Skip to main content

Table 1 Hospital profiles identified by the k-means algorithm

From: Making sense of the French public hospital system: a network-based approach to hospital clustering using unsupervised learning methods

Group

Group 1 (n = 34)

Group 2 (n = 236)

Group 3 (n = 157)

Number of stays per hospital – median [Q1 - Q3]a

82,108

[69,004 – 117,774]

18,913

[12,410 – 26,689]

2337

[1195 – 5285]

Number of beds – median [Q1 - Q3]

1129 [968 – 1431]

258 [164 - 377]

48 [30 - 80]

Regional Hospital Group classifying stays (%)b

Mean (standard deviation - SD)

2 (1)

4 (2)

11 (5)

Regional Hospital Group leader – n (%)

34 (100.0)

88 (37.3)

0 (0.0)

Absolute number of activity domains

Mean (standard deviation - SD)

24.74 (0.45)

23.61 (1.01)

19.94 (3.65)

Effective number of activity domains – Mean (standard deviation SD)

15.86 (0.70)

12.26 (1.72)

8.63 (2.73)

Degree centrality (undirected) – median [Q1 – Q3]

252 [217 - 305]

102 [68 - 134]

29 [18 - 45]

Strong edge (>  100 patients) degree centrality – median [Q1 – Q3]

19 [13 - 24]

4 [3 - 6]

2 [2 - 2]

Inbound strength (incoming mobility) – median [Q1 – Q3]

4591 [2991 – 5631]

741 [431 – 1075]

292 [203 - 410]

Outbound strength (exterior mobility) – median [Q1 – Q3]

3307 [2432 – 4111]

946 [620 – 1198]

349 [152 - 551]

  1. aHospital stays in Medicine, Surgery and Obstetrics (MSO), for year 2016 in metropolitan France (excluding oversea territories), with quality control code equal to 0 (valid stays), excluding stays with an undefined Diagnosis Related Group (Major Diagnostic Category 90) and excluding patient transfers with a length of stay < 48 h in the receiving hospital
  2. bStays that were followed by subsequent stays in another hospital of the same Regional Hospital Group within 90 days