The reset, the plan and the working week

PART FOUR — THE PROGRAMME

Everything before this page was argument. This part is scheduling.

Part One established why the plate changed. Part Two rebuilt the kitchen. Part Three gave you 120 recipes with the oil weighed and the sodium counted. None of that becomes a lower waist on its own, because the two best-designed South Asian lifestyle programmes ever run both proved that knowing is not doing. K-DPP moved diet quality, alcohol and waist circumference in Kerala and still missed its primary endpoint at two years and at nine (Thankappan et al., 2018; Jeemon et al., 2026). SAHELI improved diet quality, activity and self-efficacy in 549 South Asian Americans and moved weight by 0.07 kg (Kandula et al., 2024).

So this part is not more information. It is a sequence: one week to install four habits, four weeks to run them properly, a Sunday that makes the working week possible, a strategy for the social world that decides whether any of it survives, and an honest chapter on the situations where the plan needs a clinician's hand on it.

Nothing here is compulsory in order. If you have read this far and cooked nothing, start at Chapter 28. If you already cook this way and want the menus, start at Chapter 29. If your obstacle is a wedding season rather than a recipe, read Chapter 31 first.

Chapter 28 — The 7-Day Reset

Begin with what this week is not, because the word "reset" has been ruined by people selling things.

Nothing here is designed to purify you of anything. Your liver and your kidneys clear what needs clearing, continuously and without dietary assistance, so if you came looking for the sort of seven-day purge the wellness aisle sells, this is not one. It is not a fast either, and no meal is skipped in it. It is not a fixed-calorie prescription — you will not find a daily number here, for the reasons Chapter 5 set out, which is that at twelve months the variable that predicts outcome is whether you are still doing the thing, not the precision of the arithmetic (Gardner et al., 2018; Johnston et al., 2014). It is not a week of special food. Every dish in it is a dish somebody's grandmother made.

What it is: seven days to install four habits, one of which is the habit of measuring. Those four are the plate, the pulse, the measured oil and the weekly measurement. That is the whole content. If at the end of the week you are doing those four things without thinking about them, the week worked, whatever the scale says.

The design rule is one new thing per day. Not because you could not manage two, but because a week that asks for seven changes on Monday reliably produces zero changes by Thursday, and the cost of a failed start is not neutral — it teaches you that this does not work for you.

Before you begin: what to have in the house A digital kitchen scale reading to 1 g — the highest-return purchase in a fat-loss kitchen, as Chapter 13 argues, and cheaper than a week's fish. A 23 cm (9 inch) dinner plate, which is the plate the Kerala Plate means. A flexible tape measure. A level teaspoon you can find without looking. That is the equipment. Nothing else in this week requires anything a Kerala kitchen does not already own.

Sunday: the preparation block

Ninety minutes to two hours, before the week starts. Chapter 30 maps a full three-hour Sunday for people running the 28-day plan; this is the short version.

Shop from the list at the end of this chapter. Then, in this order: soak 200 g of vanpayar and 200 g of kadala in separate bowls of water. Make M15 · Base Shallot Masala to Freeze and freeze it in 50 g cubes — this is twenty-five of the forty minutes a weeknight curry takes, removed in advance. Make M16 · Ginger-Garlic-Green Chilli Paste and freeze it in 15 g portions. Grind M11 · Idli and Dosa Batter and set it to ferment. Cook the soaked pulses, drain them, and refrigerate them in 100 g portions.

Then take the baseline, and take it properly. Weigh yourself first thing, after the lavatory, before eating, in the same minimal clothing you will use every week. Measure your waist at the midpoint between the lowest rib and the top of the hip bone, tape horizontal, at the end of a normal breath out, not sucked in. Write both numbers down with the date. Chapter 3 explains why the waist matters more than the weight for a South Asian reader, and why the number that matters most is the ratio of waist to height, with 0.5 as the boundary (Ashwell et al., 2012; Misra et al., 2025).

The seven days

Day 1 — the plate. Today's only instruction: at lunch and at dinner, half the area of a 23 cm plate is vegetables. Not half the calories, not half the weight — half the area, which is a thing you can see. The rice occupies a quarter and the protein a quarter. Nothing is banned and nothing is counted. This works for a mechanical reason rather than a moral one: people eat a fairly consistent weight of food, so when the energy density of what is available falls, energy intake falls without conscious restraint (Rolls, 2009). Kerala cooking is unusually well supplied with the instrument — a thoran is mostly vegetable and a spoon of coconut.

Day 2 — the pulse. Keep yesterday's plate. Add one thing: at least 50 g of cooked pulse on every main plate, today and every day after. This is the single most specific edit in the book, and it comes from a Chennai trial in which brown rice with 50 g of legumes a day cut the glycaemic incremental area under the curve by 22.9% against white rice, where brown rice alone managed 19.8% (Mohan et al., 2014). The pulse goes on top of the better grain, not in place of it. Pulses also lowered LDL cholesterol by 0.17 mmol/L at a median intake of about 130 g/day across 26 trials (Ha et al., 2014). Your Sunday pulses are already cooked and portioned, so today this costs you nothing but the spoon.

Day 3 — the measured oil. Keep the plate and the pulse. Today, weigh the oil once. Put your usual pour into a bowl on the scale and read the number. Most Kerala kitchens are at 25–40 g in a single pan; the recipes in Part Three run at 5–15 g. From today, oil is added with a level teaspoon, and coconut oil in particular is added at the end rather than the start, because a half-teaspoon stirred in off the heat is perceived far more strongly than the same amount cooked in from the beginning. That is the highest-leverage trick in the book and it is free.

Day 4 — protein at breakfast. Keep the three. Add a fourth: breakfast carries protein. A Kerala breakfast of appam and a thin curry can arrive at 6 g; the target is 25–30 g per meal, on the way to 1.2–1.6 g/kg across the day (Leidy et al., 2015). Today's breakfast is R014 · Egg Roast, Rebuilt with R004 · Wheat and Coconut Puttu. The general mechanism: a pulse curry, an egg dish or fish beside the puttu or dosa, every morning, rather than a chutney alone.

Day 5 — the drink audit. Keep the four. Today, count what you drink. Two teaspoons of sugar in each of four cups of tea is roughly 32 g of sugar, which is a payasam's worth of sugar spread invisibly across a day with no meal status in your mental accounting. R106 · Chai, and the Sugar Arithmetic does the sums. Today's replacements: R111 · Sulaimani, R107 · Sambharam, R110 · Lime and Nannari Sharbat, Unsweetened. Reduce the sugar by a quarter rather than to nothing; the palate follows within a fortnight, exactly as the salt palate does in Chapter 10.

Day 6 — the meal you did not cook. Keep the five. Today you eat one meal you did not cook, deliberately and by plan, because a programme that only works when every meal comes out of your own pan is a programme that ends at the first invitation. Decide before you go what you will order — this is the implementation-intention move of Chapter 31 — and decide in advance what you will leave. Chapter 31 gives the scripts.

Day 7 — the measurement. Keep all six. Today you repeat Sunday's measurement and then put it in a diary as a permanent weekly appointment. This is the habit that carries the rest: across 449 adults in a sixteen-week programme, the proportion of days on which a person weighed themselves correlated with weight lost at r = −0.56 (Shetty and Ross, 2025), and in maintenance it was the consistency of the routine that predicted less regain rather than its raw frequency (Brockmann et al., 2020). Same day, same hour, same conditions, recorded.

The week's menus

Day 6's dinner uses the chilled-and-reheated rice method. Read its caution before you use it: chilling cooked rice for 24 hours at 4°C (39°F) and reheating it raises resistant starch and lowers the glucose response, and in 32 adults with type 1 diabetes on insulin pumps it produced 12 hypoglycaemic episodes against three because people dosed insulin for the carbohydrate count they were used to (Strozyk et al., 2022). If you use insulin or a sulfonylurea, that is a conversation with the clinician who set your dose, not a technique to adopt quietly. This book does not ask anyone to change a medicine.

The shopping list

For a household of four, seven days. Quantities assume the Sunday block above.

