The interarm blood pressure difference: Do we know enough yet?

The difference in blood pressures between arms was first described in 1900.1 Since 2001–2002, when we hypothesized that interarm difference (IAD) was associated with peripheral arterial disease and then reported the first prospective association of IAD with increased mortality,2, 3 interest in the phenomenon has grown, seeing an exponential rise in the number of publications.4 Cross-sectional and prospective cohort studies continue to appear, with at least 24 trials published since 2013 demonstrating the associations of interarm differences with cardiovascular and all-cause mortality,5-14 describing prevalence in selected populatiions,15-20 or exploring correlations with established markers of elevated cardiovascular risk.21-28 There have also been at least seven systematic reviews.29-35 So, what can a further cohort study tell us? In a new cross-sectional study of 1426 patients, Ma and colleagues have carefully examined the associations of systolic interarm blood pressure differences (IADs) with carotid intima-media thickness (IMT).36 They calculated maximum and average IMT over 36 carotid sites, avoiding plaque (which was scored separately), and achieved excellent interobserver agreement. Brachial-ankle pulse wave velocity (PWV) and ankle-brachial index (ABI) were also measured, along with IAD, using an established device (VP1000, Omron, Tokyo, Japan). Using a cutoff value ≥10 mm Hg to define absolute systolic IAD, they found it to be significantly associated with higher mean values for both mean and maximal IMT, greater PWV, and lower ABI. There was no association with the presence of carotid plaque. Body mass index, prior diagnosis or treatment of hypertension, systolic blood pressure (BP), and age were also higher with an IAD. After multivariable adjustment, a significant association with lower ABI remained, but multivariable models for IMT and PWV were not reported. ABI is an established marker for peripheral arterial disease and makes an independent contribution to cardiovascular risk prediction.37, 38 In our 2012 meta-analysis, we identified nine cross-sectional studies associating IAD with peripheral arterial disease and/or a reduced ABI, with a relative risk of 3.3 (95% confidence interval, 2.1–5.2) for ABI <0.9 with systolic IAD ≥10 mm Hg.30 We have also recently confirmed a continuous negative correlation of ABI with IAD in a large community study.39 Stroke and cognitive decline are associated with IAD.6, 30, 40, 41 Increasing carotid IMT is associated with increasing risk of stroke and may help refine risk prediction.42, 43 The large Multi-Ethnic Study of Atherosclerosis (MESA) previously identified the association of systolic IAD ≥15 mm Hg with greater carotid IMT; it included 6743 participants and the association remained after multivariable adjustment.44 Severe carotid disease has also been associated with a large IAD (mean 41 mm Hg) in one selected vascular surgery series.45 Smaller IAD cohort studies (sample sizes from 364 to 554) have failed to confirm an association of IAD and IMT16, 25, 46; therefore, the current study, with its careful evaluation of IMT, is a welcome addition to the evidence from MESA. Unfortunately, a multivariable-adjusted analysis of the association is not presented, so we cannot draw firm conclusions as to the relevance of carotid IMT to the etiology of IAD. Existing evidence suggests that larger MESA-sized studies may be needed to fully explore this association. Carotid-femoral PWV was previously found to be higher with a systolic IAD ≥10 mm Hg in 1045 participants from the Baltimore Longitudinal Study of Aging. IAD was also prospectively associated in a subgroup of 629 patients with greater unit increases in PWV over a mean interval of 4.1 years.21 We have also correlated IAD as a continuous variable with carotid-femoral PWV in an English cohort of 554 patients with and without existing vascular diseases46; however, no correlation was observed for 484 Finnish participants in the FINRISK study.25 Maeda and colleagues could not identify any association between systolic or diastolic IAD ≥10 mm Hg with brachial-ankle PWV among 364 hospital outpatients using the VP1000.16 Su and colleagues have also used this device to study differences between right and left arm brachial-ankle PWV in a cohort of 1111 Taiwanese patients attending their hospital for echocardiography. They hypothesized that differences in arterial stiffness between sides, represented by PWV differences, would correlate with BP differences between limbs. This was shown for inter-ankle pressure differences but not IAD.27 While brachial-ankle PWV is strongly correlated with carotid-femoral PWV, they are not the same. Approximately one quarter of the variance between them is attributed to leg PWV, thus brachial-ankle PWV measures central and peripheral arterial stiffness while carotid-femoral PWV reflects central arterial stiffness more accurately.47 At present, we have some evidence