Article Text

Download PDFPDF

Cardiovascular physiology
Augmented blood pressure response to exercise is associated with improved long-term survival in older people
  1. P Hedberg1,2,
  2. J Öhrvik3,
  3. I Lönnberg4,
  4. G Nilsson2
  1. 1
    Department of Clinical Physiology, Central Hospital, Västerås, Sweden
  2. 2
    Center for Clinical Research, Uppsala University, Central Hospital, Västerås, Sweden
  3. 3
    Cardiology Research Unit, Karolinska Institutet, Stockholm, Sweden
  4. 4
    Department of Cardiology, Central Hospital, Västerås, Sweden
  1. Dr P Hedberg, Department of Clinical Physiology, Central Hospital, SE-721 89 Västerås, Sweden; par.o.hedberg{at}


Objective: Studies on the prognostic importance of the systolic blood pressure (SBP) response during exercise report ambiguous results. Most research focuses on younger and middle-aged selected patient groups and rarely includes women. We investigated the prognostic value of SBP response during exercise testing in 75-year-olds.

Design: Prospective observational cohort study.

Setting: A community-based random sample of 75-year-old men and women (n = 382).

Main outcome measures: The prognostic value of SBP change from rest to peak exercise during a symptom-limited cycle test was evaluated for the endpoints all-cause mortality and cardiovascular mortality during long-term follow-up.

Results: After a median follow-up of 10.6 years, 140 (37%) of the participants had died, 64 (17%) from cardiovascular causes. The all-cause mortalities for exercise SBP changes of ⩽30 mm Hg, 31–55 mm Hg and >55 mm Hg were 5.1, 4.2 and 2.6 per 100 person-years, respectively (logrank 9.6; p = 0.008). For every 10 mm Hg increase in SBP during exercise the relative hazard for all-cause mortality was reduced by 13% (p = 0.030) and for cardiovascular mortality by 26% (p = 0.004) after adjustment for sex, smoking, waist circumference, total/HDL cholesterol ratio, prevalent ischaemic heart disease, hypertension, diabetes, cardiovascular medication, pre-exercise SBP, exercise capacity, resting left ventricular ejection fraction and left ventricular mass index.

Conclusions: Our findings suggest that an augmented SBP response during exercise is associated with an improved long-term survival among community-living 75-year-old individuals.

View Full Text

Statistics from

Blood pressure is routinely measured during clinical exercise testing. A drop in exercise systolic blood pressure (SBP) below the pre-exercise value is universally accepted as a predictor of adverse prognosis.1 Other patterns of SBP response during exercise are more controversial. An exaggerated rise in SBP during exercise predicts future hypertension,2 3 myocardial infarction,4 stroke,5 cardiovascular and all-cause mortality.6 7 In contrast, other studies associate an exaggerated SBP response during exercise with an improved prognosis. Among male patients referred for clinical reasons, a higher SBP rise during exercise is associated with lower rates of sudden death,8 cardiovascular mortality9 and all-cause mortality.10 Most studies on blood pressure response during exercise focus on younger and middle-aged patients and very few included women. Data on older people from a general population are limited. The purpose of the present study was to evaluate the long-term prognostic value of the SBP response during a maximal symptom-limited cycle exercise test in a community-based sample of 75-year-old men and women.


Study population

In 1997, there was a population of 1019 men and women born in 1922 and living in the city of Västerås, Sweden. Based on the Swedish population register, which contains the names and addresses of all Swedish residents, we invited a random sample of 618 persons from this 75-year-old population. The invitation was accepted by 433 individuals, corresponding to a response rate of 70.1%. Drop-outs have been described in detail elsewhere.11 Exercise testing was not performed in 44 cases because of logistical reasons (n = 23), locomotive impedance (n = 17), unwillingness (n = 3) or high resting blood pressure (n = 1). Data on blood pressure during exercise were missing in three participants. Exercise hypotension is a recognised predictor for an increased risk of cardiac events.12 Consequently, we excluded an additional four participants whose SBP during exercise dropped below the pre-exercise value. Thus, the final cohort comprised 382 participants, 192 men and 190 women. The study was approved by the Ethics Committee of Uppsala University, Sweden. All participants gave their informed consent.

