Prediction of health care expenditure increase: how does pharmacotherapy contribute?

Table 4 Weight analysis: Contribution of drug classes to the prediction

Contribution to prediction of increase
ATC	Name	Acc,%	N
N06DA	Anticholinesterases	77.3	88
A12CC	Magnesium	71.4	56
B03BB	Folic acid and derivatives	69.1	408
A10AE	Insulins and analogues for injection, long-acting	68.1	91
B01AA	Vitamin K antagonists	67.3	1065
A03FA	Propulsives	65.5	58
Contribution to prediction of decrease
		Without Hospitalisation		With Hospitalisation
ATC	Name	Acc,%	N	Acc,%	N
A12CC	Magnesium	78.6	28	78.8	33
B01AC	Platelet aggregation inhibitors excluding heparin	76.9	26	82.9	345
C01BD	Antiarrhythmics, class III	73.3	30	81.0	58
N06CA	Antidepressants in combination with psycholeptics	69.5	59	75.0	44
B01AA	Vitamin K antagonists	68.8	173	74.5	471
A03FA	Propulsives	65.8	146	84.6	143

All numbers were calculated on the test dataset. Drug groups contributing at least 5% to the overall positive or negative score (complete model without costs). Additionally, the drug classes must have contributed for at least N = 40 patients (increase) or N = 20 patients (decrease without hospitalisation). The top 6 drug classes are provided, arranged by descending order of accuracy
Acc = Accuracy
N = Number of patients for whom the drug class contributed at least 5% to the overall positive or negative score
ATC = “Anatomical Therapeutic Chemical” classification code
Bold data are significant

ISSN: 1472-6963