KDIGO AKI staging (pediatric)

KDIGO stage = the higher (maximum) of two independent axes. Serum-creatinine axis: Stage 1 if current creatinine is 1.5–1.9× baseline or has risen ≥ 0.3 mg/dL; Stage 2 if 2.0–2.9× baseline; Stage 3 if ≥ 3.0× baseline, or renal replacement therapy has started, or — in a patient under 18 years — estimated GFR < 35 mL/min/1.73 m². A creatinine of ≥ 4.0 mg/dL is also Stage 3, but only once the AKI definition itself is met (a rise of ≥ 0.3 mg/dL, or ≥ 1.5× baseline): with a baseline entered that is checked, so a chronically high creatinine that never rose stages on the ratio and rise criteria alone rather than jumping to 3; with no baseline entered it cannot be checked, so Stage 3 is reported with the 'stage is not settled' output set to 1 rather than either asserted or withheld. Urine-output axis: Table 2 states four (rate, duration) rows, not rate bands, and each is tested on its own with the highest satisfied row governing — < 0.5 mL/kg/h for 6 hours to under 12 hours is Stage 1; < 0.5 mL/kg/h for 12 hours or more is Stage 2; < 0.3 mL/kg/h for 24 hours or more is Stage 3; anuria for 12 hours or more is Stage 3. Anuria also counts as an output below 0.5 mL/kg/h for the first two rows, because there is no urine — so anuria for 6 hours to under 12 hours is Stage 1 — but no numeric rate is assigned to it. A rate below 0.5 mL/kg/h held for less than 6 hours satisfies no row and does not meet the AKI definition on this axis. The reported stage is the maximum of the two axes; if neither is met the stage is 0. Where the duration is not entered — or is '12 hours or more' while the rate is below 0.3 mL/kg/h, leaving the 24-hour row open — the urine-output axis cannot be resolved: the stage shown is then the highest CERTAIN stage, and the 'stage is not settled' output is 1, to be read as '≥' that stage, which is KDIGO's own notation for an unresolvable case in Chapter 2.4 Table 10. That output is set only where an open row could actually raise the stage. Where no open row could change the answer — a rate of 0.5 mL/kg/h or above satisfies no row at any window, a rate of 0.4 mL/kg/h can never reach the Stage-3 row, and nothing can exceed Stage 3 — the answer is settled and the flag stays 0. The same flag carries the un-baselined ≥ 4.0 mg/dL case described above, where the uncertainty runs the other way: the stage is the most the entered creatinine supports and a baseline could lower it. Either way the flag means the same thing to a reader — this is a bound, not a final stage, and a missing input would change it.

Not a summed score: the serum-creatinine and urine-output axes are evaluated independently and the MAXIMUM stage governs — treating it as additive is wrong. URINE OUTPUT: KDIGO Table 2's urine-output rows are (rate, duration) PAIRS, not rate bands, and all four are tested independently with the highest satisfied row governing. Branching on the rate first is the classic implementation defect: 0.25 mL/kg/h for 8 hours is Stage 1, because 8 hours sits in the 6-to-under-12-hour window and the < 0.3 mL/kg/h row needs 24 hours — reading it as Stage 3 over-stages, and the same defect makes Stage 2 unreachable. A rate alone cannot select a row at all: 0.4 mL/kg/h is no AKI at 5 hours, Stage 1 at 8 hours and Stage 2 at 13 hours, so the duration is asked for rather than assumed. A rate below 0.5 mL/kg/h sustained for less than 6 hours meets no row — that is 'the AKI definition is not met on the urine-output criteria entered', not an error and not a graded Stage 0. WHEN THE URINE-OUTPUT AXIS CANNOT BE RESOLVED the stage shown is the highest stage that is CERTAIN and the 'stage is not settled' output is set to 1 — read the stage as '≥' — but only where an open row could actually raise it. The axis is open when a rate below 0.5 mL/kg/h or anuria is entered with no duration band, and also when the band is '12 hours or more' while the rate is below 0.3 mL/kg/h, since that band does not exclude 24 hours and the Stage-3 row stays open. The flag stays 0 wherever the open rows could not change the answer: a rate of 0.5 mL/kg/h or above satisfies no row at any window, a rate of 0.4 mL/kg/h caps the axis at Stage 2, and nothing can exceed Stage 3. Reporting '≥ 2' for an answer already settled at 2 is false caution, and it erodes the flag exactly where it has to mean something. No duration is ever guessed: assuming the shortest qualifying window systematically under-stages, assuming the longest over-stages, and KDIGO supplies no default — its Chapter 2.1 research recommendations state it is not yet settled how the urine-volume criteria should be applied at all. The '≥' notation is the guideline's own: Chapter 2.4 Table 10 records a case as '≥ 1' and another as '?' rather than guessing. ANURIA is a Table 2 row of its own — anuria for 12 hours or more is Stage 3 — and it is treated here as a clinical flag rather than a number. THAT IS A CONFIRMED