| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Epidemiology and Biostatistics and Department of Radiology,
Assessment of Radiological Technology Program, Erasmus MC-University Medical
Center Rotterdam, Rm. EE21-40b, Dr. Molewaterplein 50, Rotterdam 3015 GE, The
Netherlands.
2 Massachusetts General Hospital-Institute of Technology Assessment, Harvard
Medical School, Boston, MA.
3 Department of Health Policy and Management, Harvard School of Public Health,
Boston, MA.
Received April 22, 2004;
accepted after revision October 6, 2004.
Address correspondence to J. L. Bosch
(j.l.bosch{at}erasmusmc.nl).
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. The outcomes of 1,411 consecutive procedures in 1,583 limbs recorded in a multicenter registry involving six hospitals were analyzed. Six sets of outcome criteria were evaluated: one based on symptomatic change, three based on ankle-brachial index (ABI) measurements, and two based on combining the symptomatic and ABI outcome measures. Agreement among the outcome measures was compared using the kappa statistic.
RESULTS. The ABI outcome measures alone showed good agreement
(
= 0.74-0.94). The symptomatic outcome measures yielded a
substantially higher 12-month success rate than the ABI outcome measures
(difference, 18-24%) and the agreement was only fair ( = 0.52-0.60).
The agreement between symptomatic outcome and ABI outcome measures was poor in
patients with a pretreatment ABI measurement at rest of more than 0.90
( = 0.20). Combining symptomatic outcome and the ABI outcome measures
with the logical operator "OR" showed good agreement with the
symptomatic outcome measures alone ( = 0.97) and using
"AND" showed good agreement with the ABI outcome measures alone
( = 0.87).
CONCLUSION. In patients with a pretreatment ABI measurement at rest of more than 0.90, classifying procedures using a criterion based on improvement in ABI measurements with more than 0.10 is inaccurate and underestimates the actual success rate at 12 months after percutaneous intervention. Furthermore, combining subjective improvement in symptoms and improvement in ABI measurements does not yield more information than reporting these outcome measures separately. Therefore, we suggest that improvement in symptoms and improvement in ABI measurements should be reported separately to indicate the 12-month success rate of percutaneous interventions for peripheral arterial disease.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In randomized clinical trials or studies with protocols, the proposed criteria for reporting outcomes of percutaneous interventions work well. However, in studies and registries that evaluate daily clinical practice, extensive evaluation in follow-up after an apparent successful intervention is less evident. Clinicians often rely on history and physical examination including ABI measurements and only if warranted will additional tests such as duplex sonography, standardized segmental pressure measurements, and angiography be performed.
The purposes of this study were to evaluate outcome measures of the success rate at 12 months after percutaneous interventions for peripheral arterial disease that are available in daily clinical practice and explore how these rates change when different outcome criteria are applied.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Data regarding each patient at the time of the intervention and details about the procedure performed were entered prospectively in the Vascular Intervention Registry. We recorded each patient's risk factors such as history of smoking, comorbidity, symptomatic status (claudication, rest pain, ulceration, gangrene, or threatened bypass), and procedural parameters such as ABI at rest and after exercise before the intervention, diameter of the treated artery, type of intervention, procedural medication, local and generalized complications, technical outcomes, and additional vascular interventions. Data regarding follow-up after the procedure were abstracted by two of the authors from medical reports. These data included change in symptoms, ABI measurements at rest, walking distances, peak systolic ratios measures with duplex sonography, vascular reinterventions, and late complications.
All data were recorded using standardized forms and entered into a computerized database (Access 97, Microsoft). The institutional review board approved the study; because the registry did not involve experimental tests or treatment, but included the registration of daily practice, informed consent of patients was not needed.
Based on the variation in the definitions of success reported in the studies evaluating daily practice, we constructed six outcome measures. In all outcome measures, procedures that ended in technical failure or limbs that underwent revascularization during follow-up were classified as failures. Technical failure was defined as failing to enter the vessel, cross the lesion, or improve blood flow. The following criteria of success were evaluated: the patient reported continued symptomatic improvement, the ABI was consistently more than 0.10 above the preprocedural level, the ABI was consistently more than 0.15 above the preprocedural level, the ABI was consistently above 0.90, the patient reported continued symptomatic improvement or the ABI was consistently more than 0.10 above the preprocedural level, and the patient reported continued symptomatic improvement and the ABI was consistently more than 0.10 above the preprocedural level. This last outcome measure is similar to the clinical-hemodynamic outcome proposed by the TransAtlantic Inter-Society Consensus Working Group and differs from it only in that it was not based on the patient's ability to complete a standardized treadmill exercise test but on the patient's self-evaluation of his or her symptomatic change [1].
