Comparison between placebo and each SZC treatment group (high to low) with regard to the mean S-K level during the randomized treatment phase days 8-29. Mixed-effects models were used to estimate least-squares means.

A subject was considered to be a responder if, during the evaluation period, they maintained a pre-dialysis serum potassium (S-K) between 4.0 and 5.0 mmol/L on at least 3 out of 4 dialysis treatments following the long inter-dialytic interval (LIDI) and did not receive rescue therapy. Subjects with no data during the evaluation period were classified as non-responders. The S-K levels used for this analysis were based on the measurements obtained by the central laboratory.

The sensitivity analysis assessed the impact of subjects classified as non-responders due to missing serum potassium (S-K) data. Missing central lab (c-lab) pre-dialysis values were imputed using corresponding pre-dialysis i-STAT (a portable blood analyser) measurements. In addition, a "last observation carried forward" (LOCF) approach was utilized to further impute missing values of pre-dialysis S-K during the evaluation period. This technique will replace missing c-lab S-K values with the last available non-missing pre-dialysis LIDI observation recorded for that patient (and this could be a c-lab value or an imputed c-lab value). The Primary endpoint analysis was repeated on the imputed data.

The number of patients who experienced any AE, including those which were serious (SAE), had an outcome of death, were severe, led to discontinuation of ZS or were causally related to ZS are presented for the MP. AEs of special interest are also presented, including oedema-related AEs, cardiac failure and hypertension.

The least squares means (LS-means) are derived from a linear regression model of absolute change in S-K at 4h with the following covariates: treatment group; baseline S-K; time from the start of dosing insulin to the start of dosing SZC/placebo and the dose (units/kg) of the first course of insulin. The 95% CI is associated with LS-Means.

Blood samples for determination of potassium were collected pre-dose, and at 1, 2, and 4 hours post Dose 1 on Day 1. An additional sample was collected at 90 minutes post Dose 2 on Day 1 if i-STAT potassium values at the 4-hour post Dose 1 time point was ≥ 6.1 or \<4.0 mmol/L. On Day 2 samples were analysed pre-dose, and 1 and 4 hours post Dose 1. S-K levels were analysed at the Central Laboratory. Natural logarithm of S-K from 0 to 48 hours post dose are modelled by the random coefficients model including fixed effects of intercept, time, time x treatment and patient-level random effects for time and intercept. Exponential rate of change refers to the slope estimate from the random coefficients model.

The 48- hour urine potassium excretion on Study Days 3 and 4 (baseline) will be compared with 48-hour urine potassium excretion on Study Days 7 and 8 (on study drug).