Nutrients 2023 , 15 , 2688
7of 20
2.8. Compliance and Adverse Events Participants were asked to collect and return empty IP and CaD supplement containers at each visit. Compliance was calculated from the number of IP and CaD supplements returned. Compliance for both IP and CaD was calculated, in percentage, as: (100 × total number of capsules administered)/(theoretical number of capsules per day × extent of exposure in days) at each individual visit and for the entire study period. Participants were considered non-compliant for the IP if they had (1) an overall IP compliance <80% or >120%, or (2) an overall compliance within [80%; 120%], but at least one individual visit IP compliance <70% or >130%, or (3) an overall IP compliance missing and less than three available individual visit IP compliances within [80%; 120%]. Adverse events (AEs) were collected on AE forms throughout the trial. AEs were considered as treatment emergent (TEAE) if they began or worsened from the date of the first IP administration. AEs were recorded in the eCRF at each visit; recordings included a description of the AE, the relationship to the intervention (“not related” or “related or suspected”), whether the AE was serious (i.e., resulted in death, life-threatening, required hospitalization, or resulted in persistent disability) or nonserious, and the intensity of the AE (mild, moderate, severe). 2.9. Power Calculation and Statistical Analysis The sample size was calculated to detect a difference in the change from baseline to 48 weeks in BMD measured by DXA on the L2-L4 lumbar spine region between HE and placebo (primary outcome) with consideration of the findings from previous studies [35–37]. The minimum relevant difference expected in the change in BMD (48 weeks versus baseline) between HE and placebo was 1.45% (corresponding to 0.014 g/cm 2 ), and the standard deviation expected in both groups was 2.2% (corresponding to 0.021 g/cm 2 ). A total of 74 evaluable patients (37 per group) were necessary to ensure an 80% power to detect a sig- nificant difference between treatment groups for two-sided test at the 5% level. Assuming 25% of non-evaluable patients, a total of 100 patients were randomized. Data were analyzed using SAS ® Enterprise Guide software version 8.2 (SAS ® for Windows version 9.4M6). Graphs were created by GraphPad Prism version 9 (GraphPad Software, Inc., Boston, MA, USA). Statistical analyses were performed by an independent biostatistician (Atlantstat, France) on the full analysis set (FAS) according to the analysis group (HE or placebo). FAS is defined as all randomized participants with at least one dose of study treatment (HE or placebo) and with at least one not missing post-baseline value for the efficacy criteria. Due to the COVID-19 pandemic, post-baseline efficacy and safety evaluations were assigned to visits based on time windows around the planned visit dates. Only post-baseline evaluations recorded at the planned day ± 28 days were considered for the statistical analysis. Unfortunately, at week 24, due to the low number of observations for both DXA (8% and 6% for HE and placebo group, respectively) and blood collection (10% and 9% for HE and placebo group, respectively), it was not possible to conduct the statistical analysis for these parameters at this visit. No imputation rule was applied for missing values. The primary outcome, and all other DXA parameters, were analyzed using an analysis of covariance (ANCOVA) including analysis group, baseline values, time since menopause at screening (months), and BMI at baseline (kg/m 2 ). These cofactors were selected based on previous work demonstrating their relevance to bone-related outcomes in postmenopausal women [38,39]. Comparisons within and between groups were studied. The same anal- yses were also conducted on the relative changes from baseline (changes from baseline/ baseline × 100) for all DXA parameters. Moreover, the relative change from baseline in total BMD was categorized in different groupings (<1/ ≥ 1%) and analyzed at week 48 using a logistic regression including analysis group, baseline value, time since menopause at screening, and BMI at baseline. Finally, a subgroup analysis was conducted for BMD L2-L4 and total BMD according to serum 25-OH D3 concentration at baseline (<75/ ≥ 75 nmol/L). The relative changes from baseline at week 48 were analyzed using ANCOVA including
Powered by FlippingBook