Methods

The design and selection criteria of the Original and Offspring cohorts of the Framingham Heart Study (FHS) have been described previously. In this study, we used pooled epochs from repeated examinations. Original cohort participants were eligible if they attended examination 11 (1968 to 1971, n = 2,955), examination 17 (1981 to 1984, n = 2,179), or examination 23 (1993 to 1995, n = 1,026). Offspring cohort participants were eligible if they attended examination 1 (1971 to 1975, n = 5,124), examination 3 (1984 to 1987, n = 3,873), or examination 6 (1995 to 1998, n = 3,532). Participants were excluded if they had a history of pacemaker placement (n = 9), had inadequate electrocardiograms for measurement of QRS interval duration (n = 99), were aged <20 years at examination (n = 246), were taking antiarrhythmic agents (n = 134), or were missing covariates (n = 470). After these exclusions, 17,731 observations for 8,311 participants were available for analysis. All participants gave written informed consent, and the study protocol was approved by the institutional review board at Boston University Medical Center.

At each FHS examination, all attendees underwent routine physical examinations, anthropometry, laboratory assessments of cardiovascular disease risk factors, and standard 12-lead electrocardiography at a paper speed of 25 mm/s. Physician investigators interpreted all 12-lead electrocardiograms. The QRS interval duration was defined as the interval from the onset of the Q or R wave (junction between the PR isoelectric line and the beginning of the Q or R deflection) to the end of the R or S wave (junction with the ST segment). The maximum QRS duration was measured and recorded to the nearest 10 ms, on the basis of an assessment of all 12 of the leads. During Original cohort cycle 23 and Offspring cohort cycle 6, when physician-measured and computer-derived QRS duration were available on the same electrocardiograms, the correlation between techniques was high (r = 0.81, p <0.001). The presence of QRS duration 100 to <120 ms was considered mild QRS prolongation. The presence of QRS duration ≥120 ms was considered bundle branch block, and criteria advocated by a World Health Organization working group were used to categorize the type of bundle branch block. Left bundle branch block (LBBB) was defined as the presence of QRS ≥120 ms, an absence of Q waves and the presence of wide-notched R waves in leads V ₅ and V ₆ , the presence of monophasic QS in leads V ₁ and V ₂ , and an absence of secondary R waves in lead V ₁ . Right bundle branch block (RBBB) was defined as the presence of QRS duration ≥120 ms; the presence of broad, notched R waves (rsr′, rsR′, or rSR′ patterns) in leads V ₁ and V ₂ ; and the presence of wide, deep, and notched S waves in leads V ₅ and V ₆ . Subjects with QRS durations ≥120 ms who did not meet criteria for LBBB or RBBB were categorized as “indeterminate.”

All FHS participants are under surveillance for cardiovascular events and procedures. Pacemaker implantation events were ascertained from a review of medical histories, physical examinations in the FHS, and hospitalization and personal physician records, including electrocardiograms. All participants are additionally under surveillance for death and cardiovascular events, including myocardial infarction, coronary insufficiency, stroke, and heart failure. A panel of 3 experienced investigators reviewed pertinent medical records for all suspected new events. The indication for each pacemaker implantation was reviewed by a physician investigator (SC) on the basis of the available medical records.

We used the method of pooling repeated observations to assess the relation of QRS duration with the incidence of pacemaker events over consecutive 12-year intervals. Participants became eligible to reenter the analyses if they remained free of events and met the exclusion criteria at each index visit. Original and Offspring cohort participants attended a total of 17,731 person-examinations over the course of up to 35 years of follow-up.

Multivariable Cox proportional-hazards models were constructed for each 12-year follow-up interval, a period over which the hazards of pacemaker placement were proportional. We analyzed QRS duration as a continuous and as a categorical variable, using the following 3 categories: <100 ms (referent), 100 to <120 ms (mild QRS prolongation), and ≥120 ms (bundle branch block). Multivariable models adjusted for age, gender, body mass index, hypertension, smoking, diabetes, and history of myocardial infarction or heart failure.

In secondary analyses, we adjusted all analyses for interim myocardial infarction or heart failure as time-dependent covariates; we also repeated all analyses after excluding subjects with histories of myocardial infarction or heart failure. Additionally, we adjusted analyses for baseline PR interval duration, heart rate, and electrocardiographic left ventricular hypertrophy. Given the previously reported association of PR prolongation and pacemaker implantation, we also analyzed the relation of QRS duration to pacemaker implantation in subjects with PR intervals ≤200 and >200 ms separately. Because nodal blocking medications (e.g., β blockers, calcium channel blockers, and cardiac glycosides) and certain antidepressants can potentially affect conduction times, we performed analyses with and without participants taking these medications. We also investigated whether incident pacemaker placement varied according to type of baseline bundle branch block in multivariable Cox models (adjusting for the same covariates listed above) that compared subjects with left, right, and indeterminate bundle branch block (as defined earlier) with the referent group (QRS <100 ms). Last, continuous QRS analyses were repeated after excluding participants with QRS durations ≥120 ms.

All statistical analyses were performed using SAS version 9.1.3 (SAS Institute Inc., Cary, North Carolina), and a 2-sided p value <0.05 was considered statistically significant.