacker communities, and the CGM-for-the-curious cohort (Levels, Veri, NutriSense) report near-universal CGM trace flattening after introducing a 10–20 minute post-meal walk. The signal is consistent enough that it appears in every CGM vendor's published case-study libraries and in dozens of self-tracking write-ups. Volume is in the tens of thousands of consistent reports; survivorship bias is modest because the failure case (no flattening) is easy to observe on one's own trace.

Historical / cross-cultural. The Italian passeggiata, the German "Verdauungsspaziergang" (literally "digestion walk"), the Indian post-dinner family stroll, and the Japanese shokugo no sanpo all encode a centuries-old practice of an after-meal walk. The recurrence of the same practice across unrelated cultures — and its absence as a formal pre-meal practice — is the kind of convergent evolutionary signal that often precedes formal evidence.

Protocol

Science. Reynolds 2016 used three walks of 10 minutes each, beginning within 5 minutes of meal completion, at self-selected light-to-moderate pace (about 4 km/h) Reynolds et al. 2016. DiPietro 2013 used three walks of 15 minutes at 3 mph (≈4.8 km/h) starting 30 minutes post-meal DiPietro et al. 2013. Manohar 2012 reported a near-linear glucose response across the 5–30 minute window and across moderate paces Manohar et al. 2012. The convergent picture: 10–20 minutes, starting within 30 minutes of finishing the meal, at a conversational pace. Longer is fine; faster is fine where the gut tolerates it; earlier in the window is generally better than later because the bout has to overlap the glucose appearance curve.

Practice. Diabetes educators commonly anchor the walk to a fixed cue — after lunch, walk to the next coffee; after dinner, walk the dog; after breakfast, walk part of the commute. The habit substrate (the meal itself) is what makes this protocol reliable where standalone exercise prescriptions fail.

Contraindications

Science. The intervention is low-risk for almost everyone able to walk. Three real considerations:

• Gastroesophageal reflux disease (GERD). Body position and walking effort interact with reflux events. Light upright walking tends to reduce reflux versus reclining post-meal Festi et al. 2009; vigorous walking, bending, or wearing tight waistbands can provoke reflux by raising intra-abdominal pressure. The honest read is that an unhurried walk in upright posture is generally GERD-friendly, but a brisk walk immediately after a large meal is not.
• Symptomatic gastroparesis or post-bariatric dumping syndrome. In these populations, accelerated gastric emptying — exactly the effect walking induces in healthy stomachs — can be counterproductive or symptomatic. Individual titration with a clinician is appropriate.
• Cardiovascular instability. Severe heart failure, unstable angina, recent MI within the rehab window. Standard pre-existing exercise restrictions apply; the after-meal timing does not change them.

For type 1 diabetes and insulin-using type 2 diabetes, post-meal walking can produce hypoglycaemia if insulin dosing is timed to a sedentary postprandial period and the walk is then added Manohar et al. 2012. The protocol still works; insulin dosing has to account for it.

Misconceptions

Practice. Three common errors:

• "Walking after meals will give me cramps or indigestion." The 19th-century counsel ("after dinner sit a while…") is a confusion of cause and effect — vigorous activity right after a large meal can be uncomfortable, but light walking accelerates gastric emptying and, in trials, reduces postprandial dyspeptic symptoms Franke et al. 2008.
• "Walking before a meal is just as good." Pre-meal exercise improves fasting and 24-hour glucose, but post-meal exercise specifically blunts the postprandial peak in a way pre-meal does not — because the bout has to overlap the glucose appearance window to compete with it for muscle uptake Engeroff et al. 2023 Aqeel et al. 2020.
• "You need a long, brisk walk for it to matter." Buffey 2022 showed that light, conversational-pace walking interruptions are sufficient — the contraction-stimulated GLUT4 mechanism does not require high intensity Buffey et al. 2022 Hamilton et al. 2022.

