2001 年 32 巻 6 号 p. 295-312
The aim of this study was to look at the clinical trials from the viewpoint of medicalinstitution's own data management. From source documents verification (SDV) from 33clinical trials (126 cases) carried out in Nippon Medical School Hospital between June1999 and July 2000, the verification rate, inconsistency rate and occurrence state ofdeviation were surveyed.
In the SDV of 126 cases, the mean verification rate was 97.5 ± 4.0% [mean± SD (range: 85.8-100%)]. The data with the lowest verification rate was on items for“evaluation” (82.9%) followed by “others”, “side effects/adverse events” and “withdrawal/dropout”. The major reason for the impossibility to verify was due to the factthat investigators entered the records directly to CRF without recording on to medicalcharts (no statement was made to regard data entered to CRF as source data). Otherreasons were due to the monitor's inappropriate interpretation of the medical chart, insufficient preparation of source documents for SDV. There was no case which sourcedata was missed.
Among the 26, 628 data verified, inconsistency was found between CRF and thesource documents in 333 data [1.7±2.3% (range: 0-8.1%)]. Inconsistency was causedmainly by inaccurate entering of data to CRF and by incomplete data of CRF due toomission of entry. Other reasons were miscalculation, misspelling, and contradictionbetween source documents.
In 126 cases 68 deviations were found. The majority was that laboratory tests hadnot been carried out. Serious deviations observed were those of exclusion/inclusioncriteria, inappropriate informed consent, and omission of the evaluation.
Data error observed in this survey were considered to be within the acceptable range.We conclude that by reviewing the verification rate, the inconsistency rate and theoccurrence state of deviation as measurements of quality of clinical trial data, eachmedical institution can achieve their own data management.
