Table 4.

Study objectives and statistical analysis in the ImpresU clinical trial

Objectives		Statistical analysis	Time point(s) of evaluation of this outcome measure (if applicable)
Primary objective
1	The primary objective of this study is to investigate if methenamine hippurate reduces the need for antibiotic use due to recurrent UTI (measured as number of antibiotic courses).	Standard linear regression will be used where the number of UTI antibiotic treatments will be the dependent variable. Group allocation together with the confounding variables above will be independent variables. The dependent variable will be transformed using a rank transformation in case it is not normally distributed. A p value will be delivered, but no useful effect size if a rank transformation is used.	After 6 months of treatment
Secondary objectives
2a	To investigate if methenamine hippurate will have a prolonged effect on antibiotic usage even after discontinuation.	Will be analysed using the same statistical approach as objective 1.	Six months after completing (12 months after commencing) treatment
2b	To investigate if methenamine hippurate reduces the incidence of UTI.	Will be analysed using the same statistical approach as objective 1. UTIs will be the dependent variable.	After 6 months of treatment
2c	To investigate if methenamine hippurate can reduce severity of UTI symptoms.	The outcome variable is measured at first day of UTI using a 6-grade ordinal scale. The median value of all UTIs will be used if the patient has more than one episode of UTI. The dependent variable will be transformed using a rank transformation and analysed with linear regression using the same covariates as when analysing objective 1.	After 6 months of treatment
2d	To investigate if methenamine hippurate can reduce duration of UTI episodes.	Will be analysed using the same statistical approach as objective 1. Number of days with symptoms will be the dependent variable. The dependent variable will be transformed using a rank transformation in case it is not normally distributed. A p value will be delivered, but no useful effect size if a rank transformation is used.	After 6 months of treatment
2e	To investigate if the number of complications such as pyelonephritis and hospital admission for UTI differ between methenamine hippurate and placebo.	Will be analysed using the same statistical approach as objective 1. Number of severe events such as pyelonephritis or hospital admission will be the dependent variable. The dependent variable will be transformed using a rank transformation in case it is not normally distributed. A p value will be delivered, but no useful effect size if a rank transformation is used.	Six and 12 months after commencing treatment
2f	To investigate if strain characteristics/phylogenetic subgroups of Escherichia coli found at inclusion are an effect modifier in all the above outcomes.	Phylogenetic subtype of pure cultures of E. coli in the inclusion urine culture will be analysed in a separate statistical analysis if the number of cultures available for typing is at least 60. All statistics above will be repeated adding phylotype as an extra independent variable. Purely descriptive statistics will be presented if the number is less than 60.	See above

UTI, urinary tract infection.