Table 3 Comparison of validity outcomes in the SCANDAT study and the current DTD results
From: Validation of multisource electronic health record data: an application to blood transfusion data
Outcome | SCANDAT 1/2 | DTD example |
|---|---|---|
External concordance of database and official statistics on the number of transfusions | >97% | >98.7% for products and 99.96% for transfusions |
% transfusions linked to the corresponding donor | 95% | 99.99% |
% transfusions linked to hospitalization | 88.7% | 99.2% (of which 23.6% day admissions) |
% duplicated donations and transfusions | 4.9% (donations) and 9.1% (transfusions) | 0% (donations) and 0.14% (transfusions) |
% missing or invalid values for identification number or date values | Range between 0.1% to 3.6% | 0%–0.01% |
Time patterns for donations and transfusion counts | In 1 year approximately 160,000 transfusions were missing; it took 2 years for the number of donations and transfusions to stabilize after the start of a new registration system | In 1 year the link of transfusions to products could be made for 2.2%, however this could be improved by adding donation data from the previous year |
% of recipients had records of receiving a blood transfusion in two or more local registers | 8.9% | 6% of patients are transferred to another hospital |