Patient-reported effectiveness of strategies to convert paroxysmal atrial fibrillation to sinus rhythm in the home setting – a community-based participatory research approach

Abstract

Background

Patients with paroxysmal atrial fibrillation (AF) have limited options to terminate AF episodes in their home setting. The aims of this study were to assess patient-perceived effectiveness of self-management strategies used by patients to terminate paroxysmal AF and to investigate potential confounding factors, including patient characteristics, AF burden, and anti-arrhythmic drug use.

Methods

Utilizing a community-based participatory research approach, a survey instrument was developed and deployed to AF advocacy social media groups (> 50,000 members). The reported strategies were classified based on potential physiological mechanisms into hemodynamic, autonomic, exercise, meditation/yoga, cold or heat application, breathing techniques, electrolyte/fluid/dietary supplement intake, and other interventions. An effectiveness index (EI_Adj, range: 0.0 to 1.0), adjusted for confounding variables (patient characteristics, AF burden, anti-arrhythmic medication) was calculated based on how frequently the intervention was perceived to be successful and on the perceived latency of the effect. In generally healthy volunteers (n = 8), the hemodynamic responses to the combination of 500 mL ice-cold water intake, leg raising, and a modified diving response were assessed as an initial step to explore potential mechanisms and safety concerns.

Results

606 patients from 20 countries responded to the survey. Of these, 132 patients reported at least one strategy, totaling 259 unique strategies. Fluid/electrolyte/dietary supplement intake (n = 89; EI_Adj=0.44 ± 0.20, mean ± SD) and breathing techniques (n = 46; EI_Adj=0.44 ± 0.20, mean ± SD) were perceived as less effective than hemodynamic interventions (mostly leg raising; n = 30; EI_Adj=0.56 ± 0.21, mean ± SD, P < 0.05) and exercise (mostly brisk walking, running, or cycling; n = 41; EI_Adj=0.53 ± 0.17, mean ± SD, P < 0.05). Multiple regression analysis revealed that high age, large AF burden, and β-blocker use were associated with less effectiveness (R² = 0.14, F = 7.80 on 5 and 242 DFs, P < 0.001). The combined intervention increased stroke volume, cardiac output, and left ventricular end-diastolic volume and decreased heart rate in healthy volunteers.

Conclusions

Hemodynamic interventions and exercise were perceived as most effective in terminating paroxysmal AF episodes in the home setting. However, high age, large AF burden, and β-blocker use appear to limit the effectiveness of home-based self-management strategies. Changes in cardiac autonomic tone and activation of atrial stretch-activated ion channels may contribute to the effectiveness of such self-management strategies.

Background

The symptoms of atrial fibrillation (AF), including palpitations, tachycardia, fatigue, lightheadedness, and dyspnea negatively impact quality of life [1] particularly in patients with paroxysmal AF [2]. Patients experiencing symptomatic paroxysms may wait for self-termination or undergo cardioversion which is associated with a 50% relapse rate within one year [3]. Thus, patients may undergo cardioversion multiple times a year which is costly and burdensome. In an attempt to terminate AF in the home setting, some patients have developed their own self-management strategies. Often, such self-management strategies are used without approval from cardiologists, potentially exposing patients to unknown risks. To our knowledge, effectiveness and safety of self-management strategies used by patients to terminate paroxysmal AF in the home setting have not been investigated systematically. While there are potential risks with the use of unsupervised self-management strategies, there is also the possibility that some of these strategies are safe and effective. Identifying such safe and effective self-management strategies has the potential to enhance patient safety, reduce health care costs, and improve overall quality of life of affected patients. As a first step towards this long-term goal, the primary objective of this study was to investigate the perceived effectiveness of self-management strategies used by patients in their home setting to convert paroxysmal AF episodes into sinus rhythm.

It has been demonstrated that higher age and female gender are associated with longer lasting paroxysmal AF episodes and higher ventricular rates [4]. Likewise, history of paroxysmal AF [5] and anti-arrhythmic drug use [6] are known to impact AF recurrence. Therefore, effectiveness of self-management strategies may be confounded by patient characteristics, history and severity of AF, and anti-arrhythmic drug use. Thus, a secondary objective of this study was to investigate the effect of these potential confounding factors on the perceived effectiveness of self-management strategies.

Mechanistically, home-based self-management strategies may target the autonomic nervous system, electrolytes, hemodynamic parameters affecting atrial filling, and other targets. While there have been case reports of vagal maneuvers to be effective in select patients [7, 8], vagal maneuvers, such as the Valsalva maneuver, carotid body massage, or the diving response [9] are generally considered less effective in paroxysmal AF, possibly because they can reduce atrial conduction velocity and shorten atrial effective refractory period [10]. Electrolyte disorders are associated with increased risk of AF [11, 12] and may provide a target for self-management strategies (e.g., intake of electrolyte solutions). Increased levels of natriuretic peptide [13] and elevated left ventricular filling pressures predict AF recurrence after cardioversion [14]. Thus, hemodynamic interventions altering venous return to the heart (e.g., postural changes or exercise activating the skeletal muscle pump) may potentially act on atrial stretch-activated ion channels [15, 16] to terminate paroxysmal AF. Based on these considerations, we assessed – in healthy individuals – blood pressure (BP), stroke volume (SV), cardiac output (CO), total peripheral vascular resistance (TPR), left-ventricular end-diastolic volume (LV-EDV) and heart rate (HR) responses to a combination of self-management strategies that patients perceived as highly effective. Mechanistically, an increase in LV-EDV may indicate activation of atrial stretch-activated ion channels, while a bradycardic response may indicate increased cardiac vagal tone. Furthermore, these hemodynamic measurements – even when collected in healthy individuals – may inform safety concerns associated with such self-management strategies. For example, a marked increase in LV-EDV may indicate cardiac volume overload which may be a safety concern in patients with congestive heart failure. Likewise, increases in preload (LV-EDV), afterload (TPR), and HR may be associated with increased myocardial oxygen consumption, which could be a safety concern in patients with coronary artery disease.