Read this before you buy potassium salt M07 · The 70:30 Salt is the strongest single-ingredient swap in this book: in 20,995 people over a mean of 4.7 years, a 70% sodium chloride / 30% potassium chloride substitute cut stroke by 14% and all-cause mortality by 12% (Neal et al., 2021). It is also not for everyone. Do not use it, or any potassium-based salt substitute, without your doctor's approval if you have chronic kidney disease, or take a RAAS inhibitor — an ACE inhibitor (-pril, such as ramipril) or an angiotensin receptor blocker (-sartan, such as telmisartan) — or a potassium-sparing diuretic (spironolactone, eplerenone, amiloride), or a potassium supplement. All of these raise the risk of hyperkalaemia. If you are unsure what you take, ask the pharmacist who dispenses it.

What success looks like after seven days

Honest expectations, because the alternative is a reader who concludes on day eight that this does not work.

What you should see. Four habits running without effort. A kitchen with cooked pulses in it. A record with two dated numbers in it. A palate that has already begun to notice that Friday's tea tasted very sweet.

What the scale will probably show. Between 0.5 kg and 2 kg down, and most of that will not be fat. When carbohydrate intake and total food volume fall, glycogen and its bound water go with them, along with the sheer mass of food in transit; the number on the scale moves fast in week one and then stops, and the stopping is what people misread as failure.

Do the arithmetic on the fat, and it makes the point cleanly. The defensible rate of fat loss is 0.5–1.0% of body weight per week, so a 75 kg reader is entitled to expect 0.4–0.75 kg in a week — under a kilogram, invisible in a mirror, and inside the noise of a bathroom scale. Adult body weight responds to a change in intake with a half-time of about one year (Hall et al., 2011). One week is a rounding error, by design.

What you should not expect. Anything you can see. Anything a photograph would show. And nothing resembling the seven-day liver-fat result you may have read about: liver fat did fall by roughly a third within seven days in Taylor's Counterpoint study, but that was on about 800 kcal a day of formula under clinical supervision (Lim et al., 2011, via Taylor, 2025). This week is not that, and does not pretend to be.

What the week is actually for. In POUNDS LOST, attendance at group sessions predicted weight loss at 0.2 kg per session attended while the macronutrient split predicted nothing at all (Sacks et al., 2009). Turning up is the mechanism. This week's product is not a kilogram; it is a routine that can still be running next Onam.

Chapter 29 — The 28-Day Kerala Plate Plan

Four weeks, four themes, and one job per week.

Week 1 fixes the plate. Week 2 fixes the grain. Week 3 fixes the fat and the salt. Week 4 fixes the social world, which is the one that decides whether weeks 1 to 3 survive.

Each week gives seven days of breakfast, lunch, dinner and one planned snack; a shopping list; a Sunday batch block naming the mother preparations to make; one new habit; and the specific target that week is meeting. Every dish is cross-referenced by its R- or M-number. Nothing in the four weeks is repeated wholesale from week to week, because a plan that is week one four times is a plan people stop reading in week two.

Two rules that run across all four weeks. First, the grain is weighed cooked, not raw — this matters because the PURE cohort's rice association was measured in cooked grams, and in South Asia the comparison of 450 g or more a day against under 150 g gave a hazard ratio of 1.61 (95% CI 1.13–2.30) for incident diabetes (Bhavadharini et al., 2020). Second, a pulse goes on every main plate, minimum 50 g cooked, for Mohan's 22.9% (Mohan et al., 2014).

Week 1 — The Plate

The week's new habit: half the area of a 23 cm plate is vegetables at every main meal, and you serve yourself onto that plate rather than eating from a serving dish.

The target being met: a plate's worth of vegetables and a pulse at every main meal, and protein arriving at roughly 1.2 g/kg for the first time. For a 70 kg reader that is about 84 g of protein a day, which this week's menus reach through fish five times, egg three times, chicken twice and a pulse at every plate. The mechanism is energy density: lower it and intake falls at constant food weight (Rolls, 2009).

Batch Sunday: M15 · Base Shallot Masala to Freeze (50 g cubes, 3 months); M08 · Coconut Milk, Three Extractions (first milk in an ice-cube tray at about 30 ml a cube); M02 · Sambar Podi; M16 · Ginger-Garlic-Green Chilli Paste (15 g portions). Cook and portion 500 g of vanpayar and 400 g of kadala.

Week 1 shopping list

Week 2 — The Grain

The week's new habit: the cooked grain goes on the scale before it goes on the plate. Once. After a week your eye is calibrated and you can stop. The portion this book works to is 125 to 150 g of cooked rice per main meal for most adults — 48 to 58 g dry — which is a quarter of a 23 cm plate and roughly half what a Kerala lunch usually carries. Two rice meals a day at that size come to 250 to 300 g of cooked rice, inside PURE's middle bands. Someone doing physical work can carry 200 g without apology.

The target being met: glycaemic load falling while the grain stays. Low-GI eating delivers roughly 0.3–0.5% off HbA1c across 54 randomised trials — larger than any spice in this book (Zafar et al., 2019; Ojo et al., 2018). Measured in Kerala itself, red rice puttu came in at a glycaemic index of 38 and matta rice at 38, against 45 for a whole-wheat roti, with HbA1c falling 0.93% over the trial (Pavithran et al., 2020). Polish is what matters: parboiled rice runs at a GI of 57.6 unpolished and 79.6 polished. This week rotates the grain deliberately — matta, ragi, millet, wheat, oats, the chilled rice method, and rice-with-dal — so that no single form of starch carries the whole week.

Batch Sunday: M11 · Idli and Dosa Batter; M12 · Ragi-Matta Dosa Batter; M13 · Fish Stock from Heads and Frames (freeze 3 months in 250 ml portions); M06 · The Low-Sodium All-Purpose Podi. Cook 400 g muthira and 400 g vanpayar. Portion cooked matta rice into 130 g bags and freeze.

Week 2 shopping list

Week 3 — The Fat and the Salt

The week's new habit: the oil is weighed into the pan, not poured, for all seven days — and the salt cellar is either measured or swapped.

The target being met: sodium falling and saturated fat falling, with flavour held. Kerala's own state survey measured salt intake at 6.7 g/day against WHO's 5 g target, in a state where only 12.4% of people with hypertension have it controlled (Sarma et al., 2019). The DASH pattern lowered blood pressure by 11.4/5.5 mmHg in people with hypertension with no change in salt and no weight loss at all (Appel et al., 1997), and DASH at low sodium came in 11.5 mmHg below a typical diet at high sodium (Sacks et al., 2001). Sodium reduction alone gives −3.39/−1.54 mmHg, and the gain is non-linear: most of it arrives in the step below about 2.3 g of sodium a day, which is the step most Indian households never take (Aburto et al., 2013).

Kerala already wins the other half. Raising potassium lowered systolic pressure by 3.49 mmHg, reaching 7.16 mmHg at 90–120 mmol a day, and was associated with 24% lower stroke risk, though the blood-pressure benefit appeared only in people already hypertensive (Aburto et al., 2013). Tapioca, plantain, yam, chena, chembu, drumstick, jackfruit, coconut water, every thoran green and every pulse are potassium-dense. The lever is keep the potassium, cut the salt — the achar, the unakka meen, the pappadam, the packaged masala and the parcel food, not the vegetables.

On the fat side, run Mensink's arithmetic. Replacing 1% of energy from saturated fat lowers LDL by 1.8 mg/dL with polyunsaturated fat and 1.3 mg/dL with monounsaturated (Mensink et al., 2003). Coconut oil is 90.9% saturated; groundnut, sesame and rice bran run at 16–24%. Moving 20 g a day from one to the other is an estimated 8–12 mg/dL off LDL, consistent with the 16 pooled trials in which coconut oil raised LDL by 10.47 mg/dL against non-tropical oils (Neelakantan et al., 2020). Coconut flesh stays: about 10 g of fibre and 3.8 g of protein per 100 g, where the oil carries neither.