Alternatives

Science / mechanism. The contraction-driven, insulin-independent glucose clearance mechanism is the same one targeted by several adjacent interventions; all are legitimate, with different friction profiles:

• Soleus push-ups (seated calf raises sustained at low intensity). Hamilton et al. 2022 showed prolonged seated soleus contractions produce a ~52% reduction in postprandial glycaemic excursion — larger than typical post-meal walking — with metabolic cost roughly tripling resting metabolic rate Hamilton et al. 2022. Useful when a walk is not feasible (office, plane, long meeting).
• Standing breaks alone. Standing reduces postprandial glucose modestly, but light walking outperforms it Buffey et al. 2022.
• Interrupting prolonged sitting every 30 minutes with 2–3 minute light-activity breaks. Lower-dose equivalent of the same lever, useful through the rest of the day after the post-meal walk has done its work on the immediate peak Dunstan et al. 2012.
• Pre-meal protein or fibre. Operates by a different mechanism (gastric-emptying delay, incretin response) and stacks additively with walking rather than substituting.

Failure modes

Where the intervention fails in practice:

• Wrong window. A walk taken 90 minutes after a meal misses the peak. The behaviour is happening; the glucose effect is largely gone.
• Pace too high after a large meal. Brisk walking with a full stomach provokes upper-gut discomfort, especially after a fatty or carbonated meal, and the person attributes the discomfort to "walking after meals" generally and stops.
• Intensity too low to count. A leisurely 3-minute saunter to the kitchen and back doesn't reach the duration floor. The trial-tested minimum is roughly 10 minutes.
• Walking only after one meal. Reynolds 2016 showed the largest single effect after dinner; DiPietro 2013 showed the largest 24-hour effect from three walks. A single-meal habit captures a fraction of the available benefit.
• Stacking onto insulin dosing without adjustment. For insulin-treated diabetes, the new activity changes the dose-response curve.

Practicalities

Cost is zero. Equipment is none. Time is borrowed from the post-meal sit-and-scroll window most people have anyway. The unusual feature of this protocol is the habit substrate: meals are the most reliable recurring event in adult life. Attaching a behaviour to a meal — rather than to a calendar slot — exploits one of the strongest known cues for behaviour persistence. The Reynolds 2016 protocol survives at compliance rates higher than standalone walking prescriptions because of this substrate Reynolds et al. 2016.

Stakes

Science. The long-run reason the postprandial-glucose lever matters is that postprandial glucose, independent of fasting glucose, is associated with all-cause and cardiovascular mortality. The DECODE study followed 25,000 European adults and found 2-hour post-load glucose to be a stronger predictor of cardiovascular mortality than fasting glucose, with hazard ratios rising progressively across the impaired-glucose-tolerance range — well below the diabetes diagnostic threshold DECODE Study Group 2001. Postprandial hyperglycaemia drives endothelial dysfunction, oxidative stress, and advanced-glycation-end-product accumulation; chronic exposure tracks with arterial stiffness and cognitive decline. A meaningful share of population cardiovascular risk sits inside post-meal glucose excursions in people whose fasting numbers look unremarkable. Lowering those excursions across years is the longevity lever.

Payoff

Science / community. Within a week: a flatter CGM trace, less afternoon energy collapse — the "post-lunch dip" is in part a reactive-hypoglycaemia signal following a high spike Buffey et al. 2022. Within months: lower HbA1c in prediabetic and diabetic adults (effect sizes in the range of a low-dose antidiabetic medication) Pahra et al. 2017 Reynolds et al. 2016; modest reductions in postprandial triglycerides and blood pressure when bouts accumulate Miyashita et al. 2008. Over years: the daily activity accumulation matters as much as the glucose effect — three 10-minute walks add about 30 minutes per day of light activity, ≈3,000 steps, which moves a person from "low active" to "active" without requiring a separate exercise session. The compounded effect on cardiovascular and metabolic risk over a decade is meaningful.

