|Anticoagulation clinics and patient self-testing for patients on chronic warfarin therapy: a cost-effectiveness analysis
|Lafata J E, Martin S A, Kaatz S, Ward R E
This is a critical abstract of an economic evaluation that meets the criteria for inclusion on NHS EED. Each abstract contains a brief summary of the methods, the results and conclusions followed by a detailed critical assessment on the reliability of the study and the conclusions drawn.
The management of patients on chronic warfarin therapy by either: anticoagulation clinic testing (in-clinic capillary monitors with immediate test results); patient self-testing (at-home capillary monitors with immediate test results telephoned by patients to an organised anticoagulation clinic); or usual care (venipuncture blood draw with delayed test results and no organised anticoagulation clinic).
Economic study type
The study population was a hypothetical cohort of patients initiating warfarin therapy at 57 years of age. This cohort was chosen to parallel the available clinical data from which the model was populated.
Outpatient clinic, laboratory and home. The economic study was carried out in Michigan, USA.
Dates to which data relate
The effectiveness data were derived from studies published between 1989 and 1997. Resource use was based on expert opinion, the authors' assumptions, and studies and data sources published between 1989 and 1997. 1997 prices were used.
Source of effectiveness data
Effectiveness data were based on a review of published studies, expert opinion and information available within the authors' institution.
A decision tree incorporating a Markov model was used to estimate outcomes and costs.
Outcomes assessed in the review
The outcomes assessed in the review were: the monthly probabilities of transition between health states (no prior event; prior event, disabled, continuing therapy; prior event, permanently disabled, continuing therapy; prior event, permanently disabled, discontinuing therapy; dead), the estimated percentage of time spent below and above the therapeutic range of usual care, and rates of thromboembolic and haemorrhagic events per 100 patient years by time spent below, in and above the therapeutic range. The events assessed were thromboembolic first events, thromboembolic subsequent events, haemorrhagic first events and haemorrhagic subsequent events, where each was stratified into fatal, life-threatening or serious. The rates of subsequent thromboembolic and haemorrhagic events per 100 patient years for individuals discontinuing therapy were also assessed.
Study designs and other criteria for inclusion in the review
A range of study designs was included in the review including large anticoagulation clinical trials, observational studies, and a longitudinal study of patients seen in five anticoagulation clinics. Some unpublished data were also included. No other details of study designs or inclusion criteria were given.
Sources searched to identify primary studies
Details of the sources searched to identify studies were not given.
Criteria used to ensure the validity of primary studies
Details of the criteria used to ensure the validity of studies were not given.
Methods used to judge relevance and validity, and for extracting data
Details of the methods used to judge the relevance and validity of, and extract data from studies were not given.
Number of primary studies included
At least 25 studies were included in the review.
Methods of combining primary studies
The estimated percentages of time spent below and above therapeutic range were the averages of those reported in three studies (usual care), nine large trials and four observational studies (anticoagulation clinics) and four studies (patient self-testing). Transition probabilities for event rates were derived from the longitudinal study. Event rates for patients who discontinue therapy were derived from the large, randomised anticoagulation trial. The authors did not state how the data were combined.
Investigation of differences between primary studies
There is no evidence that the authors investigated differences between primary studies.
Results of the review
Patients receiving usual care spend 33% (range: 15% - 50%) of their time below and 17% (range: 10% - 25%) of their time above the therapeutic range. Patients receiving anticoagulation clinic care spend 26% (range: 8% - 40%) of their time below and 9% (range: 1% - 17%) of their time above the therapeutic range. Patients self-testing spend 6% (range: 5% - 15%) of their time below and 5% (range: 1% - 7%) of their time above the therapeutic range. Fatal Event rates per 100 patient years by time spent below, in and above the therapeutic range are as follows:
Thromboembolic first event (fatal): Below 0.05 (range: 0.00 - 0.30), In 0.01 (range: 0.00 - 0.05), Above 0.00 (range: 0.00 - 0.05).
Thromboembolic subsequent event (fatal): Below 0.01 (range: 0.00 - 0.30), In 0.00 (range: 0.00 - 0.05), Above 0.01 (range: 0.00 - 0.05).
Haemorrhagic first event (fatal): Below 0.05 (range: 0.00 - 0.30), In 0.04 (range: 0.00 - 0.03), Above 0.20 (range: 0.00 - 0.06).
Haemorrhagic subsequent event (fatal): Below 0.02 (range: 0.00 - 0.30), In 0.02 (range: 0.00 - 0.03), Above 0.07 (range: 0.00 - 0.06).
Life-threatening event rates and serious event rates were also reported in the paper.
Methods used to derive estimates of effectiveness
The authors' institution's clinical experience was used to support the estimated times spent below, in and above the therapeutic range for patients receiving usual care. The authors' assumptions estimated the percentage of all life-threatening events that result in permanent disability, and the percentage of these that lead to a discontinuation of therapy.
Estimates of effectiveness and key assumptions
The authors assumed that 60% of life-threatening thromboembolic events and 10% of life-threatening haemorrhagic events resulted in permanent disability. They assumed that 50% of patients suffering a permanently disabling event discontinued therapy.
Measure of benefits used in the economic analysis
Thromboembolic and haemorrhagic events avoided was the benefit measure used.
The economic analyses were carried out from the perspectives of society and the medical provider. When the medical provider perspective was adopted, direct costs included equipment, supplies, nurse and physician time, and the costs of adverse events (office visits, emergency department costs, hospital admission and nursing home costs). The authors with reference to the literature estimated the duration of nurse and physician consultations. Resources were valued using prevailing wage and price data from the authors' institution. The cost of adverse events was valued using the Source Book of Health Insurance Data and cost estimates from the authors' institution. Costs refer to 1997 prices and were discounted at 3%.
