Table 3

Osmolar and acid–base temporal models

Dependent variableEffective osmolalitySerum glucoseSerum sodiumGlucose-corrected sodiumpHAnion gapBase excessPCO2
Fractional polynomial powers (P1, P2)(0.5, 2)(−1, −1)(0.5, 2)(0.5, 2)(−0.5, 0)(1, 1)(2, 2)(−0.5, 0)
Intercept variables
 Oedema group (control, early, late)0.200.510.030.080.310.860.280.10
 Blood urea0.010.110.590.030.570.630.200.001
Trajectory (slope) variables
 Time power 10.04<0.001<0.0010.001<0.001<0.001<0.001<0.001
 Time power 20.001<0.0010.070.08<0.001<0.001<0.001<0.001
 Time power 1 × group interaction<0.0010.780.001<0.0010.
 Time power 2 × group interaction<0.0010.86<0.001<0.0010.550.170.020.08
  • The first row of the table shows the fractional polynomial powers that provided the best model fit for each of the dependent variables. In no case was the change in dependent variable with time trajectory linear (this would be a model with P1=1 and absence of P2). The numbers within the table body represent the multivariable p values (significance) for each of the independent variables. As an example, interpretation for effective osmolality (first column) is as follows. The intercepts (values of effective osmolality at time zero) were not different for the three oedema groups (p=0.2); however, blood urea did influence the intercept (time zero value) for effective osmolality within individual patients (p=0.01). This was not seen for PCO2 (p=0.1). Overall, effective osmolality changed over time in a non-linear fashion (time power 1, or time0.5 p=0.04; and time power 2, or time2 p = 0.001). In addition, the rate of change for effective osmolality differed for the three oedema groups, as evidenced by the highly significant p values for both time-by-group interactions (time0.5 by group, and time2 by group, both p<0.001). The prototypical trajectories for each group are also shown in figure 1.

  • PCO2, partial pressure of dissolved carbon dioxide at presentation to hospital.