Through an ongoing clinical trial (ClinicalTrials.gov ID: NCT05944575), we established working relationships with moderators and members of several AF advocacy groups totaling over 50,000 patients. Capitalizing on these established working relationships, we utilized a community-based participatory research (CBPR) approach to design and launch a survey to investigate the patient-perceived effectiveness of home-based self-management strategies used by patients to terminate paroxysmal AF. Furthermore, we investigated potential confounding factors, including patient characteristics, history and severity of AF, and anti-arrhythmic medication, and explored the impact of potential physiologic mechanisms on patient-perceived effectiveness. As an initial step towards exploring underlying mechanisms and potential safety concerns, we assessed hemodynamic parameters in response to a combination of self-management strategies in a limited number of healthy volunteers.

Methods

Ethics approval and consent to participate

The study adhered to the Declaration of Helsinki and was reviewed by the Institutional Review Board (IRB) of Burrell College of Osteopathic Medicine (BURRELL IRB 0141_2024 and BURRELL IRB 0164_2025). An exempt determination was made by the IRB for the survey study (BURRELL IRB 0141_2024). As part of this exempt determination the IRB waived the requirement for informed consent to participate in the survey. All participants in the hemodynamic study (BURRELL IRB 0164_2025) provided written informed consent.

Methods for survey study

Study design

A CBPR approach was used to design and deploy a cross-sectional survey-based study. An internet search identified 14 AF advocacy social media groups that were screened based on availability of contact information for the group moderator, daily online activity, and focus on AF. Ten groups met these criteria, and moderators from six of these ten groups participated in a video conference to discuss the purpose of the study. Five of the six group moderators agreed to collaborate on the research project and facilitated communication with the respective group members to assist with the development and deployment of the survey instrument. The number of members in the five included AF advocacy groups ranged from 652 to 34,000 for a total number of 50,052 members. Through group moderators, comments and suggestions on an initial draft of the survey were obtained from group members. These comments/suggestions concerned (1) linguistic aspects (replacement of technical terms with lay terms); (2) units for height and weight (allow for units other than feet/inches and pounds); and (3) terminology of descriptors for Likert scale questions. Based on this feedback, the survey was revised and returned to the group members for final review. The final survey (Table 1) was implemented using the Qualtrics platform (Qualtrics, Provo, UT. https://www.qualtrics.com) and assessed exclusion criteria (Q2 and Q3), demographic and anthropometric information and medication use (Q4-Q12), AF history and severity (Q13-Q16), and self-management strategies and perceived effectiveness (Q17-Q21). The answers to the questions were coded as indicated in Table 1. The survey was deployed during the time period of November 22, 2024 and January 9, 2025 (7 weeks).

Table 1.

Survey Questions and Coding

Atrial fibrillation burden

The coding for Q14 (“How often do you have an episode of paroxysmal AF?”) was used to estimate the number of paroxysms per month (coding 1 = 0.5/mo; coding 2 = 1/mo; coding 3 = 2/mo; coding 4 = 4/mo; coding 5 = 8/mo; coding 6 = 30/mo; coding 7 = 60/mo). The coding for Q15 (“How long do your AF episodes typically last?”) was used to estimate the duration of each paroxysm (coding 1 = 7.5 min; coding 2 = 37.5 min; coding 3 = 210 min; coding 4 = 540 min; coding 5 = 1,080 min; coding 6 = 2,880 min; coding 7 = 7,200 min). AF burden (in percent) was then calculated as 100 times the number of paroxysms per month times the duration of each paroxysm in min divided by 43,200 min per month. Multiple linear regression analysis was performed using the R statistical software [17] to investigate the effects of sex, age, BMI, and years since AF diagnosis on AF burden.

Perceived effectiveness of self-management strategies

Perceived effectiveness of self-management strategies was defined as a combination of “how often” (Q20) and “how fast” (Q21) a self-management strategy converts paroxysmal AF into sinus rhythm. How often a strategy converts to sinus rhythm was coded from 0 (never) to 4 (always), while how fast a strategy converts to sinus rhythm was coded from 0 (never) to 5 (immediately). Thus, higher numbers represent more effective strategies. A strategy that converts AF more frequently is obviously more effective. In addition, the probability that a conversion from AF to sinus rhythm is caused by the self-management strategy (rather than a spontaneous conversion) is higher if the conversion occurs fast. We considered both of these aspects equally important and, therefore, weighed them equally. Weighing both variables equally, an “effectiveness index” ranging from 0 (not effective) to 1.0 (most effective), was calculated based on the following equation (Q20: coding for how often [0–4]; Q21: coding for how fast [0–5] the strategy converted AF to sinus rhythm):

Role of confounding factors

The following potential confounding factors were considered: sex (Q4), age (Q5), body mass index (BMI, Q6 and Q7), age at diagnosis of AF (Q8), frequency of AF episodes (Q14), duration of AF episodes (Q15), and Vaughan Williams [18] class I, class II, class III, and class IV anti-arrhythmic drug use (Q10). Stepwise forward and backward multiple linear regression analysis was performed to investigate the impact of these potential confounding factors on how often (Q20) and how fast (Q21) the self-management strategies converted paroxysmal AF into sinus rhythm and on the calculated effectiveness index. The patient-reported numeric values for how often (Q20) and how fast (Q21) were then adjusted for the significant confounding factors by subtracting the difference between the predicted values (based on the estimates of the regression analysis) and the mean of the predicted values. The adjusted values for how often (Q20) and how fast (Q21) were then used to calculate an adjusted effectiveness index using the same equation as listed above. These calculations were done using the R statistical software [17].

