1. Assess dehydration first

2. Home care (Plan A)

• Low-osmolarity ORS after each loose stool. The amounts below are from standard WHO guidance, not the passages I pulled from the library:
  • Under 2 years: 50-100 mL.
  • 2 to 10 years: 100-200 mL.
  • Give small, frequent sips. If the child vomits, wait 10 minutes and restart more slowly.
• Zinc: 10 mg/day if under 6 months and 20 mg/day if 6 months or older, for 10-14 days. This is WHO guidance as reported in Park's and Rosen's. It shortens the illness and lowers recurrence over the next 2-3 months.
• Keep feeding. Continue breastfeeding and give the usual diet. Avoid sugary drinks and undiluted juices, which worsen the diarrhoea.
• Some dehydration (Plan B): ORS about 75 mL/kg over 4 hours, then reassess (WHO standard).
• Severe dehydration (Plan C): IV Ringer's lactate or saline in hospital.

3. Abdominal pain

• Paracetamol for pain or fever.
• Avoid antimotility drugs such as loperamide in children. Avoid routine antispasmodics, because they can mask a surgical abdomen.
• Avoid NSAIDs if the child is dehydrated.

4. Drugs that usually are not needed

• Antibiotics: only for dysentery (blood in stool), suspected cholera, or specific high-risk situations. Shigella is treated, but Salmonella treatment generally prolongs excretion and does not shorten the illness in uncomplicated cases (Rosen's, Table 167.4).
• Probiotics and racecadotril: evidence is inconsistent. Cecil says probiotics are not beneficial in acute diarrhoea. A 2026 systematic review of racecadotril in children exists (PMID 41504802), but I did not read it and neither is a core recommendation. Treat both as optional extras.

5. Refer or see a doctor urgently if

• Blood in the stool or high fever.
• Repeated vomiting and unable to keep fluids down.
• Signs of dehydration: lethargy, sunken eyes, very little urine.
• Severe or localised abdominal pain, especially right lower quadrant, or a distended abdomen.
• Bilious or green vomit.
• Age under 3-6 months.
• Diarrhoea lasting 14 days or more.
• Severe malnutrition, or a chronic illness or immunosuppression.

Sources

• Park's Textbook of Preventive and Social Medicine, "Diarrhoeal disease control" and the IMCI classification tables.
• Rosen's Emergency Medicine, "Management" and "Rehydration" sections of the chapter on acute diarrhoea in children.
• Goldman-Cecil Medicine, "Hydration" and "Reducing Diarrhea".

Probiotics and racecadotril in children with acute gastroenteritis: evidence summary

Probiotics

• A 2025 meta-analysis of 25 RCTs (5,170 children) found probiotics shortened diarrhoea and vomiting. The confidence interval for diarrhoea duration was wide (95% CI -14.60 to -0.91). PMID 40739406
• Saccharomyces boulardii has the most data. A 2020 meta-analysis of 29 RCTs found diarrhoea about 1 day shorter (MD -1.06 days) and hospital stay about 0.85 day shorter. The authors rated this low to very low quality, with high heterogeneity (I² around 90%). Only 17% of trials adequately concealed allocation. PMID 32056266
• A 2026 umbrella meta-analysis of 36 meta-analyses concluded probiotic effects are strain-specific and depend on dose. PMID 42080229

• L. rhamnosus GG vs placebo in 971 children aged 3 months to 4 years. Moderate-to-severe illness occurred in 11.8% vs 12.6% (RR 0.96). Diarrhoea duration was the same (about 50 hours in both groups). PMID 30462938
• L. rhamnosus R0011 plus L. helveticus R0052 vs placebo in 886 children. Moderate-to-severe illness occurred in 26.1% vs 24.7%, with no difference in diarrhoea or vomiting duration. PMID 30462939

• The group found no strain with 2 high-quality RCTs showing benefit.
• It made weak recommendations for S. boulardii (low to very low certainty), L. rhamnosus GG (very low), and L. reuteri DSM 17938.
• It made a strong recommendation against L. helveticus R0052 with L. rhamnosus R0011.
• It made a weak recommendation against Bacillus clausii strains.
• This guideline predates the two NEJM trials' wider uptake and the 2025-26 meta-analyses above. Those meta-analyses are positive, but they pool the same low-quality trials.

