Tests for qualitative treatment‐by‐centre interaction using a ‘pushback’ procedure

Abstract

Statistical analysis of difference between groups in change for some variable, adjusting for initial value, is complicated by the presence of intra-individual variation in that variable. We estimate here the asymptotic bias that results from calculating the adjusted between-group difference by ordinary least squares (OLS) from observed data. We also present explicit formulae that use the OLS estimates, the difference between treatment groups in mean initial values, and a measure of the intra-individual variation to compute a corrected estimator and its variance. Alternatively, we can use OLS on transformed data to obtain unbiased estimates, in which we replace initial observed values by conditional Stein estimates of true values. We illustrate the results with data from an observational study and a clinical trial.

Problems in using incidence to analyze risk factors in follow-up studies

“…Excluding carefully all subjects with early signs of the disease from follow-up can minimize the bias. Alternatively, one can try to adjust the analysis for markers of undetected disease [16,17] or use other models, partly under development [18][19][20] or, e.g. for asthma, analyze change in prevalence [12].…”

“…However, it has been shown for continuous outcome variables that when adjusting the analyses for the baseline level of the outcome variable, measurement error can lead to bias either away or towards null, if there is an association between exposure and disease already at baseline [16,17]. This may happen also with dichotomous outcome variables using transition models [18].…”

“…This would allow better control for baseline health status and better power to analyze change in disease status than analyses of incidence [16,17]. In epidemiology of coronary heart disease this has earlier been attempted by categorizing subjects on disease severity using combinations of ECG changes and symptoms, and this has later been replaced by non-invasive measures of atherosclerosis mainly with ultrasound.…”

Detecting Qualitative Interactions in Clinical Trials: An Extension of Range Test

“…On balance, we would recommend the ERT for general uses unless it is evident that the number of subsets in which one treatment is superior to the other is nearly half. Ciminera et al (1993) considered a multi-center clinical trials to compare a drug for reducing blood pressure with a placebo. The summary data of the trial are given in Table 6, ordered by the standardized treatment difference (r).…”

“…Piantodosi and Gail (1993) proposed the standardized range test as an alternative procedure. Other important work in this field includes, for example, the "pushback" procedure proposed by Ciminera et al (1993), the test for qualitative interaction of clinical significance by Pan and Wolfe (1997), and the exact testing approach by Silvapulle (2001). Testing procedures were also developed under a non-inferiority trial setting by Wiens and Heyse (2003) and Xin (2004).…”