Discriminant Scores for Predicting Aminoglycoside Nephrotoxicity

Aminoglycosides available at the time of the study: gentamicin and tobramycin

Nephrotoxicity was defined as a decrease in creatinine clearance by 50% or more.

after Table 2, page 354

discriminant score prior to therapy =

= (0.049 * (age in years)) + (1.872 * (points for liver disease))  + (0.025 * (points for creatinine clearance))+ (1.102 * (points for sex)) – 5.098

where:

• I did not see where the discriminant score based on pretreatment data was used to predict nephrotoxicity.

after Table 2, page 354

discriminant score for 72 hours =

= (0.333 * (drug level points)) + (1.312 * (liver points)) + (0.032 * (age in years)) + (0.016 * (points for creatinine clearance)) + (0.739 * (points for sex)) + (0.897 * (points for shock)) – 5.357

where:

• I would think this data would be known sooner than 72 hours after therapy was started. It may be that drug assays had a longer turnaround time in 1984, especially if microbial growth inhibition assays were used.

after Figure 1, page 355

If this data is analyzed in JMP, the following equations can be derived:

risk over range of –2.0 to 0.5 =

= (1.562963 * ((score)^3)) + (8.1666667 * ((score)^2)) + (15.444974 * (score)) + 11.45873

risk over range of  0.5 to 1.5 =

= (12 * ((score)^2)) + (6 * (score)) + 15.5

risk over range of 1.5 to 4.0 =

= (0.3925926 * ((score)^3)) - (8.574603 * ((score)^2)) + (55.232011 * (score)) – 13.51667

A nomogram for calculating the risk is shown in Figure 3 on page 356,

Limitations:

• The data does not apply to patients treated with amikacin or netilimicin.

• The risk for nephrotoxicity increases with the creatinine clearance. I would have thought the reverse would be true.