• If a probiotic is used, pick a named strain with some support (S. boulardii CNCM I-745, L. rhamnosus GG or L. reuteri DSM 17938).
• Expect at most about a day less diarrhoea.
• Do not use it in place of ORS or zinc.
• Use caution in immunocompromised children or those with central lines, because of the risk of bacteraemia or fungaemia. I did not retrieve a source for this in this search, so it comes from general knowledge.

Racecadotril

• Cochrane 2019: 7 RCTs (1,140 children under 5). PMID 31858591
  • It may reduce rehydration failure (RR 0.41, 95% CI 0.13 to 1.23), but the interval crosses 1 and the evidence is low certainty.
  • The data on diarrhoea duration were insufficient for a conclusion.
  • Hospital stay was similar.
  • Adverse events were not increased (RR 0.90).
  • The authors concluded it is probably safe, has little benefit, and is not supported for routine use.
• 2026 systematic review of 5 double-blind RCTs (904 children): only one trial had low risk of bias, and it showed no significant or clinically meaningful effect on diarrhoea duration, stool output or hospital stay. The authors concluded there is "absence of evidence" of efficacy. PMID 41504802 This review used a novel method (REB) that discards higher-bias trials, so it is stricter than a standard meta-analysis.
• Earlier industry-linked reviews were more favourable (for example PMID 29614995, 2018). A 2014 European paediatric infectious diseases guideline listed racecadotril as an effective option (PMID 24739189), but I only saw a snippet of it. The newer reviews are more cautious.

• It is generally safe, but benefit is unproven.
• It is not recommended by WHO for routine use, to my knowledge. I did not verify this in this search.
• If a clinician uses it, it is only an add-on to ORS.
• Availability and licensing differ by country, so check local guidance.

Bottom line

1. ORS, continued feeding and zinc come first.
2. Probiotics: low-quality evidence for a modest benefit with some strains, and two large negative US/Canadian trials. They are optional and not harmful in healthy children.
3. Racecadotril: low-quality, inconsistent evidence. Not recommended routinely.
4. Neither prevents dehydration or replaces the red-flag review in my earlier answer.

1. Check for an emergency first

• Vomiting, abdominal pain, rapid or deep breathing, or fruity breath. These suggest diabetic ketoacidosis (DKA).
• Marked dehydration, confusion or drowsiness. These suggest a hyperosmolar state.
• Inability to keep fluids down, or an intercurrent infection.

2. Type 2 diabetes, stable and no ketosis (outpatient)

• The ADA Standards of Care 2026 say to consider insulin whatever the background therapy or disease duration if symptoms are present or A1C is above 10% or glucose is 300 mg/dL or higher (ADA 2026, section 9).
• Insulin is especially indicated if there is catabolism (weight loss, hypertriglyceridaemia, ketosis).
• If there is no evidence of insulin deficiency, a GLP-1 RA or dual GIP/GLP-1 RA (tirzepatide) is preferred over insulin in the ADA 2025 guidance. These agents carry less hypoglycaemia and weight gain. The ADA also notes that evidence is scarce when baseline A1C is above 10-12%.
• The usual plan is basal insulin (for example glargine or degludec) started at about 10 units/day or 0.1-0.2 units/kg/day. Titrate by 2 units every 3 days to a fasting glucose of 80-130 mg/dL. These numbers come from standard practice, not from the sources I retrieved.
• If the basal dose passes about 0.5 units/kg/day and A1C is still above target, add a GLP-1 RA or prandial insulin.

3. Check for complications and reduce risk

• Eyes: do a dilated eye exam before or soon after starting tight control. Very rapid lowering can transiently worsen retinopathy.
• Kidneys: check urine albumin/creatinine ratio and eGFR.
• Feet: examine them for neuropathy.
• Heart: check blood pressure and lipids, and start a statin per risk. Check for cardiovascular disease.
• Other: screen for depression and review any steroids or other drugs that raise glucose.

