Table 1

Criteria for assessing the coverage and accuracy of a clinical database

	Level 1	Level 2	Level 3	Level 4
*For example, timing of physiological measures or distinguishing primary from secondary diagnoses.
A. Extent to which the eligible population is representative of the country	No evidence or unlikely to be representative	Some evidence eligible population is representative	Good evidence eligible population is representative	Total population of country included
Specify country:
B. Completeness of recruitment of eligible population	Few (<80%) or unknown	Some (80–89%)	Most (90–97%)	All or almost all (>97%)
State when and how completeness was determined:
C. Variables included in the database	identifier admin info condition or intervention	identifier admin info condition or intervention short term or long term outcome	identifier admin info condition intervention short term or long term outcome major known confounders	identifier admin info condition intervention short term outcome major known confounders long term outcome
D. Completeness of data (percentage variables at least 95% complete)	Few (<50%) or unknown	Some (50–79%)	Most (80–97%)	All or almost all (>97%)
State when completeness was last determined:
E. Form in which continuous data (excluding dates) are collected (percentage collected as raw data)	Few (<70%) or unknown	Some (70–89%)	Most (90–97%)	All or almost all (>97%) or no continuous data collected
F. Use of explicit definitions for variables	None	Some (<50%)	Most (50–97%)	All or almost all (>97%)
G. Use of explicit rules for deciding how variables are recorded*	None	Some (<50%)	Most (50–97%)	All or almost all (>97%)
H. Reliability of coding of conditions and interventions	Not tested	Poor	Fair	Good
State when and how it was most recently tested:
I. Independence of observations of primary outcome	Outcome not included or independence unknown	Observer neither independent nor blinded to intervention	Independent observer not blinded to intervention	Independent observer blinded to intervention or not necessary as objective outcome (e.g. death or lab test)
J. Extent to which data are validated	No validation	Range or consistency checks	Range and consistency checks	Range and consistency checks plus external validation using alternative source
State when and how it was last determined: