Author + information
- Received February 13, 2012
- Revision received March 30, 2012
- Accepted April 3, 2012
- Published online October 16, 2012.
- Athanase Benetos, MD, PhD⁎,†,‡,§,⁎ (, )
- Sylvie Gautier, MD‡,§,
- Carlos Labat, BSc⁎,‡,§,
- Paolo Salvi, MD, PhD∥,
- Filippo Valbusa, MD¶,
- Francesca Marino, MD#,
- Olivier Toulza, MD⁎⁎,
- Davide Agnoletti, MD††,
- Mauro Zamboni, MD, PhD#,
- Delphine Dubail, MD‡‡,
- Patrick Manckoundia, MD, PhD§§,
- Yves Rolland, MD, PhD⁎⁎,
- Olivier Hanon, MD, PhD‡‡,
- Christine Perret-Guillaume, MD, PhD†,‡,§,
- Patrick Lacolley, MD, PhD⁎,‡,
- Michel E. Safar, MD†† and
- Francis Guillemin, MD, PhD‡∥∥
- ↵⁎Reprint requests and correspondence:
Dr. Athanase Benetos, Department of Geriatrics, University Hospital of Nancy, Vandoeuvre les Nancy 54511, France
Objectives The aim of the longitudinal PARTAGE study was to determine the predictive value of blood pressure (BP) and pulse pressure amplification, a marker of arterial function, for overall mortality (primary endpoint) and major cardiovascular (CV) events, in subjects older than 80 years of age living in a nursing home.
Background Assessment of pulse indexes may be important in the evaluation of the CV risk in very elderly frail subjects.
Methods A total of 1,126 subjects (874 women) who were living in French and Italian nursing homes were enrolled (mean age, 88 ± 5 years). Central (carotid) to peripheral (brachial) pulse pressure amplification (PPA) was calculated with the help of an arterial tonometer. Clinical and 3-day self-measurements of BP were conducted.
Results During the 2-year follow-up, 247 subjects died, and 228 experienced major CV events. The PPA was a predictor of total mortality and major CV events in this population. A 10% increase in PPA was associated with a 24% (p < 0.0003) decrease in total mortality and a 17% (p < 0.01) decrease in major CV events. Systolic BP, diastolic BP, or pulse pressure were either not associated or inversely correlated with total mortality and major CV events.
Conclusions In very elderly individuals living in nursing homes, low PPA from central to peripheral arteries strongly predicts mortality and adverse effects. Assessment of this parameter could help in risk estimation and improve diagnostic and therapeutic strategies in very old, polymedicated persons. In contrast, high BP is not associated with higher risk of mortality or major CV events in this population. (Predictive Values of Blood Pressure and Arterial Stiffness in Institutionalized Very Aged Population [PARTAGE]; NCT00901355)
The dramatic increase in the number of elderly people, especially those 80 years of age and older, has translated into a growing population that is increasingly prone to frailty, multiple comorbidities, and partial loss of autonomy. This is now one of the target populations for geriatric medicine, necessitating the development of specific diagnostic and therapeutic approaches (1). These approaches cannot be derived, however, from a simple extrapolation of the strategies applied in younger populations or even in very elderly robust populations. Thus, assessment of cardiovascular (CV) risk in these individuals represents a major issue.
High blood pressure (BP), especially systolic hypertension, is a major determinant of morbidity and mortality in the elderly (2) In addition, a decrease in BP with antihypertensive treatment in individuals 80 years of age or older has been shown to be beneficial (3). However, these results were obtained in community-dwelling individuals without major comorbidities. Actually, the association between BP levels and morbidity and mortality in very elderly persons with several comorbidities remains a controversial issue, with several studies showing a lack of such a relationship (4–6) or even an inverse relationship (7,8). It is therefore logical to seek alternative approaches to estimate CV risk in these individuals. Assessment of arterial mechanical properties by measuring pulse indexes such as pulsed wave velocity (PWV), pulse contour analysis, and pulse pressure (PP) amplification (PPA) may be of interest in this respect. The recent development of several noninvasive validated devices has allowed the possibility of such measurements in several populations (9,10). Clinical studies have shown in middle-age and older populations that such measurements can provide additional BP information for the prediction of CV risk (11,12). To date, no large study has evaluated the predictive value of PP and of pulse indexes for morbidity and mortality in very elderly individuals living in nursing homes.