4. Targets and follow-up

• Target an HbA1c below 7% for most adults. Relax this for the frail elderly, those with limited life expectancy, and those at high hypoglycaemia risk (Symptom to Diagnosis; Harrison's).
• Check home glucose or use CGM. Patients on multiple daily insulin doses should test several times a day.
• Recheck HbA1c every 3 months while not at goal or while changing therapy, then every 6 months once stable.
• Teach hypoglycaemia recognition and treatment, injection technique and sick-day rules. Prescribe glucagon if the patient is on insulin and at high risk.
• Expect a drop of roughly 1-2% or more over the first 3 months, but this varies. A 2026 meta-analysis looked at intensive insulin plus non-insulin agents in type 2 diabetes (PMID 42898168). I saw only its title, so I can't quote its findings.

5. If it is type 1 diabetes

What the evidence says

• Not enough on its own. Goodman & Gilman says randomized trials show that giving only short-acting insulin in response to hyperglycemia is less effective than scheduled insulin regimens. For non-critically ill inpatients it recommends basal plus bolus plus correction insulin, adjusted for oral intake (Goodman & Gilman's, "Management of Diabetes in Hospitalized Patients").
• Anesthesia texts agree. Barash's Clinical Anesthesia says sliding scale protocols "have been disappointing and are strongly discouraged" in hospitalized patients. It recommends basal plus bolus dosing and a glucose target of 80-180 mg/dL in non-critically ill patients.
• ADA guidance. The ADA Standards of Care strongly discourage prolonged correction insulin without basal insulin as the sole treatment. The one exception is type 2 diabetes in non-critical care with mild hyperglycemia. The ADA now prefers the terms "correction" or "supplemental" insulin.
• Type 1 diabetes. Dosing only from pre-meal glucose ignores basal needs and food intake. This raises the risk of both hypoglycemia and ketoacidosis. People with type 1 diabetes must always have basal insulin on board (Fuster and Hurst's The Heart, "Inpatient diabetes management").
• Key trial data. In the Basal Plus trial of 375 patients with type 2 diabetes, basal-bolus and basal plus correction regimens gave lower mean glucose and fewer treatment failures than SSI. SSI had less hypoglycemia (3% vs 13-16%). Diabetes Care 2013
• Guideline review. A 2025 systematic review of hospital diabetes guidelines exists (PMID 40117466). I saw only its title.

When SSI alone is acceptable

• Short stay, non-critically ill, type 2 diabetes, mild hyperglycemia, eating normally.
• A newly admitted patient while the total daily dose is still being worked out. Reassess within 24-48 hours.
• Patients on parenteral nutrition: Sleisenger's suggests starting regular insulin by scale, then adding two thirds of the 24-hour total to the next day's PN formula.

Better approach (non-ICU)

1. Basal insulin once daily. A common starting dose is 0.1-0.25 units/kg/day, with the low end for older patients or those with renal impairment.
2. Prandial insulin with meals if eating. Skip it if the patient is not eating.
3. Correction insulin added on top of meals, using a scale.
4. Hold oral agents such as metformin and sulfonylureas in sick patients.
5. Check glucose before meals and at bedtime, or every 4-6 hours if not eating.
6. Target about 140-180 mg/dL in ICU and non-ICU patients (Fuster and Hurst's). Some sources use 100-180 mg/dL (ADA 2023). Use IV insulin infusion protocols in critical illness.
7. Review the regimen any time glucose is below 70 mg/dL (ADA).

Example correction scale

• Low dose: elderly, renal impairment, lean, or total daily dose below 40 units.
• Usual: most patients.
• High dose: obese, on steroids, or total daily dose above 80-100 units.
• Bedtime: use the half-dose version, or only correct if glucose is above 250-300.
• Do not give correction insulin if the patient is not eating, unless glucose is high and your protocol allows it.
• Check ketones if glucose is above 300 on repeated checks. In type 1 diabetes, check ketones if the patient is ill or has persistent hyperglycemia.

Safety points

• Never give a correction dose without confirming the patient has eaten or will eat.
• Treat glucose below 70 mg/dL per protocol and review all insulin doses.
• Reduce doses with renal impairment, poor intake, steroid taper, or age over 70.
• Keep consistency between nursing shifts and the order set. Use the written protocol, and a standardized electronic order set if available.
• Regimen changes at discharge need a clear plan. Check A1C on admission if none is available from the last 3 months.