128 Chapter 7 documented in patient files as well. Bleeding events were defined according to the definitions of the ISTH as pain and/or swelling, only bleeds treated with additional FVIII/ BPA were considered in this study . Bleeds were classified as all, joint and/or muscle bleeds. The mean annualized bleeding rate (ABR) and the mean annualized joint bleeding rate (AJBR) with 95% CI were modelled by using a negative binomial regression with a log link to account for variations in follow-up times and the skewness of bleeding data. The ABRs before and during emicizumab therapy were compared using this regression model. The proportion of PwHA with 0 bleeds, 1−3 bleeds and >3 bleeds during the 24-week intervals of 0−24, 25−48, 49−72 and 73−96 weeks after starting emicizumab were estimated to enable a comparison with the data from the HAVEN studies . The individuals were only included in a 24-week interval if they completed the entire interval. Drug waste The drug waste (i.e., the difference between the prescribed dose and the dispensed drug in the vial(s)) was calculated by subtracting the label-based dose from the vial size suggestion, which were both obtained by entering every individual’s body weight in the HEMLIBRA® Dosing Calculator (http://www.hemlibra-hcp.com/dosing-administration/ dosing-calculator). This drug waste was expressed as a proportion, or was calculated as waste during follow-up or as annualized waste. For individuals with a maintenance-dosing interval of 21 days, the mean of the label-based doses at 14- and 28-day intervals was used. The closest label-based interval was taken for other alternative intervals. Statistical methods The continuous variables were presented as means with standard deviations (SD) if they were normally distributed or as medians with interquartile ranges (IQR) and categorical variables were presented as individual counts with percentages. The last emicizumab concentration that was observed for each individual during maintenance therapy (steady state condition) and the ABRs were compared across subgroups using respectively nonparametric tests and multivariable regressions. Subgroups included age categories (adults/adolescents [≥12 years] versus children [<12 years]), dosing interval categories (label-based [7, 14, 28 days] versus alternative dosing intervals), BMI categories (BMI <18 versus 18–25 versus >25 kg/m2), FVIII inhibitor status at baseline (present versus absent) and adherence categories (adherent versus non-adherent). For this last category, the individuals who self-reported skipping injections of emicizumab repeatedly were defined as non-adherent. When self-reported, non-adherence was quantitated by comparing the pharmacy expenditure records with prescribed dose . Additionally, the ABRs and AJBRs were compared across the subgroups of concentrations (<40 versus 40–80 versus >80 µg/mL). These concentration subgroups were based on the mean ±standard deviation of 6mg/kg every 4 weeks (highest peak-trough fluctuation), as stated in the drug label .