The credibility range

Optimist case

The post-meal-walking lever is the rare intervention with mechanism, RCT evidence, meta-analysis confirmation, clinical guideline support, and lay reproducibility on CGMs all aligned. It costs nothing, takes 10–20 minutes, attaches to an existing recurring event, and produces postprandial glucose reductions in the range of 12–22% across well-controlled trials Reynolds et al. 2016 DiPietro et al. 2013. The mechanism — contraction-stimulated, insulin-independent glucose disposal — is the textbook physiology, validated by isotope tracer studies decades before walking-after-meals trials were run. The behavioural substrate (meals recur reliably; activities anchored to meals persist) is among the more robust habit-formation findings in psychology. Across cultures, the practice predates the science: the Italian passeggiata and the Japanese after-dinner stroll encode the same lever, suggesting humans figured it out empirically long before glucose monitors confirmed it. For non-diabetic adults, the cumulative effect on postprandial glucose translates through a well-established epidemiological pipeline (DECODE and successor cohorts) into measurable cardiovascular risk reduction over years DECODE Study Group 2001. There is no realistic downside in the target population (ambulatory adults without GERD or insulin-dosing constraints). The closest analogue in the supplement aisle would cost $200/year, have weaker evidence, and reduce postprandial glucose less.

Skeptic case

The trials are small. Reynolds 2016 enrolled 41 participants; DiPietro 2013 ten; Manohar 2012 used CGM in a crossover with mostly healthy adults. The meta-analyses pool these small studies and arrive at moderate effect sizes that, in absolute terms, may not translate into clinically meaningful long-term outcomes — no trial has run long enough to show that the postprandial glucose reduction yields a hard cardiovascular endpoint reduction independent of total physical activity. The relationship between post-meal glucose excursions and cardiovascular mortality is observational (DECODE and similar) and confounded by general metabolic health; the inference that lowering excursions reduces mortality is a reasonable but unproven extrapolation. Some of the apparent benefit may simply be the dose of any walking — adding ≈30 minutes of activity to a person's day predictably improves metabolic markers regardless of timing; the marginal benefit of timing it after meals versus the same dose taken elsewhere is on the order of a 10–20% incremental glucose reduction, which is small in absolute terms. The CGM-forum signal carries a strong adopter-self-selection bias: people who try after-meal walks tend to be people already paying attention to their health, and their CGMs show what they expect. For healthy young adults with insulin-sensitive muscle, the postprandial excursion is small to begin with and the lever has little to flatten. The strongest skeptic position: this is not a wrong intervention, just an oversold one — the dose of walking matters more than the timing, and a 30-minute walk at any time of day captures most of the benefit.

Author's call

The intervention is real and well evidenced for the glucose effect; the dispute is not whether after-meal walking flattens postprandial glucose (it does, repeatedly, in multiple labs) but how large the long-run consequence is. The honest landing: for people with diabetes, prediabetes, or any insulin resistance, the timing is load-bearing and the intervention is one of the higher-leverage behavioural tools available — effect sizes comparable to a starting dose of metformin, at zero cost. For metabolically healthy adults, the timing matters less in absolute terms but the behaviour itself remains net-positive because of the daily-activity accumulation effect, which is independently linked to mortality and cognitive trajectory. The skeptic point about hard-endpoint trials is fair but does not unsettle the recommendation — the burden of proof for a 10-minute walk is low, the downside risk is near zero, and the upstream evidence is solid. The article should treat this as a high-evidence, low-controversy, broadly applicable intervention. The catch worth naming honestly: walking after one meal a day captures less than walking after all three; the dinner walk is the highest-yield single instance if only one can be sustained.

