********************************************************************************************** *Stata code for replication results reported in Cook JA. An introduction to clinical trials. OUP 1st Edition, 2023. *21Aug2023 *LEAP Trial analysis code used to generated results *See key publication Du Toit et al, LEAP Trial, 2015 (doi: 10.1056/NEJMoa1414850) *Dataset available from https://www.itntrialshare.org/tsstatic/Terms%20Of%20Use.html *Chapter 9 Table 9.2 p214 analysis of allergy (positive skin prick test) result ********************************************************************************************** *Set-up log file - update with address for storing log file *log using "XXXX.smcl", replace ********************************************************************************************** *Import dataset - update address to where dataset is stored (available from website listed above) *import excel "LEAP trial datatset PO_ADPOUT1_nejm_2021-10-16_03-11-42.xlsx", sheet("data") firstrow ********************************************************************************************** *Check participant number overall, by stratum, and by randomised group, and also number included in ITT analysis of the primary outcome tab ParticipantID *Fig 1 - 640 were randomised matches number of rows *Included in ITT anlaysis *Note matches Fig 1 of Du Toit et al (2015) has 263+1+266+51+6+41=628 in ITT tab IntenttotreatITTSample *Note matches Fig 1 of Du Toit et al (2015) fig 1 for randomised numbers overall tab TreatmentGroupChar tab TreatmentGroupChar, nolabel *Matches stratum Fig 1 numbers - 542 and 98 tab Stratumchar ********************************************************************************************** *Generate indicator variables to use in analysis *Generate indicator variable for treatment group gen TreatGroup_indicator=1 if TreatmentGroupChar=="Peanut Avoidance" replace TreatGroup_indicator=0 if TreatmentGroupChar=="Peanut Consumption" tab TreatGroup_indicator TreatmentGroupChar *Generate indicator variable for baseline stratum group (SPT-negative/positive) gen Stratumchar_indicator=1 if Stratumchar=="SPT-positive Stratum" replace Stratumchar_indicator=0 if Stratumchar=="SPT-negative Stratum" tab Stratumchar_indicator Stratumchar *************** *Check number randomised by baseline stratum *Randomised by baseline stratum group (SPT-negative/positive) - matches Figure 1 tab Stratumchar_indicator TreatGroup_indicator *************** *Primary outcome data check *Note - PO is OverallV60OutcomeOFCIndet not OutcomeofPeanutOFCatV60ch. tab OutcomeofPeanutOFCatV60ch OverallV60OutcomeOFCIndet tab OutcomeofPeanutOFCatV60ch TreatmentGroupChar, col tab TreatmentGroupChar tab OverallV60OutcomeOFCIndet, nolabel gen TOverallV60OutcomeOFCIndet_ind=1 if OverallV60OutcomeOFCIndet=="FAIL OFC" replace TOverallV60OutcomeOFCIndet_ind=0 if OverallV60OutcomeOFCIndet=="PASS OFC" *Check indicator tab TOverallV60OutcomeOFCIndet_ind OverallV60OutcomeOFCIndet *Matches values (17.2% and 3.2% for avoidance and consumption groups) for overall combined cohort in Fig 2. tab TOverallV60OutcomeOFCIndet_ind TreatGroup_indicator, col ********************************************************************************************** *Analysis of primary outcome overall (both baseline SPT strata together) *Risk difference with 95% confidence interval (CI) [Newcombe's method 10] rdci TOverallV60OutcomeOFCIndet_ind TreatGroup_indicator *Risk ratio with 95% CI cs TOverallV60OutcomeOFCIndet_ind TreatGroup_indicator *Odds ratio with 95% CI using logistic regressions logistic TOverallV60OutcomeOFCIndet_ind TreatGroup_indicator *Invert direction so that OR/RR<1 and RD <0 indicates less allergy events recode TreatGroup_indicator (1 = 0) (0 = 1), gen(Invert_TreatGroup_indicator) tab TOverallV60OutcomeOFCIndet_ind Invert_TreatGroup_indicator, col *Risk difference with 95% confidence interval (CI) [Newcombe's method 10] as per Table 9.2 (Cook, 2023) rdci TOverallV60OutcomeOFCIndet_ind Invert_TreatGroup_indicator *Risk ratio with 95% CI result as per Table 9.2 p214 (Cook, 2023) cs TOverallV60OutcomeOFCIndet_ind Invert_TreatGroup_indicator *Odds ratio with 95% CI result as per Table 9.2 p214, (Cook 2023) logistic TOverallV60OutcomeOFCIndet_ind Invert_TreatGroup_indicator ********************************************************************************************** *Analyses by stratum (not reported in Chapter 9 (Cook, 2023) but corresponding to analyses reported in paper). *Note analyses done separately though a different approach of using the combined dataset but including interaction tests in a regression model could have been used. The analysis of each stratum separately is adopted here which follow the approach in Du Toit et al (2015). *************** preserve keep if Stratumchar=="SPT-negative Stratum" *Risk difference with 95% CI (Newcombe's method 10) rdci TOverallV60OutcomeOFCIndet_ind Invert_TreatGroup_indicator *Risk ratio with 95% CI cs TOverallV60OutcomeOFCIndet_ind Invert_TreatGroup_indicator *Odds ratio with 95% CI using logistic regression logistic TOverallV60OutcomeOFCIndet_ind Invert_TreatGroup_indicator restore *************** preserve keep if Stratumchar=="SPT-positive Stratum" *risk difference with 95% CI (Newcombe's method 10) rdci TOverallV60OutcomeOFCIndet_ind Invert_TreatGroup_indicator *Risk ratio with 95% CI cs TOverallV60OutcomeOFCIndet_ind Invert_TreatGroup_indicator *Odds ratio with 95% CI using logistic regression logistic TOverallV60OutcomeOFCIndet_ind Invert_TreatGroup_indicator restore ********************************************************************************************** *Alternative analysis approaches to overall analysis (not reported in Chapter 9 but some commonly used alternative approaches to handling stratum structure) *************** *Mantzel-Haenzel (stratified) comparison cc TOverallV60OutcomeOFCIndet_ind TreatGroup_indicator, by(Stratumchar) *************** *Logistic regression with interaction effect *Generate treatment interaction effect gen Stratumtreat_interaction=0 replace Stratumtreat_interaction=1 if ((Stratumchar_indicator==1)&(TreatGroup_indicator==1)) tab Stratumtreat_interaction Stratumchar_indicator *Logistic regression with treatment-by-stratum subgroup interaction effect logistic TOverallV60OutcomeOFCIndet_ind TreatGroup_indicator Stratumchar_indicator Stratumtreat_interaction ********************************************************************************************** *Log close *Update with location used to store log file view "XXXX.smcl"