Batch Sunday: M09 · Varutharacha Thenga (roasted coconut paste, 30 g portions, freeze 3 months) — this is the week's flavour engine, because a roasted-coconut suspension delivers body without an emulsion and without added oil; M07 · The 70:30 Salt, if it is appropriate for you and only after reading the carve-out below; M06 · The Low-Sodium All-Purpose Podi; M14 · Vegetable Stock, Kerala Aromatics (freeze 6 months). Cook 400 g kadala and 400 g muthira.

The potassium carve-out, restated because it must appear every time M07 · The 70:30 Salt is contraindicated, or needs medical supervision, in chronic kidney disease and for anyone taking a RAAS inhibitor (ACE inhibitors ending -pril; angiotensin receptor blockers ending -sartan), a potassium-sparing diuretic (spironolactone, eplerenone, amiloride) or a potassium supplement, because all of these raise the risk of hyperkalaemia. SSaSS screened its population for renal function; you are not screened. If any of this describes you, do not make the blend, do not buy a commercial low-sodium salt, and raise the question with your doctor before you change anything. Cut ordinary salt instead, which needs no permission from anyone.

Week 3 shopping list

Week 4 — The Social World

The week's new habit: one implementation intention, written down, for every social eating occasion in the week ahead. The form is fixed: when X happens, I will do Y. Not "I will be careful at the wedding" but "when the biryani plate arrives I will eat the chicken and leave half the rice, and I will ask for the salad by name". Chapter 31 is the full treatment; this week is the rehearsal, and it deliberately contains a hosting day and a feast day rather than avoiding them.

The target being met: the whole set, held across a week that includes celebration. Protein at 1.2–1.6 g/kg, fibre high, sodium down, a pulse on every plate — and one leaf of sadya inside the four weeks, on purpose. DiRECT's remission gradient tracks sustained average weight loss, not perfection: nobody who gained weight went into remission, but losing 5–10 kg over a year bought better than one chance in three (Lean et al., 2018). One feast does not move a sustained average. Four unmanaged weeks do.

Batch Sunday: M04 · Nadan Chicken Masala; M01 · Kerala Garam Masala; M03 · Rasam Podi; M05 · Fish Masala; M10 · Appam and Palappam Batter (do not freeze the batter — the starch network does not recover; freeze cooked appams for 1 month instead). Cook 500 g kadala and 400 g vanpayar for the week and for the sadya.

Day 28's dinner is deliberately thin, and not as penance. A leaf at midday is a large meal; kanji, moru and a sprout salad in the evening is what a Kerala household actually eats after a sadya, and it has been the correct answer for about four hundred years.

Week 4 shopping list

When you fall off

You will. Plan for it now, while you are calm, because the protocol you need is the one you can follow on a bad day rather than the one you would design on a good one.

First, the arithmetic, which is not a kindness but a fact. Adult body weight responds to a change in intake with a half-time of about one year (Hall et al., 2011). A single day of eating 1,500 kcal more than usual is, at the margin, roughly 150 g of tissue if none of it is spent, and much of what appears on the scale next morning is water, sodium and food in transit. One day is inside the noise. What matters is the sustained average, and that is what DiRECT's gradient was built on: nobody who gained weight over twelve months went into remission, and losing 5–10 kg bought a 34% chance of it (Lean et al., 2018).

Second, the thing readers systematically get backwards. A single day matters far less than you fear. A fortnight of drift matters far more than you think. Two weeks of not-quite-doing-it does not show up as a dramatic event; it shows up as a plan that quietly stopped being the plan, and the National Weight Control Registry's maintainers are notable precisely for keeping the same eating pattern from day to day and from weekday to weekend. The failure mode is not the wedding. It is the eleven days after the wedding in which nobody cooked a pulse.

So: on the day itself, do three things and no others.

Eat the next meal on the plan. Not a smaller meal, not a skipped meal, not compensation. The next meal is a normal meal from the current week's table. Compensating with restriction is how one day becomes a cycle, and it has the additional cost of making the plan feel punitive, which is how plans die.

Do not weigh yourself tomorrow. Weigh on your usual day, at your usual hour. A measurement taken to confirm a feeling is not a measurement; it is a mood with a decimal point.

Write one line. What happened, and what would have to be true for it not to happen the same way next time. That is not journalling for its own sake; it is how you discover that four of your five difficult days were the same situation — Friday evening, the office canteen, the third day after a long flight — which converts an apparently moral problem into a scheduling problem, and scheduling problems are solvable.

Then the framing, stated once and plainly. This programme has no moral register. There is no dish in Kerala cooking that is a lapse, and nothing here that you can fail at. There is a plate that produces a particular result when it is what you mostly eat, and arithmetic about how much of the time "mostly" has to mean. Anyone selling you shame is selling you something, and the evidence says it does not work: what predicted weight loss in POUNDS LOST was attendance, at 0.2 kg per session (Sacks et al., 2009). Turning up, not resolve.

One case is different, and it deserves saying separately. If the drift has lasted a month or more and you notice that you cannot get started again, that is not a willpower question and it is not what this book can fix. Low mood, disordered eating and a relationship with food that has become distressing are all common, all treatable, and all better addressed by a clinician than by a menu plan. Say it to your doctor.

What to measure, and when

Four numbers, on a fixed schedule, and no others.

The waist is the primary instrument in this book, and Chapter 3 explains why: waist-to-height ratio improved discrimination of cardiometabolic risk by 4–5% over BMI across more than 300,000 adults, and beat waist circumference alone for diabetes, hypertension and cardiovascular disease in both sexes (Ashwell et al., 2012). For a South Asian reader the thresholds are ≥90 cm in men and ≥80 cm in women, with the ratio above 0.5 (Misra et al., 2025). The waist also moves early, because the visceral and hepatic depots empty first — which is why the tape moves before the mirror does and the blood tests move before the tape.

Measure the waist properly or do not bother. Standing, feet together, at the end of a normal exhalation, tape horizontal all the way round and snug without compressing. Same time of day, because it varies. Take it twice and use the mean. Expect 0.5 to 1.5 cm over four weeks if the plan is running; for a floor, Kerala's peer-led community programme moved mean waist from 89.5 to 87.5 cm over twelve months with no formal diet at all (Ravindranath et al., 2020).

Now the warning, which matters more than the routine. The self-weighing evidence is favourable — r = −0.56 between the proportion of days weighed and weight lost across 449 adults (Shetty and Ross, 2025) — and it is regularly misread as an instruction to weigh daily and mind the result. In maintenance, what predicted less regain was the consistency of the routine, while raw frequency predicted nothing (P=0.141) (Brockmann et al., 2020). Consistency is the active ingredient; frequency is the packaging.

So daily weighing is optional, and for some readers it is harmful. A bathroom scale fluctuates by 1–2 kg with hydration, sodium, the menstrual cycle, bowel content and last night's salt, none of which is fat. If reading that noise every morning lowers your mood or changes what you eat, the measurement costs more than it returns. Weekly, on a fixed day, is enough. If you find yourself weighing twice a day, or a number decides whether the day was good, stop weighing, keep the tape, and tell someone.

One last measurement, and it is the honest one: can you still do this? At the end of four weeks, the question that predicts the next twelve months is not how much you lost. It is whether the four habits are still running without effort, and whether you would be willing to do another four weeks. If the answer is yes, the number will come. If the answer is no, the plan was too strict and the fix is to loosen it rather than to abandon it.

Chapter 30 — Batch Cooking and the Working Week

The reason people stop cooking Kerala food on weekdays is not that it is difficult. It is that a nadan curry involves twenty-five minutes of shallots before anything recognisable as cooking has begun. Move those twenty-five minutes to Sunday and the same curry is a fifteen-minute dish. That is the whole idea, and the payoff is larger than convenience: in an inpatient crossover trial with meals matched for presented calories, energy density, macronutrients, sugar, sodium and fibre, people ate 508 kcal more per day on the ultra-processed arm and gained 0.9 kg in two weeks, losing 0.9 kg on the unprocessed arm (Hall et al., 2019). Cooking from whole ingredients is itself the intervention. A Sunday that makes weekday cooking possible is therefore not housekeeping. It is the mechanism.

Sunday's three hours, mapped

Two and a half to three hours, in this order, with the slow things started first. Elapsed minutes, not clock times, so you can start when you like.

