Patterns of systolic blood pressure response at the end of exercise and mortality and morbidity in patients referred for exercise testing

Study cohort

From a clinical database of consecutive cycle ergometer exercise tests in 13 763 patients, we included subjects aged ≥18 years with maximal tests (rating of perceived exertion (RPE) ≥17 on the Borg scale) and exercise duration of 6–14 min. Tests with no SBP recordings during the last two minutes of exercise and fewer than two SBP readings during the last third of the test were excluded. We also excluded subjects with a body mass index (BMI) >40 kg/m² or arrhythmia during the test, where reliable SBP measurements could be difficult to obtain, as well as those with an SBP at rest <80 or >180 mm Hg (figure 1).

The data were cross-linked with nationwide registries on mortality, in- and outpatient hospital diagnoses, and current medications, as detailed elsewhere,3 20 used for cohort characterisation and outcomes during follow-up.

Exercise testing

All exercise tests were performed on an electrically braked bicycle (Rodby Inc, Karlskoga, Sweden). The testing procedure has been described elsewhere.20 In brief, a ramp protocol was used, in most cases starting at 30 W in women and 50 W in men, followed by a continuous and consistent increase by 10 W/min in women and 15 W/min in men until maximal exertion, aiming at a total exercise time of 8–12 min.1 The peak work rate in W (Wpeak) was recorded for each individual and was compared with normative values (Wpeak%pred), based on age, sex and height.21

Blood pressure (BP) measurements

Resting SBP and diastolic blood pressure (DBP) were measured in the supine position, before exercise after a few minutes of rest, with a stethoscope over the brachial artery and manual cuff inflation/deflation. Sitting on the bicycle, SBP was measured in the right arm using a Doppler probe over the radial artery, with manual cuff inflation/deflation, before the start of exercise and every 2–3 min during cycling. The highest SBP measured during exercise was defined as SBPpeak. We also calculated the SBP/W-slope over the entire exercise phase, by dividing the difference between SBP at supine rest and at peak exercise (last SBP measurement), by the increase in W between the same time points.5

SBP response patterns

The SBP response at the end of exercise was calculated as the difference between the two last SBP measurements, standardised to the difference in W between the same time points, mm Hg/W (figure 2). If there was less than 1 min between the two last SBP measurements, the third-to-last SBP measurement was used instead of the second-to-last. The SBP response patterns at the end of exercise were characterised as: drop (negative change), plateau (no change in SBP) or increase (positive change) in SBP. Any SBP increase was further divided into tertiles labelled slow, intermediate and steep increase (figure 2).

Ethics

The study complied with the Declaration of Helsinki and was approved by the regional Ethical Review Board in Linköping, Sweden (2018/141-31, 2019–05739). Informed consent was waived for this analysis of already collected data.

Statistics

The data were analysed using IBM SPSS Statistics (V.29). Comparisons of continuous data between multiple groups were performed using one-way analysis of variance. Where significant differences were found across groups, Tukey post hoc tests were performed to analyse differences compared with the reference group (slow SBP increase). Proportions were compared between groups using Pearson’s χ² tests. Associations between SBP response pattern and outcomes were analysed using Cox proportional hazards regression. Hazard ratios (HRs) were presented with 95% confidence interval (CIs). Model 1 was unadjusted. Model 2 was adjusted for age, sex, BMI, hypertension, diabetes mellitus, history of CVD (heart failure (HF), ischaemic heart disease (IHD) or cerebrovascular disease) and medication with beta-blockers. In model 3, additional adjustment was made for exercise capacity (Wpeak). The SBP response at the end of exercise was analysed as a continuous variable and as the difference in risk between the SBP response categories, using a slow increase as the reference.

The primary outcome was all-cause mortality. The secondary outcome was incident CVD, defined as a new diagnosis of IHD (diagnoses I20–I25 according to the International Classification of Diseases (ICD)-10),22 HF (ICD-10 diagnose I50) or CV death (any circulatory disease coded as the underlying cause of death in the mortality registry). For analyses of the secondary outcome, patients with a baseline diagnosis of IHD, HF or valvular heart disease were excluded (figure 2).

A two-sided significance level of 0.05 was used.

Equity, diversity and inclusion statement

The author group consisted of junior and senior researchers from three different countries. The first author is a woman, while the rest of the team consists of men. The study cohort included real-world patients referred for exercise testing with both sexes represented in similar proportions and a wide age range. We did not examine the effects of ethnicity or socioeconomic status.

Patient and public involvement statement

In this study, including retrospectively collected data, patients and the public were not involved in the planning of the research.

