Author + information
- Julio A. Chirinos, MD⁎ ( and )
- Raymond Townsend, MD
- ↵⁎University and Woodland Avenue (Room 8B111), Philadelphia, Pennsylvania 19104
We read with interest the paper by Benetos et al. (1) regarding pulse pressure amplification (PPA) as a predictor of cardiovascular risk. We fully agree that measuredPPA is an important cardiovascular risk marker, because it reflects the level of central systolic/pulse pressures for any given level of brachial pressures. However, we would like to challenge the investigators' conclusion that computed PPA based on standard risk factors, without the use of actual central pressure measurements, predicts cardiovascular risk and that “results were independent of any other confounding factors.” The investigators generated a regression equation from a sample of adults (n = 834) to predict carotid pulse pressure (CPP). They identified age, sex, height, and glycemia as predictors of CPP after adjustment for brachial pulse pressure (BPP). The investigators present a model that predicts CPP with an impressive R2(85.8% predicted variability). However, BPP alone predicted 83.5% of the CPP variability, all the other terms providing as little as a 2.3% increase in R2(corresponding to <10% of the CPP variability not explained by BPP). The investigators provide no evidence of the external validity of their equation, proceeding to apply it to a large population (n = 125,151). They concluded that the estimated carotid/brachial pulse pressure ratio (C/B ratio) was predictive of cardiovascular and all-cause death and that “results were independent of any other confounding factors.” To the degree that the estimated C/B ratio is a simple function of BPP and classic cardiovascular risk factors, it cannot possibly have true independent value beyond those factors. The calculated C/B ratio is necessarily correlated with BPP, hence its prognostic value after adjustment for age and sex. The marginal increase in adjusted standardized hazard ratios over BPP is likely due to the prognostic effect of body height, which is used to estimate the C/B ratio (and therefore bears an independent association with it). Body height itself is associated with increased cardiovascular risk (2) and was not included in proportional hazards models. Furthermore, the investigators provided no evidence of statistical superiority of models including the estimated C/B ratio over those including BPP or any data reflecting actual model performance (such as Akaike's information criteria or C-statistics). Had the investigators compared a model including age, sex, BPP, body height, and glycemia with a model containing the computed C/B ratio, prediction of cardiovascular death would have been superior with the former.
There is a need to continue to gain understanding regarding the mechanisms and the predictive value of central hemodynamics, but the benefit of PPA as a marker of cardiovascular risk lies in the large variability that cannot be predicted by standard risk factors, therefore requiring actual assessments via pulse wave analysis, rather than on the relatively small variability that can be predicted with the reported regression equation.
- American College of Cardiology Foundation
- Benetos A.,
- Thomas F.,
- Joly L.,
- et al.
- Langenberg C.,
- Marmot M.