ABSENCE, NOT AN UNANSWERED QUESTION: KDIGO gives no numeric definition of anuria anywhere in Table 2 or the Chapter 2.1 rationale, and nephrology has no single agreed numeric definition to borrow — the term is deliberately left to clinical judgement. Searching for one again will not produce one, and any millilitre figure attached to the word here would be this calculator's invention rather than the guideline's, so none is attached. What anuria does establish without any number is that the output is below every positive cutoff: the absence of urine is necessarily below 0.5 mL/kg/h, so it satisfies the Stage 1 row (6 hours to under 12 hours) and the Stage 2 row (12 hours or more) as well, and the highest satisfied row governs as usual. Treating anuria as satisfying nothing below 12 hours returned Stage 0 for a child with no urine for 8 hours, while a recorded rate of 0 over the same window returned Stage 1 — the same patient under-staged for being described in words rather than millilitres, which is the dangerous direction. The < 0.3 mL/kg/h row is the one place that entailment is deliberately NOT applied: it fires only at 24 hours or more, and every window establishing 24 hours also establishes the 12 hours at which the anuria row is already Stage 3, so it could not change an answer. Enter the measured rate as well when one is available. WEIGHT BASIS: KDIGO does not state which weight the mL/kg/h is indexed to [NEEDS SOURCE for a KDIGO-endorsed weight basis], so the rate is applied exactly as entered and this calculator takes no position. KDIGO also records that the urine-output criteria are less well validated than the creatinine criteria, that drugs (ACE inhibitors are its example), fluid balance and diuretics need clinical judgement, and that in very obese patients the criteria can capture patients with normal urine output. AGE: age is required because the estimated-GFR < 35 mL/min/1.73 m² route to Stage 3 exists only 'in patients < 18 years' (Table 2) — an eGFR entered for a patient aged 18 or over does not stage on that branch. There is no pediatric modification of the urine-output thresholds: 0.5 and 0.3 mL/kg/h at 6 / 12 / 24 hours apply to children unchanged. The predecessor pediatric system pRIFLE (Akcan-Arikan 2007) uses different durations, is a separate instrument, and is neither reproduced nor blended in here. BASELINE CREATININE is the hardest input — KDIGO does not fix a single pediatric baseline-creatinine method [NEEDS SOURCE for a KDIGO-endorsed pediatric baseline rule]. That marker is about GUIDELINE ENDORSEMENT and is still open; it is not the same question as which surrogate performs best, which now has a paediatric answer (Lee 2022, below) and is closed. The baseline supplied here drives the ratio-based and ≥ 0.3 mg/dL-rise stages, and the ≥ 0.3 mg/dL rise is applied as (current − baseline) rather than a timed 48-hour delta. THE ≥ 4.0 mg/dL ROUTE TO STAGE 3 IS NOT STANDALONE, and this is where the baseline earns its keep. KDIGO's Chapter 2.1 rationale requires the Rec 2.1.1 creatinine-change definition (≥ 0.3 mg/dL within 48 h, or ≥ 1.5× baseline) to be satisfied FIRST — a deliberate 2012 departure from AKIN's wording, made to bring definition and staging into parity. A chronically elevated creatinine that never rose is therefore not Stage 3 AKI; it is not AKI. Three cases, and only the third is a compromise. (1) A baseline is entered and the rise qualifies: ≥ 4.0 mg/dL is Stage 3, settled. (2) A baseline is entered and the rise does NOT qualify: the route does not fire at all and the ratio and rise criteria stand on their own — a chronic elevation stages as what it is, which is the whole point of asking for the baseline. (3) NO baseline is entered: the definition cannot be assessed either way, so Stage 3 is reported but the 'stage is not settled' output is set. Gating case 3 strictly would return Stage 0 for every patient entered without a prior value, and in a PICU that under-staging is the more dangerous error by a wide margin — reporting it as a bound neither under-stages nor claims more than the entered number supports. A patient who is already Stage 3 by a settled route (RRT, eGFR < 35 under 18 years, ≥ 3.0× baseline, or a closed Stage-3 urine-output row) is NOT flagged, because nothing about that answer is open. WHAT TO ENTER WHEN THERE IS NO PRIOR CREATININE: the LOWEST creatinine measured during this admission. It is a value the bedside already has, and THE EVIDENCE BEHIND IT IS NOW PAEDIATRIC RATHER THAN BORROWED FROM ADULTS. Lee 2022 (Kidney Res Clin Pract 2022;41(3):322-331, DOI 10.23876/j.krcp.21.120) took 710 critically ill children aged 1 month to 18 years who each had a real measured baseline on file, and asked which surrogate reproduces it. The lowest creatinine within 7 days of PICU admission won: agreement with the true baseline ICC 0.62, AKI detected with sensitivity 87.8% and specificity 71.0%, misclassification 