|Home | About | Journals | Submit | Contact Us | Français|
Aims: To evaluate the evidence of therapeutic international normalised ratio (INR) control reporting and to provide recommendations for future reporting, particularly for research and audit purposes.
Methods: A systematic review of literature published over a five year period describing therapeutic INR control. Papers were identified from the Medline electronic database, and those that met the quality criteria were reviewed independently by an academic general practitioner and a consultant haematologist.
Results: Fifteen papers were identified that met the quality criteria for review. The sample size of studies ranged from 53 to 2545 (mean, 483.9) patients. Follow up ranged from three months to 13 years. Twelve studies reported results from secondary care only, one from primary care only, and two from both primary and secondary care. Seven of the 15 papers reported percentage time in range, five of 15 papers reported mean INR, six of 15 papers reported the proportion of tests in range, and five of 15 papers reported mean warfarin dose. Additional methods of presenting INR results were: dose changes each month, distribution of INR results, deviation of INR value from mean, percentage dose changes, time between visits, and median INR value. Six papers reported only one outcome measure, six reported two outcomes, two papers reported three outcomes, and one paper reported five outcomes.
Conclusions: It is recommended that at least two outcome measures should be reported and measures should be selected so that both the INR determinations and dosing advice are monitored.
The past few years have seen an increase in interest in the therapeutic management of oral anticoagulation. Principally, this has been driven by the increasing numbers of patients receiving oral anticoagulation treatment as a result of trials demonstrating the effectiveness of oral anticoagulation in preventing strokes in patients with atrial fibrillation.1 One criticism of the original studies demonstrating the effectiveness of oral anticoagulation in preventing stroke was that although target international normalised ratios (INRs) for treatment were stated, the achieved intensity of anticoagulation achieved was not. This is an important issue because achieved intensity of anticoagulation is related to the benefit derived from the treatment. Similarly, subsequent studies have used a wide range of outcome measures to determine performance levels of both individuals and anticoagulation clinics. Our own research has shown that using different outcome measures of anticoagulant clinic effectiveness produced different results on the same data sets.2,3 The need for standardisation has previously been highlighted.4
“The aims of our study were to establish the nature of outcome measures being reported for the therapeutic effectiveness of oral anticoagulation and to provide recommendations for international normalised ratio reporting in the future”
To our knowledge, there have been two attempts to introduce some standardisation into the process, with the description of point prevalence5 (or last look in the book) and the per cent time spent in range.6 It is now widely accepted that per cent time in range should be calculated assuming a linear change in INR values over time.7 However it is not clear how widely these outcome measures are used. We have undertaken a systematic review of papers reporting INR data published in the past five years, subsequent to these recommendations. The aims of our study were to establish the nature of outcome measures being reported for the therapeutic effectiveness of oral anticoagulation and to provide recommendations for INR reporting in the future.
A literature search from January 1995 to December 1999 was carried out using Medline. No attempts were made to access grey literature. The keywords used for the search were “anticoagulation” and “INR”. Of the papers identified using the Medline search, DF excluded inappropriate papers based on their titles. Copies of the remaining papers were identified and the abstracts reviewed separately by EM and DF. Papers were excluded if they: (1) did not present results in terms of INR, (2) included less than 50 patients, (3) reported induction of warfarin only, (4) had a follow up period of less than three months, or (5) were written by members of the research team. Papers that were excluded by one reviewer only were reassessed by the team (DF, EM, and KG) and a consensus was taken on their eligibility for our study. Eligible papers were reviewed in full following a proforma devised by the study team (table 11)) by a haematologist (PK) and a clinical research fellow (RMc) to establish the variety of outcome measures used to report therapeutic oral anticoagulation control.
Three hundred and fifty four papers were identified using a Medline search. Of these, 314 were excluded based on their titles. Of the remaining papers, 15 were included in the formal review. The sample size of the reviewed studies ranged from 53 to 2545 (mean, 484) patients. The duration of follow up ranged from three months to 13 years. Twelve studies reported results from secondary care only, one from primary care only, and two from both primary and secondary care. All but one of the studies reported results from adult populations. In terms of the presentation of INR results, seven of 15 papers reported percentage time in range, five of 15 papers reported mean INR, six of 15 papers reported proportion of tests in range, and five of 15 papers reported mean warfarin dose. Additional methods of presenting INR results were: dose changes each month, distribution of INR results, deviation of INR value from mean, percentage dose changes, time between visits, and median INR value. None of the papers reviewed reported point prevalence. Six papers reported only one outcome measure, six reported two outcomes, two papers reported three outcomes, and one paper reported five outcomes (table 11).). The paper reporting five outcome measures was a randomised trial of computerised anticoagulant dosage, which reported time in range, mean INR, proportion of tests in range, the percentage of dose changes, and the mean time between visits.16
In total 339 papers were excluded. Three hundred and fourteen papers were excluded on the basis of title alone. A further 25 papers were excluded following review by DF and EM for reasons given in table 22.. Informal review of these papers by DF was undertaken following exclusion to identify any further parameters that had not been noted in the formal review. No further parameters were noted. Excluded papers are listed in table 33.
There has been increasing interest in the therapeutic management of oral anticoagulation following evidence for its beneficial thromboprophylactic effect in non-rheumatic atrial fibrillation. However, one of the difficulties in interpreting research findings has been the lack of consistency in expressing INR data. This inconsistency makes the comparison of findings between different centres very difficult. In a sense, this mirrors problems encountered within laboratory comparisons before the introduction of the INR system. Our own data have shown that there are differences in the observed efficacy depending upon the parameters chosen, with differences of up to 10% found in terms of INR control (table 44).). This is important because INR control gives a proxy measure for clinical outcomes, which are relatively infrequent. Thus, improved INR control should correlate with improved clinical outcome.
“It is unclear why point prevalence has not been more widely reported because this is a relatively easy statistic to generate”
The most surprising finding of our study was that none of the papers reviewed used one of the recommended methods of reporting, namely point prevalence. A large minority of papers reported only one outcome measure. The four most widely reported parameters were:
The first three of these measures relate to therapeutic control, whereas warfarin dose is only likely to be discrepant if there is a problem with the INR estimation between centres. The paper that reported five outcome measures centred around computerised dosing. The routine use of computerised data collection should allow the production of therapeutic control data in a variety of formats and should be encouraged.
In conclusion, we recommend that at least two outcome measures should be reported and that these should be selected from the four measures stated above. The measures should be selected so that both the INR determinations and dosing advice are monitored. It is unclear why point prevalence has not been more widely reported because this is a relatively easy statistic to generate. Our study highlights the need for standardisation of reporting of INR. This will become increasingly important with the development of differing models of oral anticoagulation management if clinical governance is to be taken seriously.
Take home messages