Stakeholder + incentive map

• Diabetes-care professionals (endocrinologists, certified diabetes educators) — actively promote the practice; the lever is in their first-line behavioural toolkit and reduces the population's medication needs. Incentive is patient-outcome and guideline-aligned, not commercial.
• Continuous glucose monitor vendors (Dexcom, Abbott; consumer CGM channels Levels, Veri, NutriSense) — promote it heavily because the on-trace flattening is a vivid product demonstration. Commercial alignment with the recommendation; the recommendation itself is not commercially distorted (CGMs sell themselves on showing the flattening, not on selling the walk).
• Public-health and cardiology bodies (ADA, ESC, AHA lifestyle subgroups) — endorse "movement snacks" and post-meal walks as guideline-consistent.
• Behavioural-design / habit-formation community — uses post-meal walking as a worked example of cue-anchored habit formation (BJ Fogg's tiny habits, James Clear's habit stacking).
• Wellness influencer ecosystem — has popularised "the 10-minute walk after meals" hard, sometimes with overstated claims ("it cures insulin resistance"). The overstatement is the failure mode; the core practice survives the noise.
• Skeptic / counter-incentive corner — limited. Sports-medicine commentators occasionally note that the timing effect is smaller than the social-media hype, but no one is materially against post-meal walking. There is no "anti-walking-after-meals" industry.

Population variability

• Type 2 diabetes and prediabetes. Largest absolute effect; insulin resistance is precisely what the contraction-stimulated route bypasses. Reynolds 2016 and Pahra 2017 populations Reynolds et al. 2016 Pahra et al. 2017.
• Older adults with impaired glucose tolerance. DiPietro 2013 demographic; substantial 24-hour glycaemic improvement, and an additional benefit on nocturnal hyperglycaemia after a dinner walk DiPietro et al. 2013.
• Metabolically healthy adults. Smaller absolute effect because the peak being flattened is smaller; the practice remains net-positive through activity accumulation and triglyceride attenuation Miyashita et al. 2008 Buffey et al. 2022.
• Type 1 diabetes. Practice works mechanistically; insulin dosing must be adjusted. Manohar 2012 included T1D and healthy participants and observed comparable contraction-stimulated glucose disposal Manohar et al. 2012.
• Pregnancy (including gestational diabetes). Light post-meal walking is widely recommended in obstetric practice for glycaemic control; no special restriction beyond standard pregnancy-exercise guidance.
• GERD. Mixed. Light upright walking generally helps; brisk activity or bending immediately post-meal may worsen.
• Gastroparesis, post-bariatric dumping syndrome. Individual titration; the accelerated-emptying effect may be counterproductive.
• Children and adolescents. Evidence base is thin (most trials in adults), mechanism is identical; family practice in many cultures already includes the after-dinner walk.
• Shift workers / night-meal eaters. Night-time meals produce larger glycaemic excursions due to circadian insulin-sensitivity decline; post-meal walking is plausibly more impactful but trials are sparse.

Knowledge gaps

• Hard-endpoint trials. No randomised controlled trial has powered for cardiovascular or all-cause mortality on a post-meal-timed activity protocol versus an equal-dose untimed protocol. The mortality inference rides on the established postprandial-glucose-to-mortality observational link DECODE Study Group 2001. A multi-year hard-endpoint trial is unlikely to be funded.
• Optimal window precision. The trial evidence collapses the window to "within 30 minutes" with the strongest signal in the first 15. Whether 5 minutes is meaningfully better than 20 is not well resolved.
• Dose-response below 10 minutes. Buffey 2022's "light activity break" literature suggests 2–3 minute interruptions every 30 minutes have measurable effects; the threshold at which a post-meal "walk" becomes too short to count as such is not crisp Buffey et al. 2022.
• Pace-by-meal-size interaction. Few trials systematically vary pace by meal-glycaemic-load. The clinical intuition (slower after larger meals, brisker after lighter ones) is unstudied.
• Children, adolescents, and pregnancy. Mechanistically the same; trial evidence is thinner than in adults with type 2 diabetes.
• Long-term adherence determinants. The behavioural-design hypothesis (meal-anchored habits persist) has strong general support but limited direct study in the post-meal-walking case.
• What would change the call. A large RCT showing no incremental benefit of timing once total walking dose is matched would soften the timing recommendation toward "any walking will do." A trial showing harm in a sizable GERD subgroup would tighten the contraindication. Neither is on the horizon.