0–10 minutes. Put water on to boil. Weigh and rinse 500 g of vanpayar and 400 g of kadala into separate pans, cover with water, bring to the boil and leave them. Set 200 g of muthira to soak — it needs eight hours, so today it soaks for tomorrow. Clear a fridge shelf and a freezer drawer.

10–25 minutes. Peel 800 g of shallots, and do not resent it — this is the highest-value twenty minutes of the week. Peel 150 g of ginger and 4 heads of garlic. Grate the coconuts, or weigh out frozen grated coconut.

25–55 minutes. Start M15 · Base Shallot Masala to Freeze. The shallots go into a wide pan with 15 g of oil over a medium flame and cook down slowly while you do everything else. Meanwhile make M16 · Ginger-Garlic-Green Chilli Paste in the mixie, and portion it into 15 g blobs on a tray for the freezer. Check the pulses; they will be 20–30 minutes from done.

55–75 minutes. Extract coconut milk — M08 · Coconut Milk, Three Extractions — and get the first milk into an ice-cube tray at about 30 ml a cube. Second and third extracts go into a jar for tonight and tomorrow. Drain the pulses, spread them on a tray to cool fast, then portion into 100 g bags. Stir the shallots.

75–105 minutes. The masala is now soft and going brown. Finish it, cool it, and portion it into 50 g cubes. Grind the batter you need — M11 · Idli and Dosa Batter or M12 · Ragi-Matta Dosa Batter — and set it somewhere warm to ferment overnight. If it is a varutharacha week, roast and grind M09 · Varutharacha Thenga now, while the pan is already hot, and portion it into 30 g pieces.

105–135 minutes. Cook the week's rice. R015 · Matta Rice, Properly Cooked in a large batch, spread shallow, chilled within one hour, portioned into 130 g bags. Simultaneously, make one main protein dish that improves with keeping: R010 · Kadala Curry freezes for 3 months and is better on day two; R042 · Pepper Chicken freezes for 2 months and is the best freezer dish in its chapter; R048 · Beef Ularthiyathu, Traditional and Rebuilt braises now and roasts in twelve minutes on a weeknight.

135–165 minutes. The vegetable work. Cut and bag what keeps raw — beans, cabbage and carrot for 2 days, muringayila leaves stripped and bagged for 2 days, okra washed and dried for 1 day. Parboil chena and koorka, which both freeze better boiled than glazed and glaze better dry. Make one thoran that improves with keeping: R067 · Beetroot Thoran holds 4 days in the fridge and 2 months in the freezer, and is better on day two.

165–180 minutes. Label everything with contents and date, because an unlabelled bag of orange paste is a bag you throw away in March. Wash up. Write the week's menu where whoever cooks can see it.

What freezes, and what does not

The rule underneath the table is simple: suspensions freeze, emulsions do not. A roasted-coconut paste is a suspension of solids and survives; first-extract coconut milk is an emulsion stabilised by protein and splits on thawing. Curd splits worse. Boiled egg white turns rubbery and weeps. Leafy greens thaw as grey water.

The mother preparations in rotation

You do not make seventeen mother preparations every Sunday. You make four, on a rotation, and the freezer carries the rest. Here is what each unlocks, which is how to decide what this Sunday needs.

Containers, and the discipline of a fixed weight

Two container rules, both boring and both load-bearing.

Shallow beats deep. Cooked rice must get from 60°C (140°F) to 4°C (39°F) fast, and a hot deep box in a crowded fridge does not manage it. Use flat containers or flat bags, spread the food, and refrigerate within one hour of cooking. This is a food-safety rule before it is a texture rule.

Portion the rice by weight, once, and then stop weighing. Weigh cooked matta rice into fixed portions and freeze them flat: 125 g for a light meal, 150 g for a standard main plate, 190 g for the reworked ghee rice, 290 g for khichdi where rice and dal share the portion. After a fortnight your eye knows what 150 g looks like on a 23 cm plate and the scale can live in the drawer. The reason to bother sits in PURE, where the exposure that carried a South Asian hazard ratio of 1.61 was measured in cooked grams, not cups or servings (Bhavadharini et al., 2020).

Glass or steel for anything acidic, non-metal for kudampuli and tamarind curries, and a running list on the freezer door — because a freezer you cannot inventory is a cupboard you are heating.

Reheating, by type of dish

The chilled-rice method, and its caution repeated

Chapter 7 and R016 · The Chilled-and-Reheated Rice Method carry the full account. In short: chilling cooked rice for 24 hours at 4°C (39°F) raises resistant starch from about 0.64 to 1.65 g per 100 g and lowers the glucose area under the curve by roughly 18%. It is the only free glycaemic lever in the book and it fits a batch-cooking Sunday perfectly, because freezing after the 24-hour chill preserves the retrograded starch and resets the safety clock.

And the caution, briefly, because it must travel with the technique every time. In 32 adults with type 1 diabetes on insulin pumps, chilled and reheated rice produced 12 hypoglycaemic episodes against three (P=0.0039), because participants dosed insulin for the carbohydrate count they were used to and the food behaved differently (Strozyk et al., 2022). Sulfonylureas — glimepiride, gliclazide, glibenclamide — carry the same logic, since they raise insulin whether glucose is rising or not. If you use either, this method changes the glucose response of a food whose dose is already calibrated, and that is a conversation with the clinician who set the dose. Nothing here is an instruction to alter a medicine.

Two food-safety rules that are not negotiable: cooked rice goes into the fridge within one hour, spread shallow, and it is reheated once, to steaming hot right through. Cooked rice held warm is where Bacillus cereus lives, and no glycaemic benefit is worth that trade.

The limits of meal prep

An honest section, because a chapter that claims everything keeps is a chapter that produces a fridge full of food nobody wants to eat.

Worse on day three, so cook it fresh. Any leafy thoran, cheera above all — it darkens by day two and is grey by day three. Okra, which softens into slime. Appam, dosa and puttu, all fifteen-minute foods if the batter or crumb is ready. Any first-extract coconut-milk curry: moilee, olan, the finishing stage of a stew. Every curd dish. Liver. Rasam, at its best within ten minutes of the tempering and a sour soup within the hour. Prawns and squid, tender to rubber on the second heating. Fried fish, which is a made-to-order food and always was.

What genuinely improves. Kadala curry, muthira curry, kalan — the make-ahead dish of the sadya and the reason it exists — beef ularthiyathu, ulli theeyal, pavakka theeyal, sambar for two days, beetroot thoran, Kerala chana masala, and meen vevichathu, unambiguously.

So the workable division is not "cook everything on Sunday" but cook the slow half on Sunday and the fast half on the night. Sunday makes the pulses, the masala base, the batter, the stock, the coconut paste, the braise and the rice. Tuesday makes the thoran, eight minutes with pre-cut cabbage, and warms the curry that has been waiting for it. That is a twenty-minute weeknight dinner with four components on the plate, and it is the only version of this people are still running in November.

Chapter 31 — The Social World

This is the chapter that decides whether the rest of the book survives contact with a Malayali life.

Start with the fact everything else has to be built around, because getting it wrong is why most dietary advice fails in this culture. In Kerala, feeding is love. Not a metaphor for love, not a vehicle for it — the act itself. A mother who has cooked for four hours and is holding a ladle over your leaf is not offering you calories; she is offering you the thing she has to give. When you decline, the message received is not "I am watching my carbohydrate." It is "I do not want what you have for me." And the second-helping pressure — onnu koodi, one more — is not a lapse in manners. It is the grammar of hospitality working exactly as designed.

So the strategy cannot be built on refusal, and cannot be built on willpower. Serving in a Malayali household is done by the host, not the eater, so portion size is socially determined rather than individually determined. That sentence reorganises the whole chapter. If the ladle is in someone else's hand, the intervention has to operate on what is offered, not only on what is declined — which means the highest-leverage position in a Kerala household, for anyone trying to change what they eat, is the position of the person cooking.

The supporting evidence is not decorative. SAHELI's process evaluation found behaviour change diffusing to close family members who were never enrolled in the trial, and also found participants meeting resistance from some community members, with the pattern differing by gender and by English proficiency (Lam et al., 2025). Change moves through households. It also gets pushed back on. Both facts argue for working on supply.

