Atrial Fibrillation, Venous Thromboembolism, and Risk of Pulmonary Hypertension: A Swedish Nationwide Register Study

Abstract

Background

Atrial fibrillation (AF) is suggested to be associated with venous thromboembolism (VTE) and might thereby play a role in the development of chronic thromboembolic pulmonary hypertension (PH). By elevating left‐sided filling pressure and promoting atrial myopathy, AF may also play a role in the development of postcapillary PH. We aimed to investigate the association between AF, with or without incident VTE, and the occurrence of PH.

Methods

A total of 521 988 patients diagnosed with AF between 1987 and 2013, without a previous diagnosis of VTE or PH, were identified from the Swedish National Patient Register and matched for age, sex, and county with 1 017 277 population controls without AF, VTE, or PH.

Results

The mean age of the patients with AF was 71.1 (SD ±10.1) years and 42.8% were women. During a median follow‐up period of 11 (interquartile range 5.1–17) years, 4454 (0.9%) patients with AF, and 1855 (0.2%) controls were diagnosed with PH, hazard ratio 4.7 (4.4–5.0). The AF group had a significantly higher comorbidity burden at baseline, with a mean CHA₂DS₂‐VASc of 2.9 compared with 2.1 in controls. In the absence of intercurrent VTE, the hazard ratio of PH was 4.2 (3.9–4.5) among patients with AF compared with controls. Intercurrent VTE increased the hazard ratio of PH a further 1.9‐fold (1.7–2.1) and 3.5 (3.1–4.0), among patients with AF and controls, respectively. The hazard ratio for PH in patients with AF with incident VTE was 8.1 (7.3–9.1).

Conclusions

AF was associated with a markedly increased risk of developing incident PH, and this risk was further increased by incident VTE.

Nonstandard Abbreviations and Acronyms

NPR

Swedish National Patient Register

PAH

pulmonary arterial hypertension

PH

pulmonary hypertension

Clinical Perspective.

What Is New?

• This study demonstrates that atrial fibrillation is associated with the risk of developing pulmonary hypertension, with incident venous thromboembolism further amplifying this risk.

What Are the Clinical Implications?

• Clinically, these findings highlight the need for closer monitoring and early intervention in patients with atrial fibrillation to mitigate the progression to pulmonary hypertension, especially in those with concurrent venous thromboembolism.

Atrial fibrillation (AF) has been found to create a prothrombotic state and contribute to endothelial injury by triggering inflammation and oxidative stress ¹ and is a well‐established cause of arterial thromboembolic events, primarily ischemic stroke. It has been suggested that AF might also increase the risk of venous thromboembolism (VTE) ² , ³ and pulmonary embolism (PE). ⁴

An association between AF and VTE has been found in previous studies. ⁵ , ⁶ , ⁷ However, this association may be attributed to the numerous risk factors they have in common. ⁸ Recent data from our research indicate that the risk of VTE, and particularly PE, is comparable to the risk of ischemic stroke in the initial months following an AF diagnosis, and is especially pronounced among younger patients and women. ⁹ It is speculated that the increased risk of VTE in AF may be mediated by multiple mechanisms, in accordance with Virchow's triad, ¹ including stasis in the caval and lung circulation with secondary in situ thrombosis, as well as dislodged thrombus from the right side of the heart. Impaired thrombus resolution, potentially influenced by genetic factors that alter fibrin structure and increase fibrinolytic resistance, ¹⁰ , ¹¹ is believed to be a key factor in the development of chronic PE and secondary pulmonary hypertension (PH). ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁶ Thus, it is reasonable to assume that AF may enhance the risk of developing PH over time. In a recent study, ¹⁷ Lutsey et al. reported that the incidence of PH within 2 years is 3.5% after a VTE event and 6.2% following a PE event. In many cases, however, incident PE was never diagnosed or clinically apparent.

AF can also lead to increased filling pressures in the left atrium, backward stasis into the pulmonary veins, and elevated pulmonary artery pressure, contributing thereby to the development of postcapillary PH. Moreover, by promoting atrial matrix reconstruction ¹⁸ and fibrosis, as well as ventricular dysfunction, longstanding AF may enhance the risk of postcapillary PH. ¹⁹

The relationship between AF and the development of PH is thus intricate, since PH in its turn can increase the risk of AF by inducing right atrial dilation and electrical disruption, forming a vicious circle. Prognosis in PH is ominous, and especially with concomitant AF. ²⁰ , ²¹ Furthermore, recent results ²² , ²³ show that PH independently increases the incidence of all types of strokes in patients with AF.

