Authors' response

4.85 to 8.61 respectively, both P<0.001).Seemingly this is exactly the sort of evidence base to justify measuring both arms, but all studies included in the analysis were of populations at existing high vascular risk, for example, referrals to angiography services.We cannot discover an evidence base that permits extrapolation of the guideline statement to the general population of which it is aimed.Parker and Glasziou also raise the important issue of how to measure an inter-arm difference.We have found that prevalence of an inter-arm difference is over-estimated without a robust measurement technique.2 While this is of epidemiological importance we have found repeated simultaneous measurements to be a barrier to recruitment in primary care 5 and this approach has been criticised as impractical.6 To overcome this we have compared the use of a single sequential pair of measurements to our gold standard simultaneous technique in 187 subjects in primary care with type 2 diabetes.Preliminary findings in 187 subjects have shown a high negative-predictive value of 0.97 in excluding a systolic inter-arm difference >10 mmHg.7 Consequently, the vast majority of subjects who do not have an inter-arm difference can be identified within a single consultation, and only the 10-20% remaining will need further assessment.The validity of this approach, and the clinical implications of detecting an inter-arm difference in subjects at low cardiovascular risk, both require further study.