Table 4 Analysis of secondary outcomes

Recruitment rate per month

We will present recruitment numbers by month and compute the average monthly recruitment figure.

Proportion of patients followed-up at 30 days

We will present the overall proportion of patients followed-up at 30 days and the proportion of patients followed-up at 30 days by treatment arm.

Compliance and adherence with heating protocol

We will present the frequency and proportion of patients complying with the allocated intervention as not adhering, one session, two sessions or fully adhering. We will present reasons for non-adherence where available. For each of the three heating sessions, we will present measures of central tendencies (mean and median) and variability (SD and IQR) for the time of occurrence of the heating application, the temperature of the water and duration of the heat supplication.

Length of stay in hospital following SSM/NSM (days)

Kaplan–Meier curves will be plotted by treatment arm for length of stay in hospital, where length of stay is defined as the difference in days between the date the patient was discharged from hospital, and the date they were admitted. A Cox proportional hazards model will be fitted to estimate the intervention effect on length of stay, adjusting for the minimisation variables. Adjusted time to event curves will be plotted. If the proportional hazard assumption is not deemed a reasonable assumption then an alternative method for adjusting the curve will be sought either through use of a different time-to-event model or stratification of the Kaplan–Meier curves [24, 25].

Proportion of patients with necrosis requiring surgical intervention

The proportion of patients with necrosis requiring surgical intervention will be estimated by treatment arm with 95% CI. The surgical intervention required will be summarised using frequencies and proportions by treatment arm.