The three tools, and their honest evidential status

Implementation intentions. Convert an intention into a specific if-then plan: not "I will be careful at the reception" but "when the biryani plate is put in front of me, I will eat the chicken and the raita and leave half the rice." Made in advance, in writing, in the form when situation X arises, I will do behaviour Y. Its power is that at the moment of decision there is nothing left to decide.

Environment design. Change what is within reach rather than what you are resisting. This tool has the cleanest mechanism behind it, and it is not psychology: people eat a fairly consistent weight of food, so when the energy density of what is available falls, intake falls without conscious restriction, and adding water-rich foods produced substantial weight loss in trials where nobody was told to restrict anything (Rolls, 2009). Applied to a Kerala house: what is on the table when people sit down decides most of the outcome, and a table with two thorans and a pulse does the work that a table with one thoran and three fried items asks willpower to do. Note that the popular laboratory literature on plate size and food cues includes retracted work; this book does not cite it.

Habit stacking. Attaching a new behaviour to a reliable existing one — the pulse goes on to soak when the evening tea goes on; the tape comes out on the morning you weigh. Presented as craft rather than evidence: useful, cheap, and with published effect sizes thinner than its popularity suggests.

And the fourth tool, which comes from the anthropology of the room rather than the behavioural literature: change what is offered.

Weddings and the sadya

Chapter 11 gives the plate-level tactics and Chapter 27 the full sadya architecture. What belongs here is the social mechanics.

At a sadya, the leaf has no edges, so bring your own. Decide before you sit that you will take one rice course rather than three — the rice is the largest item on the leaf and the easiest to reduce, because nobody counts rice courses, they count enthusiasm. Then the two moves that work better than any refusal: leave a little rice on the leaf, because empty space is an invitation and half an inch of rice is not; and ask for the vegetable dishes by name, loudly and often. A guest who wants a third helping of thoran is a guest the host is delighted with, and nobody is counting thoran.

At a wedding reception where the default is biryani, parotta, a fry and a payasam, the plate arrives pre-assembled. Eat the meat and leave a substantial part of the rice or the parotta. This is socially invisible in a way that declining a dish is not — a half-finished plate reads as a small appetite, while a refused plate reads as a judgement. Find the raita, the salad or the pickle table and use it. And if there are two receptions in a week, the intervention is the days between them.

Church, temple and the ritual calendar

Kerala's eating calendar is dense and it is not optional: Christmas and Easter lunches, the perunnal feast, temple festival meals, Vishu, Onam, and in Karkidakam the medicinal-gruel month in which R108 · Karkidaka Kanji is a ritual rather than a recipe. These are not occasions to manage down. They are the reason there is a calendar.

The structural answer is the one DiRECT's gradient licenses: risk reduction tracks the sustained average, not the perfect record (Lean et al., 2018). Six feast days a year, eaten fully, do not move a sustained average. Six feast weeks do — because a feast day surrounded by four days of leftovers and three days of "well, it is the season" is not a feast day, it is a fortnight.

So: eat the ritual meal properly, without arithmetic. Then do two things. Give the leftovers away on the day, which is culturally normal, actively welcomed, and removes four days of high-energy-density food from the house without a single act of restraint. And put the next ordinary meal on the calendar before the feast happens, so the return is a scheduled event rather than a decision made in a state of satiety.

Ramadan and nombu kanji

For Muslim readers, and anyone eating in a Mappila household during Ramadan, three structural facts. The eating window inverts. Iftar is the largest meal and arrives after a long fast, which is the condition under which portion control is hardest. And the traditional foods of the table sit at both ends of the energy-density range.

The good news is that nombu kanji is structurally excellent and needs almost nothing done to it. Rice, coconut milk, a pulse and often meat or fish, cooked wet and eaten hot, is a low-energy-density, pulse-containing dish arriving exactly when a low-energy-density dish is most useful. Break the fast with dates and water, then the kanji, and let it do what a soup preload does: in Rolls's synthesis a low-energy-dense soup preload reduced intake at the following meal by about 26% (Rolls, 2009). Whether that transfers to an iftar table has not been tested, and this book will not claim it has.

The adjustments are small. Fry fewer items and steam more — R104 · Savoury Kozhukkatta and R103 · Unnakkaya, Rebuilt both belong on this table. Put the arithmetic of R106 · Chai, and the Sugar Arithmetic to work on the sharbat, usually the day's largest source of sugar. And if you take any glucose-lowering medication, fasting changes both its timing and its risk, which is a conversation to have with your doctor before Ramadan rather than during it. Nothing in this book adjusts a dose.

Onnu koodi: declining without refusing

The specific problem: somebody who loves you is holding a ladle. Saying no costs them something, saying no repeatedly costs the relationship something, and "I am on a diet" is the worst available answer, because it converts a warm act into a health lecture.

What works, in rough order of usefulness.

Redirect rather than refuse. "Not the rice, but give me more of that thoran — it is very good." The most effective line in the chapter, because it satisfies the actual transaction. The host wanted to feed you. You have been fed. Nobody has been declined.

Accept and slow down. Take the ladle's worth, put the bowl down, let it sit. In most Kerala households a half-eaten second helping produces no comment, while a refused one produces a negotiation.

Praise the specific thing. "Ammachi, the meen curry is exactly right this time." Naming the dish completes the exchange the offer was part of; a great deal of onnu koodi is a request for acknowledgement wearing the clothes of a request to eat.

Locate the limit in your appetite. "I have already had two helpings, I cannot do justice to a third" is both true and unarguable, and blames no rule.

Tell one person the real reason. Not the table — one person, usually whoever cooks. A blood result, a waist measurement, a doctor's sentence. This converts your most persistent feeder into your most effective ally, which is the diffusion effect SAHELI observed (Lam et al., 2025), and it means the next sadya at that house has two thorans more on it. That is worth more than every refusal you will ever manage.

Hosting: feed generously and eat your own way

The hosting problem is real and it is not solved by cooking less. A Malayali host who serves modest quantities has failed at hosting, and knows it.

So do not cook less. Cook differently, in the direction guests will not notice. Serve more items, not larger ones — abundance in Kerala is read as variety, so three vegetable dishes and two pickles reads as more generous than two vegetables and a mountain of biryani, and it is also the Kerala Plate. Put the vegetables in the biggest bowls at the front, because what is nearest gets eaten. Serve the rice from the kitchen in a smaller vessel refilled as needed, which is normal practice and removes the largest item from arm's reach. Portion the payasam into small glasses in the kitchen, as R120 · Sadya Payasam, Portioned does, rather than putting a jug on the table. Put out R107 · Sambharam and water rather than soft drinks.

For the sweets, Chapter 26's position applies: R112 · Palada Payasam, Rebuilt and R116 · Chakka Pradhaman are lower-load builds of real dishes, R113 · Ada Pradhaman, with the Arithmetic Shown prints what the traditional version costs, and R115 · Elayada is modest by construction. Serve one, well made, in a small glass.

The result is a table nobody can criticise and a plate you can eat from — environment design doing the work of willpower, for every person at the table including the ones who never asked for help.

Restaurants, thattukada and the parotta-and-beef default

Maida parotta with beef fry is the default evening social meal for a large part of young Malayali Kerala, and it is the highest energy-density combination in the state: refined flour laminated with fat, beside a dry roast finished in more fat. There is no version of this book that bans it, because a book that bans the social meal gets closed.

The structural moves: order one parotta instead of two, which is the entire intervention and costs no willpower once it is the order you always place; ask for extra of whatever vegetable is on the board; and take the fish if there is fish, since a Kerala restaurant's fish is usually better than its beef. Notice too that the shared plate is the traditional format here, which is a gift — one parotta and one beef between two, with rice and a vegetable each, is a genuinely good meal.

At a thattukada at midnight, the hour is more of a problem than the food, because a late meal is rarely accounted for and is frequently the day's second dinner. Eat something before you go, so the stall is not being asked to solve real hunger. And if the choice is between going and not going, go. The social meal is not the problem. The second parotta is.