The aim of the present study was to evaluate the prognostic value of arterial mechanical parameters (PPA and PWV) and BP on total mortality and major CV events in very elderly individuals living in nursing homes.
The PARTAGE (Predictive Values of Blood Pressure and Arterial Stiffness in Institutionalized Very Aged Population) study is a multicenter, longitudinal study aimed at determining the 2-year predictive value of BP and arterial functional parameters on total mortality (primary endpoint) as well as major CV outcomes and cognitive decline (secondary endpoints) in a large population of individuals 80 years of age and older living in nursing homes. The rationale and baseline parameters of this study were previously described (13).
Participants were enrolled in 4 French (Nancy, Dijon, Paris, Toulouse) and 2 Italian (Cesena and Verona) university hospital centers between January 2006 and June 2008. A total of 72 nursing homes participated in this study in France and Italy (13).
Participants were included if they were 80 years of age and over, living in nursing homes, and signed the informed consent. Subjects were excluded if they had severe dementia (Mini-Mental Status Examination score <12 out of 30), a low level of autonomy (Activity of Daily Living [ADL] scale score ≤2 out of 6) or were under guardianship or some measure of legal protection. No other exclusion criteria were applied. The family and/or the physician of the patient were informed of the study and gave their approval.
According to the inclusion criteria, 1,259 individuals living in the nursing homes participating in this study were eligible. Among them, 1,130 (89%) agreed to participate and were enrolled in this study. Four subjects were excluded from the present analysis because they did not have self-measured BP. Therefore, 1,126 subjects (874 women and 252 men) were analyzed.
This study was approved by the respective regional ethics committees in France (Comité de Protection des Personnes) and in Italy (Comitato Etico Area Vasta Romagna and Comitato Etico della Provincia di Verona), and all participants gave written informed consent before the study.
Patients were included from January 2007 to June 2008 and were followed for 2 years. Adverse outcomes were recorded every 3 months from inclusion to the end of the study, using a questionnaire addressed to the physicians at each nursing home. In addition, 2 visits were conducted by the study investigators at the end of the first and second year of follow-up.
Clinical data collection
All geriatric assessment instruments and arterial measurements were performed in the nursing homes by a trained medical research investigator present at each university hospital center participating in the study. Clinical data collection was performed during face-to-face interviews and acquired from the patients' medical records.
Arterial functional parameters
Central BP values and aortic pressure waveforms were obtained directly from the common carotid artery using an applanation tonometer (9,14). Arterial tonometry was performed on right common carotid artery and femoral artery using a PulsePen device (DiaTecne srl, Milan, Italy) (10). As previously demonstrated, the pressure waves recorded noninvasively by the PulsePen tonometer at the site of the common carotid artery are similar to pressure waveforms obtained invasively by means of an intra-arterial catheter (10). Moreover, several studies demonstrated that central BP values and pulsed wave analysis recorded in the common carotid artery are a reliable surrogate of analysis recorded in the aorta by invasive methods (9,12,15). Central BP values were obtained by the carotid BP curve integral after calibration with brachial mean and diastolic BP (DBP) measured noninvasively by a validated oscillometric sphygmomanometer (16) at the brachial artery (Omron 705IT, Omron Co., Kyoto, Japan). The PPA was the percentage of increase of PP in the brachial artery (PPB) relative to central PP (PPC), according to the formula: PPA = 100 · (PPB − PPC)/PPC. The augmentation index (AI) was measured from the analysis of carotid pulsed waves according to the previously reported method (10).
The PulsePen device was also used for measuring carotid-femoral PWV, which is considered the gold standard for measuring arterial stiffness (10,11,13). The procedure was described in detail previously (13). For technical reasons, arterial measurements were not obtained in 66 (for PPA and AI) and 56 (for PWV) subjects.
BP and heart rate measurements were performed at the brachial artery level using the validated automated oscillometric device Colson DM-H20 (Dupont Médical, Frouard, France). The midarm circumference was measured and the cuff width adapted accordingly.
Self-measurements of BP were performed following the “rule of 3” (3 measurements with intervals of 1 min in the morning and evening for 3 consecutive days) according to the protocol proposed by the French Society of Hypertension (17). The procedure was described previously (13). In the present study, the average of the BP values of the 3 days (morning and evening) was used for the different analyses.
Mean arterial pressure (MAP) was calculated as: DBP + 1/3 PP.
The primary endpoint was overall mortality during the 2-year follow-up period.
