Skip to main content

Table 5 Examples of subgroups derived from the decision tree

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

 

Examples of subgroups for increase

N

PI

Gain

 #1

Patients younger than 35 years with at least 1 prescription for folic acid (B03BB), no more than two outpatient office visits in the second quarter of the year, and fewer than 12 drug prescriptions

634

0.88

0.23

 #2

Patients with at least 1 prescription for magnesium (A12CC), no hospitalisation (≤ 1 day), at least 5 outpatient office visits with a gynaecologist during the year, no more than 1 outpatient visit in the first quarter of the year overall, and at least 4 visits in the third quarter of the year

265

0.86

0.16

 #3

Patients with at least 1 prescription for iron (trivalent, oral preparations, B03AB), a deductible > 1000 Swiss francs for 2014, no change in this deductible for 2015, at least 6 prescribed drugs, and no more than 5 prescriptions*

114

0.78

0.21

 #4

Patients with at least 1 prescription for anticholinesterases for dementia (N06DA), no home care (≤ 1 day), no more than 1 prescription in February, and few prescriptions filled by pharmacies

276

0.70

0.13

 

Examples of subgroups for decrease

N

PD

Gain

 #5

Patients with at least 2 prescriptions for anticholinergics for ophthalmologic use (S01FA), no concomitant therapy with Vitamin K antagonists, no more than 11 prescriptions, and no more than 6 outpatient physician visits in the third quarter of the year

303

0.89

0.45

 #6

Patients with at least 4 prescriptions for platelet aggregation inhibitors (excluding heparin, B01AC), who were hospitalised (cardiac-related major disease category) and had frequent home care (mean interval < 3.3 days)

3777

0.83

0.04

 #7

Patients with at least 2 prescriptions for any other general anaesthetics (N01AX), with a mode of administration ‘intravenously’ and fewer than 2 prescriptions for sulfonamides (C03CA), and no hospitalisation in the first year

4237

0.74

0.18

 #8

Patients with at least 1 prescription for both a beta blocking agent (S01ED) and a corticosteroid and anti-infective (S01CA) for ophthalmologic use within one year, who had a maximum of 2 outpatient visits in December

2229

0.67

0.08

  1. N = size of subgroup; PI, PD = conditional probability of increase or decrease for the cuts, P(increase | cut); Gain = |difference P(increase | cut) − P(increase | cut without drug class)|; Maximal number of cuts = 5, *defined as different purchase dates
  2. Bold data are significant