In this trial, there are various statistical techniques used to analysedata in order to determine if result of these test were statisticallysignificant. Consequently, this allows us to identify if we can be confidentenough to apply the results to patient groups suffering from CLL and SLLoutside of this study. Groupsequential testing provides us with a stopping criterion to evaluate whether weaccept or reject the null hypothesis at each interim stage with conclusiveevidence of the efficacy of the treatment. In this test, the initial samplesize is not fixed, which is useful as there are new cohorts of patientsentering the trial at different points. As there are multiple analysis to bemade, the p value needs to be altered for each analysis to avoid inflating Type1 error. This allows trials to either be stopped early or allows patients transferinto the control group.
These decisions are made during interim analysis (Chen et al., 2017). The purpose of interimanalysis is to analyse data before the completion of the data collection occursin the study. This is useful to researchers as it is often the case that clinical trialsenroll patients in a staggered manner as it is continuous. As a result, theyare able to use the data that has been collected so far to assess thedifference between the treatment group and the placebo group.
This in turn,allows an early termination or changing of the study based on whether there arebeneficial or harmful factors identified post analysis (Chakraborty and Kumar, 2016). The hazard ratio measures theeffect of an intervention on an outcome over a period of time. In this case thehazard ratio was used to measure the effect of Ibrutinib group compared to thetreatment group and the outcome was progression free survival. The ratio can bedefined as Hazard in the Ibrutinib group/hazard in the placebo group. In the interimanalysis, the hazard ratio was 0.203 and this means that 20% of patients weremore likely to experience PFS compared to the placebo group (Sedgwick and Joekes, 2015).
Kaplan Meir method is usedwhen there are incomplete observations in cases where a patient discontinues orif they joined the study later. Therefore, having shorter observation time andso may or may not experience the event within the follow-up time. We cannotexclude these subjects otherwise our sample size will be too small. (Kishore et al., 2010). Therefore,Kaplan Meir method is used to estimate survival over time in spite of theseissues.
At each interval, a survival probability was calculated. After thefirst interval, 91% of the treatment group did not experience PFS. The twocurves can be compared to see if the results are statistically significant andthis is done by long Rank test. In the log rank test, we generate a chi-squaredvalue. The p-value was 0.0001 > 0.05, therefore there is a significantdifference between the survival times of both groups (Rich et al.
, 2010). The exposureadjusted incidence rate (EAIR) can be calculated by dividing the total numberof subjects experiencing a specific event by the time under which the patientwas exposed to risk (Liu et al., 2006) (Siddiqui, 2009).
In relation to theHelios trial, the EAIR can be applied to adverse events for example, majorhaemorrhage can be used to see whether the treatment was effective or not overdifferent durations. The researchers found that 11 patients in the ibrutinibgroup suffered from a haemorrhage following a median duration of 4.21 months(EAIR= 2.61). In the placebo group, there were 5 patients who experienced thesame event in a median duration of 2.3 months (EAIR=2.
17); thereby,highlighting the higher incidence rate of a haemorrhage following ibrutinibtreatment. In efficacyanalysis when there are issues with the compliance of treatment, the resultsobtained from those who have followed protocols are analysed in order todetermine whether the experimental therapy is better suited than current/alternativetherapies. The trial also utilised intention-to-treat analysis as a basis forthe efficacy analysis. This is based upon the theory that all patients shouldbe put into randomly allocated groups regardless of which experimental groupthey were initially assigned to or whether they the complete the treatment(Sphweb.bumc.bu.edu, 2016).
Due to compliance issues to the protocols theefficacy analysis for ibrutinib was required. The sample for the analysisexcludes those who did not follow the regime, patients that were lost viasample attrition and patients from the crossover group. However, an issuearises due to the smaller sample size attained for the efficacy analysis. As aresult, the comparison drawn between the ibrutinib treatment and current therapyis less representative when generalised to the population.
The Inverseprobability of censoring weighting (IPCW) is used to assess the clinicalbenefit of the experimental and control groups. It attempts to reduce biascaused by treatment change. It was used to ensure that the outcome of the studywas not affected by crossing over. For example, if a patient in the controlgroup was experiencing symptoms of relapse, then the patient was crossed overto the treatment group (90 patients crossed over).
This allowed for thepatient’s interests to be put above the studies interests. The patients that werecrossed over were weighted zero and those who remained in the placebo groupwere given a higher weighting. It was assumed that there were no measures ofconfounding variables and that randomisation was not preserved (Ascopubs.org,2017) TheFisher Exact Test was used to analyse negative response to Minimal ResidualDisease (MRD). This allowed us to see whether there was an association betweenexperiencing negative MRD and receiving Ibrutinib. It uses contingency table to display the probability of different outcomes.
Therows were the outcome and the column is the exposure. The data in the contingency table was entered into theSAS programmed and a p-value was generated (Biostathandbook.com,2017).