Prevalent ischaemic heart disease was defined as either a history of myocardial infarction or angina pectoris, or abnormal Q-waves (Minnesota codes 1.1–1.2) on a resting 12-lead electrocardiogram (ECG). Hypertension was judged to be present if the participant had received this diagnosis and had been prescribed antihypertensive medication from their physician. Cardiovascular medication was defined as regular medication with ACE inhibitors, Ca inhibitors, diuretics, beta-blockers, lipid-lowering drugs or acetylsalicylic acid (ASA). Smoking was defined as current or previous daily smoking within a year before the inclusion date.

Exercise stress testing

Rodby RE 830/990 electrically braked cycle ergometers (Rodby Innovation AB, Vänge, Sweden) were used for the upright symptom-limited exercise test. A 12-lead ECG was registered before, continuously during and during 4 min after the exercise test. The workload was started at 30 W and was increased in steps of 10 W per minute. The pedalling rate was close to 60 per minute. Participants were encouraged to exercise until exhaustion. Reasons for discontinuation of the exercise test other than exhaustion were chest pain (n = 6), pain in hip or knee (n = 7), ST abnormalities (n = 6), frequent ventricular ectopic beats (n = 2) and a drop in SBP of >10 mm Hg but not below the pre-exercise value (n = 2). Exercise capacity (W), was calculated as metabolic equivalents (METs) for maximal oxygen consumption ((13×workload (W)/weight (kg) +3.5)/3.5).13 The maximum predicted heart rate was estimated as 208–0.7×age.14 In participants with interpretable ST-segment changes, an ischaemic response was considered present if there was a ⩾1 mm horizontal or downsloping ST-segment depression 80 ms after the J point. The SBP before, during and after exercise was measured manually with an arm-cuff sphygmomanometer. The pre-exercise SBP was obtained sitting on the cycle ergometer immediately prior to the start of the exercise test. During exercise, SBP was measured in 2–3 min intervals. When approaching the end of exercise, efforts were made to obtain one SBP measurement as close as possible to the end of the exercise phase before the subject stopped pedalling. SBP rapidly drops after cessation of exercise,15 and so we defined peak exercise SBP as the last SBP value obtained during the exercise phase. Exercise SBP change was defined as the difference between peak exercise SBP and sitting pre-exercise SBP.


The echocardiographic studies were performed using an Acuson XP 128 system (Acuson Co, Mountain View, California) with the participants in the left recumbent position. A single physician, who was blinded to the participants’ clinical data, carried out all studies.

Linear dimensions of interventricular septal thickness (IVS), left ventricular posterior wall thickness (LVPW) and left ventricular cavity diameter (LVD) in diastole were obtained from the two-dimensional parasternal long-axis view.16 The left ventricular mass was calculated as 0.8×(1.04[(LVD+IVS+LVPW)3−(LVD)3]+0.6 and indexed for body surface area.17 Left ventricular hypertrophy was defined as an LV mass index >115 g/m2 in men and >95 g/m2 in women.18

Left ventricular wall motion was assessed by a semiquantitative method by dividing the left ventricle into nine segments examined in standard projections.19 Each segment was assigned one score of +3 for hyperkinesia, +2 for normokinesia, +1 for hypokinesia, 0 for akinesia and –1 for dyskinesia. The average score for the nine segments constituted the wall motion score index (WMSI). The left ventricular ejection fraction (LVEF) was assessed by the biplane disc summation method according to the modified Simpson rule (Simpson-LVEF).18 Left ventricular systolic dysfunction was defined as LVEF <50%. In participants in whom Simpson-LVEF could not be determined (n = 127), LVEF was derived from the WMSI according to: 18×e(0.6×WMSI).20 In the present study population, this surrogate measurement was strongly correlated with Simpson-LVEF (r = 0.83; p<0.001) and detected a Simpson-LVEF of <50% with sensitivity, specificity, negative and positive predictive values of 100%, 98%, 100% and 85%, respectively.20


Outcomes were all-cause and cardiovascular mortality. Cardiovascular mortality was defined as death caused by myocardial infarction, heart failure, stroke or ruptured aortic aneurysm. Follow-up information was based on the Swedish population register (all-cause mortality) and the Causes of Death Register (cardiovascular mortality). The tenth revision of the International Statistical Classification of Diseases (ICD) was used to identify causes of death. The registers were linked to the participants by the unique personal identification number assigned to each Swedish resident. Participants were followed from the index examination in 1997 until any end-point or, at the latest, the censoring date of 31 December 2007. Assessment of follow-up survival status was complete in all participants. Causes of death were missing in three of the deceased. Participants with non-cardiovascular or unknown cause of death were treated as censored at the time of death when analysed for cardiovascular mortality.

