Background The demand for anticoagulant treatment is increasing. We co
mpared the benefits of computer generated anticoagulaant dosing with t
raditional dosing decided by experienced medical staff in achieving ta
rget international normalised ratios (INRs). Methods In five European
centres we randomly assigned 285 patients in the stabilisation period
and stabilised patients to the computer-generated-dose group (n=137) o
r traditional-dose group (n=148). Centres had a specialist interest in
oral anticoagulation but no previous experience with computer-generat
ed dosing. The computer program calculated doses and times to next vis
it. Our main endpoint was time spent in target INR range (Rosendaal me
thod). Findings For all patients combined, computer-generated dosing w
as significantly beneficial overall in achieving target INR (p=0.004).
The mean time within target INR range for all patients and ail ranges
was 63.3% (SD 28.0) of days in the computer-generated-dose group comp
ared with 53.2% (27.7) in the traditional-dose group. For the stabilis
ation patients alone, computer-generated doses led to a non-significan
t benefit in all INR ranges (p=0.06), whereas in the stable patients t
he benefit was significant (p=0.02). Interpretation The computer progr
am gave better INR control than the experienced medical staff and at l
east similar standards to the specialised centres should be generally
available. Clinical outcome and cost effectiveness remain to be assess
ed.