Five out the of the six algorithms compared in this manuscript contain a coefficient for age. Curiously, the Le Gal et al. derivation cohort had a higher mean therapeutic dosage requirement to our validation cohorts, despite age being a factor that increases maintenance dose, shown in Table 1. This suggests that even though age has an effect on warfarin maintenance dose requirements, the relationship may not be MK-4827 linear in nature. This is reinforced by Moreau et al. were elderly patients are shown to require lower induction and maintenance doses. For warfarin dosing in peadiatrics, a number of specialised algorithms have been produced, further showing that age has a complex relationship with warfarin dosing requirements. In the more general population algorithms investigated in this manuscript, the age to dose relationship may not be as prevalent, however, further research is recommended to assess the linear relationship assumed by linear regression methods. Hamberg et al. provide an excellent overview of current dosing algorithms available for paediatric populations, comparatively coefficients attributed to age are higher than those in this manuscript. Currently, paediatric algorithms use similar covariates to adult algorithms however future sources of variability found in either subpopulations are recommeneded to be investigated for translational impact. The validation cohorts utlised in this manuscript consisted of various Caucasian populations. There have been a number of recent algorithms derived in non-Caucasian populations and a study of the effect that the VKORC1 polymorphism across 3 racial groups that show dose requirements vary with ethnicity. The inconsistent replication observed in manuscript, which has also been observed previously, leads us to hypothesise that regression modelling may not be the most optimal approach for developing a warfarin MD algorithm. Linear regression models may not be able to fully draw on the variability that can be explained by potential factors. Alternatively, potential factors may be important in different stages of warfarin therapy, but less consequential in the determination of a dose to prescribe a patient in the maintenance phase of therapy. This latter suggestion is hypothesised for pharmacogenetic factors in Horne et al.. Different genetic biomarkers have been hypothesized to affect warfarin dose requirements; these include the genes, CYP4F2, CALU and GGCX. The implementation of these new genetic biomarkers into dosing algorithms has been seen only in non-Caucasian derivation cohorts. The validation cohorts utlised in this manuscript consisted of various Caucasian populations so we were unable to independently validate non-Caucasian algorithms in this manuscript. Further research of alternative genes hypothesised to affect warfarin dosing could be beneficial in Caucasian populations. This research should be in consideration of the current literature that shows the VKORC1 polymorphism causes differing warfarin dose requirements across 3 racial groups, potentially investigating epistasis effects.