The Gulf work canteen

For the Gulf reader — and the Kerala Migration Survey is a reminder of the stakes, with return migrants coming home at a mean age of 51, 55% reporting poor health and 70% of those carrying at least one chronic disease (Paul et al., 2023) — the canteen is the whole diet five or six days a week, on a menu you did not design.

Three levers, effective because they operate daily. Ask for half rice — canteens comply without comment, and this is the largest single change available. Take the dal or the chana every day without exception, which puts a pulse on a plate you did not cook. Take the salad even when it is indifferent, because it is the only thing on the counter with the energy density you need.

Then the shift-work problem. Where the day's largest meal lands late at night, move volume earlier where you have any choice. This is not a chrononutrition claim: in TREAT, sixteen-hour fasting produced a between-group difference of −0.26 kg against three structured meals (P=0.63), and the fasting arm lost slightly more lean mass (Lowe et al., 2020). It is a practical claim — a large meal eaten at one in the morning after twelve hours of nothing is a large meal. And keep R105 · Roasted Kadala Mix in a locker: six weeks in a fridge, four months in a freezer, no cooking, a portable pulse.

Air travel and long flights

Airline food is engineered for shelf stability, which makes it ultra-processed almost by definition, and Hall's 508 kcal at matched macronutrients is why that matters (Hall et al., 2019). The practical package: order the vegetarian Indian meal in advance, which usually arrives as a dal and a vegetable rather than a sauce and a roll; eat properly before boarding so the tray is not the day's only food; skip the bread and the dessert, which nobody is watching; drink water rather than juice; and treat the trolley's second pass as optional, which it is.

The bigger travel problem is not the flight. It is the fortnight of arrival — the visiting, the feeding, four households who each have something for you, and no kitchen of your own. Handle it as a supply problem: ask the house you are staying in for one thing you actually want cooked, which they will be pleased to make, and which changes what is on the table for everybody.

The diaspora Indian restaurant

The specific, measurable problem here is oil. A kitchen cooking for volume, for a customer expecting richness, uses far more fat than any home kitchen would, in dishes that are also larger; cream and ghee do structural work there that they do not do in Kerala; and the "Kerala" section of the menu is often three dishes long. There is also the reused-frying-oil question, worth stating honestly: repeated heating drives lipid oxidation, and the mechanistic and animal evidence points consistently in a bad direction, though there is no human outcome trial (Ng et al., 2014).

None of that makes a restaurant meal a problem to be solved. The workable order: one curry between two people, with rice and a vegetable side each, rather than a curry each. Ask for it without cream where cream is not traditional, which is most Kerala dishes. Take the dal. And accept that the answer here is frequency, not a cleverly modified order.

One closing observation, because it reframes the chapter. In MASALA, South Asian Americans carried an age-adjusted diabetes prevalence of 23% against 6% in white Americans, and the gap widened after adjustment (Kanaya et al., 2014) — yet the same programme's acculturation work found bicultural individuals, keeping traditional practices while adopting some Western ones, clustered at the lowest carotid intima-media thickness, with both very strong and very weak cultural ties associated with higher risk. The evidence does not say eat like your grandmother, and it does not say eat like your neighbours. It says the pattern that does best is a traditional food framework run with modern nutritional discipline — which is, precisely, this book.

Chapter 32 — Special Situations

Everything in this chapter has the same shape: what the evidence says, what the food can contribute, and then the sentence that ends every item, which is that the decision belongs to a clinician who has seen your results. That is not legal padding. It is the correct division of labour. A cookbook can change what is on a plate. It cannot examine you, order a test, read a scan, know what you take, or bear the consequences of getting it wrong.

Nothing in this chapter is a treatment protocol. Nothing in it tells anyone to start, stop or change a medicine.

PCOS and insulin resistance in South Asian women

Polycystic ovary syndrome affects roughly 8–13% of women of childbearing age worldwide (Yasmin et al., 2025). Indian figures run higher and vary with criteria and population: among 1,164 college-going women aged 18–25 in the Delhi region, Rotterdam criteria with ultrasonography found 17.40%, against a pooled Indian estimate of 8.41% in that age group (Sharma et al., 2025). An Assam case-control study found 9.18%, with biochemical hyperandrogenism in 92.45% of cases and type 2 diabetes in 52.8% of them (Yasmin et al., 2025).

Two facts make this Kerala-specific rather than general. South Asians show higher insulin resistance and lower beta-cell function than every other group studied after adjusting for age and adiposity (Kanaya et al., 2014). And abdominal obesity runs at 72.6% in Kerala women (Sarma et al., 2019). The population most insulin-resistant at any given body weight overlaps almost exactly with the population carrying the most abdominal fat.

What the plate contributes is this book's core prescription with the emphasis moved from weight to waist: a bounded grain, a pulse at every meal, on the strength of Mohan's brown-rice-plus-legumes arm at 22.9% against white rice (Mohan et al., 2014), protein at 1.2–1.6 g/kg (Leidy et al., 2015), half a plate of vegetables. Low-GI eating is worth roughly 0.3–0.5% off HbA1c in glucose-intolerant populations (Zafar et al., 2019), and 5–10% weight loss has the most consistent effect on insulin resistance in the literature.

PCOS is diagnosed, monitored and treated by a clinician. Fertility decisions, hormonal treatment and metformin are theirs, not this book's. If you think you have it, or you have it and your numbers are drifting, that is a referral, not a recipe.

Gestational diabetes

Pooled across 110 studies yielding 117 prevalence estimates, gestational diabetes affects about 13% of pregnancies in India (95% CI 9–16%), at 12% urban and 10% rural, with the authors noting that the lack of consensus on screening and diagnostic criteria drives much of the between-state variation (Mantri et al., 2024). Roughly one pregnancy in eight, in a population that also develops type 2 diabetes at a mean BMI of 25.8 rather than 31 (Ramachandran et al., 2006).

Pregnancy is where the general advice bends most, and it bends in one direction: pregnancy is not the time for an energy deficit, and nothing in Part Four should be run as a weight-loss programme while pregnant. What the cuisine offers, within whatever plan a clinician sets, is naturally lower-glycaemic staples — Kerala's own measured values put red rice puttu at a glycaemic index of 38 and matta rice at 38 against 45 for whole-wheat roti (Pavithran et al., 2020) — plus a pulse-forward structure, generous vegetables, fish, and dishes like R012 · Kanji with Vanpayar and R023 · Kerala Khichdi that pair a bounded grain with a pulse by construction. Protein and iron requirements rise: R024 · Meen Vevichathu, R058 · Parippu Curry and R065 · Cheera Thoran, with lime or tomato at the same meal for the iron.

Two cautions. Mercury guidance in pregnancy restricts some large predatory fish, and which species and how much is a national-guidance question that differs by country. And fenugreek in therapeutic quantities is contraindicated territory in pregnancy.

Gestational diabetes is screened for, diagnosed and managed by the team looking after your pregnancy, with glucose targets and insulin decisions that are theirs alone. Take this book's food to that appointment as a question, not as a plan.

MASLD and fatty liver

This is the chapter Chapter 4's central finding earns.

The distinctive South Asian lesion is liver fat, not visceral fat. Pooling imaging data from eleven datasets, South Asian men showed +0.56 standardised mean difference in liver fat (95% CI 0.14–0.98) despite a BMI 0.5–0.7 kg/m² lower than white European men, with visceral fat essentially identical (−0.03 SMD, −0.24 to 0.19); South Asian women showed +0.31 SMD liver fat (0.14–0.48) at a BMI 0.9 units lower (Iliodromiti et al., 2023). In MASALA, South Asians had the highest intrahepatic fat of five US ethnic groups (Garg et al., 2016).

The Indian prevalence figures should stop a reader. In a general-population screening study of 1,243 adults using transient elastography, MASLD was present in 43.7% overall, and in 21.3% of lean subjects against 66.7% of non-lean (Prasad et al., 2024). Among 345 Indian IT-sector employees, FibroScan found 84.06% (Bhargava et al., 2025). And among 688 healthy South Indian medical students, mean age 20.5, with no significant alcohol intake, 23.1% already had hepatic steatosis, BMI being the only independent predictor (Desai et al., 2025).

