Table 6: Risk of bias assessment of Wildier, et al.'s study.

A. Selection bias	Evaluation	Low risk
A. Selection bias	Justification	Stratification has been done in randomization, which prevents imbalance between treatment groups for known factors that influence prognosis or treatment responsiveness (Kernan, et al. [19]). The baseline characteristics of participants such as age and hormone receptor positivity, that may influence outcome are distributed fairly between treatment and comparator. This would ensure any difference in outcome to be attributable to intervention (Roberts & Torgerson, et al. [20]).
B. Performance bias	Evaluation	Unclear risk
B. Performance bias	Justification	The open-label trials could potentially lead to biased results. However, since the patients in the intervention arm were given daily cyclophosphamide in addition to comparator arm, it was impossible to mask the participants. Moreover, it is noteworthy that even if the participants were unblinded, the nature of the outcome (overall response rate) would not allow them to influence the outcome despite being aware of the treatment allocation. Hence unmasked patients would not really increase the bias of the results as such.
C. Attrition bias	Evaluation	Low risk
C. Attrition bias	Justification	In this study all the participants in intervention, as well as comparator group, were followed up after every 9 weeks to evaluate tumour response. Outcome measurement of patients in both arms after certain fixed interval would certainly reduce the bias and attribute any effect to given treatment. Even though participants were lost in between trials, at the time of outcome measurement, both the arms had equal number of patients. Hence there was no important or systematic difference between the groups.
D. Detection bias	Evaluation	Low risk
D. Detection bias	Justification	The participants in the study were followed up for tumour evaluation every 9 weeks. Since patients from both groups are followed up at certain fixed intervals, it will ensure uniformity and reduce risk of bias. The authors also have clearly stated definitions of outcomes supposed to be measured along with methodology. Investigators were not kept blind to participants’ exposure to intervention. But this might not significantly add bias to study due to objective outcome.
E. Internal validity	Rating	++
E. Internal validity	Justification	The findings of this study are trustworthy because of its well-designed research method with no apparent systematic bias.
F. External validity	Rating	++
F. External validity	Justification	With true stratification and different subtypes of breast cancer considered while randomization, and tumour response evaluated at regular intervals the findings of this study can be generalizable.