Sixty studies were included (15,323 patients): 41 were the original Cochrane review and 19 from the updated searches. Nine studies (1,901 patients) treated anaemia of cancer. Four studies (1,314 patients) were of radiotherapy. Forty-seven studies (12,108 patients) were of chemotherapy, of which 20 (8,145 patients) had long-term follow-up.
Mortality: There was no evidence of a difference between ESA treatment and controls for mortality (OR 1.06, 95% CI 0.97 to 1.15; 60 studies) and no heterogeneity (I2=0%). A similar lack of difference was seen for all chemotherapy studies combined and when the chemotherapy studies were split into subgroups based on mean entry haemoglobin levels, and for radiotherapy and anaemia of cancer studies. There was no evidence of a difference between ESA treatment and controls in the 20 chemotherapy studies with long-term follow-up data and in the 16 chemotherapy studies with individual patient data available. For those studies where both could be calculated, hazard ratios and odds ratios were very close.
Disease progression: There was no evidence of a difference between ESA treatment and controls for disease progression for all 26 studies that reported this outcome and the 21 studies of chemotherapy alone.
Venous thromboembolic events: ESA treatment increased the risk of a thromboembolic event (OR 1.48, 95% CI 1.28 to 1.72; 44 studies). There was no heterogeneity (I2=0%). Similar results were seen for the 35 chemotherapy studies alone and the 18 chemotherapy studies with long-term follow-up.
Further results that repeated previously reported meta-analyses were presented in the paper.