In all but the first outcome measure, information obtained with duplex scanning or angiography was also considered. Duplex sonography or angiography was performed in 32.6% of the patients. The results of the intervention were classified according to the peak systolic velocity ratio or the severity of stenosis, respectively, without considering the ABI measurement. Peak systolic velocity ratios greater than 2.5 or angiographic stenosis of 50% or more was defined as procedural failures. In, for example, 212 (13.4%) of 1,583 limbs, the outcome based on these imaging tests differed from the outcome based on the ABI measurement if success was defined as consistently more than 0.15 above the preprocedural ABI level (14.0% if success was defined as consistently more than 0.10 above the preprocedural level). To report success rate at 12 months after percutaneous interventions and evaluate the impact of outcome criteria in daily practice, we included these patients in the analyses.
To determine whether the reported outcomes of procedures would be classified differently when different outcome measures were applied, we constructed 2 x 2 tables and calculated kappa values to determine the level of agreement among the outcome measures. Subgroup analyses were performed to assess whether the estimates for the kappa value depended on localization of the treated lesions (iliac vs femoropopliteal), clinical indication (claudication vs critical ischemia [defined as ulceration, gangrene, or pain at rest, stages III and IV of the Fontaine classification]), or pretreatment ABI measurement at rest (ABI 0.90 vs ABI > 0.90, because this latter group would not have received treatment based on ABI alone).
In addition, because some patients had a longer symptomatic follow-up than ABI follow-up or vice versa, a subgroup analysis was performed including those patients with at least 1 year of symptomatic and 1 year of ABI follow-up. The mean symptomatic follow-up was 18 months, and the mean ABI follow-up was 15 months. We used the SPSS statistical package (version 10.0.7, Statistical Package for the Social Sciences) for our analyses.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Table 2 shows the variation in results when applying different outcome criteria. The success rate at 1 year ranged from 37% to 61%. The highest success rates were achieved by applying the patients' subjective self-evaluations of symptomatic status. The differences in results when the criteria based on ABI were applied were small (range for 1-year success rates, 37-41%).
|
To assess whether these different outcome measures classified the same
cases as successes or failures, we calculated the kappa statistic
(Table 3). The agreement
between the outcome measures based on ABI measurements was good ( =
0.74-0.94; SE, 0.02). In particular, the two outcome measures based on a
change in ABI showed a high level of agreement ( = 0.94). However, the
agreement between symptomatic improvement measures and the ABI measures was
only moderate ( = 0.52-0.60; SE, 0.020).
|
Combining symptomatic improvement and improvement in ABI of more than 0.10
in one outcome measure using the logical operator "AND" showed
good agreement with the outcome measure based on the same improvement in ABI
alone ( = 0.87; SE, 0.01). Combining symptomatic improvement and change
in ABI using the logical operator "OR" showed good agreement with
the outcome measure based on symptomatic improvement alone ( = 0.97;
SE, 0.02).
Subgroup analyses exploring the influence of localization and differences
in follow-up duration did not have a substantial impact on the agreement among
outcome measures (maximum absolute change in , 0.06). In the subgroup
of patients with critical ischemia, however, all kappa values increased with a
maximum change of 0.19. Furthermore, subgroup analyses in patients with a
pretreatment ABI measurement at rest of 0.90 or less versus those with a
pretreatment ABI higher than 0.90 showed a large difference in kappa values
(Table 4) and in hemodynamic
success rates. Restricting the population to patients with a low ABI resulted
in only a moderate increase in kappa values
(Table 4), no influence on the
symptomatic success rate, and an increase in hemodynamic success rate from 40%
to 47% at 1-year follow-up.
