Reply to ‘Comment on ‘Genotype‐guided warfarin dosing <i>vs.</i> conventional dosing strategies: a systematic review and meta‐analysis of randomized clinical trials’ by Tse <i>et al</i>.’

We read with great interest the letter by Sridharan and Sequeira 1 about our recent meta-analysis of randomized controlled trials on genotype-guided warfarin dosing versus conventional dosing strategies 2. We would like to thank both authors for their positive comments on, and for their critical analysis of, our manuscript. Specifically, they suggested that the different variations between the different parameters of the algorithms used in some of the trials included in the meta-analysis have contributed to the heterogeneity observed in some of the endpoints described in our previous article. Whilst we feel that our original aim was to make a general comparison between genotype- and conventionally guided warfarin dosing rather than to examine in detail the different genotype-guided algorithms per se, we have performed the additional analyses suggested by Sridharan and Sequeira. Moreover, we would like to point out that even these four studies contain a degree of heterogeneity, in that, only three of the four studies used fixed-dosing algorithms in the control arm 3-5, whereas the remaining study used a clinically guided dosing regimen 6. Therefore, the analysis by Sridharan and Sequeira and the additional subgroup analyses presented here should be interpreted with these considerations in mind. Our additional analysis by pooling only those trials that used the International Warfarin Pharmacogenetics Consortium (IWPC) algorithm in the genotype-dosing arm did not significantly alter the mean difference (MD) in the time-to-first therapeutic International Normalized Ratio (INR) or its heterogeneity (Table 1). By contrast, it reduced heterogeneity for the time-to-first stable INR but not its MD. For percentage of time in therapeutic range (TTR), the MD was no longer statistically significant whilst heterogeneity remained the same. For INR ≥ 4, both the risk ratio and heterogeneity remained unaltered, as was the case for bleeding risks. Finally, regarding their query whether replacing the Taiwan algorithm cohort with the IWPC algorithm cohort by the Wen ,et al. study altered our results, we would like to clarify the following points. This altered the data in the genotype-guided group only for two endpoints: %TTR and INR ≥ 4. However, it did not significantly alter the pooled %TTR (MD: 4.4, 95% CI: 1.1–7.6; I2 = 80% for IWPC algorithm vs. MD: 3.1, 95% CI: 0.7–5.5; I2 = 80% for Taiwan algorithm) or INR ≥ 4 (risk ratio [RR]: 0.87; 95% CI: 0.77–0.98; I2 = 0% for IWPC algorithm vs. RR: 0.87; 95% CI: 0.78–0.98; I2 = 0% for Taiwan algorithm). We thank both authors and the editor for the opportunity to elaborate on our previous work. There are no competing interests to declare.