The secondary pre-specified endpoints were major CV events. Major CV events (CV morbidity and mortality) included both nonfatal CV events that led to hospitalization or a specific long-term new treatment as well as death from cardiac, cerebrovascular, and other vascular causes.
Information on the cause of death and adverse cardiac events were reported by the physicians at the nursing homes every 3 months according to the procedure detailed previously.
According to the reported information, 2 investigators in our group (S.G., P.S.) classified subjects with or without CV events. At the end of the study, all events were re-evaluated in a blinded manner by a third investigator (A.B.). In few cases, in which the 2 classifications differed, a third consensus review was conducted with the 2 investigators. Less than 5% of such differences were observed, and in all cases, consensus was reached during the consensus meetings.
Descriptive values are expressed as mean ± SD or number and percentage. For the comparisons of men and women (Table 1), the Wilcoxon rank sum test was used for continuous variables, the chi-square test for discrete variables, and the log-rank test for 2-year mortality and major CV events. Univariate correlations were made with Pearson's parametric test. The occurrence of total mortality and major CV events according to tertiles of each hemodynamic parameter was estimated using Kaplan-Meier curves for graphic representation and compared by the log-rank test. Cox regression multivariate models were used to assess the relative risk (hazard ratio and 95% confidence interval) of total mortality and of major CV events, according to each hemodynamic parameter as a continuous variable. According to the univariate analyses, the following variables were associated at the 0.10 level with total mortality and subsequently included in the multivariate Cox models: sex, age, ADL, body mass index, Charlson comorbidity index, and history of CV disease. For major CV events, by using the same procedure, age, ADL, a history of CV disease, and the presence of antihypertensive treatment were included. For PPA and PWV, additional adjustments for MAP and heart rate were made. This was necessary to ascertain whether the possible effect of PPA on endpoints was independent of BP and heart rate. The interaction between arterial parameters and current antihypertensive treatment on mortality and major CV events was tested with the Cox model by including the interaction term in the model. The term relative risk is used for hazard ratio throughout this paper. The proportional hazards assumption was assessed on the basis of a test of Schoenfeld residuals with the Cox regression using NCSS 2000 software (NCSS, LLC, Kaysville, Utah). A p value <0.05 was considered statistically significant. Statistical analyses were performed using NCSS 2000 statistical software package.
Among the 1,126 subjects of a mean age of 88 ± 5 years, 78% were women. Table 1 shows the main demographic and clinical characteristics of the men and women as well as the rates of total mortality and major CV events.
Table 2 shows that among arterial functional parameters, PWV showed the strongest relationship with MAP, whereas PPA showed the weakest relationship with MAP. Interestingly, PWV was not correlated with PPA. Multivariate analysis showed that PPA was positively associated with heart rate (p < 0.001) and negatively associated with age (p < 0.002) and AI (p < 0.02). Sex, MAP, and PWV were not associated with PPA on multivariate analysis.
Relationships of PPA and PWV with total mortality and major CV events
Among the 1,126 patients enrolled in the study, 839 completed the 2-year follow-up (Fig. 1). Among the remaining 287 subjects, 247 died and 40 were lost to follow-up. During the 2-year follow-up, 228 subjects experienced major CV events. At the 6 centers, no significant differences were observed for total mortality (20% to 28%; p = 0.34) and major CV events (19% to 33%; p = 0.10).
Figure 2 shows the survival curves for total mortality and major CV events according to the tertiles of PPA and PWV. A lower PPA was significantly associated with both higher total mortality (p = 0.003) and more major CV events (p = 0.004). By contrast, PWV was not associated with total mortality or with major CV events.
The role of PPA and PWV in total mortality or major CV events was also assessed by considering these parameters as continuous variables. Figure 3 depicts the relationships between PPA and total mortality (Fig. 3A) or major CV events (Fig 3B). On both univariate and multivariate analyses, the higher the PPA was, the lower the total mortality and major CV events. In this multivariate model, a 10% increase in PPA was associated with a significant 24% decrease in total mortality and 17% decrease in major CV events. Older age, low ADL scale score, and low body mass index were determinants of total mortality (Fig. 3A). In addition, male sex (p = 0.0003), high Charlson morbidity index score (p = 0.045), low MAP (p = 0.026), and high heart rate (p = 0.0007) were independent determinants of total mortality. A low ADL scale score was associated with major CV events (Fig. 3B).
