Jump to comment:

• Changes to Statistical Plan
  Iain Bressendorff, Ditte Hansen, Morten Schou, Charlotte Kragelund and Lisbet Brandi
  Published on: 13 December 2019

• Published on: 13 December 2019

  Changes to Statistical Plan

  • Iain Bressendorff, Nephrology Fellow Nordsjællands Hospital, Hillerød, Denmark
  • Other Contributors:
    • Ditte Hansen, Consultant Nephrologist
    • Morten Schou, Consultant Cardiologist
    • Charlotte Kragelund, Consultant Cardiologist
    • Lisbet Brandi, Consultant Nephrologist

  We hereby write to inform the readers of changes to the statistical plan for our clinical trial MAGiCAL-CKD, which we have previously published in BMJ Open. At the time of writing, the trial is still on-going and the data set has not been unblinded. Thus, any changes to the statistical plan at this time will not compromise the integrity of the trial design.
  The purpose of the MAGiCAL-CKD trial is to examine the effect of magnesium (Mg) supplementation on coronary artery calcification (CAC) score in patients with chronic kidney disease. The original statistical plan was to analyse the change in CAC score from week 0 to week 52 (delta CAC score) and compare the delta CAC score between the two treatment groups. The choice of delta CAC as the primary endpoint (and not the between-group difference in CAC score at week 52) was made due to the potential for an imbalance in CAC score between the two treatment groups at week 0. The delta CAC was thought to better account for any baseline imbalance. However, since the publication of the trial protocol we have become aware that this methodology is flawed and that the correct analysis is to perform an analysis of covariance (ANCOVA) of CAC score between the two treatment groups at week 52 adjusted for CAC score, age and prevalent diabetes mellitus (yes/no) at week 0. Therefore, the ANCOVA test will be applied to analyse the primary endpoint. Essentially, we are examining the same scientific question, but using better methodology...

  Show More

  We hereby write to inform the readers of changes to the statistical plan for our clinical trial MAGiCAL-CKD, which we have previously published in BMJ Open. At the time of writing, the trial is still on-going and the data set has not been unblinded. Thus, any changes to the statistical plan at this time will not compromise the integrity of the trial design.
  The purpose of the MAGiCAL-CKD trial is to examine the effect of magnesium (Mg) supplementation on coronary artery calcification (CAC) score in patients with chronic kidney disease. The original statistical plan was to analyse the change in CAC score from week 0 to week 52 (delta CAC score) and compare the delta CAC score between the two treatment groups. The choice of delta CAC as the primary endpoint (and not the between-group difference in CAC score at week 52) was made due to the potential for an imbalance in CAC score between the two treatment groups at week 0. The delta CAC was thought to better account for any baseline imbalance. However, since the publication of the trial protocol we have become aware that this methodology is flawed and that the correct analysis is to perform an analysis of covariance (ANCOVA) of CAC score between the two treatment groups at week 52 adjusted for CAC score, age and prevalent diabetes mellitus (yes/no) at week 0. Therefore, the ANCOVA test will be applied to analyse the primary endpoint. Essentially, we are examining the same scientific question, but using better methodology. The ANCOVA test will likewise be used to examine the between-group differences at week 52 in blood levels of Mg, phosphate, ionised calcium, parathyroid hormone, fibroblast growth factor 23, estimated glomerular filtration rate (eGFR), potassium, and calcification propensity; urine levels of Mg and phosphate; carotid-femoral pulse wave velocity and pulse wave analysis; and bone mineral density. Linear mixed models with repeated measures will be used to analyse changes in blood Mg, phosphate, ionised calcium, parathyroid hormone, eGFR, and potassium; as well as urine Mg and phosphate.

  In addition, we wish to investigate whether certain subgroups affect the treatment effect of Mg supplementation on CAC score. The prespecified subgroups are: 1) prevalent diabetes mellitus at week 0; 2) CAC score = 0 versus CAC score > 0 at week 0; and 3) tertiles of serum calcification propensity at week 0. These subgroups will be analysed for heterogeneity of treatment effect and will be presented as Forrest plots with p-values for treatment interaction.

  Lastly, in order to examine whether Mg supplementation affects various clinical outcomes stratified Cox proportional hazards models and Kaplan-Meier time-to-event analyses will be used to examine the incidence of major adverse cardiovascular events (cardiovascular death, non-fatal myocardial infarction, non-fatal stroke, new-onset heart failure, hospitalisation for heart failure, limb amputation or revascularisation due to peripheral arterial occlusion) after 1 and 5 years follow-up with treatment group as a fixed factor and additional covariates at week 0 (age, gender, CAC score, prevalent diabetes mellitus, and eGFR). Additional models will be used to examine the incidence of all-cause mortality with treatment group as a fixed factor and additional covariates at week 0 (age, gender, CAC score, prevalent diabetes mellitus, prevalent cardiovascular disease, and eGFR) as well as the incidence of end-stage kidney disease (dialysis treatment for more than 3 months, kidney transplantation or death from renal disease) with treatment group as a fixed factor and additional covariates at week 0 (age, gender, 24-hour urine protein, prevalent diabetes mellitus, and eGFR).

  Show Less

  Conflict of Interest:
  None declared.

  • Back to top