Then the hopeful part, which is why this belongs in a cookbook at all: liver fat is the depot that responds fastest. In Counterpoint, liver fat fell by roughly one-third within seven days of a large energy deficit, with fasting glucose normalising at day 7, before most of the weight had gone (Lim et al., 2011, via Taylor, 2025). In ReTUNE, in people with type 2 diabetes at a BMI of 21–27, liver fat was elevated about threefold at baseline and normalised on a weight loss of 5.5–10.2%, median 6.5%, with 70% achieving remission at one year (Taylor et al., 2023).

The plate contributes the book's programme with three emphases sharpened: total energy down, refined carbohydrate and free sugars down, alcohol down. On alcohol, the honest quantitative position is that cutting from six drinks a day to three buys real millimetres of mercury — −5.50 mmHg systolic in the heaviest drinkers halving intake — while cutting from one to zero does not measurably move blood pressure (Roerecke et al., 2017). WHO's position is nonetheless that no level is risk-free, and heavy drinking on top of high liver fat is the worst available combination for this population.

Fatty liver is diagnosed by imaging and blood work, staged for fibrosis, and monitored over time. Whether you have simple steatosis or something that needs a hepatologist is not something a menu can tell you. Ask for the scan, ask what stage it showed, and take the answer to the person who ordered it.

Hypothyroidism

Both halves of this need saying, and most writing on it says only one.

The first half: hypothyroidism is common in this population and under-treated. Pooled across 60 Indian studies in pregnancy, prevalence was 17% (95% CI 14–19%) — subclinical 15%, overt 3% — with the authors warning that a quarter of studies used non-standard TSH thresholds ranging from 4.0 to 10.0 mIU/L and that India-specific cut-offs are lacking (Manna et al., 2026). In 282 post-menopausal women near Delhi, 27.3% were hypothyroid, and among those already diagnosed and on treatment only 46.4% had thyroid hormones in the normal range (Sharma and Verma, 2024). If you have symptoms, a family history, or you are treated and have not had a recent test, get tested.

The second half, said without flinching: treated hypothyroidism is not why most readers are carrying twelve kilograms of excess fat. The weight attributable to an underactive thyroid is modest, is substantially fluid rather than fat, and largely resolves when the biochemistry is corrected. A thyroid problem is a real problem to fix and a poor explanation to rest on. Both are true, and a reader told only the first will not act on their plate, while a reader told only the second will feel dismissed.

The food contribution is small. Iodine matters and iodised salt is the main source, which interacts directly with Chapter 10's sodium reduction — cutting salt also cuts iodine, which is a reason to take iodine from fish and dairy rather than to abandon the salt reduction. Selenium comes from fish and egg. The goitrogen question is smaller than the internet suggests: brassicas are not a practical problem at culinary quantities in an iodine-sufficient person.

There is one goitrogen in this cuisine that deserves more attention than the cabbage does, and it is kappa. Cassava carries cyanogenic glycosides, and the systematic review this book relies on lists thyroid dysfunction, alongside konzo and tropical ataxic neuropathy, among the chronic disorders documented where cassava is poorly processed (Forkum et al., 2025). Keep the size of that right. The disorders in that literature come from populations eating badly processed cassava as a dominant staple, often with iodine deficiency alongside; nobody has shown a thyroid effect from properly boiled kappa eaten twice a week. The practical response is the one Kerala already worked out and Chapter 12 sets out in full: peel thoroughly, cut, boil in plenty of water, discard that water, cook through. Boiling for 30 minutes removes 96.3% of the cyanogen; soaking alone removes about 20% and sun-drying alone about 30%, so dried kappa that has only been sun-dried is the form to be careful with, and bitter varieties carry more than sweet. If you eat kappa often and you are hypothyroid, that is a reason to be exact about the boiling water, not a reason to give up the dish.

Timing, if you take levothyroxine. This is the one place where a change of breakfast interacts with a tablet, and it needs saying plainly because this book's central prescription is a substantial high-fibre breakfast eaten early. Levothyroxine absorption is affected by food in general, and by dietary fibre, calcium and iron in particular. A reader who moves from tea and toast to a 7 g-fibre puttu breakfast, and swallows the tablet with it, has changed the conditions under which the drug is absorbed. That does not mean the breakfast is wrong. It means the two need separating in time, and how far apart is a question for the person who prescribed it. Take the timing question to your doctor or pharmacist before you change your breakfast, tell them what the new breakfast contains, and ask them what interval they want between the tablet and the food, and between the tablet and any calcium or iron supplement. What this book will not do is give you a number and let you adjust around it.

Thyroid dosing is a blood-test-and-prescription matter, adjusted at intervals your clinician chooses. Nothing here changes a levothyroxine dose, and nothing here should delay a test.

Sarcopenia and protein in older adults

For readers over 65, and for the diaspora reader's parents, this section points in the opposite direction from the rest of the book.

The PROT-AGE recommendations are specific: adults aged 65 and over need at least 1.0–1.2 g of protein per kg per day to maintain and regain lean mass and function; at least 1.2 g/kg if exercising and otherwise active; and 1.2–1.5 g/kg with acute or chronic illness, with both endurance- and resistance-type exercise recommended at individually safe levels. The stated exception is severe kidney disease (eGFR below 30 ml/min/1.73 m², not on dialysis), which may require restriction (Bauer et al., 2013). Older adults need more rather than less because splanchnic extraction rises and the anabolic response to a given dose of protein declines with age.

A rice-forward Kerala diet is very easy to eat at 0.6–0.8 g/kg, and in a seventy-year-old that is a sarcopenia risk factor. The dishes that fix it are already here: R024 · Meen Vevichathu, R026 · Mathi Curry and R031 · Kozhuva Peera for fish; R046 · Mutta Curry and R014 · Egg Roast, Rebuilt for egg; R010 · Kadala Curry, R053 · Vanpayar Thoran, R056 · Muthira Thoran and R058 · Parippu Curry for pulses; R107 · Sambharam and curd for dairy. Aim at 25–30 g of protein in each of three meals rather than one large evening portion; keep food soft and moist where chewing is a problem — kanji with a well-cooked pulse, fish in a wet curry, dal — and note that appetite falls with age, so energy density is the one place in this book where an older reader may need it to go up. Two further notes: unintentional weight loss in an older adult is a symptom to report rather than a success, and resistance training matters at least as much as the protein.

Protein targets in the presence of kidney disease, and any exercise prescription after a cardiac event or with significant frailty, are clinical decisions. Take these numbers to a doctor or dietitian who knows the kidney function, and let them set the target.

Vitamin B12, vitamin D and iron

A rice-and-vegetable-forward diet has three predictable gaps, and Kerala's food supply closes two of them easily.

Vitamin B12 comes only from animal foods and fortified products. Among older North Indian adults, deficiency below 200 pg/mL ran at 17.1% with a further 17.1% borderline (Biswas et al., 2024), and among 835 hospitalised Indian adults it was 26.1% (Priyadarshi et al., 2026). A largely vegetarian Malayali reader has no dietary route to sufficiency without fish, egg, dairy or a supplement. Kerala's advantage is under-used: R026 · Mathi Curry, R031 · Kozhuva Peera, R034 · Kallummakkaya Roast and R036 · Unakka Meen Thoran are all B12-dense, mussels and small oily fish being among the richest sources in the food supply. Metformin and long-term acid suppression are both associated with lower B12, which is a reason to ask for the test rather than to change anything.

Vitamin D. In the same cohort, 28.9% were deficient below 20 ng/mL with a further 31.9% insufficient, with seasonal variation and lower levels in women (Biswas et al., 2024). Say the uncomfortable thing plainly: this is common even in a state with a great deal of sun, because of sun avoidance, indoor work and skin pigmentation, and it is not solvable by diet alone in most people. Oily fish and egg yolk contribute a little — R027 · Ayala Varuthathu, Baked and R028 · Ayala Varuthathu, Fried and Measured — and that is the honest ceiling of what a cookbook offers.

