On 2014 Feb 23, Valeria Fadda commented:

Incremental benefit of warfarin over placebo in patients with atrial fibrillation

Valeria Fadda, Roberta Gatto, Dario Maratea - HTA Unit, Regional Health System, 50100 Firenze (Italy)

One important section of the study by Hart and co-workers (1999) is represented by the meta-analytical comparison between warfarin and placebo in patients with atrial fibrillation (6 trials shown in Table 2 of Hart's article; end point = stroke). These data have in fact been used in recent times to determine the sample size enrolled in the three pivotal trials that have compared dabigatran or rivaroxaban or apixaban with warfarin (trials RE-LY, ROCKET AF, and ARISTOTLE, respectively). While the model of Hart's meta-analysis was based on the Peto method, most meta-analyses published thereafter have employed the random-effect model, which has in fact become the most widely recognised standard. Another point that deserves to be re-examined is that at least two outcome measures can be considered in analyzing these data (relative risk, RR, which is a relative outcome measure; risk difference, RD, which is an absolute outcome measure), but Hart and co-workers considered only RR and not RD. We have therefore re-analysed the above mentioned 6 trials by conducting two separate random-effect meta-analyses which employed, respectively RR and RD as outcome measures. Our re-analysis generated the two Forest plots shown in Figure 1.

Figure 1. Forest plot: repetition of the meta-analysis of Hart and co-workers. The random-effect model has been used according to a relative outcome measure (RR, Panel A) and an absolute outcome measure (RD, Panel B). The two Forest plots show the outcome measures (RR in Panel A, RD in Panel B) calculated for individual trials (squares = outcome measure; horizontal bars = 95%CIs of the outcome measure) and for the pooled analysis (diamond in blue and vertical dotted line in red). I² is a measure of heterogeneity. Statistical calculations were performed by the OMA software (Open Meta-Analyst version 4.16.12, Tufts University, U.S., url http://tuftscaes.org/open_meta/). The complete references for the 6 randomized trials can be found in the article by Hart et al. (1999). Abbreviations: Ev, number of events; Trt, number of patients receiving treatment.

NOTE: Figure 1 can be downloaded from the following link: http://www.osservatorioinnovazione.net/papers/aim1999reanalysis.jpg

References

Hart RG, Benavente O, McBride R, Pearce LA (1999) Antithrombotic therapy to prevent stroke in patients with atrial fibrillation: a meta-analysis. Ann Intern Med. 1999 Oct 5;131(7):492-501.

On 2014 Feb 23, Valeria Fadda commented:

Incremental benefit of warfarin over placebo in patients with atrial fibrillation

Valeria Fadda, Roberta Gatto, Dario Maratea - HTA Unit, Regional Health System, 50100 Firenze (Italy)

One important section of the study by Hart and co-workers (1999) is represented by the meta-analytical comparison between warfarin and placebo in patients with atrial fibrillation (6 trials shown in Table 2 of Hart's article; end point = stroke). These data have in fact been used in recent times to determine the sample size enrolled in the three pivotal trials that have compared dabigatran or rivaroxaban or apixaban with warfarin (trials RE-LY, ROCKET AF, and ARISTOTLE, respectively). While the model of Hart's meta-analysis was based on the Peto method, most meta-analyses published thereafter have employed the random-effect model, which has in fact become the most widely recognised standard. Another point that deserves to be re-examined is that at least two outcome measures can be considered in analyzing these data (relative risk, RR, which is a relative outcome measure; risk difference, RD, which is an absolute outcome measure), but Hart and co-workers considered only RR and not RD. We have therefore re-analysed the above mentioned 6 trials by conducting two separate random-effect meta-analyses which employed, respectively RR and RD as outcome measures. Our re-analysis generated the two Forest plots shown in Figure 1.

Figure 1. Forest plot: repetition of the meta-analysis of Hart and co-workers. The random-effect model has been used according to a relative outcome measure (RR, Panel A) and an absolute outcome measure (RD, Panel B). The two Forest plots show the outcome measures (RR in Panel A, RD in Panel B) calculated for individual trials (squares = outcome measure; horizontal bars = 95%CIs of the outcome measure) and for the pooled analysis (diamond in blue and vertical dotted line in red). I² is a measure of heterogeneity. Statistical calculations were performed by the OMA software (Open Meta-Analyst version 4.16.12, Tufts University, U.S., url http://tuftscaes.org/open_meta/). The complete references for the 6 randomized trials can be found in the article by Hart et al. (1999). Abbreviations: Ev, number of events; Trt, number of patients receiving treatment.

NOTE: Figure 1 can be downloaded from the following link: http://www.osservatorioinnovazione.net/papers/aim1999reanalysis.jpg

References

Hart RG, Benavente O, McBride R, Pearce LA (1999) Antithrombotic therapy to prevent stroke in patients with atrial fibrillation: a meta-analysis. Ann Intern Med. 1999 Oct 5;131(7):492-501.