Table 2.

Follow-up methods, including weight assessments and AF case definition

First author (year)	Follow-up (years)	When weight change was assessed	How weight change was assessed	AF case definition	Relative timing of weight and AF assessment	AF incidence per 1000 person years
Rosengren et al 29	Maximum 34.3	Single midlife evaluation (at 12 years after screening appointment) where weight was measured and participants reported their recalled weight at age 20 years.	Weight change between recalled weight at age 20 and midlife evaluation (approximately 30-year interval).	Hospital discharge code for AF	Weight change calculated at single midlife evaluation visit (1970–1973). AF incidence recorded over subsequent study follow-up to 31 December 2004. Median time between midlife evaluation and discharge for AF was 25 years.	7.5
Tedrow et al 24	Mean 12.9	Self-reported weight on the 24, 36, 60, 72and 108 month questionnaires during the randomised trial, at the beginning of the observational phase and yearly thereafter. Self-reported height used to calculate BMI.	BMI change over the first 60 months of the study (5-year interval).	Participant self-reporting, triangulated against electronic health record	Weight change in first 60 months of study and AF incidence after that time period through to the end of the study.	1.9
Grundvold et al 23	Maximum 35	Measured weight at baseline visit and recalled weight at age 25 years recorded at that time. Baseline visits 1972 to 1975.	Weight change between recalled weight and baseline (approximately 25 year interval).	ECG at study visit and hospital discharge codes	Weight change calculated at baseline visit based on change from recalled weight aged 25 years. AF incidence over subsequent follow-up, up to 31 December 2007.	4.5
Huxley et al 30	Maximum 11	Weight change between baseline and visit four (9-year interval)	Weight was measured at baseline and at 3, 6 and 9 years later.	ECG at study visit and codes from hospital discharge and death certificates	Weight change and AF incident cases across same study follow-up period.	6.9
Alonso et al 21	Mean 9	Weight was measured at baseline and annually thereafter. Measurements done in duplicate using a digital scale and stadiometer.	Weight change was between baseline and first annual visit (1-year interval)	ECG at biannual follow-up visit and hospital discharge codes	Weight change in first year of study with AF incidence calculated across follow-up, including the first year.	6.4
Grundvold et al 22	Mean 4.6	Weight was measured at baseline and at a second visit within 18 months.	Weight change was between the two visits (average intervals were 421, 396 and 402 days for the weight gain, stable weight and weight loss categories, ie, around 13–14 months)	Codes from national patient registries and primary care electronic database	Follow-up for AF started after the second visit and continued until new AF diagnosis, death or 31 December 2009.	9.8*
Johnson et al 25	Mean 22.3 (from rescreening) 28.3 (from baseline)	Weight was measured at two screening examinations, separated by an average of 6 years (first 1974–1984, second 1981–1985).	Average annual weight change was calculated.	Hospital inpatient and outpatient codes	Weight change between first and second screening examinations, AF incidence between second screening examination and study endpoint, 31 December 2010.	8.5*
Berkovitch et al 28	Mean 6.4	Weight was measured at annual visits.	The weight change between visits was integrated into the Cox model as a time-dependent variable.	ECG at study visit, Holter monitor or primary care electronic records	Weight change and AF incidence both reported across same study follow-up period.	2.5*
Diouf et al 27	Mean 5	Weight was measured at baseline (in 1999/2000) and a second visit (in 2004/2005).	Weight change between the two visits (approximately 5-year interval)	Single study follow-up ECG	Weight change and AF incidence both reported across same study follow-up period.	2.0
Ball et al 26	Mean 15.7	BMI was measured at each study survey. Participants visited a study centre and were weighed in light clothing with no shoes for consistency.	Change in BMI between the third (1986–1987) and fourth (1994–1995) survey	Hospital inpatient and outpatient records, linked with National Causes of Death Registry	Weight change between third and fourth survey used to calculate BMI change. Incidence of AF calculated over subsequent follow-up after fourth survey to 31 December 2013.	3.6

AF, atrial fibrillation; BMI, body mass index.