However, AF has not been studied as an independent risk factor for PH and the relative contribution of VTE/PE to the development of PH in patients with AF is not elucidated.

We hypothesized that a diagnosis of AF increases the risk of developing incident PH and that this risk may be influenced or exacerbated by the presence of VTE.

The primary aim of this study was to assess the occurrence of incident PH in patients with a diagnosis of AF compared with a matched control population without AF; a secondary aim was to investigate the association of incident VTE to PH in patients with AF versus control subjects.

METHODS

The data that support the findings of this study are available from Martin Adiels (author responsible for statistical analyses) upon reasonable request.

Study Cohort

The study cohort was derived from the Swedish National Patient Register (NPR), which records all primary and secondary inpatient diagnoses in Sweden since 1960. The register has had full national coverage since 1987. In addition, since 2001, all outpatient visits to all hospitals in the country are also recorded in the NPR. Sweden provides universal health care to all citizens at a low cost, which results in complete coverage of the Swedish population in NPR. Hospital diagnoses of major cardiovascular conditions in Sweden have been shown to have high validity.

From the NPR, we retrieved data on all reported first cases of AF, as a primary or as a secondary diagnosis, (ie, AF diagnosis identified in the NPR in any position), between January 1, 1987 and December 31, 2013. A first case was defined as no prior diagnosis of AF during the last 7 years. AF was defined as International Classification of Diseases (ICD) codes 427D, and I48 for ICD ‐ 9, and ICD ‐ 10, respectively. All cases were matched by sex, age, and county to 2 control individuals without AF from the Swedish Population Register.

Data on several baseline comorbidities were also retrieved from the NPR and a comorbid condition was defined as present at baseline if there was at least 1 primary or secondary diagnosis in the 7 years preceding the AF diagnosis. ICD codes for baseline comorbidities are listed in Table S1. The Cause of Death register was used to record dates and causes of death.

AF cases were excluded together with their matched controls if vital data were inconsistent or if death occurred during the hospital stay. Control individuals were assigned the same inclusion date as their case.

The final data set was obtained by excluding individuals with prior (up to the same day as AF diagnosis) diagnosis of VTE and PH. See flow chart (Figure S1) for details on selection criteria for the study population.

The National Prescribed Drug Register

Since mid‐2005 all drug prescriptions in Sweden are recorded in the Prescribed Drug Register. For the cohort of patients with AF and controls with inclusion dates between January 1, 2006 and December 31, 2013, data were retrieved from the prescribed drug registry on filled prescriptions of platelet inhibitors (aspirin, clopidogrel, prasugrel), warfarin, and direct‐acting oral anticoagulants, as well as other relevant drugs (angiotensin enzyme inhibitors/angiotensin receptor blockers, Class I and III antiarrhythmic drugs, β‐blockers, digitalis, diuretics, lipid‐lowering drugs, and chronic obstructive pulmonary disease [COPD] medications). Patients and controls were defined to be on treatment at baseline if there were at least 2 filled prescriptions for the drug within 1 year before the AF diagnosis. See Table S2 for anatomical therapeutic chemical codes.

Definitions of Outcomes

Pulmonary arterial hypertension (primary PH) was defined as a main diagnosis of 416A (ICD9) or I27.0 (ICD10). Secondary PH was defined as any diagnosis of 416C, 416W, 416X (ICD9) or I27.2, I27.8, I27.9 (ICD10).

Statistical Analysis

Patients with AF and matched control subjects were followed until (1) a first event of PH, (2) death, or (3) end of study (December 31, 2013). Events of VTE were recorded for time‐updated models, as explained below.

Crude incidence rates and 95% CI were calculated using Poisson regression. Cox regression was used to assess the risk for PH following AF diagnosis. Unadjusted (only adjusted for age and sex) and adjusted (age, sex, and baseline comorbidities: diabetes, hypertension, ischemic heart disease, and heart failure [HF]) models were tested. To improve accuracy and clinical interpretability of the model, effect of incident VTE was included in the model as a time update (ie, at the time‐point when a VTE occurred). This implies that the model reflects the natural progression of risk associated with the occurrence of VTE. Ignoring its timing would misrepresent the risk profile.

