Table 1

Inclusion and exclusion criteria

Inclusion criteria Studies were included if they:	Exclusion criteria Studies were excluded if they:
Assessed the diagnostic accuracy of automated algorithm of a handheld ECG device (ie, a mobile instrument that produces a rhythm strip) in detecting paroxysmal, persistent or permanent AF. Performed a gold standard test for comparison (12-lead ECG or Holter monitor) within the same screening session. Recruited participants aged 18 years and older. Specified the setting in which participants were recruited and allowed categorisation in either a community setting (including general practice, pharmacy and community centres) or a hospital setting (including specialist centres and hospitals). Reported primary data (true positive, false negative, true negative, false positive) of ECG autointerpretation provided by the device or sensitivity and specificity results that allowed back calculation for the primary data.	Involved more than one person in interpreting the same ECG rhythm strips without reporting the consensus results among the interpreters*. Used alternate devices to detect AF, such as photoplethysmography or blood pressure devices.

Inclusion criteria
Studies were included if they:

Exclusion criteria
Studies were excluded if they:

Assessed the diagnostic accuracy of automated algorithm of a handheld ECG device (ie, a mobile instrument that produces a rhythm strip) in detecting paroxysmal, persistent or permanent AF.
Performed a gold standard test for comparison (12-lead ECG or Holter monitor) within the same screening session.
Recruited participants aged 18 years and older.
Specified the setting in which participants were recruited and allowed categorisation in either a community setting (including general practice, pharmacy and community centres) or a hospital setting (including specialist centres and hospitals).
Reported primary data (true positive, false negative, true negative, false positive) of ECG autointerpretation provided by the device or sensitivity and specificity results that allowed back calculation for the primary data.

Involved more than one person in interpreting the same ECG rhythm strips without reporting the consensus results among the interpreters*.
Used alternate devices to detect AF, such as photoplethysmography or blood pressure devices.