Table 2.
Predictor variables | Change in Hct (%) | 95% CI | P-value |
---|---|---|---|
Time, per 5 years of follow-up | −1.4 | (−1.9, −0.8) | <0.001 |
Change in eGFR, per 10 mL/min/1.73 m2 decrease for baseline eGFR | |||
20 mL/min/1.73 m2 | −3.7 | (−4.2, −3.2) | <0.001 |
30 mL/min/1.73 m2 | −2.5 | (−2.8, −2.2) | <0.001 |
40 mL/min/1.73 m2 | −1.3 | (−1.5, −1.2) | <0.001 |
45 mL/min/1.73 m2 | −0.8 | (−1.0, −0.5) | <0.001 |
50 mL/min/1.73 m2 | −0.6 | (−0.8, −0.5) | <0.001 |
60 mL/min/1.73 m2 | −0.5 | (−0.6, −0.3) | <0.001 |
eGFR, estimated glomerular filtration rate; Hct, hematocrit; CI, confidence interval.
Data presented as absolute change in hematocrit for six sample baseline eGFR values.
In addition to the listed predictor variables, the joint model included additional terms to control for age at the time of randomization, gender, baseline proteinuria (urine protein-to-creatinine ratio >0.22), interaction terms between baseline eGFR (linear spline with a knot at 45 mL/min/1.73 m2) and time-varying eGFR, randomized drug assignment, randomized blood pressure target group, and interaction terms between treatment assignments and time.