Level 1:The potential data
Level 2:The captured data
Level 3:The extracted data
Level 4:The analysed data
Level 5:The interpreted data
There are several levels to electronic audit in which an aspect of the practice population is modeled at increasing degrees of sophistication. Errors can creep in at any of these stages to invalidate the interpretation put on the data.
Failure to capture data as it presents to the clinician or arrives in letters.
Patient behaviour (eg DNAs) preventing data capture.
Deliberate misleading of the clinician eg on smoking habit.
Changes in the practice population.
Failure to accurately interpret clinical situations.
Solutions include: staff training in consultation technique/diagnosis and computer use / feedback / easy access to computer console
Entering the wrong codes by mistake.
Contamination of agreed coding conventions by non standard coding.
Using codes that are not audited instead of the correct ones.
Inconsistent or inadequate data entry.
Data corruption/software problems.
Solutions include: training in and use of templates/protocols/coding schemes /routine housekeeping data consistency checks/computer security and backup/agree coding at start.
Faults in query interpreter.
Faults in query logic or syntax.
Faults in query submission egg wrong dates.
Solutions include: careful checks by query writers/feedback to software developers.
Faults in response file handling.
Faults in spreadsheets.
Solutions include: as per Level 3
Faults in interpretation of data through lack of local knowledge.
Failure to identify errors carried through to higher levels.
Faults in audit criterion choice rendering accurate interpretation difficult.
Solutions include: use of independent experts/comparisons with external and historical data/open feedback to practice level for comment on local circumstances