Effect of Serum Albumin Changes on Mortality in Patients with Peritoneal Dialysis: A Joint Modeling Approach and Personalized Dynamic Risk Predictions
Table 3
Joint modeling results—415 patients used for training the model and 2 patients were used as test samples for dynamic risk predictions.
(a)
Longitudinal part (linear mixed effects)
Variable
Estimate (95% CI)
value
Serum creatininea
0.028 (0.012, 0.042)
0.004
Serum calciuma
0.279 (0.198, 0.358)
<0.001
BUNa
-0.0001 (-0.004, 0.004)
0.968
Age at PD initiation
-0.005 (-0.009, 0.001)
0.081
Number of diseasesb
0.0006 (-0.063, 0.057)
0.998
Peritonitis rate
-0.051 (-0.123, 0.018)
0.163
WBC
-0.029 (-0.058, 0.014)
0.103
Transportation characteristic (high)†
-0.270 (-0.414, -0.207)
<0.001
(b)
Survival part (Cox proportional hazard)
Variable
Estimate (95% CI)
HR (95% CI)
value
Serum creatininea
-0.212 (-0.336, -0.096)
1.24 (1.10, 1.40)†††
<0.001
BUNa
0.004 (-0.016, 0.025)
1.01 (0.98, 1.03)
0.668
Age at PD initiation
0.016 (-0.001, 0.34)
1.02 (0.99, 1.40)
0.069
WBC
0.061 (-0.073, 0.183)
1.06 (0.93, 1.20)
0.323
HD history (yes)†
1.165 (0.665, 1.740)
3.21 (1.94, 5.70)
<0.001
Number of diseasesb
0.345 (0.09, 0.661)
1.41 (1.09, 1.94)
0.009
BMI
0.071 (0.011, 0.148)
1.07 (1.01, 1.16)
0.010
Peritonitis rate
0.556 (0.140, 0.891)
1.74 (1.15, 2.44)
<0.001
Transportation characteristic (high)†
-0.095 (-0.467, 0.276)
0.91 (0.63, 1.32)
0.622
Albumin ()††
-0.889 (-1.425, -0.392)
2.43 (1.48, 4.16)†††
<0.001
HR: hazard ratio; BMI: body mass index (kg/m2); GFR: glomerular filtration rate. †Model parameters were obtained for the group given in parenthesis. ††Albumin levels are estimated from longitudinal part of joint model. †††Hazard ratios were estimated for 1-unit decrease in corresponding predictors. aAveraged over the follow-up period. bThe total number of comorbid and renal diseases observed in a patient.