to associate elevated carotid-femoral PWV with IAD, but only Ma and colleagues’ study, the largest cohort yet reported, supports an association for brachial-ankle PWV, in contrast to the two smaller negative studies. Again, an adjusted analysis might have helped us interpret this finding further, and we cannot be sure how much their finding depends on study size alone. The physical and pathological basis of an IAD is not fully understood. The PWV studies discussed lend support to the emerging “stiffness hypothesis,” which attributes observed differences in brachial BP to asymmetrical or localized variations in arterial stiffness.21, 48 This is gaining traction over the older “stenotic” hypothesis that saw an IAD ≥15 mm Hg labelled as subclavian stenosis.44, 49 Undoubtedly, some IADs are attributable to fixed arterial stenosis or disease, but measurable stenoses usually result in larger IADs than those observed in IAD studies.30, 35 A frequent criticism of the utility of IAD as a marker of arterial disease is the lack of reproducibility.50 As long ago as 1939, Allen and Hines recognized that IAD was usually transient rather than constant and suggested that persistent differences should lead to suspicion of organic disease.51 This has not been extensively studied but evidence from small numbers of cases suggests that a reproducible repeated finding of IAD can indicate an underlying stenosis requiring further arterial imaging.30, 52, 53Arterial stiffness is a plausible explanation for IAD, since the observed brachial BP relies on the interplay of soft tissue factors, hand dominance, and forward and backward reflected pulse waves affected in turn by vascular tone, white-coat effects, and stiffness.17, 54-56 Stiffness develops early in the evolution of arterial disease and is commonly asymmetric.57, 58 There are two other caveats to consider with the current study. First, prevalence of an IAD ≥10 mm Hg was reported as 10.5%. We recently reported comparable pooled prevalence data from good-quality community studies, finding figures of 11.2% (95% confidence interval, 9.1–13.6) in hypertensive cohorts, 7.4% (5.8–9.2) in diabetic cohorts, and only 3.6% (2.3–5.0) in a general adult population. Given the relatively young mean age of 51 years in Ma's study, with only 34% of participants having hypertension, a lower IAD prevalence than reported was expected in population terms. In addition, we have observed from previous studies that the VP1000 shows a trend towards lower prevalences than other methods of simultaneous IAD assessment.59 The prevalence of diabetes was high in this study (15%), questioning how the proportional sampling technique employed was applied. A second unexpected finding is that the mean pulse pressure is higher in the left than the right arm. In our recent meta-analysis of 20 studies, we found that mean systolic BP was 0.9 mm Hg (0.6–1.1) higher in the right arm with low heterogeneity (I2=14%) between studies.55 The current study reports by contrast a mean of −2.0 mm Hg (−1.3 to −2.7; P<.0001 for difference). These two differences from previously published studies raise concern that either the population studied may not be fully representative of their community or that the measurement technique may not have been robust. Simultaneous measurement methods, as used by Ma and colleagues, are recommended to reduce overestimation of IAD.31, 59, 60 The authors refer to the VP1000 as being validated but no protocol exists against which to validate measurements of IAD–this only applies to BP measurement itself.61 For this study, a single simultaneous pair of BP measurements were obtained over 10 seconds,62 whereas multiple readings are preferred to avoid overestimation of IAD.21, 31, 56, 59 So, do we know enough about IAD now? Ma and colleagues have added some support for the stiffness hypothesis of IAD and demonstrate that future studies will need larger sample sizes to usefully add to current knowledge. Hypertension guidelines advocate recognition of increased arterial stiffness (identified on pulse pressure or PWV measurement) as target organ damage, requiring aggressive cardiovascular risk management.63 If we accept the stiffness hypothesis, then we should add IAD to the list of cardiovascular risk markers and act on it when detected. There is some evidence that addition of IAD to the Framingham score alone can improve precision of cardiovascular risk estimation,8, 64 and we are aiming to develop a prediction tool that takes account of IAD in our current international Inter-arm Blood Pressure Difference Collaboration (INTERPRESS) individual patient data meta-analysis.65 Adoption of BP measurement in both arms is increasing with time,66 so practical evidence-based guidance to inform clinical interventions upon detecting an IAD is required. Studies of interventions based on detection of an IAD will need to be large and are not on the horizon. For now, we may have enough observational evidence to start to exercise clinical judgment. None.