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Tests that are used routinely in daily clinical practice include the ABI and the patient's self-evaluation of symptoms. From the patient's perspective, and probably also the treating physician's perspective, the most relevant outcome is the symptomatic outcome. Symptomatic change, however, is highly subjective. Symptoms may be affected by the patient's attitude about his or her health status, the placebo effect, and the patient-doctor relationship. The ABI, on the other hand, is not affected by these potential biases. A major disadvantage of the ABI, however, is that it is not of primary interest—neither to the patient nor to the physician performing the intervention. Furthermore, the precision of an ABI measurement is limited [9, 10]. Clinicians, however, still frequently use the ABI to assess a patient's status because it is an easy and inexpensive tool that yields results that are unbiased by the patient's attitude. The limited precision of one ABI measurement may result in misclassification but, because of the random nature of the measurement error, it does not introduce a bias.
Findings from the current study highlighted another disadvantage associated with an outcome based on change in ABI: In patients with a pretreatment ABI at rest higher than 0.90, the agreement between a change in ABI by more than 0.10 and symptomatic improvement was poor, whereas in patients with an ABI of 0.90 or lower, the agreement was good. Although symptomatic improvement cannot be considered a gold standard, of all patients with a high pretreatment ABI measurement who reported symptomatic improvement, a large proportion did not have a change in ABI of more than 0.10. This finding can easily be explained by the fact that the potential gain in ABI that can be achieved in patients with a high ABI measurement before treatment is limited and does not reflect the potential gain in symptomatic status.
Possible explanations for high pretreatment ABI measurements in the presence of significant ischemic symptoms are the existence of calcified, noncompressible vessels or measurement error so the actual ABI was lower. Unfortunately, we did not measure vessel calcification and were therefore unable to adjust for this variable. However, irrespective of the cause, in patients with a high ABI before treatment, assessing outcome by assessing the change in ABI seems inaccurate, and excluding patients with a high ABI before treatment resulted in a higher hemodynamic success rate.
To overcome the problems associated with symptomatic outcome and ABI measurements, some authors propose combining these outcomes into one outcome measure [5]. In the current study, we combined symptomatic outcome and ABI measurement using the logical operator "AND" and found that it showed good agreement with the outcome measure based on a change in ABI by more than 0.10 alone. Together with the finding that the curves associated with these two outcome measures did not deviate much, this indicated that combining the outcomes using the "AND" operator does not yield more information than applying an outcome measure based on change in ABI alone. Similarly, combining symptomatic improvement with change in ABI using the logical operator "OR" showed good agreement with symptomatic improvement alone.
These results suggest that combining symptomatic outcome with hemodynamic outcome with either the logical operator "AND" or "OR" does not provide more information. Reporting symptomatic outcome and change in ABI separately, on the other hand, increases the clarity of the methods by giving a direct answer to the questions: What proportion of patients had symptomatic improvement, which is subjective but extremely relevant to the patient and the treating physicians, and what proportion had objective proof that the treatment had effect, which is not affected by the patients' and physicians' attitudes?
This study was limited by the lack of a reference standard. As we discussed previously, imaging studies during follow-up were performed in only a small proportion of the patients; therefore, an outcome measure based on imaging studies could not be used. The authors of a previous study [2] that also examined the impact of varying outcome criteria, albeit in a small patient population, reported similar results as found in the current study. They reported that applying outcome criteria based on symptomatic outcome resulted in higher success rates than applying an outcome based on an improvement in ABI. Furthermore, the authors reported similar success rates after applying an outcome based on improvement in ABI alone and after combining symptomatic and ABI improvement using the logical operator "AND." Although the authors did not use kappa calculations or other statistics to examine the agreement among outcome measures and although their study population was small (106 patients), their findings support our finding that combining symptomatic improvement and improvement in ABI does not provide more information than reporting them separately.
Another limitation was that we could not explore the influence of combining objective symptom improvement measured by a standardized treadmill test and ABI measurements in one outcome. To our knowledge, no studies exploring the influence of combining objective symptom improvement and ABI measurements into one outcome or reporting them separately have been published.
In conclusion, in patients with an ABI higher than 0.90 before treatment, classifying procedures using a criterion based on improvement in ABI with more than 0.10 is inaccurate to assess the outcome of intervention of peripheral arterial disease. Applying an ABI criterion in this group of patients may underestimate the actual treatment effect. Furthermore, our results indicate that combining subjective symptomatic improvement and improvement in ABI does not yield more information than reporting them separately. To preserve the distinct perspective that each represents and to ensure clarity of the results, we recommend reporting subjective symptomatic improvement and ABI improvement separately.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