Classification of Self-Management strategies

Three of the authors (HMS, AM, JP) individually reviewed the reported self-management strategies (Q19) and classified them into the following categories: (1) autonomic maneuvers (mostly Valsalva maneuver and diving response); (2) breathing techniques (mostly slow breathing); (3) intake of electrolytes, fluids, or dietary supplements; (4) exercise (mostly walking, running, cycling); (5) hemodynamic maneuvers (mostly leg raising and other interventions associated with an increase in cardiac preload); (6) yoga or meditation; (7) application of hot or cold temperature (mostly hot or cold showers, but also ice-cold water ingestion); and (8) other interventions. In case of inconsistencies between the classifications by the three investigators, consensus was achieved by discussion. It is important to note that some self-management strategies consisted of combinations of different categories. Such self-management strategies were assigned to each category involved. As an example, a strategy that consists of the combination of leg raising plus the Valsalva maneuver would be assigned to both the hemodynamic and autonomic categories.

Statistics

Data are presented as arithmetic means ± standard deviation (SD). Comparisons of patient characteristics (Table 2) and effectiveness measures (Fig. 1) between the eight self-management strategies were done by one-way analysis of variance (ANOVA) for independent measures with post-hoc t-tests. For binary variables (sex, patient satisfaction with physician-prescribed treatment, medication use) Chi Squared tests were used. Statistical analysis was performed using the R software [17].

Table 2.

Patient Characteristics by Category of Self-Management Strategies

Cat: Category of self-management strategies; N: Number of patients; BMI: Body Mass Index; Y of AF: Years of atrial fibrillation (Q8 and Q5); N of AF: Number of AF episodes per month (Q14); D of AF: Duration of AF episodes in hours (Q15); AF Burden: estimated from N of AF and D of AF; Satisfied: Satisfied with physician-prescribed treatment (Q16); A: Autonomic maneuvers; B: Breathing techniques; EFS: Electrolyte/Fluid/dietary Supplement intake; Ex: Exercise; H: Hemodynamic interventions; YM: Yoga/Mediation; T: hot/cold Temperature application; O: Others; All: All patients who responded to the survey; SMS=0: patients who did not report any self-management strategies; SMS³1: patients who reported at least one self-management strategy. Data are means±SD. There were no statistically significant differences in any patient characteristics between the eight categories of self-management strategies

Fig. 1.

Effectiveness of various categories of self-management strategies. Top left: patient-perceived percentage of successful AF conversions (Q20) adjusted for confounding variables; Top right: patient-perceived latency of AF conversion (Q21) adjusted for confounding variables; Bottom left: patient-perceived effectiveness index as reported by patients (not adjusted); Bottom right: patient-perceived effectiveness index adjusted for confounding variables. H: Hemodynamic interventions; Ex: Exercise; A: Autonomic maneuvers; YM: Yoga/Mediation; T: hot/cold Temperature application; B: Breathing techniques; EFS: Electrolyte/Fluid/dietary Supplement intake; H&A: combined Hemodynamic and Autonomic maneuvers. Data are shown as combined Violin and Box-Whisker plots. The shapes of the “violins” represent the distribution of the data. The “box-whiskers” show the median, the 95% confidence interval, and the lowest and highest values. Blue lines above the “violins”: P < 0.05 vs. EFS; Red lines above the “violins”: P < 0.05 vs. B

Methods for hemodynamic study

Study participants

Young adult (age > 18 years) and generally healthy individuals of both sexes were recruited from the local Las Cruces, NM region. Exclusion criteria included: Pregnancy or nursing mothers; any cardiac conditions (e.g., cardiac arrhythmia, heart failure, coronary heart disease); any other chronic conditions (e.g., hypertension, diabetes, rheumatic conditions); any acute illnesses (e.g., fever, common cold, etc.); any prescription medications other than contraceptives; any other conditions that may interfere with the experimental protocol.

Experimental protocol

After written informed consent was obtained, study participants completed an online questionnaire that assessed age and sex and confirmed absence of any exclusion criteria. Then, upper arm blood pressure and heart rate (Model BP786N, OMRON, Lake Forest, IL) and height and body weight (Seca 700, Seca, Hamburg, Germany) were measured. Participants assumed the supine position and were instrumented for non-invasive continuous blood pressure monitoring (Finapres, Model 2300 BP Monitor, Ohmeda, Madison, WI) and impedance cardiography (Minnesota Impedance Cardiograph Model 304 B) [19, 20].

The experimental protocol (Fig. 2) started with a 30-min rest period to establish stable baseline conditions. Then, participants were asked to drink 500 mL of ice-cold water (bottles were kept in a refrigerator set at a temperature of 4 °C) as fast as possible (typically 5–10 min). Thereafter, both legs were raised (LR in Fig. 2) using a wedge-shaped memory foam leg elevation pillow (height 20 cm). Then, participants performed three modified diving responses. An ice-cold face mask (Znöcuetöd Cold Face Eye Mask, Amazon) that was stored in a freezer at a temperature of −20 °C was placed on the face of the study participants covering the whole face with the exceptions of the eyes, nose, and lips. With the ice-cold face mask in place, study participants conducted three inspiratory breath holds for as long as they could (range 10–45 s). The three inspirator breath holds (I-BH and I1, I2, I3 in Fig. 2) were separated by 2-min rest periods. The three breath holds were followed by a 15-min rest period.

Fig. 2.

Experimental protocol for the hemodynamic study. A 30-min baseline recording was divided in three consecutive 10-min segments (B1, B2, B3). Thereafter participants drank 500 mL of ice-cold water (H₂O intake) and their legs were raised (LR). Then, three modified diving responses consisting of inspiratory breath holds (I-BH) while wearing an ice-cold face mask were elicited with 2 min rest periods in between (I1, I2, I3). The protocol ended with a 15-min post-intervention recording, that was divided in three consecutive 5-min segments (P1, P2, P3)

Data analysis

The ECG (provided by the Minnesota Impedance Cardiograph), the blood pressure waveform (provided by the Finapres device), as well as the impedances Z0, ΔZ, and ΔZ/Δt (provided by the Minnesota Impedance Cardiograph) were recorded using the WinAD module of the freely available HemoLab software [21] at a sampling rate of 500 Hz. Using the Analyzer module of the HemoLab software, beat-by-beat heart rate was extracted from the ECG and systolic, mean, and diastolic blood pressures were derived from the arterial blood pressure waveform. Mean heart rate, systolic, mean and diastolic blood pressure values were obtained from three consecutive 10-min sections of the 30-min baseline recordings (B1, B2, B3 in Fig. 2), for the three inspiratory breath holds (I1, I2, I3 in Fig. 2), and for three consecutive 5-min sections of the 15-min rest period (P1, P2, P3 in Fig. 2). The same time segments were used for impedance cardiograph analysis.