To assess the interaction between AF and comorbidities, and the risk of PH we included interactions between AF and the comorbidities and used contrasts to compute specific hazard ratios (HRs).

In a separate cohort from January 1, 2006 to December 31, 2013, the models were further adjusted for the use of warfarin/direct‐acting oral anticoagulants, aspirin, or clopidogrel at baseline.

The study was approved by the Regional Ethics Review Board in Gothenburg (Dnr. 104‐15) and because the results are based on anonymized data, the need for patient consent was waived. M. Adiels had full access to all the data in the study and takes responsibility for its integrity and the data analysis.

RESULTS

Baseline

A total of 521 988 first AF cases (age 35–85 years) without a previous PH or VTE diagnosis were identified and matched for age and sex with 1 017 277 non‐AF controls; the mean age was 71.1 (SD ±10.1) and 71.0 (SD ±10.1) years for AF and controls, respectively. Heart failure, COPD, ischemic heart disease, adult congenital heart disease, previous myocardial infarction, valvular heart disease, and most other cardiopulmonary diagnoses and risk factors were more prevalent at baseline in the AF group, as was cancer (Table 1). The AF population had higher median CHA₂DS₂‐VASc score compared with the control group, 3 [2, 4] and 2 [1, 3], respectively.

Table 1.

Baseline Characteristics in Patients With Atrial Fibrillation and Age‐ and Sex‐Matched Controls from the General Population

	Atrial fibrillation n=521 988	Population controls n=1 017 277
Women	223 337 (42.8)	436 181 (42.9)
Age (y), mean±SD	71.1±10.1	71.0±10.1
Heart failure	119 889 (23.0)	34 251 (3.4)
Chronic obstructive pulmonary disease	30 886 (5.9)	25 537 (2.5)
Valvular disease	38 768 (7.4)	11 690 (1.2)
Prior myocardial infarction	60 046 (11.5)	40 251 (3.9)
Ischemic heart disease	135 577 (26.0)	96 231 (9.5)
Congenital heart disease	2471 (0.5)	873 (0.1)
Ischemic stroke	57 610 (11.0)	40 473 (4.0)
Transitory ischemic attack	21 100 (4.0)	18 396 (1.8)
Hemorrhagic stroke	4582 (0.9)	6485 (0.6)
Hypertension	158 112 (30.3)	110 241 (10.8)
Diabetes	70 482 (13.5)	65 068 (6.4)
Renal failure	14 790 (2.8)	9055 (0.9)
Asthma	18 443 (3.5)	16 889 (1.7)
Peripheral artery disease	4367 (0.8)	3942 (0.4)
Obesity	7711 (1.5)	4394 (0.4)
Cancer at baseline	61 670 (11.8)	96 303 (9.5)
CHA2DS2VASc, median [IQR]	3 [2–4]	2 [1–3]

Baseline characteristics of the whole study population; data shown as n (%) if not otherwise specified. For definitions (International Classification of Diseases [ICD] codes), see Table S1. All analyzed clinical characteristics (except for age and sex used for group matching) showed significant differences (P [lt]0.001) between patients with atrial fibrillation and controls. IQR indicates interquartile range.

PH During Follow‐Up

During a median follow‐up of 11 ((interquartile range 5.1–17) years, 4454 (0.9%) PH events were recorded in the AF group compared with 1855 (0.2%) in the control group. Of these, 308 (0.06%) and 141 (0.01%) were primary PH (pulmonary arterial hypertension [PAH]), while 4146 (0.8%) and 1714 (0.2%) consisted of secondary or unspecified PH in the AF and control cohorts, respectively (Table 2). Thus, of all PH cases recorded during the study period, 6.9% and 7.6% consisted of PAH in the AF and control cohorts, respectively, while the remaining cases had secondary/unspecified PH.

Table 2.