Systolic blood pressure (SBP) at the end of exercise

The last SBP measurement was recorded on average 14.2±22.4 s before the end of the work phase, with a mean time difference between the two last SBP recordings of 100.5±36.5 s.

In 5311 patients (83.9%), there was an increase in SBP between the two last measurements, whereas in 950 patients (15.0%), there was a plateau, and in 68 patients (1.1%), there was a drop (table 1).

Compared with patients with a slow SBP increase at the end of exercise, resting SBP before exercise was higher in patients with a drop, plateau or a steep SBP increase, while SBPpeak was higher in patients with a steep SBP increase. The exercise capacity (Wpeak) was highest in patients with a slow SBP increase, although % of predicted Wpeak was similar in patients with a slow increase and a plateau (table 1). Online supplemental table 1 shows baseline comorbidities in the whole sample.

Systolic blood pressure (SBP) response pattern and mortality

There were 402 (6.4%) deaths during follow-up, with an incidence rate of 7.3 per 1000 person-years. There was a positive association between SBP increase at the end of exercise and mortality in the unadjusted model but no significant association in the fully adjusted model (HRadjusted 1.06 (95% CI 0.85 to 1.32) per 1 mm Hg/W increase) (online supplemental table 2). In a sensitivity analysis including only patients on beta-blockers, the risk estimates were further attenuated (online supplemental table 3).

In the unadjusted model, steeper SBP increases at the end of exercise were associated with increased mortality risks, as compared with a slow SBP increase. Similarly, after adjustment for demographics and baseline comorbidity (model 2), intermediate and steep SBP increases were associated with all-cause mortality (HRs 1.36–1.39, p <0.05), while after additional adjustment for exercise capacity, there were no significant differences in mortality risks between any of the SBP response categories (table 2).

Systolic blood pressure (SBP) response pattern and incident cardiovascular disease

In the 5394 patients with no cardiac disease at baseline (figure 1), there were 715 incident cases of IHD, HF or CV death during follow-up of 8.0±3.7 years, distributed between the SBP response categories as follows: drop n=16 (34.0%), plateau n=111 (13.3%), slow increase n=179 (10.8%), intermediate increase n=168 (12.3%) and steep increase n=241 (16.2%).

There was a positive association between SBP increase at the end of exercise and incident CVD in the unadjusted model, but no significant association in the fully adjusted model (HRadjusted 0.88 (0.74–1.05) per 1 mm Hg/W increase) (online supplemental table 2).

A drop in SBP at the end of exercise was associated with CVD (HRadjusted 3.10 (95% CI 1.85–5.19)). There was a trend towards a risk increase for increasing (intermediate or steep) SBP at the end of exercise and incident CVD, as well as for a SBP plateau, but in the adjusted models, there were no significant differences in risks for incident CVD between a plateau, intermediate or steep SBP increase, when compared with a slow SBP increase (table 2).

Survival curves and the unadjusted and adjusted risks of mortality and CVD for the different SBP response categories are seen in figure 3.

Slow systolic blood pressure (SBP) increase at the end of exercise

During exercise, CO must increase to meet the greater metabolic demands of the working muscles. The augmentation in CO is normally generated by a combined increase of stroke volume (SV) and heart rate (HR). At the same time, there is an active vasodilation in the arteries of the working skeletal muscles, generating a minor reduction in TPR. We anticipated that a lack of SBP increase during the final phase of incremental exercise, manifesting as either a drop, a plateau or perhaps even a slow SBP increase, would indicate inadequate increase of the CO and inability to ‘fill’ the dilated circulation during exercise, which could imply risk of future CVD or death.

Rather, given the low mortality rate in patients with a slow SBP increase, as well as the higher absolute exercise capacity and the lower mean age as compared with patients with an SBP plateau, it could be argued that a slow SBP increase at the end of exercise reflects a healthy vascular response. By being able to appropriately reduce TPR, a healthy vascular system could accommodate the increase in BP generated by increasing CO. In the current study, patients with a slow SBP increase at the end of exercise tended to have the lowest prevalence of hypertension and hyperlipidaemia, while the prevalence was substantially higher in patients with a steep SBP increase (table 1), supporting a healthier vascular state with a levelling off of the SBP.

Importantly, a slow SBP increase at the end of exercise, as studied here, should be distinguished from a low absolute SBPpeak, which has been found to be associated with increased mortality risk.11 16 In this study, the absolute SBPpeak did not differ significantly between patients with a plateau, slow or intermediate SBP increase, while patients with a steep SBP increase at the end of exercise had a higher SBPpeak.