Iron. About two-thirds of Indian children aged 6–59 months are anaemic (Kashyap et al., 2026), and anaemia in women of reproductive age is a major Indian public-health problem. Plant iron is non-haem and poorly absorbed, and absorption rises substantially with vitamin C at the same meal, which Kerala has in lime, tomato, gooseberry, guava and raw mango. So: R065 · Cheera Thoran and R060 · Parippu with Cheera with lime squeezed at the table; R052 · Liver Fry, the densest source in the book; and R089 · Kerala Kitchen Salad or R085 · Mango Curry beside a pulse meal. Tea and coffee taken with a meal reduce non-haem iron absorption, which is a reason to move the chaya to between meals rather than give it up.

All three are blood tests with treatment thresholds, doses and durations that differ by person. Do not self-supplement a suspected deficiency, and do not assume a supplement is harmless — iron in particular is dangerous in overload conditions and in children. Ask for the test, and let whoever ordered it decide what follows.

Medication and food interactions worth knowing

Four interactions matter enough to name. This section exists so that you can raise them with a professional, not so that you can act on them.

Fenugreek in therapeutic amounts, with insulin or a sulfonylurea. Fenugreek has the strongest glycaemic evidence of any spice in this book, at HbA1c reductions of 0.63% to 0.85% in the pooled trials (Neelakantan et al., 2014), reachable through soaked seeds, methi leaves and batters at 10–25 g a day — far above a seasoning quantity and genuinely achievable through food. That is precisely why it needs a caution. A food that lowers glucose, added on top of insulin or a sulfonylurea (glimepiride, gliclazide, glibenclamide) which lowers glucose whether or not glucose is rising, can produce hypoglycaemia. This applies to R109 · Jeera and Methi Water, and What It Does Not Do taken daily, and to any deliberate therapeutic use. Tell the clinician who set your dose before you start, and expect to monitor.

Potassium salt substitutes, with RAAS inhibitors, potassium-sparing diuretics, or chronic kidney disease. M07 · The 70:30 Salt is the strongest single-ingredient swap in the book — stroke down 14% and all-cause mortality down 12% across 20,995 people over a mean 4.7 years (Neal et al., 2021) — and it is contraindicated or requires supervision in chronic kidney disease, and for anyone on an ACE inhibitor (-pril), an angiotensin receptor blocker (-sartan), a potassium-sparing diuretic (spironolactone, eplerenone, amiloride) or a potassium supplement, all of which raise the risk of hyperkalaemia. SSaSS screened its participants' renal function; a shopper does not. This applies equally to commercial low-sodium salts, which are the same product with a label. If any of this describes you, ask your doctor or pharmacist before buying it, and cut ordinary salt in the meantime, which is safe for everyone.

Ginger and garlic, with anticoagulants and antiplatelet drugs, and before surgery. Both have antiplatelet activity, and concurrent use with warfarin is a standard precaution in drug-interaction references. Be precise about the evidence grade: what exists is precaution-level rather than outcome-level, and human evidence of a clinically significant interaction at culinary quantities is weak. Garlic's cardiovascular effect is real at doses overlapping cooking — across 108 trials in 7,137 participants, −3.7/−2.0 mmHg blood pressure, LDL −5.9 mg/dL and total cholesterol −10.2 mg/dL (Behrouz et al., 2026) — which is why this book uses garlic generously. The caution attaches to gram-level concentrated supplements, and particularly to the perioperative period. If you take warfarin, a direct oral anticoagulant, clopidogrel or aspirin for a cardiac indication, or you have surgery scheduled, mention any supplement use to the prescriber and the anaesthetist. Cooking with garlic and ginger is not the issue; a capsule is a different object.

High-dose turmeric supplements, and the liver signal. Turmeric as a cooking spice is safe and this book uses it constantly. Turmeric as a concentrated supplement is a different product with a different risk profile. The Drug-Induced Liver Injury Network reported ten cases of turmeric-associated liver injury, including one death, with seven of ten carrying HLA-B*35:01, and specifically implicated increasing co-formulation with black pepper extract (piperine), present in three of the seven products chemically analysed (Halegoua-DeMarzio et al., 2023). Turmeric carries a LiverTox likelihood score of A for clinically apparent liver injury. The "add pepper to boost curcumin absorption by 2,000%" line traces to one small pharmacokinetic study measuring blood levels over an hour, co-authored by the founder of a company selling piperine, with no clinical outcome. The food is fine. The capsule is not risk-free, particularly in anyone with existing liver disease — which, given the prevalences above, is a large fraction of this readership.

And the general rule covering everything not listed: no spice in this book replaces a medicine. Nothing here supports substituting any food for metformin, a statin or an antihypertensive. If your numbers improve on this plate — and they may improve faster than you expect — the person who changes a prescription is the person who wrote it, and the change is made with your results in front of them. Bring the food. Leave the dosing to them.

A Year From Now

Twelve months is the honest unit, because body weight responds to a change in intake with a half-time of about a year (Hall et al., 2011). So here is what a year of this looks like, specifically, with no numbers promised.

Your waist is smaller. How much smaller depends on where you started and how consistent the year was, and the range across the trial literature is wide enough that a promise would be a lie. But the tape moved first and moved most, because the visceral and hepatic depots empty early, and Kerala's own peer-led community programme shifted mean waist from 89.5 to 87.5 cm in twelve months with no formal diet at all (Ravindranath et al., 2020). You are running an actual plate. You should be ahead of that.

Your blood work has changed in a direction your doctor noticed without being told why, and some of it moved before the tape did. If your HbA1c was in the prediabetic range, the levers you pulled — the bounded grain, the pulse at every plate, the vegetables at half the area — are worth roughly 0.3 to 0.5 percentage points on their own (Zafar et al., 2019). If your blood pressure was high, the pattern is worth up to about 11 mmHg before you touch the salt cellar (Appel et al., 1997), and roughly another millimetre per kilogram lost (Neter et al., 2003). If your LDL was high, the oil bottle by the stove did more than any supplement would have.

The shape of a weeknight is different, and this is the change you will actually notice. There are cooked pulses in the fridge in labelled bags, shallot-masala cubes in the freezer, a jar of varutharacha paste, portioned rice at a fixed weight. Dinner takes twenty minutes and has four things on the plate. Sunday has a rhythm that is not a chore, because two and a half hours buys back an hour every weekday and you have done the arithmetic. Nobody in the house has asked what happened to the food, because nothing happened to the food. There is more thoran on the table and less oil in the pan, and the fish curry is the same fish curry it always was.

Your children have learned to eat vegetables as the largest thing on a plate, which is not something you taught them and not something you announced. They learned it because that is how the table looked. That transfer is documented — behaviour change diffuses to close family members who were never enrolled in anything (Lam et al., 2025) — and it is the part of this book with the longest reach. A child who grows up thinking half a plate is vegetables and a pulse comes with rice does not have to unlearn anything at forty-five.

You have had a bad fortnight, possibly two. Four weddings in a season, none of them managed well, and a stretch in the monsoon when nobody cooked and everything came in a parcel. The difference between this year and the previous attempts is that the fortnight ended, because you had written down what to do on the day and it turned out to be small: eat the next meal on the plan, weigh on the usual day, write one line. What predicted the outcome, across every trial in this book, was turning up (Sacks et al., 2009).

And Onam happened. You ate a full leaf, with two payasams if you wanted them, and it did nothing to the year, because a sustained average is not disturbed by one meal (Lean et al., 2018).

That is the realistic picture. Not a transformation. A different default, running quietly, for long enough to show up in a blood test.

One last thing, and it is where this book began. Kerala has a life expectancy of 75 and an infant mortality rate of 6 per 1,000, and it also has abdominal obesity in 72.6% of its women and only one hypertensive in eight with their blood pressure controlled (Sarma et al., 2019). That paradox is not a verdict on thoran, or coconut, or rice, or your grandmother. It is a verdict on what replaced them: polished imported rice in doubled portions, oil by the glug instead of the spoon, maida parotta, the parcel, the packet, and a plate that lost its vegetables somewhere between 1969 and now.

You have spent a year putting them back. The food was never the problem.