Statistical analysis

Data are presented as means with standard deviations (SD) or medians with interquartile ranges [IQR] depending on distributions. Categorical data are presented as proportions with 95% confidence intervals (CI) calculated using the Clopper-Pearson interval. Groups were compared using the Fisher’s exact test, t-test or Mann–Whitney U-test depending on distributions. We performed no formal sample size calculation.

For the main analysis (to assess the incidence of bioaccumulation in patients receiving intermediate dose nadroparin) we calculated the proportion (95% CI) of patients with at least one trough anti-Xa level > 0.2 IU/mL in two groups: those who had anti-Xa measurements while having any episode of AKI, and those without AKI. Analyses were not further stratified according to AKI stage due to the low sample size. To quantify the association between trough anti-Xa levels and renal function, we used a linear multilevel model with trough anti-Xa levels modelled as a function of time and AKI, adjusted for body mass index (BMI). Analyses were limited to patients with a stable dose course of intermediate dose nadroparin. AKI was entered into the model as a time-dependent binary variable to allow for non-linear change over time. We assessed interaction terms between the covariates (AKI and BMI) and time. Clustering in the data was accounted for by a random intercept and random slope at patient level. We selected the best fitting model based on likelihood ratio tests. Model assumptions were checked and judged to have been sufficiently met. Since trough anti-Xa values below the limit of detection are reported as ‘ < 0.04 IU/ml’ we performed the analyses twice: with imputed values of 0.0 IU/ml (main model) and 0.02 IU/ml (sensitivity analysis 1). To explore the effect of any recent nadroparin dose change we did a second sensitivity analysis including all patients regardless of recent dose changes (sensitivity analysis 2).

For the secondary analysis, we compared trough anti-Xa levels in patients with AKI to anti-Xa levels in patients without AKI using a Mann–Whitney U test. A two-sided P-value < 0.05 indicated statistical significance. All analyses were performed using R software version 4.0.4.