Skip to main content
. 2021 Jul 1;4:103. doi: 10.1038/s41746-021-00474-9

Table 1.

Statistics for the two datasets on DRG.

Train set Test set Test cohort
MS-DRG
 Patient 14,836 1648 1648
 Hospital stay 17,815 1977 1648
 Unique DRG 595 369 349
 DRG weight 3·045 (2·747) 3·049 (2·790) 3·147 (2·835)
 Stay of MDC 05 4633 (26·0%) 522 (26·4%) 456 (27·7%)
 Stay of MDC 01 2765 (15·5%) 320 (16·2%) 293 (17·8%)
 Stay of MDC 04 2115 (11·9%) 219 (11·1%) 169 (10·3%)
APR-DRG
 Patient 20,266 2252 2252
 Hospital stay 24,667 2747 2252
 Unique DRG 893 517 480
 DRG weight 3·265 (3·166) 3·291 (3·211) 3·300 (3·153)
 Stay of MDC 05 6473 (26·2%) 733 (26·7%) 613 (27·2%)
 Stay of MDC 01 3556 (14·4%) 424 (15·4%) 379 (16·8%)
 Stay of MDC 04 2881 (11·7%) 319 (11·6%) 239 (10·6%)

Data is reported in count (%) or mean (SD). DRG is assigned on the basis of hospital stay, and a patient can have more than one stay in both datasets. Each unique DRG code in MS-DRG corresponds to the DRG group, whereas in APR-DRG the DRG code combines clinical group and patient severity. All DRG codes have a relative weight that is connected to reimbursement. Three most common MDCs: diseases of disorders of the circulatory system (05), nervous system (01), and respiratory system (04).

DRG diagnostic-related group, SD standard deviation, MS-DRG Medicare severity-DRG, APR-DRG all patient refined-DRG, MDC major diagnosis category.