Table 4.
Bivariate and multivariable analysis of the association between natremia levels and lengths of stay or 30-day unplanned readmissions: Lausanne University Hospital, 2012–2018.
| Length of stay (days) | Readmissions | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Bivariate | Multivariable | Bivariate | Multivariable | ||||||
| Median [IQR] | p value | Mean ± SE | p value§ | No | Yes | p value | OR (95% CI) | p value | |
| Natremia levels | <0.001 | <0.001 | <0.001 | ||||||
| Normal | 8 [5–13] | 10.7 ± 0.2 | 2,946 (69.4) | 1,470 (64.1) | 1 (reference) | ||||
| Decreased | 10 [7–17] | 13.4 ± 0.2 | 1,301 (30.6) | 822 (35.9) | 1.15 (1.01–1.31) | 0.031 | |||
| Natremia levels | <0.001 | <0.001 | <0.001 | ||||||
| Normal | 8 [5–13] | 10.6 ± 0.2 | 2,946 (69.4) | 1,470 (64.1) | 1 (reference) | ||||
| Mild (130–135 mEq/L) | 10 [7–16] | 12.6 ± 0.3 | 910 (21.4) | 557 (24.3) | 1.14 (0.98–1.31) | 0.080 | |||
| Moderate (125–129 mEq/L) | 12 [8–20] | 15.2 ± 0.5 | 253 (6.0) | 175 (7.6) | 1.22 (0.96–1.54) | 0.106 | |||
| Severe (<125 mEq/L) | 12 [8–21] | 16.1 ± 0.7 | 138 (3.3) | 90 (3.9) | 1.14 (0.82–1.56) | 0.438 | |||
| p value for trend | <0.001 | 0.381 | |||||||
| Persistent hyponatremia | <0.001 | 0.003 | <0.001 | ||||||
| Normal | 8 [5–13] | 10.7 ± 0.2 | 2,946 (69.4) | 1,470 (64.1) | 1 (reference) | ||||
| Non-persistent | 11 [7–19] | 14.2 ± 0.3 | 890 (21) | 517 (22.6) | 1.04 (0.89–1.20) | 0.644 | |||
| Persistent | 10 [6–15] | 11.9 ± 0.4 | 411 (9.7) | 305 (13.3) | 1.41 (1.17–1.71) | <0.001 | |||
| p value for trend‡ | 0.003 | <0.001 | |||||||
CI, confidence interval; IQR, interquartile range; OR, odds ratio; SE, standard error.
Multivariable model.
Based on log-transformed data.
Results are expressed as median [interquartile range] or multivariable-adjusted mean ± standard error for continuous variables and as number of hospitalizations (percentage) or multivariable-adjusted odds ratio (95% confidence interval) for categorical variables. Between-group bivariate analysis performed using Kruskal–Wallis test for continuous variables and chi-square for categorical variables. Multivariable analysis conducted using multilevel mixed-effects linear regression for continuous variables and multilevel mixed-effects logistic regression for categorical variables. Multivariable analyses were performed adjusting for age (four categories), gender, main cause for hospitalization (eight categories), and number of associated comorbidities.