Statistical methods

Continuous variables were summarised by means (SD) and categorical variables by counts and percentages. The Wilcoxon–Mann–Whitney rank sum test was used to compare two groups and the Kruskal–Wallis test to compare three groups for continuous variables. For categorical variables, the Fisher exact test was used.

Univariate Cox regression was used to identify factors potentially associated with survival (sex, smoking, body mass index, waist circumference, total/HDL cholesterol, hypertension, diabetes, ischaemic heart disease, cardiovascular medication, pre-exercise heart rate, pre-exercise SBP, exercise capacity, exercise chest pain, ischaemic exercise ECG response, LV mass index and LVEF). The relation of exercise SBP change to outcome was evaluated by multiple Cox regression models adjusting for covariates which were significant at p<0.05 in univariable analyses (supplementary table). Pre-exercise SBP was used as a covariate despite a lack of univariable association with outcome. Exercise SBP change was analysed first as a categorised (tertiles) and then as a continuous variable. Three levels of adjustment were used: sex and pre-exercise SBP (Model A); covariates in Model A plus ischaemic heart disease, hypertension and cardiovascular medication (Model B); and covariates in Model B plus smoking, waist circumference, total/HDL cholesterol, diabetes, METs, LVEF and LV mass index (Model C). Further, sex-stratified Cox regression analyses were performed. However, because of the reduced number of endpoints, adjustments were restricted to the covariates ischaemic heart disease, hypertension and medication in the sex-stratified models.

To determine whether exercise SBP change adds significantly to the prediction of mortality risk beyond traditional risk factors, we conducted log-likelihood ratio tests by comparing two nested Cox models, one model with traditional risk factors (Model C except exercise SBP change) and the other model with traditional risk factors plus exercise SBP change. For adequate control for confounders, covariates were retained in the models regardless of their statistical significance. The assumption of proportional hazards was confirmed by including time-dependent covariates in the models. Survival curves were generated by means of Kaplan–Meier estimates. Differences in survival were compared by the logrank test. All tests were two-tailed, and a p value <0.05 was considered statistically significant. STATA version 10 (StataCorp, College Station, Texas) was used for all analyses.


Baseline characteristics according to follow-up status

The median follow-up was 10.6 years, during which time 140 deaths were observed (rate 4.0 per 100 person-years). The cause of death was cardiovascular in 64 cases (47%, rate 1.8 per 100 person-years). At baseline, those who subsequently died were significantly more likely to be male smokers and to have a higher prevalence of hypertension, diabetes and ischaemic heart disease than those who survived (table 1). The proportions of subjects on medication with ACE inhibitors (13% vs 5%; p = 0.005), Ca inhibitors (20% vs 7%; p<0.001) and diuretics (26% vs 11%; p<0.001) were higher among non-survivors than among survivors. Medication with beta blockers (25% vs 17%), lipid-lowering drugs (5% vs 2%) and ASA (24% vs 16%) did not differ significantly between the groups.

Table 1 Baseline clinical and echocardiographic characteristics of the 75-year-old participants according to follow-up survival status

Exercise test characteristics

No major complications occurred during the exercise testing. There was no significant difference between survivors and non-survivors regarding sitting heart rate and SBP immediately prior to the exercise test (table 2). The mean exercise duration was significantly shorter in non-survivors than in survivors (8.6 vs 9.5 min; p = 0.009). Non-survivors had a significantly lower exercise capacity, peak heart rate, peak SBP and SBP change during exercise than survivors. The distribution of peak exercise SBP is shown in fig 1.

Figure 1

Distribution of peak exercise systolic blood pressure (SBP) for non-survivors and survivors.

Table 2 Findings in the baseline exercise test according to follow-up survival status

Baseline characteristics according to exercise blood pressure response

An attenuated SBP change during exercise was significantly associated with female sex, higher prevalence of hypertension and use of cardiovascular medication (table 3). As expected, pre-exercise SBP was higher in those with an attenuated exercise SBP response. There were no significant differences between tertiles of exercise SBP change concerning ischaemic heart disease, diabetes, LV mass index or LVEF.