Brief coverage. The topic brief named five consequences — post-meal blood glucose, insulin response, digestion and reflux, triglycerides, and daily activity accumulation. The article covers all five. Reflux is treated honestly as direction-dependent (light upright walking helps; brisk or bent-over walking with a full stomach hurts) rather than silently dropped or oversold.

Rating call: evidence at 4, not 5. The glucose-flattening effect itself is multiply replicated (Reynolds 2016, DiPietro 2013, Pahra 2017, van Dijk 2013) and meta-analysed (Engeroff 2023, Buffey 2022), with a guideline statement (ADA 2016). What's missing for a 5 is a hard-endpoint trial — cardiovascular mortality on the post-meal-timed protocol versus a dose-matched untimed control. The mortality inference rides on the established postprandial-glucose-to-CV-mortality observational link (DECODE 2001). A 5 would require RCT mortality data; this is the conservative call.

Rating call: beauty_cumulative and mood scored 0, not 1. Both are theoretically real (AGE/glycation reduction over years; mood lift from daylight + steadier glucose), but neither earns a paragraph in the article body, and forcing one in would be padding the case beyond the substance's honest reach. The cross-coverage rule (every non-zero dimension gets a home in the prose) is what kept these honest at 0. Walking after meals is not a beauty intervention or a mood intervention; it incidentally nudges both, by a margin too small to recommend the practice on.

Rating call: sleep at 1. Real but modest. The dinner-specific bout reduces nocturnal glucose and aids gastric emptying before bed; effect is meaningful but limited to the evening walk, so 1 rather than 2. Covered in the payoff section.

Rating call: focus at 2. Tight call between 1 and 2. Argued for 2 on the grounds that the reactive-hypoglycaemia mechanism (Buffey 2022) is the same one driving the post-lunch dip, which is a real and felt cognitive effect for many adults. Smaller than the energy effect, hence 2 rather than 3. Covered glancingly in the stakes section ("3pm distraction" line).

Separate-entry candidates.

• Soleus push-ups — the Hamilton 2022 seated-calf-raise protocol has a distinct enough evidence base, mechanism story, and use-case (desk-bound, plane, long meeting) to warrant its own entry. Mentioned here as the walk-impossible fallback; flagged for the backlog.
• Breaks in prolonged sitting / movement snacks — Dunstan 2012 / Buffey 2022 cover an adjacent but distinct intervention pattern (any-time light movement vs meal-anchored walks). Distinct entry.
• Pre-meal fibre / protein order for glycaemic flattening — different mechanism (gastric-emptying delay, incretin response), different choice point (food rather than activity). Distinct entry.
• Continuous glucose monitoring for non-diabetics — the on-trace feedback loop that has popularised this practice deserves its own entry on cost / signal-quality / who-it's-for grounds.

Future links. Once they exist: soleus-pushups, breaks-in-prolonged-sitting, pre-meal-fibre, continuous-glucose-monitoring, standing-desks, oral-glucose-tolerance-test, and adjacent metabolic entries (postprandial-triglycerides if it lands as its own entry).

Voice / dream-tier call. Overall score lands ≈ 49 (after the beauty / mood reset), still within the obligatory-dream-narrative band (40+). The dek leads with the 2pm slump (highest-leverage immediate hook), and projects out to the decade-scale vascular hinge. The tagline compresses to the single felt-experience line — afternoons coming back rather than vascular age, on the grounds that the immediate hook is the one a casually-scrolling reader can verify in a week. The mechanism section drops its <h2> to flow directly out of the dek per the editorial pattern in the article spec.

Honest framing on "comparable to metformin." This phrasing appears in the dek and the evidence callout. It is hinged: Reynolds 2016 and Pahra 2017's effect sizes on postprandial glucose are in the ballpark of a starting metformin dose (≈500 mg). It is not a claim that the long-run mortality benefit is equivalent — that comparison would not be evidence-supported. The phrasing names the magnitude honestly without overpromising the downstream.