Statistical analysis of costs
Costs were included as point estimates with no statistical analysis.
When the societal perspective was adopted, costs of time and travel to the patient and their caregiver were included. The length of time spent travelling, waiting, training and testing, and the percentages of patients accompanied by family members and receiving help with home testing, was estimated by the authors with reference to the literature. Time was valued using the prevailing national wage rate for individuals aged between 55 and 64 years of age. Costs referred to 1997 prices and were discounted at 3%.
A one-way simple sensitivity analysis and a multi-way simple sensitivity analysis was conducted on the underlying assumptions of the model: estimated time below, in and above the therapeutic range for each management alternative, the risk of adverse events, the risk of disability following an adverse event, the risk of discontinuing anticoagulation therapy following permanent disabling, and testing frequency for each management alternative. The range used was that of the data from which the baseline assumption was formed. A Monte Carlo simulation was also performed to test the sensitivity of the results to variation in all uncertain parameters.
Estimated benefits used in the economic analysis
Over the five year period of the model, 3.7 adverse events were avoided under anticoagulation clinic testing as compared to usual care. Over the same period, 4.9 adverse events were avoided with patient self-testing as compared to anticoagulation clinic testing. Side-effects of treatment were neither considered nor relevant to the study.
The five-year cost per 100 patients was reported for each management alternative, where costs had been discounted at 3%.
Under the medical provider perspective total costs were $419,514 (usual care), $405,560 (anticoagulation clinic testing) and $526,014 (patient self-testing).
Under the societal perspective total costs were $529,737 (usual care), $645,671 (anticoagulation clinic testing) and $622,727 (patient self-testing).
Synthesis of costs and benefits
Benefits and costs were combined by calculating the cost per event avoided.
For anticoagulation testing compared to usual care, this value was a cost-saving (medical provider perspective) and $31,327 (societal perspective).
For patient self-testing compared to anticoagulation testing, this value was $24,818 (medical provider perspective) and a cost-saving (societal perspective).
The saving in medical care costs when switching from usual care to anticoagulation clinics disappeared after only small changes (i.e. 4% - 6%) in the time spent below and above the therapeutic range.
The results were also sensitive to annual testing frequency. The cost-saving when switching from anticoagulation testing to self-testing disappeared if the annual frequency of self-testing increased from 52 to 57.
The results of the Monte Carlo simulation showed that moving from usual care to anticoagulation clinic testing resulted in fewer thromboembolic and haemorrhagic events (88% certainty), more life-years (79% certainty), fewer disabled life-years (80% certainty) and, under the medical provider perspective, a cost-saving (80% certainty).
Moving from anticoagulation clinic testing to self-testing resulted in fewer adverse events (91% certainty), more life-years (85% certainty) and fewer disabled life-years (95% certainty) and, under the societal perspective, a cost-saving (48% certainty).
From the societal perspective, self-testing is the most cost-effective alternative to usual management of warfarin treatment.
CRD COMMENTARY - Selection of comparators
Although no explicit justification was given for the comparators used, "usual care" would appear to represent current practice, whilst the comparison of self-testing with anticoagulation clinic testing hypothesises that the latter is accepted. You, as a user of the database, should decide if these comparators represent current practice in your own setting.
Validity of estimate of measure of benefit
The authors did not state that a systematic review of the literature had been undertaken. Effectiveness estimates were dependent on the time spent below, in and above the therapeutic range of care, and this differed by management alternative. These estimates were derived by taking an unweighted average of reported values, which does not reflect difference in sample size or study populations/settings. Elsewhere, the methodology was not specified and it was unclear how the authors derived their estimates. Where the authors justified their assumptions, this was with reference to their institution's clinical experience. The authors stated that expert opinion was used, although it was not clear from the study where this input occurred. Estimates were investigated by sensitivity analysis, with the range implicitly being that derived from the literature studied. The estimation of benefits was modelled. The instrument used to derive a measure of health benefit, a decision analysis with an embedded Markov model, was appropriate.
Validity of estimate of costs
All categories of cost relevant to the perspective adopted were included in the analysis. Some relevant costs were omitted from the analysis. The cost of lost earnings due to hospitalisation or institutionalisation following an adverse event was not included, although travel/testing time costs were valued using wage rates. Inclusion of this cost may affect the cost-effectiveness estimate. Costs and quantities were reported separately. A sensitivity analysis of quantities, i.e. frequency of tests, was conducted using apparently appropriate ranges. No sensitivity or statistical analyses of prices or the discount rate were performed.
The authors did not make appropriate comparisons of their findings with those from other studies and did not address the issue of generalisability to other settings. The authors did not present their results selectively. The study was based on a hypothetical cohort of patients initiating therapy at 57 years of age, yet the authors generalised their conclusions without reference to the scope of their study. The authors pointed out that the cost-effectiveness of self-testing is dependent on the inclusion of societal costs, which may not be considered by provider organisations.
Implications of the study
The authors imply that there is no reason not to organise anticoagulation management in a clinic setting. Self-testing is the most cost-effective option, although providers may not consider patient and caregiver costs in the organisation of warfarin management.
Lafata J E, Martin S A, Kaatz S, Ward R E. Anticoagulation clinics and patient self-testing for patients on chronic warfarin therapy: a cost-effectiveness analysis. Journal of Thrombosis and Thrombolysis 2000; 9(1 Supplement S): S13-S19
Subject indexing assigned by NLM
Anticoagulants /economics /therapeutic use; Drug Costs /statistics & Drug Monitoring /economics; Hemorrhage /economics /prevention & Humans; Markov Chains; Outpatient Clinics, Hospital /economics; Self-Examination /economics; Thromboembolism /economics /prevention & Warfarin /economics /therapeutic use; control; control; numerical data
Date bibliographic record published
Date abstract record published