Skip to main content
. 2020 Dec 21;12(12):3868. doi: 10.3390/cancers12123868

Table 2.

Linear regression analysis for LOS (log-transformed). The univariate models include CCI or Clavien-Dindo as the only covariate. The multivariate models are adjusted for the following covariates: type of surgery (major vs. minor), open vs. laparoscopy, age (per year), American Society of Anesthesiologists (ASA) score (1–2 vs. 3–4), Child grade (A vs. B), duration of surgery (>4 h vs. ≤4 h), center volume (high vs. medium/low). Patients died during hospital stay are excluded from all samples.

Sample Model Derivation Set Validation Set
Mean Change (95% CI) R2 RMSE R2 RMSE
Overall sample
Derivation set n = 1331.
Validation set n = 1304.
CCI, per 10 unit increase
Univariate model
0.290 (0.273;0.307) 45% 0.41 36% 0.49
Clavien-Dindo, per category increase
Univariate model
0.312 (0.292;0.332) 41% 0.43 34% 0.49
CCI, per 10 unit increase
Multivariate model
0.267 (0.251;0.282) 57% 0.37 44% 0.50
Clavien-Dindo, per category increase
Multivariate model
0.291 (0.273;0.309) 54% 0.38 43% 0.51
Subgroup of patients with at least two complications
Derivation set n = 177. Validation set n = 154.
CCI, per 10 unit increase
Univariate model
0.240 (0.197;0.283) 41% 0.47 15% 0.59
Clavien-Dindo, per category increase
Univariate model
0.254 (0.186;0.323) 24% 0.54 12% 0.59
CCI, per 10 unit increase
Multivariate model
0.219 (0.176;0.262) 47% 0.45 23% 0.59
Clavien-Dindo, per category increase
Multivariate model
0.223 (0.157;0.289) 32% 0.52 21% 0.63

LOS: Length of stay; CCI: Comprehensive complication index; CI: Confidence Interval; RMSE: Root mean squared error.