With regard to PWV, the multivariate analyses confirmed the results of the univariate analyses (i.e., the absence of any relationship between PWV and total mortality or major CV events) (Table 3). A strong interaction was found between antihypertensive treatment and PWV in major CV events (p = 0.001) and was still significant after adjustment for covariates (p = 0.002). After these results, we conducted separate analyses to assess the influence of PWV on major CV events in subjects with and without antihypertensive therapy. In subjects not receiving antihypertensive treatment (n = 231), an increase in PWV of 1 m/s was associated with a 9% higher risk of major CV events (p = 0.003) (Fig. 4). By contrast, in subjects receiving antihypertensive treatment (n = 895), no such an association was found. No association was found between the AI and total mortality or major CV events in the entire population and in the subgroups with or without treatment (data not shown).
Relationships between BP and total mortality and major CV events
Figure 5 shows the survival curves for total mortality and major CV events according to the tertiles of self-measured BP. Higher mortality was significantly associated with lower DBP (log-rank, p = 0.021). The association between SBP or PP tertiles and total mortality did not reach significance (p = 0.057 and p = 0.15, respectively). No association was found between any of the BP measurements and major CV events. Central PP (not shown) was not associated with total mortality or major CV events. The results of the multivariate analyses are shown in Table 3. A 10-mm Hg increase in SBP, DBP, or MAP was associated with a significant decrease in the risk of total mortality by 9%, 16%, and 15%, respectively. BP levels were not associated with major CV events (Table 3).
With regard to BP levels, similar results were observed when BP measured by a clinician, instead of self-measured BP, was considered (data not shown).
This longitudinal study performed in subjects 80 years of age and older and living in nursing homes provides 2 major original findings. 1) The increase in PP from central (carotid) to peripheral (brachial) arteries is a good predictor of total mortality and major CV events in this population. The lower the PPA is, the higher the total mortality and major CV events. 2) Self-measured BP is either not associated or inversely associated with total mortality and major CV events, confirming previously reported data obtained with standard BP measurements in very elderly subjects.
Influence of PPA on total mortality and major CV events
Several reports have indicated interest in assessing PPA to predict CV complications and benefits of antihypertensive treatment (12,18,19). We recently reported in a large middle-age population study (20,21) that calculated PPA was strongly associated with both CV and total mortality and that this association was stronger than the association between mortality and central or peripheral PP, each taken separately.
The present longitudinal analysis shows that an increase of 10% in PPA (corresponding to ∼1 SD) was associated with a decrease of 24% and 17% in total mortality and major CV events, respectively. Of note, the association between PPA and the endpoints was markedly significant after adjusting for several confounders, including history of CV disease. This adjustment was necessary because in a cross-sectional analysis of the baseline data of the present study (22), we demonstrated a link between a history of heart disease and PPA levels. We also tested this association after adjusting for heart rate because the latter is a significant determinant of PPA and may also influence CV events. Again, the association between PPA and mortality or CV events remained highly significant.
Influence of SBP, DBP, and PP on total mortality and major CV events
This study provides significant new information showing that PPA has a prognostic value in an elderly population, whereas BP measurements do not. In fact, in this population, the association between BP and endpoints was either absent (for major CV events) or even negative (for total mortality). Previous studies showed that the respective roles of SBP, DBP, and PP are modified with advanced age with a weakening in the impact of DBP and an increasing role of SBP and PP (23,24).
However, in the present study, similar results were also observed for the other 2 components of BP (i.e., SBP and PP). Previous relatively small studies in very elderly subjects have reported the absence of any prognostic value of BP (4–6) or even negative associations between BP levels and morbidity and mortality (7,8). These paradoxical results can be explained by the fact that in these very elderly frail individuals, a low SBP may not simply be a sign of so-called good arterial health but often of malnutrition and comorbidities such as heart failure, neurological disorders, and other comorbidities associated with poor prognosis. Irrespective of the underlying explanation, the present results indicate that the BP levels in very elderly frail individuals are evidently not reliable, whereas PPA provides more pertinent information about the patient's prognosis. Interestingly, PPA is a ratio of 2 pressures, and therefore this parameter is weakly or even not related to absolute BP levels.
