Minimization in interventional trials: great value but residual vulnerability

When randomisation is not good enough: Matching groups in intervention studies

Randomised assignment of individuals to treatment and controls groups is often considered the gold standard to draw valid conclusions about the efficacy of an intervention. In practice, randomisation can lead to accidental differences due to chance. Researchers have offered alternatives to reduce such differences, but these methods are not used frequently due to the requirement of advanced statistical methods. Here, we recommend a simple assignment procedure based on variance minimisation (VM), which assigns incoming participants automatically to the condition that minimises differences between groups in relevant measures. As an example of its application in the research context, we simulated an intervention study whereby a researcher used the VM procedure on a covariate to assign participants to a control and intervention group rather than controlling for the covariate at the analysis stage. Among other features of the simulated study, such as effect size and sample size, we manipulated the correlation between the matching covariate and the outcome variable and the presence of imbalance between groups in the covariate. Our results highlighted the advantages of VM over prevalent random assignment procedure in terms of reducing the Type I error rate and providing accurate estimates of the effect of the group on the outcome variable. The VM procedure is valuable in situations whereby the intervention to an individual begins before the recruitment of the entire sample size is completed. We provide an Excel spreadsheet, as well as scripts in R, MATLAB, and Python to ease and foster the implementation of the VM procedure.

Pulmonary Metastasectomy in Colorectal Cancer: updated analysis of 93 randomized patients - control survival is much better than previously assumed

AIM

Lung metastases from colorectal cancer are resected in selected patients in the belief that this confers a significant survival advantage. It is generally assumed that the 5-year survival of these patients would be near zero without metastasectomy. We tested the clinical effectiveness of this practice in Pulmonary Metastasectomy in Colorectal Cancer (PulMiCC), a randomized, controlled noninferiority trial.

METHOD

Multidisciplinary teams in 14 hospitals recruited patients with resectable lung metastases into a two-arm trial. Randomization was remote and stratified according to site, with minimization for age, sex, primary cancer stage, interval since primary resection, prior liver involvement, number of metastases and carcinoembryonic antigen level. The trial management group was blind to patient allocation until after intention-to-treat analysis.

RESULTS

From 2010 to 2016, 93 participants were randomized. These patients were 35-86 years of age and had between one and six lung metastases at a median of 2.7 years after colorectal cancer resection; 29% had prior liver metastasectomy. The patient groups were well matched and the characteristics of these groups were similar to those of observational studies. The median survival after metastasectomy was 3.5 (95% CI: 3.1-6.6) years compared with 3.8 (95% CI: 3.1-4.6) years for controls. The estimated unadjusted hazard ratio for death within 5 years, comparing the metastasectomy group with the control group, was 0.93 (95% CI: 0.56-1.56). Use of chemotherapy or local ablation was infrequent and similar in each group.

CONCLUSION

Patients in the control group (who did not undergo lung metastasectomy) have better survival than is assumed. Survival in the metastasectomy group is comparable with the many single-arm follow-up studies. The groups were well matched with features similar to those reported in case series.

A method for early evaluation of a recently introduced technology by deriving a comparative group from existing clinical data: a case study in external support of the Marfan aortic root

OBJECTIVE

During the early phase of evaluation of a new intervention, data exist for present practice. The authors propose a method of constructing a fair comparator group using these data. In this case study, the authors use the example of external aortic root support, a novel alternative to aortic root replacement.

DESIGN

A matched comparison group, of similar age, aortic size and aortic valve function to those having the novel intervention, was constructed, by minimization, from among patients having conventional aortic root replacement in other hospitals during the same time frame.

SETTING

Three cardiac surgical units in England.

PATIENTS

The first 20 patients, aged 16-58 years with aortic root diameters of 40-54 mm, having external support surgery were compared with 20 patients, aged 18-63 years and aortic root diameters of 38-58 mm, who had conventional aortic root replacement, between May 2004 and December 2009.

INTERVENTIONS

A pliant external mesh sleeve, customised by computer-aided design, encloses the whole of the ascending aorta. The comparator group had conventional aortic root replacement, 16 valve-sparing and four with composite valved grafts.

MAIN OUTCOME MEASURES

Duration of cardiopulmonary bypass (CPB), myocardial ischaemic time, blood loss and transfusion of blood, platelets and clotting factors.

RESULTS

Comparing total root replacement and customised aortic root support surgery: CPB (median (range)) was 134 (52-316) versus 0 (0-20) min; myocardial ischaemia 114 (41-250) versus 0 (0-0) min; 4 h blood loss was 218 (85-735) versus 50 (25-400) ml; and 9/18 had blood transfusion, 9/18 platelets and 12/18 fresh frozen plasma after root replacement versus 1/20, 0/20 and 0/20, respectively, for the novel surgery.

CONCLUSIONS

Avoidance or large reductions in CPB, myocardial ischaemia and blood product usage were achieved with the novel surgery. These data are of use in decision analysis and health economic evaluation and are available early in evaluation before randomised trial data are available.