Pulmonary Hypertension Event Rates Registered for Patients With Atrial Fibrillation and Age‐ and Sex‐Matched Controls from the General Population

	Atrial fibrillation n=521 988	Population controls n=1 017 277
Pulmonary hypertension
Events, total	4454 (0.9)	1855 (0.2)
Event rate	75.2 (73.0–77.4)	16.0 (15.3–16.7)
Follow‐up, median [IQR], y	11 [5.1, 17]	11 [5.2, 17]
Number of deaths*	637 (0.1220)	396 (0.0373)
Pulmonary arterial hypertension (group 1)
Number of events (%)	308 (0.06)	141 (0.01)
Event rate	5.2 (4.6–5.8)	1.2 (1.0–1.4)
Follow‐up, median [IQR], y	11 [5.1, 17]	11 [5.2, 17]
Number of deaths*	24 (0.00460)	26 (0.00245)
Pulmonary hypertension (group 2–5)
Number of events (%)	4146 (0.8)	1714 (0.2)
Event rate per 100 000	70.0 (67.9–72.2)	14.8 (14.1–15.5)
Follow‐up, median [IQR], y	11 [5.1, 17]	11 [5.2, 17]
Number of deaths*	613 (0.1174)	370 (0.0349)

Crude events, event rates per 100 000 person y (95% CI), and median follow‐up time (y).

IQR indicates interquartile range.

^*Death reported within 30 days of diagnosis.

Plots of cumulative events showed a greater risk for PH among AF cases compared with controls (Figure 1). In a Cox proportional regression analysis, the HR for PH in AF cases versus non‐AF controls was 4.7 (4.4–5.0), in the model adjusted for age and sex, and 3.6 (3.4–3.9), in the model adjusted for relevant comorbidities, as shown in Figure 2.

Figure 1. Plots of cumulative events of pulmonary hypertension in atrial fibrillation cases and matched controls.

Y‐axis represents cumulative events of PH in patients without prior VTE. AF indicates atrial fibrillation; PH, pulmonary hypertension; and VTE, venous thromboembolism.

Figure 2. Hazard ratios for pulmonary hypertension in atrial fibrillation cases and controls.

The unadjusted models were adjusted for age and sex only. *Models adjusted for COPD/bronchial asthma, diabetes, systemic hypertension, ischemic heart disease, heart failure, valvular disease, and CKD at baseline. Subjects were censored if a venous thromboembolism, VTE‐event, occurred. AF indicates atrial fibrillation; COPD, chronic obstructive pulmonary disease; CTRL, control; PH, pulmonary hypertension; and VTE, venous thromboembolism.

VTE as Risk Factor for PH

To elucidate the association of incident VTE to PH in patients with AF and controls, we developed a time‐dependent Cox regression model. The model was updated when a subject had an incident VTE, thus allowing us to estimate the HR for PH with and without an incident VTE. There was a significant interaction between AF and incident VTE (beta −0.59, P value [lt]0.001) (Table 3). Unadjusted results are shown in Table S5.

Table 3.

VTE as a Risk Factor for Pulmonary Hypertension, Adjusted for Relevant Comorbidity (COPD/Bronchial Asthma, Diabetes, Systemic Hypertension, Ischemic Heart Disease, Heart Failure, Valvular Disease, and CKD)

	Multivariable	Multivariable analysis with interaction	P value for interaction
Main effects
AF vs controls	3.68 (3.46–3.92)	3.97 (3.72–4.24)*	[lt]0.0001
VTE vs No‐VTE	2.35 (2.17–2.55)	3.48 (3.05–3.97)†
Within controls
VTE vs No‐VTE	2.35 (2.17–2.55)	3.48 (3.05–3.97)
Within patients with AF
VTE vs No‐VTE	2.35 (2.17–2.55)	1.96 (1.77–2.17)
Contrasting groups
Controls w/o VTE	1.00 (1.00–1.00)	1.00 (1.00–1.00)
Controls with VTE	2.35 (2.17–2.55)	3.48 (3.05–3.97)
AF w/o VTE	3.68 (3.46–3.92)	3.97 (3.72–4.24)
AF with VTE	8.66 (7.84–9.57)	7.78 (6.97–8.69)

The multivariable model without interaction assumes an additive risk (on the log scale) for PH of HR 2.4 for incident venous thromboembolism, VTE (same for AF, and matched controls) and HR 3.7 for AF (same with and without incident VTE) adding up to HR 8.7 for AF patients with an incident VTE. The multivariable model with interaction allows for estimation of different HR for incident VTE in AF and in matched controls.

