A retrospective cohort study evaluating patients with sick sinus syndrome who received a first permanent dual-chamber pacemaker has demonstrated that a five-variable landmark model can predict the risk of device-detected new-onset atrial fibrillation over three years of post-implantation follow-up. Conducted by investigators at Sanming First Hospital Affiliated to Fujian Medical University, the research focused exclusively on individuals programmed with managed ventricular pacing who remained entirely free of atrial tachyarrhythmias throughout their initial three months after device placement.
Dual-chamber pacemakers represent a standard therapy for patients diagnosed with sick sinus syndrome, a condition characterized primarily by sinoatrial node degeneration and extensive atrial electrical remodelling. Prior risk assessment tools were frequently derived from heterogeneous cohorts that combined individuals treated for sick sinus syndrome with those receiving pacemakers for atrioventricular block. These distinct patient subsets exhibit fundamentally different underlying pathophysiology, because atrioventricular block stems from structural failure within the specialized conduction system rather than primary atrial disease, often requiring much higher cumulative ventricular pacing percentages that confound statistical risk estimates.
To overcome these limitations and isolate clinical predictors specific to sinus node dysfunction, the investigators implemented a standardized clinical protocol using a single pacemaker model, the Medtronic SEDR01, with managed ventricular pacing and rate-response functions uniformly activated. Under this programming configuration, the lower rate was set to 60 beats per minute, the upper tracking rate to 130 beats per minute, the sensed atrioventricular delay to 150 milliseconds, and the paced atrioventricular delay to 180 milliseconds.
The study derived its development cohort from 322 consecutive adult patients implanted between January 2015 and December 2021 who satisfied rigorous inclusion criteria and completed evaluable follow-up beyond the three-month landmark visit. Patients presenting with prior atrial fibrillation or flutter, concomitant valvular or congenital heart disease, uncontrolled thyroid dysfunction, or who experienced atrial high-rate episodes during the initial three-month landmark window were excluded. A subsequent temporal validation cohort comprising 121 consecutive patients implanted between January 2022 and December 2025 under identical institutional criteria was reserved exclusively for model validation. Standardized device interrogations conducted at the three-month landmark provided the objective atrial pacing percentage and ventricular pacing percentage values used as baseline predictors.
Twenty-six candidate predictors, including thirteen continuous variables and thirteen categorical demographic and clinical characteristics, were evaluated through univariable Cox proportional hazards regression. Backward variable selection based on the Akaike information criterion identified five independent predictors of subsequent device-detected atrial fibrillation: recipient age, maximum P-wave duration measured across simultaneous 12-lead electrocardiograms, left atrial volume index quantified by echocardiography, an atrial pacing percentage exceeding 50 percent, and a ventricular pacing percentage exceeding 40 percent. In the multivariable survival model, each additional year of patient age conferred an adjusted hazard ratio of 1.043, while each 10-millisecond increment in maximum P-wave duration yielded a hazard ratio of 1.200. Each 1 milliliter per square meter increase in left atrial volume index was associated with an adjusted hazard ratio of 1.051.
Pacing burden thresholds measured at the landmark interrogation also exhibited substantial prognostic associations with the primary arrhythmic outcome. An atrial pacing percentage greater than 50 percent was associated with an adjusted hazard ratio of 1.845, whereas a ventricular pacing percentage exceeding 40 percent was associated with an adjusted hazard ratio of 1.898. During a median follow-up of 41.6 months in the development cohort, new-onset atrial fibrillation occurred in 90 of the 322 patients, representing an incidence rate of 28.0 percent. The five selected parameters were subsequently incorporated into an absolute risk equation and a graphical nomogram designed to calculate individualized probabilities of device-detected atrial fibrillation at one, two, and three years following the three-month landmark.
The biological mechanisms underlying these five retained predictors reflect complementary dimensions of atrial arrhythmogenic vulnerability. Advancing age is associated with progressive cardiomyocyte hypertrophy, apoptosis, and interstitial myocardial fibrosis that slows conduction velocity and exacerbates electrical dispersion. The maximum P-wave duration captures intra-atrial and interatrial conduction delays, including structural impairment across the Bachmann bundle, which predispose cardiac tissue to local re-entrant circuits. Similarly, the left atrial volume index provides an established three-dimensional echocardiographic metric of chronic hemodynamic loading and structural remodelling. In contrast, elevated atrial and ventricular pacing percentages reflect both intrinsic disease severity within the cardiac conduction system and the potential electrophysiological consequences of non-physiological electrical activation, though investigators emphasized these device parameters serve strictly predictive rather than proven causal roles.