Furthermore, although both HR and SV in healthy subjects may continue to increase throughout exercise until reaching VO₂max,23 at maximal exercise intensity, there is a physiological plateau of the VO₂ despite further increase in intensity. An absent or slow SBP increase at the end of such maximal exercise may therefore in some subjects indicate an ability to continue exercising at, or above, maximal aerobic capacity, with a plateau in both VO₂ and SBP. Studies including cardiopulmonary exercise test data would further elucidate this hypothesis.

It should also be noted that the proportion of women in patients presenting with a slow SBP increase in this study was lower than within the other SBP response categories. It has previously been shown that healthy women have a steeper SBP/W-slope when calculated over the whole exercise phase.3 Sex-specific cut-offs may be needed in future studies of the SBP response at the end of exercise, to better characterise potential sex differences in SBP response patterns.

Steep systolic blood pressure (SBP) increase at the end of exercise

A steep SBP increase at the end of exercise was associated with mortality and CVD in the unadjusted models and was more often seen in older patients with a higher prevalence of baseline comorbidity, including IHD, hypertension, diabetes and hyperlipidaemia. This is in line with previous studies showing that an exaggerated peak BP is more often seen in subjects with impaired endothelial vasodilator function.24

In the current study, patients with a steep SBP increase at the end of exercise also reached the highest absolute SBPpeak among all studied SBP response categories. This, as well as the finding of a low exercise capacity in comparison with the other studied groups, may be explained by both age3 and CV comorbidity. A recent meta-analysis summarised the higher occurrence of CV risk factors (including arterial structure, lipid, metabolic, inflammatory and kidney function markers) among patients with an exaggerated exercise SBP response compared with normal responders, supporting a hypertensive response to exercise being an indicator of CV risk.25 Our results suggest that the same may also be true for the SBP trajectory at the end of exercise, where a steep SBP increase reflects the underlying CV risk factor profile.

Systolic blood pressure (SBP) drop at the end of exercise

A drop in SBP at the end of a maximal exercise test was infrequent (1%) but independently associated with CVD. Hence, our results are consistent with previous assumptions that a drop in SBP is a strongly negative marker that may indicate pathology such as severe ischaemia26 27 and in line with guidelines advocating discontinuation of an incremental exercise test in case the SBP drops >10 mmHg.19 We found no significant association between a SBP drop and mortality, yet the absolute number of patients and events in that group was low and the CIs wide. Of note, this study included only maximal exercise tests (RPE ≥17 and a duration of ≥6 min). Consequently, patients with a drop in SBP at submaximal exercise were excluded per protocol.

A plateau in systolic blood pressure (SBP) at the end of exercise

Our initial assumption was that a ‘flat response’ could be a milder manifestation of the same mechanisms underlying an SBP drop. The results were somewhat in line with this hypothesis, since a trend towards increased mortality and CVD risk was observed in patients with a SBP plateau in comparison with patients showing a slow SBP increase, yet the associations were not statistically significant in the fully adjusted models (table 2). This trend suggests that there might be a small-effect size association and that the a priori hypothesis merits further research in other cohorts to confirm or contrast these findings.

Clinical implications

SBP is routinely measured during clinical exercise testing, and an increased understanding of how to interpret different SBP responses is of great relevance to clinicians and exercise physiologists. Our findings support that only a drop in SBP had a strong and robust association with an increased risk of incident CVD, beyond that of traditional CV risk factors and exercise capacity. Further, we observed a mildly and non-significantly increased risk in patients experiencing a plateau in SBP at the end of the exercise test, which should be further investigated. We believe that this study contributes to the conceptual understanding of different SBP responses during exercise, beyond evaluation of the SBPpeak and their relevance in a clinical exercise testing context.

Limitations

We lack data on smoking habits in the current cohort, which may have an impact on both the SBP response and the outcome. Furthermore, the cut-offs for a slow and a steep SBP increase at the end of exercise are derived from the current clinical population of patients referred to exercise testing, which limits their generalisability. As the prevalence of women in the group with slow SBP increase at the end of exercise was low, the findings related to a slow SBP increase should be interpreted with caution in women. To further address this, sex specific cut-offs may be needed in future studies.

Conclusion

In this study of different trajectories of the SBP during the final phase of a maximal exercise test in patients referred to exercise testing, a drop in SBP was the only pattern independently and strongly associated with incident CVD. There was a trend towards higher mortality and CVD risk also for a plateau in the SBP response, although the effect size was small and did not reach statistical significance in the adjusted models. No statistically significant differences were observed for mortality risk among the different SBP trajectories, yet the lowest risk estimate was observed in those with slow SBP increase. Our findings illustrate the importance of interpreting the SBP response in the context of patient characteristics including comorbidities, age, sex and exercise capacity.