Impedance cardiography analysis

For impedance cardiography analysis, the functions implemented in the Analyzer module of the HemoLab software was used. The impedance signal Z0 was calibrated from 0.0 to 25.5 Ohms, and the ΔZ/Δt signal was calibrated from 0.0 to 1.0 Ohms/s based on the calibration signals provided by the device. For each time segment, the ECG and impedance signals were time averaged triggered by the R-wave of the ECG with 0.35 s before and 1.0 s after the R-wave. This step essentially averages all heartbeats into an “averaged single heartbeat”, which markedly improves the signal to noise ratio. This “averaged heartbeat” is then used by the software to calculate stroke volume (SV) based on the Kubicek [19], Sramek-Bernstein [22], and BMI-adjusted Sramek-Bernstein [22] equations. The Analyzer software also calculates the left ventricular ejection fraction (EF) based on the Capan [23], Judy [24], and Van der Meer [25] equations. Cardiac output (in L/min) was calculated as heart rate (in bpm) times stroke volume (in mL) divided by 1000. Total peripheral vascular resistance (TPR, in mmHg*min/L) was calculated as mean blood pressure (in mmHg) divided by cardiac output (CO, in L/min). Left ventricular end-diastolic volume (LV-EDV in mL) was calculated as 100 times stroke volume (in mL) divided by ejection fraction (in %).

Statistical analysis

Data are presented as means ± standard deviation unless otherwise noted. Statistical analysis was conducted using the free R-software [17]. For the data shown in Fig. 3a Friedman repeated measures ANOVA was calculated to compare the nine different time points at baseline (B1, B2, B3 in Fig. 2), during the modified diving responses (I1, I2, I3 in Fig. 2), and following the intervention (P1, P2, P3 in Fig. 2). The Friedman repeated measures ANOVA was also calculated for the time points B3, the average of I1, I2 and I3 (I123), and the average of P1, P2, and P3 (P123). These results are provided in the text. If the Friedman ANOVAs revealed significant changes over the time course of the experiment, post-hoc paired Wilcoxon tests were conducted to compare individual time points. Statistical significance was assumed for P < 0.05.

Fig. 3.

Hemodynamic responses to the combination of ice-cold water intake, leg raising, and the modified diving response. The time courses of systolic (black circles), mean (grey circles), and diastolic (white circles) blood pressure (BP), heart rate (HR), stroke volume (SV), cardiac output (CO), total peripheral vascular resistance (TPR), and left-ventricular end-diastolic volume (LV-EDV) are shown. B1, B2, B3 (baseline values), I1, I2, I3 (inspiratory breath holds), and P1, P2, P3 (post-intervention values) refer to time points as illustrated in Fig. 2. Data are means ± standard errors of the mean. *: P < 0.05 vs. time point B3 (highlighted)

Results

Results for survey study

Survey responses

606 patients from 20 countries (Algeria, Australia, Belgium, Canada, Cyprus, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Kenya, New Zealand, Philippines, South Africa, Spain, Turkey, The United Kingdom of Great Britain and Northern Ireland, and The United States of America) participated in the survey. Of those, 132 patients from 9 countries (Australia, Canada, Germany, Ireland, Kenya, New Zealand, Turkey, The United Kingdom of Great Britain and Northern Ireland, and The United States of America) reported between 1 and 8 (1.96 ± 1.15, mean ± SD) different self-management strategies for a total number of 259 unique self-management strategies.

Patient characteristics (Table 2)

The majority (80.2%) of patients (486/606) were female, 62.9 ± 11.4 years old, overweight (body mass index 28.9 ± 6.9 kg/m²), and had a history of paroxysmal AF of 5.2 ± 6.6 years. In average, patients experienced paroxysmal AF episodes several times a month (coding Q14: 2.6 ± 1.8) and the duration of the AF episodes ranged from 1 to 12 h (coding Q15: 3.6 ± 1.6), resulting in an average estimated AF burden of 5.8 ± 13.0%. The majority of patients (68.9%) were satisfied with the treatment prescribed by their physicians. The estimated AF burden in patients who were satisfied with their treatment (n = 276, AF burden: 4.3 ± 8.6%) was less than half of those who were not satisfied with their treatment (n = 122, AF burden: 9.1 ± 19.3%, t = 2.6 on 142.5 DFs, P = 0.011). Furthermore, multiple regression analysis demonstrated that higher age (P = 0.04) and larger BMI (P = 0.005) were associated with higher estimated AF burden (F = 2.9 on 4 and 365 DFs, P = 0.021). There were no statistically significant differences in any of the patient characteristics between the eight categories of self-management strategies.

Medication use (Table 3)

Table 3.

Medication use by Category of Self-Management Strategies

Data are presented as percentages and absolute number of patients in parenthesis. N: Number of patients. A: Autonomic maneuvers; B: Breathing techniques; EFS: Electrolyte/Fluid/dietary Supplement intake; Ex: Exercise; H: Hemodynamic interventions; YM: Yoga/Mediation; T: hot/cold Temperature application; O: Other strategies; All: Patients with at least one reported self-management strategy. C-I: class I antiarrhythmics; C-II: class II antiarrhythmics; C-III: class III antiarrhythmics; C‑IV: class IV antiarrhythmics; AC: anti-coagulants; CR: cardio-renal drugs (not including antiarrhythmic drugs); AL: anti-lipidemic drugs; AD: anti-diabetic drugs; PSY: drugs used for psychiatric conditions

