Skip navigation

RCT of computer-aided anticoagulation

Clinical bottom line

Computer aided management of anticoagulation meant faster stabilisation of patients newly anticoagulated, and more time in the desired INR range for those already on anticoagulation.


C Manotti et al. Effect of computer-aided management on the quality of treatments in anticoagulated patients: a prospective, randomized, multicenter trial of APROAT (Automated PRogram for Oral Anticoagulant Treatment). Hemostasis and Thrombosis 2001 86: 1060-1070.


This study was carried out over one year in Italy in five clinics federated to the Italian Federation of Anticoagulation Clinics, each with more than 1,000 patients and staffed with doctors who had been through structured training in the management of anticoagulation and had at least five years of clinical experience in the field. Extensive quality control procedures were in place for INR measurements, which were done using the same technology.

Patients in the five centres were randomised (method not stated) to computer database with computer aided dosing in which the computer algorithm suggested doses of oral anticoagulant and scheduled follow-up appointments, or to the use of the same computer database but without the algorithm, and where the physician had to decide dose and follow up. In the former case, physicians could over-ride the computer suggestions.

Patients were in two groups. The first were those starting oral anticoagulation, in whom the primary end point was the time taken to reach a stable condition, three consecutive INR values within the scheduled therapeutic range. The second were those on established oral anticoagulation, in whom the primary end point was the percentage of time patients were in the scheduled therapeutic range. Oral anticoagulants were warfarin and acecoumarol.

Analysis was planned for those with a target INR of 2.0 to 3.0, and those with a target range of 3.0 to 4.5.


Starting anticoagulation were 335 patients, 145 on computer and 190 on manual decisions. On maintenance therapy were 916 patients, 458 on computer and 458 on manual decisions. The groups were well matched for age and sex, by centre, and for indications for anticoagulation.

Starting anticoagulation

Significantly more patients reached a stable anticoagulated condition in months one and two with the computer than with manual decisions (Figure 1).

Figure 1: Cumulative percentage of patients reaching a stable state in first three months of anticoagulation with computer algorithm and manual decision-making

Established anticoagulation

Significantly more time was spent in the scheduled therapeutic range using the computer algorithm than with manual decisions (Figure 2) for both low and high INR target ranges.

Figure 2: Percentage of time spent in the scheduled therapeutic range.

As well as being more often in the target range, the computer algorithm aid resulted in about one fewer appointment per patient per year.


What is interesting and important about this randomised test of a computer algorithm to help in clinical decision-making is not just that it helps. It is that it helps in clinics run to high standards and staffed by experienced clinicians. When even experienced clinicians find that it makes a difference, then even more of a difference may be expected for those less experienced.

There is a management lesson here too. Each clinic had over 1,000 patients. The computer algorithm saved about one appointment per patient per year. This was about 6% of the clinical and laboratory load, no mean saving for the clinic, the laboratory, and the patient.