Table 3 Baseline characteristics according to tertiles of systolic blood pressure (SBP) response during exercise

Outcome in relation to exercise SBP change

The cumulative overall survival as a function of exercise SBP change categorised in tertiles is shown in fig 2. The all-cause mortalities for the lower, middle and upper tertile of exercise SBP change (⩽30 mm Hg, 31–55 mm Hg and >55 mm Hg, respectively) were 5.1, 4.2 and 2.6 per 100 person-years, respectively (logrank 9.6; p = 0.008). The corresponding cardiovascular mortalities for the tertiles of exercise SBP change were 2.6, 2.2 and 0.7 per 100 person-years, respectively (logrank 11.8; p = 0.003). After adjustment for sex and pre-exercise SBP, the association between exercise SBP change and outcome was strengthened (table 4). Further adjustment for ischaemic heart disease, hypertension and medication and, subsequently, other clinical and echocardiographic confounders attenuated the relationship between exercise SBP change and outcome, but it remained statistically significant.

Figure 2

Cumulative survival of the 75-year-old participants according to tertiles of systolic blood pressure change during baseline exercise testing.

Table 4 Relative risk of all-cause and cardiovascular mortality according to tertiles of systolic blood pressure (SBP) change during cycle exercise testing

When exercise SBP change was analysed as a continuous variable, it was significantly associated with outcome (table 5). After adjustment for important clinical and echocardiographic confounders, every 10 mm Hg increase in SBP during exercise was associated with a 13% (p = 0.030) reduction in the relative hazard of all-cause mortality and a 26% (p = 0.004) reduction in the relative hazard of cardiovascular mortality. Results were similar if either Simpson-LVEF or LVEF derived from WMSI was used as a covariate instead of the combined LVEF variable.

Table 5 Relative risk of all-cause and cardiovascular mortality according to systolic blood pressure (SBP) change during exercise testing

When comparing nested Cox models, the results indicated that the exercise SBP change added significantly to the prediction of all-cause (χ2 for adding exercise SBP change = 4.4, p = 0.037) and cardiovascular mortality risk (χ2 for adding exercise SBP change = 9.3, p = 0.002).

In the sex-stratified analysis, the association between exercise SBP response and all-cause death was significant in both men (unadjusted hazard ratio (HR) 0.79 per 10 mm Hg increment in exercise SBP change; 95% CI 0.71 to 0.88; p<0.001) and women (unadjusted HR 0.86; 95% CI 0.75 to 0.99; p = 0.034). After adjustment for pre-exercise SBP, ischaemic heart disease, hypertension and cardiovascular medication, the corresponding HR for men was 0.80 (95% CI 0.71 to 0.91; p<0.001) and for women 0.85 (95% CI 0.73 to 1.00; p = 0.046). For cardiovascular mortality, the unadjusted HR was 0.70 (95% CI 0.59 to 0.83; p<0.001) for men and 0.81 (95% CI 0.66 to 0.99; p = 0.043) for women. After adjustment for the above-mentioned covariates, the corresponding HRs were 0.71 (0.59 to 0.87; p = 0.001) for men and 0.79 (0.63 to 1.00; p = 0.05) for women.


The principal finding in the present study was that a higher SBP response during exercise was associated with an improved long-term survival in 75-year-old men and women. This association was independent of important clinical and echocardiographic risk factors, pre-exercise SBP and exercise capacity.

While systemic systolic hypertension is a well-established risk factor for cardiovascular morbidity and mortality,21 the prognostic significance of an exaggerated SBP increase during exercise is less clear. Our findings add further insight to the results of previous studies that focus on male patients. In a study on 1586 men referred for symptom-limited treadmill testing, Irving et al reported an annual rate of sudden death decreasing from 9.8% to 2.5% to 0.7% as the range of maximal exercise SBP increased from <140 to 140–199 to ⩾200 mm Hg, respectively.8 Morrow et al demonstrated that a greater SBP increase during a standard treadmill exercise test was an independent predictor of cardiovascular survival among 2546 male patients referred for routine clinical exercise testing.9 Recently, Gupta et al showed an incremental benefit in overall survival with a greater increase in SBP during treadmill exercise testing among 6145 consecutive male patients.10 During a mean follow-up of 6.6 years, they found a 23% improvement in survival for patients with an exercise SBP change above a median value of 44 mm Hg compared with those below.