PWV, total mortality, and major CV events
In the present study, we found no relationship between PWV and total mortality or major CV events in the entire population. In middle-age and elderly populations, high PWV, a strong indicator of high aortic stiffness, is a determinant of CV events (11,24–27). We previously showed in a smaller population of subjects younger than 70 years of age living in a long-stay geriatric rehabilitation department that high PWV was associated with increased CV mortality but not with total mortality (6). In the present study, high PWV was associated with major CV events only in individuals without antihypertensive treatment, probably because those who were treated had more comorbidities. This result may be related to the fact that although PWV is determined by arterial structure and function, it is also influenced by BP levels. Therefore, any comorbidity that tends to decrease BP can also decrease PWV and therefore one can have a relatively low PWV despite alterations in arterial mechanical properties. However, contrary to what was observed with BP levels, high PWV was never associated with lower total mortality.
The main limitation of the present study is the absence of biological markers of frailty to better understand the underlying conditions explaining these findings. However, under the present conditions, the drawing of blood in nursing homes could have potentially resulted in the nonadherence of a large number of subjects with the subsequent risk of selection bias.
Our results provide new information concerning the assessment of risks in a dramatically growing elderly population living in nursing homes and representing a major challenge for geriatric care. The important finding is that in this frail population in which blood pressure measurements may be misleading, PPA constitutes a noninvasive appropriate method for assessing prognosis.
Of particular interest, these are polypathological, polymedicated individuals and are, in the large majority, treated for hypertension. Although the design of the present study does not allow the possibility to conclude on the previously reported interest in controlling high BP with drugs (3), the present findings do raise the issue of the utility of BP levels as an indicator of protection in these individuals. This is of major interest because iatrogenic-induced problems are also a major issue in the elderly.
In summary, low PPA from central to peripheral arteries strongly predicts mortality and CV adverse events. Assessment of this parameter could help in risk assessment and improve diagnostic and therapeutic strategies in very elderly, frail, and polymedicated persons. In contrast, high BP is not associated with a higher risk of mortality and major CV events in very elderly individuals living in nursing homes.
The authors thank Severine Buatois, Francesca Vaienti, Sara Capelli, Alexandra Zervoudaki, Sophie Gauthier, Adrian Klapouszczak, and the Association of Physicians of Lorraine Area (AMCELOR). The authors also thank all the directors, physicians, and especially the personnel of the 72 nursing homes for contributing to the realization of the PARTAGE study. And they thank Pierre Pothier for language review and stimulating discussions.
Primary financial support from PHRC of the French Ministry of Health (Registered AFSSAPS: 2006-A00042-49). Supplementary financial support from the French Society of Hypertension, the FRM (DCV20070409250), Boehringer Ingelheim France Laboratories, Inserm, and the PPF of the French Ministry of Research. The study was conducted with the logistic support of the Centre of Clinical Investigations (INserm CIC-P) and the Centre of Clinical Epidemiology (Inserm CIC-EC) of the University Hospital of Nancy. Dr. Salvi is a consultant for DiaTecne, Milan, Italy. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- activity of daily living
- augmentation index
- diastolic blood pressure
- mean arterial pressure
- pulse pressure
- pulse pressure amplification
- pulsed wave velocity
- systolic blood pressure
- Received February 13, 2012.
- Revision received March 30, 2012.
- Accepted April 3, 2012.
- American College of Cardiology Foundation
- Boshuizen H.C.,
- Izaks G.J.,
- van Buuren S.,
- Ligthart G.J.
- Meaume S.,
- Benetos A.,
- Henry O.F.,
- Rudnichi A.,
- Safar M.E.
- Laurent S.,
- Cockcroft J.,
- Van Bortel L.,
- et al.
- Reference Values For Arterial Stiffness' Collaboration
- Nichols W.W.,
- O'Rourke M.F.
- Hashimoto J.,
- Imai Y.,
- O'Rourke M.F.
- Williams B.,
- Lacy P.S.,
- Thom S.M.,
- et al.,
- Anglo-Scandinavian Cardiac Outcomes Trial Investigators; CAFE Steering Committee and Writing Committee
- Benetos A.,
- Thomas F.,
- Joly L.,
- et al.
- Regnault V.,
- Thomas F.,
- Safar M.E.,
- et al.
- Franklin S.S.,
- Larson M.G.,
- Khan S.A.,
- et al.
- Benetos A.,
- Thomas F.,
- Bean K.,
- Safar M.,
- Guize L.
- Cruickshank K.,
- Riste L.,
- Anderson S.G.,
- Wright J.S.,
- Dunn G.,
- Gosling R.G.
- Mattace-Raso F.U.,
- van der Cammen T.J.,
- Hofman A.,
- et al.