AF indicates atrial fibrillation; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; HR, hazard ratio; PH, pulmonary hypertension; and VTE, venous thromboembolism.

^*Main effect reported in subjects without incident VTE.

^† Main effect reported in matched controls.

Using the model with an interaction between AF and VTE we could estimate the risk for PH in controls and patients with AF. In controls with an incident VTE, the HR for PH was 3.48 (3.05–3.97), whereas patients with AF experiencing an incident VTE had a HR of 1.96 (1.77–2.17) for developing PH. However, AF by itself was associated with a HR of 3.97 (3.72–4.24). Thus, relative to controls without incident VTE, patients with AF with incident VTE had a HR of 7.78 (6.97–8.69) (Table 3).

Use of Medication at Baseline and Follow‐Up

No data on drug prescriptions were available before the National Prescribed Drug register came into operation in mid‐2005. Therefore, to assess the effect of medication at baseline, we investigated a subgroup diagnosed with a first episode of AF between 2006 and 2013, resulting in 197 177 patients with AF and 383 012 controls. Baseline characteristics did not differ from the total study population, except for hypertension and cancer, which were more prevalent in both the AF and control groups (Table 4). At baseline, medication with oral anticoagulants was more common in the AF group (Tables S3 and S4).

Table 4.

Baseline Characteristics of Subgroup With Available Data on Baseline Medication

	Atrial fibrillation	Population controls
	n=197 177	n=383 012
Women	81 068 (41.1)	157 951 (41.2)
Age (y), mean±SD	70.4±10.3	70.3±10.3
Heart failure	34 938 (17.7)	10 429 (2.7)
Chronic obstructive pulmonary disease	13 964 (7.1)	11 500 (3.0)
Valvular disease	16 768 (8.5)	5950 (1.6)
Prior myocardial infarction	19 980 (10.1)	13 576 (3.5)
Ischemic heart disease	48 192 (24.4)	37 651 (9.8)
Congenital heart disease	1236 (0.6)	500 (0.1)
Ischemic stroke	18 223 (9.2)	13 214 (3.4)
TIA/stroke	8126 (4.1)	7857 (2.0)
Hemorrhagic stroke	1769 (0.9)	2366 (0.6)
Hypertension	91 604 (46.5)	69 704 (18.2)
Type 2 diabetes	31 512 (16.0)	31 416 (8.2)
Renal failure	8350 (4.2)	4946 (1.3)
Asthma	7941 (4.0)	7852 (2.0)
Peripheral artery disease	1805 (0.9)	1470 (0.4)
Obesity	5545 (2.8)	3244 (0.8)
Cancer at baseline	32 037 (16.2)	49 073 (12.8)
CHA2DS2VASc, median [IQR]	3 [2–4]	2 [1–3]
Treatment with anticoagulants at baseline	24 641 (12.5)	3540 (0.9)

Data shown as n (%) if not otherwise specified.

IQR indicates interquartile range; and TIA, transient ischemic attack.

The pattern for PH in this subgroup was similar to the one observed in the total study population. Further adjusting the models for the use of medications at baseline provided marginally changed risk estimates for PH, compared with the total population; the most noticeable change was observed in the effect of incident VTE on PH (Table 5). Unadjusted results are shown in Table S5.

Table 5.

VTE as a Risk Factor for PH, Adjusted for Comorbidity (COPD/Bronchial Asthma, Diabetes, Systemic Hypertension, Ischemic Heart Disease, Heart Failure, Valvular Disease, CKD) and for Treatment With Anticoagulants^*

	Multivariable	Multivariable with interaction	Multivariable with interaction, adjusted for treatment
Main effects
AF vs controls	3.62 (3.24–4.06)	4.03 (3.58–4.55)†	3.22 (2.81–3.68)†
VTE vs No‐VTE	3.07 (2.64–3.57)	5.76 (4.52–7.35)‡	5.66 (4.44–7.21)‡
Within controls
VTE vs No‐VTE	3.07 (2.64–3.57)	5.76 (4.52–7.35)	5.66 (4.44–7.21)
Within patients with AF
VTE vs No‐VTE	3.07 (2.64–3.57)	2.32 (1.91–2.81)	2.41 (1.99–2.92)
Contrasting groups
Controls w/o VTE	1.00 (1.00–1.00)	1.00 (1.00–1.00)	1.00 (1.00–1.00)
Controls with VTE	3.07 (2.64–3.57)	5.76 (4.52–7.35)	5.66 (4.44–7.21)
AF w/o VTE	3.62 (3.24–4.06)	4.03 (3.58–4.55)	3.22 (2.81–3.68)
AF with VTE	11.14 (9.25–13.41)	9.35 (7.56–11.56)	7.75 (6.24–9.64)

^*See Table S5 for full description.