Considering the haemodynamic components of blood pressure, a weaker SBP rise during exercise could be a consequence of an attenuated increase in the cardiac output or of a more powerful reduction in systemic vascular resistance, or both causes combined. Failure to reduce systemic vascular resistance during exercise is a strong indicator of poor prognosis.22 Thus, the weaker SBP increase during exercise among non-survivors in the present study may mainly reflect an attenuated cardiac output response rather than an exaggerated reduction in systemic resistance. Ischaemic or hypertensive heart disease may cause such an inadequate cardiac output response. In populations with a high prevalence of ischaemic heart disease, an exaggerated SBP response during exercise has been associated with a lower prevalence of myocardial perfusion abnormalities23 and lower mortalities.24 Failure to increase SBP by more than 30 mm Hg is an independent predictor of worse outcome in patients after myocardial infarction.25 26 Interestingly, in a meta-analysis of more than 8000 heart failure patients with a mean LVEF of 28%, Raphael et al reported a 13% reduction in relative mortality risk for every 10 mm Hg increase in SBP.27 They evaluated resting instead of exercise SBP, but the findings highlight the prognostic impact of SBP and its physiological relation with cardiac function.

In the present study, the SBP change during exercise remained a strong predictor of prognosis, even after adjustment for prevalent ischaemic heart disease, hypertension, left ventricular mass and left ventricular systolic function. However, the prevalence of occult cardiac disease is probably considerable in this aged population. Notably, the occurrence of ischaemic exercise ECG responses was not significantly associated with outcome in the present study. This finding is supported by a recent study reporting that exercise induced ST-depression is a weak predictor of outcome in older people.28

Age itself has profound effects on the cardiovascular response to maximal exercise. Exercise capacity, cardiac output, heart rate and LVEF are reduced, whereas left ventricular volumes, blood pressure and systemic vascular resistance are increased during aerobic exercise in elderly compared with younger individuals.2931 It can be speculated that the SBP response during exercise has a stronger positive correlation with cardiac output in older than in younger subjects because of a weaker attenuation of blood pressure caused by less reduction in systemic resistance.

In contrast to our findings, Filipovsky et al found that an exaggerated SBP change from rest to a workload of 164 W during bicycle exercise was associated with a significantly higher all-cause and cardiovascular mortality in 4907 men with a mean age of 47.5 years.6 An association between higher SBP at a workload of 600 kpm/min (approximately 100 W) and increased cardiovascular mortality was shown by Mundal et al in 2014 apparently healthy men aged 40 to 59 years.7 Furthermore, in a population-based study of 1731 middle-aged men free from cardiovascular disease at baseline, Laukkanen et al demonstrated a 2.5-fold increased relative risk of future myocardial infarction in those with a maximum SBP during exercise of >230 mm Hg.4 Discrepancies between study results may be explained by differences in characteristics of the study population and methodology. The above-mentioned studies showing that a higher SBP during exercise is a detrimental factor for outcome differ from our study in that they examined younger subjects in whom cardiovascular disease had been excluded at baseline. In addition, some of the studies found significant relations between exercise SBP and outcome only when submaximal exercise SBP, in contrast to peak exercise SBP, was evaluated.6 7 In the present study, we measured the SBP in intervals varying between 2 and 3 min during the exercise test, making the submaximal SBP response difficult to evaluate in a standardised manner.

There are several limitations to be considered. The fact that all participants were of the same age limits the findings’ generalisability to other age groups. On the other hand, it eliminates the need to adjust for this strong confounder and is therefore a strength of the study. Non-participation bias is a possible limitation. From the original sample of 618 invited individuals, 38% were missing in the final cohort. Most of those who were not studied rejected the offer to participate. Non-participants were significantly more likely to be women than the participants (62% vs 50%, p = 0.024). Further, after the median follow-up of 10.6 years, the all-cause mortality was significantly higher among non-participants than participants (62% vs 37%; p<0.001) most likely representing a selection towards healthier subjects in the participating cohort. Another possible limitation in the present study is that measurement of SBP may be inaccurate during exercise. However, measurement error would tend to create conservative bias, unlikely to explain the main findings of the study. The data analysed in the present study are from cycle ergometer tests, and the results could not be directly extrapolated to treadmill tests. Finally, data on therapeutical interventions during the follow-up were lacking. If therapy with effect on survival was initiated during the follow-up time, and this therapy in some way was associated with exercise SBP change, it may have been a source of bias.


Our findings suggest that an exaggerated SBP change from baseline to peak exercise is a strong independent predictor of survival in community-living 75-year-olds.


We are indebted to M-L Ojutkangas, P Wahlén, T Wiklund and M-L Engström for their help in collection and processing of the data.


View Abstract


  • Funding: The Västmanland Research Fund against Cardiovascular Disease.

  • Competing interests: None.

  • Ethics approval: Ethics approval was provided by the Ethics Committee of Uppsala University, Sweden.

  • Patient consent: Obtained.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.