There were no statistically significant differences in the use of different drug classes between categories of self-management strategies. As expected, the most frequently used anti-arrhythmic drugs were Vaughen Williams [18] class II drugs (β-blockers, 52% of patients). The least prescribed antiarrhythmic drugs were class III drugs (amiodarone and dofetilide, 3% of patients). Class Ic drugs (mostly flecainide, one patient propafenone) were used by 17% of patients, and Class IV drugs (mostly diltiazem) were used by 12% of patients. Anticoagulants were used by 46% of patients. These patients mostly used target-specific oral anticoagulants. Only two patients were treated with warfarin. The use of cardio-renal drugs (mostly antihypertensives including diuretics, but not including anti-arrhythmic drugs), anti-lipidemic drugs, anti-diabetic drugs, and drugs used to treat psychiatric conditions generally reflected the prevalence of the respective conditions in the general population. Other drugs used by some patients included non-steroidal and steroidal anti-inflammatory drugs, anti-asthmatic drugs, pain medications, anti-allergic drugs, antibiotics and anti-viral drugs, gastrointestinal drugs (mostly proton pump inhibitors), Parkinson medications, hormone replacement therapy, letrozole, drugs for benign prostatic hyperplasia, thyroid medications (levothyroxine, thioamides), bisphosphonates, allopurinol, and iron.

Relationship between confounding variables and effectiveness of self-management strategies

Multiple linear regression analysis revealed that patient characteristics (age and BMI), AF burden (frequency and duration of paroxysms), and anti-arrhythmic medication (β-blocker use) significantly affects the patient-perceived effectiveness of self-management strategies (Table 4). Specifically, higher age was associated with a longer latency (negative estimate value) and lower BMI was associated with less frequent conversions to sinus rhythm (positive estimate value). Thus, high age and low BMI were significantly associated with a lower effectiveness index. Self-management strategies were perceived to be less often successful and to have longer latencies in patients with more frequent AF episodes (negative estimate values). Likewise, the latency of conversion was perceived to be longer in patients with longer lasting AF episodes (negative estimate value). Thus, the effectiveness index was perceived to be lower in patients with higher AF burden. Finally, the use of class II anti-arrhythmic drugs (β-blockers) was associated with less frequent successful conversions and with a lower effectiveness index (negative estimate values). Thus, self-management strategies are perceived to be most effective in young patients with high BMI and low AF burden who are not using β-blockers. Even though, statistically significant (P < 0.001), the R² values of the regression models were low (0.089 to 0.148). Thus, the relationship between confounding variables and the perceived effectiveness of self-management strategies should be interpreted with caution.

Table 4.

Effect of confounding factors on perceived effectiveness of self-management strategies

Effect of potential confounding factors on how often a strategy was perceived to be effective [Q20 (How Often)], on the perceived latency of the effect [Q21 (How Fast)], and on the perceived effectiveness index [Effectiveness Index]. The following potential confounding factors were included in the multiple linear regression analyses: sex (Q4), age (Q5), body mass index (BMI, Q6 and Q7), age at diagnosis of AF (Q8), frequency of AF episodes (Q14), duration of AF episodes (Q15), Vaughan Williams classes I, II, III, and IV anti-arrhythmic drug use (Q10) and how satisfied patients were with their physician-prescribed treatment (Q16). Q-numbers refer to questions listed in Table 1

Perceived effectiveness of different categories of self-management strategies

Figure 1 summarizes the perceived effectiveness of the various categories of self-management strategies. There was no statistically significant difference between categories in how often a self-management strategy was perceived to be successful (unadjusted: F = 1.3 on 6 and 314 DFs, P = 0.269; adjusted for confounding variables: F = 1.5 on 6 and 306 DFs, P = 0.183). However, there was a significant difference between the categories for the perceived latency of conversion to sinus rhythm (unadjusted: F = 6.5 on 6 and 314 DFs, P < 0.001; adjusted for confounding variables: F = 5.0 on 6 and 306 DFs, P < 0.001). Post-hoc analyses of the data adjusted for confounding variables revealed that self-management strategies categorized as electrolyte, fluid, and/or dietary supplement intake had a longer latency than all other categories (P < 0.05), except yoga/meditation (P = 0.10). Furthermore, breathing techniques had a longer latency than hemodynamic maneuvers (P = 0.04). The patient-perceived effectiveness index also differed significantly between the various categories of self-management strategies (unadjusted: F = 3.1 on 6 and 314 DFs, P = 0.006; adjusted for confounding variables: F = 2.0 on 6 and 306 DFs, P = 0.064). Post-hoc analyses revealed that the effectiveness index adjusted for confounding variables was larger for hemodynamic interventions (EI_Adj=0.56 ± 0.21, P = 0.006) and exercise (EI_Adj=0.53 ± 0.17, P = 0.018) compared to electrolyte, fluid, and/or dietary supplement intake (EI_Adj=0.44 ± 0.20). Furthermore, the adjusted effectiveness index for hemodynamic interventions (EI_Adj=0.56 ± 0.21, P = 0.016) and exercise (EI_Adj=0.53 ± 0.17, P = 0.042) were also larger than the effectiveness index for breathing techniques (EI_Adj=0.44 ± 0.20). Because of low number of subjects (n = 5), we did not include combined hemodynamic and autonomic maneuvers (labeled H&A in Figure 1) in the statistical analysis. Based on the limited data available, it seems like this combination may potentially be highly effective.

Description of strategies

Table 5 lists representative descriptions of self-management strategies as they were provided by patients (not edited for spelling errors etc.) together with the respective effectiveness indices (EI).

Table 5.

Description (unedited, as reported by the patients) and effectiveness index (EI, not adjusted for confounding variables) of representative self-management strategies from the autonomic (A), hemodynamic (H), exercise (Ex), and combined hemodynamic plus autonomic (H plus A) categories

Results for hemodynamic study

Characteristics of study participants

The study population consisted of five female and three male volunteers (n = 8 total). The age was 25.9 ± 1.7 years and body mass index was 29.2 ± 8.8 kg/m². Upper arm blood pressure was 109 ± 17 mmHg (systolic) and 71 ± 11 mmHg (diastolic). Heart rate was 75 ± 15 bpm. Left ventricular ejection fraction at baseline was 64.3 ± 7.2%.