Rigorous internal validation was conducted using 2,000 bootstrap resamples of the complete development cohort, repeating every step of the variable selection pipeline within each resample to account for potential statistical optimism. While the initial apparent concordance index in the development cohort was 0.7354, the full bootstrap pipeline revealed an average optimism of 0.0362, resulting in an optimism-corrected C-index of 0.6992. Variable selection stability remained high for left atrial volume index, patient age, and maximum P-wave duration, which were retained in 99.8 percent, 97.3 percent, and 96.5 percent of resamples, respectively. By comparison, binary thresholds for atrial pacing above 50 percent and ventricular pacing above 40 percent demonstrated more moderate selection frequencies of 63.9 percent and 57.9 percent across the bootstrap iterations.
When applied without modification to the temporal validation cohort of 121 patients, among whom 28 developed new-onset atrial fibrillation over a median follow-up of 30.5 months, the five-variable nomogram achieved an overall Harrell’s C-index of 0.769. Cumulative dynamic time-dependent area under the receiver operating characteristic curve values were 0.753 at one year, 0.881 at two years, and 0.916 at three years from the landmark.
Assessment of calibration demonstrated close concordance between predicted probabilities and observed event rates over extended follow-up, recording calibration slopes of 0.90, 0.99, and 0.98 at one, two, and three years, respectively. Calibration-in-the-large yielded values of -0.17 at 12 months, 0.24 at 24 months, and 0.35 at 36 months. However, the authors noted that one-year calibration estimates exhibited wide uncertainty intervals because only five qualifying arrhythmic events occurred within the temporal validation group during the initial twelve months post-landmark. Competing risk sensitivity analyses employing Fine-Gray subdistribution hazard models produced subdistribution hazard ratios consistent in direction and magnitude with the primary Cox model, confirming that the infrequent occurrence of non-arrhythmic death did not distort observed predictor associations.
Exploratory decision-curve analyses across representative risk thresholds between 5 percent and 20 percent revealed variable net clinical benefit across different time horizons, indicating that the nomogram should not yet serve as an absolute threshold for clinical interventions. The authors emphasized several key design constraints, noting that the single-centre retrospective framework, the deliberate exclusion of patients experiencing early post-operative arrhythmias within three months, and the reliance on a single pacemaker manufacturer with uniform programming restrict immediate clinical extrapolation. Consequently, while the five-variable landmark model presents a promising instrument for stratifying late arrhythmic risk in sick sinus syndrome, prospective multicentre external validation across diverse pacing platforms remains necessary before formal clinical adoption.
Subject of Research: Prediction model for device-detected new-onset atrial fibrillation after dual-chamber pacemaker implantation
Article Title: Device-detected new-onset atrial fibrillation after dual-chamber pacemaker implantation in patients with sick sinus syndrome under managed ventricular pacing: a risk nomogram from the 3-month landmark
Article References: Xie, J.-M., Shan, C.-H., Xiao, Y.-J., Ma, J., Lu, R.-R., & Zheng, Z.-J. (2026). Device-detected new-onset atrial fibrillation after dual-chamber pacemaker implantation in patients with sick sinus syndrome under managed ventricular pacing: a risk nomogram from the 3-month landmark. Frontiers in Cardiovascular Medicine, 13, Article 1989202. https://doi.org/10.3389/fcvm.2026.1989202
Image Credits: AI Generated
DOI: 10.3389/fcvm.2026.1989202
Keywords: atrial fibrillation, dual-chamber pacemaker, sick sinus syndrome, managed ventricular pacing, risk prediction, nomogram, landmark analysis, cardiac electrophysiology, P-wave duration, left atrial volume index, atrial pacing burden, ventricular pacing burden, cardiovascular medicine, cardiac pacing, temporal validation, cardiac arrhythmias
News Source: Ophelia Keating. (November 14, 2026). Five-Variable Model Predicts New-Onset Atrial Fibrillation After Dual-Chamber Pacemaker Implantation. Scienmag.



