Table 2.

Multivariable model for association between maximum acute kidney injury stage and post-discharge eGFR results.

	% change in eGFR (95% CI)	p value
All patients (n = 1301)
Age (per 1 year increase)	− 0.7 (− 0.8, − 0.5)	< 0.001
Female gender	0.1 (− 3.5, 3.9)	0.939
Baseline eGFR (per 10 increase)	10.5 (9.5, 11.5)	< 0.001
Pre− existing health conditionsa	− 3.7 (− 8.8, 1.7)	0.173
Change from 0 to 6 months
No AKI	− 9.5 (− 11.5, − 7.4)	< 0.001
Mild AKI	− 12.3 (− 15.1, − 9.4)
Moderate− severe AKI	− 4.3 (− 7.0,− 1.4)
Annual rate of change 6 months–7 years
No AKI	− 1.4 (− 2.2, − 0.6)	0.955
Mild AKI	− 1.5 (− 2.6, − 0.3)
Moderate− severe AKI	− 1.6 (− 2.7, − 0.5)

Log (eGFR) was modelled using linear mixed effects models with a split slope for time, allowing the rate of change to differ in the first 6 months after discharge as compared to longer term. Exponentials of model co-efficient were calculated to provide estimates of the effect of the dependent variables on the % change in eGFR. There was no evidence in of a difference in the effect of recovery by AKI level up to 6 months or from 6 months to 7 years (p for interaction = 0.256 and 0.218 , respectively).

AKI acute kidney injury ICU, eGFR estimated glomerular filtration rate.

^aRecorded according to the APACHE score chronic organ insufficiency.