This thesis presents statistical modelling approaches and analyses for the filtration function of the kidney in patients with cancer, using widely available clinical data.

The filtration function of the kidney is reported as the glomerular filtration rate (GFR). Many clinical decisions, for example dose calculations for chemotherapy drugs, are based on a patient’s GFR. The most accurate way to determine a patient’s GFR involves costly, time-consuming, and consequently frequently unavailable methods. Therefore, GFR is commonly estimated using statistical models based on routine clinical data. Although these models are used in patients with cancer, most of them have been developed using data from non-cancer patients with impaired kidney function.

Here we present a new statistical model for GFR in patients with cancer. We initially collected data from 2,579 patients managed at two cancer centres. Using regression modelling based on patient de- mographics and serum creatinine levels, a new model, CamGFR, was developed that is more accurate than other currently available models for GFR.

Following this, data from a further 7,944 patients were collected across nine cancer centres. These data enabled the assessment of the impact of different creatinine measurement methods, such as standard- ising creatinine to isotope dilution mass spectrometry (IDMS). Given that the method for creatinine measurement affects the reported laboratory value, the CamGFR model was adjusted to allow use for creatinine, which was or was not IDMS standardised.

Further analyses included the search for clinical correlates with renal function and assessment of whether such correlates can improve the estimation of GFR; the effect of treatment on GFR and esti- mated GFR; the examination of the effect of unstable creatinine on GFR estimation using longitudinal data; and a comparison of renal function between patients with and without cancer.

In summary, this thesis presents results that may help improve the estimation of kidney function and thereby care for patients with cancer.