Table 5.
Quality of data in 105 clinical databases with full entry in DoCDat, August 2003
| Quality criteria | No (%) of databases |
|---|---|
| Completeness of recruitment: | |
| Few (<80%) or unknown | 36 (34) |
| Some (80-89%) | 9 (9) |
| Most (90-97%) | 16 (15) |
| All or almost all (>97%) | 44 (42) |
| Completeness of data: | |
| Few (<80%) or unknown | 44 (42) |
| Some (80-89%) | 19 (18) |
| Most (90-97%) | 23 (22) |
| All or almost all (>97%) | 19 (18) |
| Use of explicit definitions of variables: | |
| None | 41 (39) |
| Some (<50%) | 10 (10) |
| Most (50-97%) | 16 (15) |
| All or almost all (>97%) | 38 (36) |
| Independence of observations of primary outcome: | |
| Outcome not included | 11 |
| Observer neither independent nor blinded | 23/94 (24) |
| Independent observer not blinded | 11/94 (12) |
| Independent observer blinded or outcome is objective | 60/94 (64) |
| Extent of data validation: | |
| No validation | 3 (3) |
| Range or consistency checks | 5 (5) |
| Range and consistency checks | 69 (66) |
| Range and consistency checks plus external check | 28 (27) |