Hemodynamic responses (Fig. 3)

Compared to baseline (B3), the combination of ice-cold water imbibement, leg raising, and the modified diving response (I123: I1, I2, I3 averaged) did not alter systolic, mean, or diastolic blood pressure, but it did cause marked increases in SV (χ²F=7.0, df = 2, n = 8, P = 0.030, B3 vs. I123: P = 0.016, Cohen’s D = 1.00), CO (χ²F=9.0, df = 2, n = 8, P = 0.011, B3 vs. I123: P = 0.039, Cohen’s D = 0.76), and LV-EDV (χ²F=10.8, df = 2, n = 8, P = 0.005, B3 vs. I123: P = 0.016, Cohen’s D = 0.77) that were associated with a marked decrease in heart rate (χ²F=12.2, df = 2, n = 8, P = 0.002, B3 vs. I123: P = 0.008, Cohen’s D = 0.96). Within 15 min following the intervention (P123: P1, P2, P3 averaged), SV, CO, and LV-EDV returned to baseline levels, but TPR tended to increase (χ²F=4.75, df = 2, n = 8, P = 0.093, B3 vs. P123: P = 0.023, Cohen’s D = 0.89) and HR remained below baseline levels (χ²F=12.2, df = 2, n = 8, P = 0.002, B3 vs. P123: P = 0.016, Cohen’s D = 0.45).

Discussion

The findings of this study demonstrate that patients are highly creative in developing interventions to terminate paroxysmal AF episodes in their home setting. Patients perceived self-management strategies in the categories hemodynamic interventions and exercise as most effective. Interestingly, electrolyte, fluid, and/or dietary supplement intake was used by most patients (89/259 = 34.4%) but was also perceived as the least effective (Fig. 1). Through multiple linear regression analyses (Table 4), we found that the perceived effectiveness of self-management strategies is highest in young patients who are not on β-blockers but have high BMI and low AF burden.

The purpose of this study was to establish an inventory of self-management strategies used by patients in an attempt to terminate atrial fibrillation paroxysms and to estimate their patient-perceived effectiveness. The importance of this inventory is that it can inform the design of follow-up studies aimed at objectively assessing the effectiveness and safety of select self-management strategies. Based on pathophysiologic considerations, such self-management strategies are likely to be effective if they result in activation of atrial stretch activated ion channels through an increase in cardiac preload [15, 16] and if they shift cardiac autonomic balance to parasympathetic dominance [8, 10]. Based on these considerations and the outcome of our survey study, we designed an intervention that we expected to result in a marked increase in cardiac preload and cardiac parasympathetic tone. Specifically, we combined ice-cold water intake with leg raising and a modified diving response. Besides the expected physiologic responses, the selection of these interventions was based on practical considerations, i.e., patients would be able to perform these interventions in their home setting without specialized equipment. Heart rate variability analysis has demonstrated that cooling of the gastrointestinal tract by ingestion of cold fluids causes an increase in cardiac parasympathetic modulation [26, 27] in addition to a potential effect on cardiac preload. Hemodynamic maneuvers which mainly included leg raising were perceived as highly effective and may act through venous return to the heart and atrial stretch [28]. Several patients in our survey study reported that blowing cold air in their face using an electric fan converted atrial fibrillation into sinus rhythm. It is likely that blowing cold air in the face stimulates facial receptors of the trigeminal nerve and elicits a diving response that is known to cause apnea, bradycardia, and peripheral vasoconstriction [29]. Thus, we also included a modified diving response in our hemodynamic study. As expected from these considerations, the combination of ice-cold water intake, leg raising, and a modified diving response elicited strong bradycardia, suggesting an increased in cardiac parasympathetic tone as well as a marked increase in LV-EDV, suggesting atrial stretch. These hemodynamic responses provide a strong rationale for a follow-up study designed to objectively test the effectiveness of this specific combination of interventions in patients with paroxysmal atrial fibrillation.

The patients included in the survey study reported a total of 259 unique self-management strategies. Thus, it was necessary to aggregate the reported strategies in a smaller number of categories. The eight categories were broadly selected based on potential underlying physiologic mechanisms of action. However, it is obvious that each category contains widely varying strategies. For example, the autonomic category included the classical Valsalva maneuver, multiple modifications of the Valsalva maneuver, various modifications of the diving response, or heavy coughing while massaging the ear that receives afferent vagal innervation [30]. Aggregation of self-management strategies into the eight categories, may have obscured differences in effectiveness between specific interventions.

Interestingly, 31.1% of patients who participated in our study indicated that they are not satisfied with their physician-prescribed treatment (Table 2). Furthermore, dissatisfied patients had twice the estimated AF burden (9.1 ± 19.3%) compared to patients who were satisfied with their physician-prescribed treatment (4.3 ± 8.6%, P < 0.05). It is possible that high AF burden in combination with dissatisfaction with physician-prescribed treatment motivates patients to develop their own home-remedies. Indeed, the patients in our study were highly creative in developing such home-remedies. The number of unique self-management strategies reported by each patient ranged from one to eight with an average of two different strategies per patient for a total number of 259 unique strategies. The types of interventions used by the patients included some of the expected interventions, such as the Valsalva maneuver and leg raising, but also included some unexpected strategies, such as “Back massage by my partner” (effectiveness index [EI]: 0.55, n = 1), “I use my right hand to aggressively rub my left chest above my left breast but below the collar bone. Sometimes it leaves sore/bruised spots later” (EI: 0.55, n = 1), “Acupuncture” (EI: 0.267, n = 3), or “sexual orgasm” (EI: 0.625, n = 1). It is noteworthy that our patients were recruited from social media groups, suggesting that our population of patients was highly engaged and “tech-savvy”. This selection bias may explain why a high percentage (132/606 = 21.8%) of our study population utilized self-management strategies. Furthermore, it may also have influenced the specific types of self-management strategies reported. Indeed, some of these strategies included devices, such as “blowing into closed syringe”, “riding a stationary bike”, “cold air blowing in face from two hand-held fans”. Due to the potential selection bias caused by recruiting exclusively from social media groups, generalizability to the broader paroxysmal AF population, including older individuals, those with limited digital literacy, or patients who avoid social media, may be limited. Patients presumably developed these strategies by trial and error and resorted to the ones that they perceived as most effective. The purpose of this study was to learn from these patient experiences and systematically explore the perceived effectiveness of these self-management strategies using a community-based participatory research approach [31]. From the inception of the study, we were in constant communication with moderators and members of initially 14 and finally five AF-advocacy social media groups that reviewed the survey instrument and provided important comments and suggestions that improved the overall clarity of the survey questions. The direct involvement of patients into the planning and design of the research study certainly made a positive impact on the study and contributed to the worldwide interest with over 600 responses from 20 countries.