^† Main effect reported in subjects without incident venous thromboembolism, VTE.

^‡ Main effect reported in matched controls.

AF indicates atrial fibrillation; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; PH, pulmonary hypertension; and VTE, venous thromboembolism.

DISCUSSION

This national Swedish register study of patients with AF revealed 2 key findings: (1) AF was independently associated with a 4.7‐fold increase in the risk of developing PH over time, regardless of any prior VTE events, suggesting a potential role of AF in the pathogenesis of PH; and (2) a history of incident VTE in patients with AF doubled the risk of developing PH, highlighting the involvement of both postcapillary and precapillary mechanisms in the development of PH.

Even when adjusting for age and sex, the comorbidity burden in the AF population was notably higher compared with controls. This reinforces the notion that AF may serve as an indicator of overall poor health. Consequently, given the more pronounced cardiovascular and pulmonary comorbidities, it is expected that patients with AF would have a higher risk of developing secondary PH. However, the risk for PH remained significantly elevated even after adjusting for relevant comorbidities, indicating that AF itself is associated with the occurrence of PH.

The hemodynamics of patients with AF is characterized by increased diastolic filling pressures, atrial enlargement, and fibrosis, as well as fluctuating diastolic filling time, systolic ventricular stroke volume, and ejection fraction from beat to beat. Furthermore, lack of synchrony between the atria and ventricles results in fluctuating pressures, volume, and blood flow through the caval and pulmonary circulation. ²⁴ Also, mitral‐ and/or tricuspid insufficiency of varying size is common in AF. AF per se also induces atrial and ventricular enlargement, which adds to the above. These complex hemodynamic changes could lead to increased pericardial restraint and ventricular interdependence, abnormal right ventricle–pulmonary artery coupling, and negative effects on pulmonary vasculature. ¹⁹ AF may be both a cause and an effect of HF, irrespective of ejection fraction, but the exact contribution of different types of HF to the development of PH in this patient cohort was not possible to assess.

The second main finding, namely that an incident diagnosis of VTE doubled the risk for PH in patients with AF, is in line with the discussion above. It is crucial to highlight that this is in addition to the 4.7‐fold increased risk associated with AF alone. Therefore, the combination of AF and VTE results in an almost 9‐fold increased risk for developing PH.

AF induces a prothrombotic state by potentiating all 3 mechanisms of Virchow's triad, as described in detail by Watson et al. ¹ Abnormal changes in blood flow (blood stasis), abnormalities in vessels and myocardial walls (endothelial dysfunction), as well as in blood constituents (hemostatic and platelet activation, inflammation and growth factor changes) increase the risk for thromboembolism, ¹ , ²⁵ , ²⁶ thereby contributing to VTE development. ⁹ Furthermore, in a patient with AF, a pulmonary embolism may be misdiagnosed as a pulmonary infection, an asthma or COPD exacerbation, or HF, and therefore might go undetected. Less than half of the patients diagnosed with pulmonary embolism receive a systematic screening for deep vein thrombosis, while a significant portion of patients with deep vein thrombosis have an undetected PE. ²⁷ Thus, the prothrombotic effects of AF, combined with its detrimental impact on cardiac and pulmonary hemodynamics, may significantly increase the risk of PH occurrence.

Contemporary management of AF includes early initiation of anticoagulation based on CHA₂DS₂‐VASc score and other relevant prothrombotic risk factors. However, because this study encompassed a broad time frame, there was significant variability in treatment indications and therapy choices for AF across different periods. To tackle this issue and ensure a more homogeneous study population, a subgroup analysis was conducted specifically on patients treated with anticoagulants, yielding comparable findings.

Thus, we found that AF is associated with an increased risk of incident PH. As expected, most cases were classified as PH, not PAH. Achieving a deeper understanding of the complex interplay between AF and PH, particularly in the presence of other cardiopulmonary comorbidities, is essential. Strategies are needed for the prevention and early detection of PH in this patient group, alongside the aggressive management of AF to potentially reduce its downstream complications. This could include closer monitoring for early signs of PH development, and advancements in diagnostic tools and biomarkers to improve patient outcomes. Timely treatment of AF by adequate rhythm and/or rate control, including catheter ablation, and anticoagulation when indicated, is considered essential in reducing the risk of thromboembolic complications and/or HF according to current recommendations. Our findings underscore the need for further studies to explore the mechanisms linking AF and PH and to evaluate whether aggressive treatment of AF (eg, ablation/rhythm control strategies, anticoagulation) reduces the risk of PH development. If this is the case, guidelines for AF management might incorporate recommendations for assessing pulmonary pressures or symptoms indicative of PH.

Strengths and Limitations

Major strengths of the present study include a large number of AF cases and population‐based controls, near‐complete coverage of the Swedish population, and well‐validated data on diagnoses, medication, outcome, and confounders. Some limitations are the following: the control cohort, although age‐ and sex‐matched, is overall healthier, highlighting that AF is indeed a marker of significant comorbidity. Consequently, despite statistical adjustments and modeling, determining the precise contribution of AF to the development of PH remains challenging; data were obtained from several databases and were not primarily collected for research purposes; the incidence of PH is likely to be underestimated to an unknown but probably substantial extent, since PH may be underdiagnosed and/or less often recorded in patient notes; the use of routine echocardiography was much less common during the initial time of our data collection than currently, which might have contributed to underestimation of the PH diagnosis; additionally, some PH diagnoses might have been erroneously reported as PAH, since the incidence of PAH in our population is quite high in comparison to other published epidemiological data.

The accuracy of ICD coding can vary across healthcare settings. Misclassification or coding errors might have led to over‐ or underestimation of cases. Furthermore, ICD codes may not capture the full clinical picture, such as disease severity or temporal relationships.

We lack data on adherence to oral anticoagulant treatment, which is a limitation since there is a potential link between poor adherence and risk for thromboembolism. Unknown confounders, such as silent/undiagnosed AF among controls, may also be present, but the interference risk would be low, given the very large sample size.

Additional points to consider include that our data were collected between January 1, 1987 and December 31, 2013. During this period, large changes in the treatment of AF based on validated risk scores (CHA₂DS₂ and CHA₂DS₂‐VASc) were widely implemented, and mostly during the “pre‐direct‐acting oral anticoagulants” era. It is highly likely that the initiation and prevalence of oral anticoagulant treatment in these patients have significantly increased compared with previous periods. However, there are no data indicating that the effect or the bleeding risk of direct‐acting oral anticoagulants is different from warfarin in Sweden. This is attributed to Sweden's well‐established national system for monitoring warfarin treatments, which has been in place for many decades.

Conclusions

In this study, which included a large population of patients with AF and age‐ and sex‐matched controls, we observed that AF alone is linked to a 4.7‐fold higher risk of developing incident PH, in the absence of prior VTE. Additionally, the occurrence of new VTE events significantly elevates this risk to nearly 9‐fold. Established causes of PH, such as HF, valvular heart disease, and COPD further increase the risk in patients with AF.

This association is probably mediated by several mechanisms where AF‐related hemodynamic consequences and AF‐induced prothrombotic effects may play important roles. Further research is needed to elucidate the mechanisms of AF‐induced pulmonary vascular disease and to identify which patients with AF are most susceptible to developing PH.

Sources of Funding

This work was supported by grants from the following: the Swedish state under an agreement between the Swedish Government and the County Councils Concerning Economic Support of Research and Education of Doctors [grant number ALFGBG‐427301, ALFGBG‐966211, ALFGBG‐971608, ALFGBG‐979104]; the Swedish Heart and Lung Foundation [grant number 2015‐0438, 2018‐0589, 2021‐0345, 2023‐0724]; and the Swedish Research Council [2018‐02527, VRREG 2019‐00193].

Disclosures

None.

Supporting information

Tables S1–S5

Figure S1