This lesson is about an honest divergence between documents, not about "choose the one you like".
KDIGO 2022 (Kidney Int. 2022;102(5S):S1–S127) for a person with T2DM and CKD:
- an SGLT2 inhibitor at eGFR ≥20 mL/min/1.73 m² (Recommendation 1.3.1, 1A) — for the kidney and heart, not for glucose;
- once started, it is reasonable to continue an SGLT2 inhibitor even if eGFR falls below 20, as long as the drug is tolerated and kidney replacement therapy has not been initiated (practice point 1.3.6);
- metformin at eGFR ≥30 (4.1.1, 1B);
- a GLP-1 agonist is the preferred next glucose-lowering class if glycaemic targets are not achieved on metformin and/or an SGLT2 inhibitor.
ADA 2026 agrees with a threshold of about 20 mL/min/1.73 m² for starting an SGLT2 inhibitor in the CKD population (9.10) and separately prefers GLP-1 at eGFR <30 for glycaemia (9.11). The ADA/KDIGO 2022 consensus recorded the alignment of the SGLT2i threshold down to ≥20.
The MoH RK 2022 protocol says something different in Table 10. For empagliflozin and dapagliflozin: discontinue at a sustained eGFR below 45 mL/min/1.73 m²; for canagliflozin — below 30. This reflects the drug labels and the evidence base as of the year the protocol was written (based on ADA 2021, the ADA/EASD 2018 consensus, ESC/EASD 2019, and the 2019 Kazakh consensus). CREDENCE had already been published by then (2019), but the starting threshold of 20 mL/min was not yet in KDIGO: it was the KDIGO 2022 edition that introduced it, lowering the bar from 30 to 20.
How the student should answer in two rooms.
In an exam on international standards: a person with T2DM, eGFR 28, UACR 400 mg/g, not on dialysis — KDIGO 2022 recommends an SGLT2 inhibitor (start at ≥20), metformin is no longer started (eGFR <30), and GLP-1 is useful both for glycaemia and, after FLOW 2024, for kidney-cardiovascular outcomes in a similar population.
In an exam on the MoH RK 2022 protocol: the same person — metformin is contraindicated (GFR ≤30); according to Table 10, empagliflozin and dapagliflozin are discontinued at a sustained GFR <45; for canagliflozin the discontinuation threshold is <30. Yet the protocol's management text recommends an SGLT2i or GLP-1 RA in CKD "to reduce the risks of CKD progression and cardiovascular events" — the general stance and the table of contraindications within one document diverge on GFR. This must be seen, not papered over.
What does not count as a basis. The KDIGO 2026 draft (Public Review Draft, March 2026) expands the role of long-acting GLP-1-based therapies to 1A for people with T2DM and CKD, at risk of MACE or not achieving individual glycaemic targets. Until final publication this is not a threshold. FLOW (2024) is an already published RCT, and ADA 2026 takes it into account in the living standard without waiting for the final KDIGO 2026.
The practical minimum that does not depend on the divergence. Calculate eGFR and UACR. Do not keep a person with A3 "on a sulfonylurea alone". Do not continue metformin at eGFR <30. Do not confuse the weakening of the glucose-lowering effect of an SGLT2 inhibitor at low GFR with the disappearance of organ protection.