Among the most effective self-management strategies were hemodynamic interventions and exercise, followed by autonomic maneuvers. The human heart, including the atria, receives a dense autonomic innervation [32, 33] which likely contributes to the pathophysiology of atrial fibrillation [34]. It has been suggested that a short-term increase in vagal activity precedes most AF events [35], while an activation of the sympathetic nervous system may promote arrhythmogenesis through calcium-dependent triggered activity [36]. Thus, it may seem surprising that our study identified exercise-based interventions that are associated with an increase in sympathetic tone as some of the more effective self-management strategies. However, several patients reported that conversion to sinus rhythm does not occur during exercise, but upon termination of the exercise bout (e.g., “As my pulse slows, I usually convert back” or “Once I stop [running] and HR lowers, it is typically back in NSR”). Thus, it is possible that the conversion to sinus rhythm in response to exercise depends on the transition from high sympathetic/low parasympathetic tone during exercise to low sympathetic/high parasympathetic tone upon termination of exercise.

Even though, vagal maneuvers are generally considered less effective in converting paroxysmal AF into sinus rhythm and may actually induce AF [37], there are case reports suggesting efficacy in AF [7, 8]. Furthermore, non-invasive vagus nerve stimulation has been reported to reduce AF burden in patients with paroxysmal AF [38, 39]. In addition, our study identified autonomic maneuvers among the more effective self-management strategies. The interventions included in the category of autonomic maneuvers were mostly vagal maneuvers, including the Valsalva maneuver (e.g., “Blowing into a syringe trying to move plunger”), coughing (e.g., “cough quite heavily to convert to nsr”), and the diving response (e.g., “Cold facecloth on back of neck and face”). Common to the Valsalva maneuver and coughing, which resembles the Valsalva maneuver, is a rapid transition from high sympathetic/low parasympathetic tone during phase 2 to low sympathetic/high parasympathetic tone during phase 4 of the maneuver. The diving response is associated with high cardiac parasympathetic and high sympathetic vasomotor tone [9]. Thus, upon initiation of the diving response a strong increase in cardiac parasympathetic tone occurs, just as with the transition from phase 2 to phase 4 of the Valsalva maneuver. In this regard, the autonomic responses to the Valsalva maneuver, coughing, and the diving response are similar to that experienced after an exercise bout, further supporting the idea that a switch in autonomic balance from high sympathetic/low parasympathetic to low sympathetic/high parasympathetic tone facilitates the conversion of paroxysmal AF to sinus rhythm. This idea is also supported by the finding that self-management strategies were generally less effective in patients using β-blockers, because these drugs would diminish the cardiac response to changes in autonomic tone. Finally, the hemodynamic responses to the combination of ice-cold water intake, leg raising and the modified diving response was associated with a decrease in heart rate by about 10 bpm (Fig. 3), which is in line with a transition in autonomic balance from sympathetic to parasympathetic dominance during the intervention.

The perceived effectiveness index of hemodynamic interventions was highest among all categories of interventions (Fig. 1). Most of these interventions involved leg raising (e.g., “Lie flat with legs raised up the wall”, “lie back and raise legs for 15 seconds”, “Lying down and elevating my legs”), which increases cardiac preload. Assuming a right lateral decubitus position was also mentioned frequently (e.g., “Lay in bed on right side with a cool fan till I get chilled”, “If I lean to the right soon after it starts, it will sometimes convert”, “Sleeping on right side”, “Lie on right side while breathing deeply”). In contrast, many patients reported that the left lateral decubitus position can trigger AF episodes (e.g., “Don’t lie on left side“, “Lying flat, prone, supine, or on left side is usually an episode trigger”, “I avoid all caffeine and laying on my left side”). Just as elevating the legs, assuming a right or left decubitus position has been demonstrated by Doppler echocardiography to elicit pronounced hemodynamic affects including changes in cardiac preload [40–42]. It is possible that such changes in cardiac preload facilitate conversion of paroxysmal AF to sinus rhythm through atrial stretch-activated ion channels [15, 16, 43]. Indeed, our hemodynamic study demonstrates that the combination of ice-cold water intake, leg raising and the modified diving response causes a marked increase in LV-EDV by about 40 mL. This strong increase in cardiac filling is likely to be associated with an activation of atrial stretch-activated ion channels.

Interestingly, electrolyte, fluid, and dietary supplement intake was perceived as the least effective of all categories of self-management strategies. The low effectiveness index of this category is largely related to the long latency between the intervention and the conversion to sinus rhythm (Fig. 1). Because this intervention depends on intestinal absorption, the longer latency compared to some of the other categories is not surprising. However, the importance of this finding is that it strongly suggests that the perceived effectiveness reported by the patients is not just random but appears to be consistent with the underlying physiology. Likewise, the finding that self-management strategies are generally less effective in patients with more frequently occurring and longer-lasting AF episodes (Table 4) is plausible, because such patients are likely to have more advanced structural and electric atrial remodeling [44], contributing to the reduced effectiveness of self-management strategies. Finally, the finding that AF burden was significantly correlated with higher age and higher BMI is consistent with the literature [45] and further supports the validity of the survey data. However, these correlations (Table 4) need to be interpreted cautiously, because the multiple linear regression model – while statistically significant – had only modest explanatory power with R² values in the range of 0.089 to 0.148.

