Accepted answer
An 11-point fall from 85 is 13%, which sits comfortably inside the range conventionally treated as an expected haemodynamic dip rather than injury. But the more interesting arithmetic is that your creatinine change may not even be statistically distinguishable from noise, and the mechanism question can largely be settled from analytes that were almost certainly on the same requisition.
First, verify the eGFR arithmetic
The 2021 CKD-EPI creatinine equation:
eGFR = 142 × min(Scr/κ, 1)^α × max(Scr/κ, 1)^(−1.200) × 0.9938^age × 1.012 [if female]
where κ = 0.9 (male), 0.7 (female)
α = −0.302 (male), −0.241 (female)
For you at baseline, Scr 1.05 mg/dL, male, 52:
- Scr/κ = 1.05 ÷ 0.9 = 1.1667, which is above 1, so the min term is 1 and only the max term applies.
- 1.1667^(−1.200): ln(1.1667) = 0.15415; × (−1.200) = −0.18498; exp(−0.18498) = 0.8311.
- 0.9938^52: ln(0.9938) = −0.006219; × 52 = −0.32339; exp(−0.32339) = 0.7237.
- eGFR = 142 × 0.8311 × 0.7237 = 142 × 0.6014 = 85.4 mL/min/1.73 m².
At the follow-up, Scr 1.18:
- Scr/κ = 1.18 ÷ 0.9 = 1.3111. ln = 0.27092; × (−1.200) = −0.32510; exp = 0.7224.
- eGFR = 142 × 0.7224 × 0.7237 = 142 × 0.5228 = 74.2 mL/min/1.73 m².
So the reported numbers are right: a fall of 11.2 units, 13.1% of baseline.
The part that should give you pause about over-reading it
Your creatinine rose from 1.05 to 1.18 mg/dL, a change of 12.4%. The reference change value for creatinine, combining an analytical CV of about 3% with a within-person biological CV of about 5%, is 2.77 × sqrt(9 + 25) = 2.77 × 5.83 = 16%.
Your 12.4% is below that, so formally a single pair differing by 12% does not establish that anything changed. The eGFR difference of 13 percentage points looks more alarming because the equation is non-linear and amplifies small creatinine movements into larger-looking eGFR movements — worth knowing generally, since it is the source of most eGFR panic.
For contrast, had the follow-up been 1.30 mg/dL, that is a 23.8% rise, which clears the RCV: 1.30 ÷ 0.9 = 1.4444; ln = 0.36772; × (−1.2) = −0.44127; exp = 0.6433; eGFR = 142 × 0.6433 × 0.7237 = 66.1. A fall of 22% — still inside the conventional tolerance band, but now genuinely measured.
The threshold, and where it comes from
The widely used heuristic is that a fall of up to about 30% of baseline eGFR, occurring early and then stabilising, does not warrant stopping the drug, whereas a larger or progressive fall warrants investigation.
Be clear about the provenance: that figure was established in the renin-angiotensin blockade literature and carried across to SGLT2 inhibitors, where continuing through the dip is associated with better long-term outcomes than stopping. It has not been formally established for GLP-1 receptor agonists, where the acute dip is smaller and less consistent. Borrowing it is reasonable and it is a borrowing.
Mechanism: what the dip actually is
In hyperfiltration — common with obesity and with diabetes — single-nephron GFR is elevated by raised intraglomerular pressure, which is damaging over years. Reduce that pressure and measured GFR falls immediately while the nephron is being protected. Contributory mechanisms in this class include natriuresis, inhibition of proximal tubular sodium-hydrogen exchange increasing distal sodium delivery, activation of tubuloglomerular feedback with afferent arteriolar constriction, plus reduced systemic blood pressure and reduced glomerular hypertrophy as weight falls.
So a dip is a marker of the drug acting on the relevant lever, not of harm. This is why slope analyses in this class are usually presented both from baseline and from a post-initiation reference point — the acute dip drags the from-baseline figure down and understates the chronic benefit [1].
Telling it apart from a prerenal insult, from data you probably already have
A haemodynamic dip and volume depletion both raise creatinine. They differ in what else moves. Check these on the same draw:
- Urea (BUN) relative to creatinine. In volume depletion urea rises proportionally more, because low tubular flow increases passive urea reabsorption. In US units a BUN:creatinine ratio above roughly 20:1 points prerenal; below 15:1 argues against. In SI, urea in mmol/L divided by creatinine in µmol/L times 1000 gives a comparable index, above about 100 suggesting prerenal. A pure haemodynamic dip barely disturbs the ratio.
- Haematocrit, haemoglobin and albumin. All three rise with haemoconcentration. If albumin went from 41 to 45 g/L and haematocrit rose alongside the creatinine, you were dry.
- Sodium and bicarbonate. Vomiting produces a hypochloraemic, hypokalaemic picture with raised bicarbonate; diarrhoea does the opposite to bicarbonate. Either points at GI fluid loss.
- Time course. A haemodynamic dip appears in the first few weeks then plateaus. A prerenal insult tracks the GI episode and resolves within a week or two of it ending.
Your week-five vomiting is what makes this worth checking rather than assuming. If the draw was within a fortnight of that episode, repeat creatinine and electrolytes after two to three weeks of normal eating and drinking. If it returns towards 1.05 you had a transient prerenal component; if it sits at 1.18 you have a new steady state, which is what a haemodynamic dip looks like.
What would change the answer: a continuing rise across three draws, a fall beyond 30% of baseline, new albuminuria, hyperkalaemia, or symptoms. Those belong with a clinician, not a spreadsheet.
edited 2 Dec 2024 by cold_lane — removed a claim I could not source
The observation that the eGFR equation amplifies small creatinine changes into scarier-looking eGFR changes explains a lot of panic. – Dr_Signe_Baldursdottir 26 days ago Pointing out that the 30% rule is borrowed from the RAS-blockade and SGLT2 literature rather than established for this class is the honest caveat. – juan_esquivel 9 months ago 3Urea:creatinine ratio plus albumin and haematocrit on the same draw is a free volume assessment. – Dr_Nadia_Farsi 8 months ago add a comment