19.2%, kappa 0.60, and an AKI incidence of 63.5% against a true 58.7% — a slight OVER-estimate. Back-calculating a baseline from an assumed eGFR, which is what KDIGO's own appendix suggests via MDRD from an assumed GFR of 75 mL/min/1.73 m², was far worse in the same children and worse in the direction that matters: sensitivity 31.5% (specificity 98.3%, misclassification 40.3%), and an AKI incidence of 19.1% against the same true 58.7% — roughly two thirds of the AKI in the cohort simply not seen. DO NOT BACK-CALCULATE. THE DIRECTION REVERSES BETWEEN ADULTS AND CHILDREN, AND ANYONE REASONING FROM ADULT PAPERS WILL GET THIS BACKWARDS: Lee 2022 contrasts its own finding with adult reports in which back-calculation OVER-estimates AKI, whereas in these children it UNDER-estimated severely. In a PICU that is the dangerous direction — an over-call gets reviewed and dropped at the next creatinine, a miss is never looked at again — so an adult paper's reassurance that back-calculation errs toward over-diagnosis must not be carried across. 'The adult literature' is not one voice on this either: Cooper 2021 (PMID 33732979), the adult cohort this score cited before the paediatric study existed, found assumed-GFR-75 methods MISSING more than half of all AKI, which is the same direction as the paediatric result rather than the opposite one. Whichever adult report a reader starts from, the instruction here is identical — enter a measured creatinine, do not compute one. THE SCr-MIN WINDOW IS NOT STANDARDISED, and the disagreement is worth seeing rather than smoothing over: published definitions of the 'lowest' creatinine run from 3 days, to 7 days, to the whole hospitalisation. Lee 2022 chose 7 days from PICU admission, so the operating characteristics above are 7-day characteristics; a different window is a different surrogate and does not inherit them. This calculator enforces no window — it asks for a number — so record which window the value entered came from. NO BACK-CALCULATION IS OFFERED HERE, deliberately: those equations need sex and race inputs this score does not collect, and race-based eGFR is contested in its own right, so the calculator asks for a number rather than manufacturing one. THE ADULT EVIDENCE IS KEPT ONLY FOR WHAT THE PAEDIATRIC STUDY DOES NOT TEST: Cooper 2021 (247 adults with Plasmodium knowlesi malaria in Malaysian Borneo) compared MDRD against CKD-EPI, an assumed GFR of 100 as well as 75, and age/sex-standardised reference tables, finding CKD-EPI at an assumed GFR of 100 tracked overall incidence best while still misassigning stages, and the lowest admission creatinine over-calling AKI by about a fifth while correlating with the reference value more closely than any estimate. It is single-infection adult evidence and is no longer the primary support for anything on this page. The pediatric eGFR branch is contested for young children — GFR rises developmentally and the bedside Schwartz equation was validated ~1–16 y, so do not extrapolate to neonates without a neonatal-specific estimator [NEEDS SOURCE for a neonatal eGFR method]. Creatinine SI↔conventional conversion reuses this project's shared clinical factor, 1 mg/dL = 88.42 µmol/L (creatinine's molar mass, 113.12 g/mol), whereas KDIGO itself prints the rounder 88.4. THIS IS SETTLED, NOT OPEN. The two differ by about 0.02% and cross no published threshold in either direction: KDIGO's own 4.0 mg/dL Stage-3 cutoff is 353.60 µmol/L at 88.4 and 353.68 at 88.42, and its 0.3 mg/dL rise is 26.52 against 26.53. Both resolve the guideline's printed SI thresholds to the same two-decimal mg/dL figure — 353.6 µmol/L → 4.00, a 26.5 µmol/L rise → 0.30 — so no stage, on this score or any other, turns on the choice. It is recorded as a documented and clinically immaterial implementation choice; the difference is disclosed here because the numbers should not appear to disagree silently, not because anything remains to be decided. The mg/dL cutoffs remain the authoritative ones, but a creatinine entered in µmol/L is converted and then rounded to the two decimal places creatinine is reported to in mg/dL, so KDIGO's own printed SI equivalents stage as the guideline intends: 353.6 µmol/L resolves to 4.00 mg/dL and meets the ≥ 4.0 Stage-3 criterion (unrounded it is 3.999095 and would miss it), and a 26.5 µmol/L rise resolves to a 0.30 mg/dL rise. Values genuinely below a cutoff are unaffected — 349.3 µmol/L resolves to 3.95 mg/dL and does not stage. A value entered in mg/dL is used exactly as typed. Higher KDIGO stage is associated with higher mortality and RRT risk in the outcome literature, but the staging itself is a classification, not a treatment threshold — keep any display descriptive. The per-input plausible min/max are input-validity guardrails, not published KDIGO thresholds; the age range deliberately extends past 18 years so that an adult stages correctly rather than being silently treated as a child.