Swain 1997a(088001).

Study characteristics
Methods	Block randomisation according to a prospectively prepared randomisation list (a separate list was prepared for each investigational site and within each site, the assignments were stratified relative to the presence or absence of cardiac risk factors and on the basis of measurable versus nonmeasurable disease).
Participants	349 participants (aged 25 to 84 years; all females) with breast cancer (stage III or IV) treated with doxorubicin (cumulative dose < 100 to 2700 mg/m2; peak dose (i.e. maximal dose received in 1 week) 50 mg/m2; infusion duration nm), fluorouracil and cyclophosphamide. No prior anthracycline therapy. Prior cardiac radiotherapy in 20 participants in the dexrazoxane group and 14 participants in the control group (dose nm). No prior cardiac dysfunction (defined as a LVEF < lower limit of normal percentage for the participating institution, obtained within 4 weeks before study entry; documented history of congestive heart failure or cardiomyopathy within 6 months before entry).
Interventions	Dexrazoxane (10:1 ratio of dexrazoxane to doxorubicin; slow IV push or rapid‐drip IV infusion between 15 and 30 minutes before doxorubicin) (N = 168) versus placebo (N = 181).
Outcomes	Heart failure (i.e. clinical heart failure defined as 2 or more of the following: cardiomegaly established by radiography, basilar rales, S3 gallop or paroxysmal nocturnal dyspnoea, orthopnoea, or significant dyspnoea on exertion; subclinical myocardial dysfunction defined as 1) decline in LVEF as measured by MUGA from baseline of ≥ 10% below the institution's LLN, 2) a decline in LVEF as measured by MUGA of at least 20% from baseline or 3) decline in LVEF as measured by MUGA to at least 5% below the institution's LLN). Tumour response rate (according to ECOG criteria). Overall survival (defined as time from randomisation to death). Progression‐free survival (defined as time from randomisation to progression either on or off treatment). Adverse effects (according to ECOG criteria).
Notes	Length of follow‐up: in the intervention group median 532 days (range 1 to 1863); in the control group median 511 days (range 1 to 1652). Median age in intervention group: 58 years; median age in control group: 56 years. Cumulative anthracycline dose in intervention and control group nm. Gender: all females in both treatment groups. Stage of disease per treatment group nm. Funding sources: Pharmacia & Upjohn Inc (Pharmacia Inc; Adria laboratories), Kalamazoo MI. Declaration of interests: nm, but some authors affiliated with funding source. The Weisberg 1992, Tonkin 1998, Rosenfeld 1992 and Bates 1997 publications are duplicate publications of this study; we did not use information from these publications.  The same was true for the Swain 1997 publication which was not the primary publication of this study; in addition this publication also reported on a non‐randomised controlled trial which was not eligible for our review.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	A prospectively prepared randomisation list was used.
Allocation concealment (selection bias)	Low risk	All assignments to treatment were made under double‐blind conditions.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	All assignments to treatment were made under double‐blind conditions.
Blinding of outcome assessors (detection bias) ‐ clinical heart failure	Low risk	All subsequent clinical assessments were made under double‐blind conditions.
Blinding of outcome assessors (detection bias) ‐ clinical heart failure and subclinical myocardial dysfunction combined	Low risk	All subsequent clinical assessments were made under double‐blind conditions.
Blinding of outcome assessors (detection bias) ‐ overall survival/overall mortality	Low risk	All subsequent clinical assessments were made under double‐blind conditions.
Blinding of outcome assessors (detection bias) ‐ tumour response rate	Low risk	All subsequent clinical assessments were made under double‐blind conditions.
Blinding of outcome assessors (detection bias) ‐ progression‐free survival	Low risk	All subsequent clinical assessments were made under double‐blind conditions.
Blinding of outcome assessors (detection bias) ‐ toxicities other than cardiac damage (diagnosed by laboratory tests)	Low risk	All subsequent clinical assessments were made under double‐blind conditions.
Blinding of outcome assessors (detection bias) ‐ toxicities other than cardiac damage (not diagnosed by laboratory tests)	Low risk	All subsequent clinical assessments were made under double‐blind conditions.
Incomplete outcome data (attrition bias) ‐ clinical heart failure	Low risk	Clinical heart failure evaluated in 100% of all participants in both treatment groups.
Incomplete outcome data (attrition bias) ‐ clinical heart failure and subclinical myocardial dysfunction combined	Low risk	Clinical heart failure and subclinical cardiac damage both evaluated in 100% of all participants in both treatment groups.
Incomplete outcome data (attrition bias) ‐ overall survival/overall mortality	Unclear risk	Unclear in how many participants overall survival was evaluated
Incomplete outcome data (attrition bias) ‐ tumour response rate	High risk	Tumour response evaluated in 90% of participants with measurable disease in the dexrazoxane group and 89% of the control group (127/141 in the dexrazoxane group and 136/152 in the control group).
Incomplete outcome data (attrition bias) ‐ progression‐free survival	Unclear risk	Unclear in how many participants PFS was evaluated
Incomplete outcome data (attrition bias) ‐ toxicities other than cardiac damage with the exception of SMN	Unclear risk	Unclear in how many participants toxicities other than cardiac damage were evaluated
Selective reporting (reporting bias)	Low risk	All expected outcomes were reported.
Other bias	Unclear risk	Block randomisation in unblinded trials: not applicable Baseline imbalance between treatment groups related to outcome (prior cardiotoxic treatment (anthracyclines and cardiac irradiation), age, gender, stage of disease and prior cardiac dysfunction): unclear risk (no prior cardiac dysfunction and no prior anthracyclines, stage of disease unclear; all other factors comparable between treatment groups) Difference in length of follow‐up between treatment groups: low risk (comparable follow‐up in both treatment groups)