2.7 Data and Statistical analysis
The data and statistical analysis comply with the recommendations on
experimental design and analysis in pharmacology (Curtis et al., 2018).
Urinary output is expressed as mean ± SD. All available urine samples
were analyzed. Urinary biomarkers were standardized for urinary volume
by calculating the total amount excreted in each 24-h period.
Differences between groups were assessed using mixed-effects, restricted
maximum likelihood estimation regression, with repeated measures
occurring over days; measures were repeated at the level of the
individual rat (STATA 17.0 BE, StataCorp LLC, College Station, Texas).
The primary analysis reports the
simple effects from joint tests of drug treatment group within each
level of treatment day. Margins were calculated for a full factorial of
the variables, i.e. main effects for each variable and interactions.
Referent groups were pre-treatment baseline values and saline as a
treatment, except where noted. Secondarily, to remain agnostic to
outcome variable relationship over time (and assess treatment groups
over time), locally weighted scatterplot smoothing (LOWESS) trendlines
with 95% confidence intervals were generated. LOWESS graphs were
produced in R version 4.2.2 using ggplot2 (”H. Wickham. ggplot2: Elegant
Graphics for Data Analysis. Springer-Verlag New York, 2016.,”). No data
were excluded as outliers. Vancomycin kidney concentrations were
expressed as median (interquartile range) and compared by the
non-parametric Kruskal Wallis test. Multiple Hill-type models were
explored to describe the relationship between vancomycin kidney
concentrations and urinary KIM-1. All tests were two-tailed, a
p<0.05 was required for statistical significance. Graphs were
generated using GraphPad version 9.3.1 unless specified otherwise.