In recent years, a “pill in the pocket” approach has been advocated to terminate AF in the home setting [46–48]. Markman et al. [46] reported success rates of 73% with the “pill in the pocket” approach, while the patients in our study reported perceived success rates of non-prescription strategies (“how often successful”, Q20) in the range of 25–50%. The “pill in the pocket” approach is based on oral application of class IC antiarrhythmic drugs, including flecainide or propafenone, which can be self-administered and, therefore, may prevent hospitalization. The high reported success rate of 73% [46] is appealing. However, class IC drugs are contraindicated in patients with structural heart disease. For such patients, non-prescription self-management strategies, as reported by patients in our survey study, may potentially provide options. However, there may be unknown risks associated with the unsupervised use of such non-prescription self-management strategies. For example, patients with concomitant coronary artery disease may be at risk for developing acute coronary syndrome when engaging in unsupervised exercise. For example, our hemodynamic data in response to the combined intervention of ice-cold water intake, leg raising, and the modified diving response (Fig. 3) shows an increase in cardiac preload (LV-EDV) during the intervention and an increase in afterload (TPR) after the intervention. These responses may be associated with increased myocardial oxygen consumption and may place patients with coronary artery disease at risk. Likewise, patients with concomitant heart failure may develop signs of volume overload and potentially decompensate when using hemodynamic maneuvers that increase cardiac filling (e.g., leg raising). The increase in LV-EDV observed in our hemodynamic study suggests that this potential risk should not be ignored in patients with congestive heart failure. Autonomic maneuvers that increase intrathoracic and arterial pressure and induce bradycardia (e.g., Valsalva maneuver) may be associated with the risk of aortic dissection, stroke or TIA, and syncope in certain vulnerable patients. Due to these potential risks, we do not advocate for unsupervised use of any self-management strategies without prior medical clearance by the patients’ cardiologists.

There are some limitations of our study. Due to the demographics of the AF social media groups (69% female), the majority of our patients (80.2%) were female, while the prevalence of paroxysmal AF is generally higher in male individuals [44]. Thus, the gender distribution in our study cohort may limit the external validity of the study because sex-related differences in AF symptom perception, health behaviors, coping strategies, and response to interventions could bias outcomes, including perceived effectiveness of self-management strategies. The retrospective survey design is inherently associated with the risk of recall bias which may have favored the report of self-management strategies that are perceived as effective. The recall bias may also have affected other parameters calculated based on patient-reported data, such as AF burden. Another potential bias is efficacy expectancy bias. With this bias, perceived effectiveness may reflect beliefs rather than physiologic impact, especially for strategies with long latency, such as electrolyte intake. Participants who believe strongly in their intervention are more likely to report success, regardless of objective efficacy. Thus, the effectiveness indices derived from patient recollection may have overestimated the actual conversion rates. Furthermore, the retrospective survey design of our study lacks an objective confirmation of rhythm conversion (e.g., using wearable devices). Furthermore, the reported conversion success may reflect spontaneous AF termination, misinterpretation of symptoms, or a placebo effect. It is likely that efficacy expectancy bias is larger in patients with less satisfaction with their physician-prescribed treatment. Therefore, we included satisfaction with the physician-prescribed treatment in the multiple linear regression analysis shown in Table 4. This potential confounding variable was not found to significantly affect perceived effectiveness. While this finding does not exclude the possibility of efficacy expectancy bias, it does suggest that this type of bias did not have a major impact on the perceived effectiveness of the reported self-management strategies. Nevertheless, recall bias and efficacy expectancy bias could have led to an overestimation of the true effectiveness of some interventions. Recruitment of patients from AF advocacy groups may have resulted in selection bias of more engaged and health-literate patients, which could have affected the types of self-management strategies reported. Finally, our survey did not assess safety (e.g., vagal maneuvers may cause syncope), which appears to be critically important considering that 132/606 = 21.8% of patients engaged in presumably unsupervised self-management strategies.

Conclusions

The results of our survey in mostly female (80.2%) and highly engaged social media group members who are likely “tech-savvy” suggest that patients perceive home-based self-management strategies as highly effective. Based on patient perception, such strategies have the potential to convert paroxysmal AF to sinus rhythm in about 25–50% of paroxysmal AF episodes (Fig, 1 top left). The perceived success rate may even be higher if hemodynamic and autonomic interventions are combined. While it is possible, that these results are specific to our select study population, the high patient-perceived success rates are promising and provide a strong scientific rationale for follow-up studies investigating the safety, effectiveness, and mechanisms of action of select home-based self-management strategies. Specifically, studies focusing on the combination of strategies from different categories - as explored in our hemodynamic study - may be particularly promising. Once such follow-up studies have established safety and effectiveness, the use of home-based self-management strategies may potentially be encouraged in select patients. Ultimately, home-based self-management strategies may reduce the need for clinical conversions and improve the patients’ quality of life.

Authors’ contributions

HS conceived the study; HS, AM, and JB communicated with social media group moderators, analyzed the survey data; HS, BE, SM, EP, and GT designed and conducted the hemodynamic experiments; UB provided clinical expertise for the interpretation of the data. All authors contributed to drafting, revising, and critically reviewing the manuscript.

Funding

This research was supported by the Office of Research and Sponsored Programs at Burrell College of Osteopathic Medicine, Las Cruces, NM.

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The study adhered to the Declaration of Helsinki and was reviewed by the Institutional Review Board (IRB) of Burrell College of Osteopathic Medicine (BURRELL IRB 0141_2024 and BURRELL IRB 0164_2025) An exempt determination was made by the IRB for the survey study (BURRELL IRB 0141_2024). As part of this exempt determination the IRB waived the requirement for informed consent to participate in the survey. All participants in the hemodynamic study (BURRELL IRB 0164_2025) provided written informed consent.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.