Featured Article

Accurate ABPI measurement is fundamental to the community management of suspected peripheral arterial occlusive disease. This study investigates the correlation and agreement of GP performed ABPI measurements in the community compared to those measured in the vascular laboratory (VL) in the setting of a UK district general hospital. Records of 127 patients attending vascular clinic were reviewed, of whom 25 patients (19.69%) were referred by GPs with ABPI measurements performed. From these, 47 ABPI measurements were available and had limbs assessed by both GPs and VLs. For GP ABPI assessment: mean=0.748, median= 0.780, range=0.39 to 1.14, SD=0.200. For VL ABPI assessment: mean= 0.875, median=0.880, range=0.50 to 1.40, SD=0.245. GP performed measurement were an average of -0.13 (95% CI: -0.19 to -0.06, p = 0.0003), confirmed by Bland-Altman analysis. Using a diagnostic threshold of 0.9 for the GP performed ABPI, sensitivity=96 %, specificity=45.5%, PPV=66.7%, NPV=90.9% and accuracy=72.34%. Correlation coefficient, r=0.516 (p < 0.001). Using an ABPI value of 0.9 as the threshold for peripheral vascular occlusive disease, moderate agreement was seen between GP and VL diagnoses (Î = 0.416 ± 0.245). This study demonstrates that GPs perform ABPI measurements with only moderate degree of agreement with experienced VL technicians. Evidence exists for systematic under-estimation of ABPI by GPs however due to extremely high sensitivity the test lends itself to community based screening. Training and repeated measurement may increase accuracy of community based tests.

Introduction
The ankle brachial pressure index (ABPI) is the ratio of the highest ankle systolic pressure to the highest brachial systolic pressure. An accurate ABPI measurement is fundamental to the community management of suspected peripheral arterial occlusive disease (PAOD). It is a cheap test that provides hard evidence of vascular disease. Using a threshold ratio of 0.9, sensitivity equals 95% and specificity equals 100%, compared to an angiogram.¹

This study investigates the correlation and agreement of general practitioner (GP)-performed ABPI measurements in the community compared to those measured in the vascular laboratory (VL) in the setting of a UK district general hospital.

Methods
Consecutive patients with suspected PAOD from a wide range of general practitioners (GP) who were referred to our vascular surgical department over a three-month period were studied. Patients with known or suspected venous lower limb ulcerations were excluded from this study.

Referral letters from GPs were examined for ABPI measurements. All patients with suspected PAOD were investigated in the vascular laboratory (VL) and ABPI measurements were performed.

Data were analyzed using the Pearson correlation coefficient, the Bland and Altman’s method² and the Kappa method of assessing agreement. A paired student’s t-test was used for significance testing and taken at the 95% level.

Records of 127 patients were reviewed, and 25 patients (19.69%) were referred by GPs who had performed ABPI measurements. From these 25 patients, 47 ABPI measurements were available and limbs were assessed by both GPs and VL. For GP assessment: mean = 0.748, median = 0.780, range = 0.39 to 1.14, SD = 0.200. For VL assessment: mean = 0.875, median = 0.0.880, range = 0.50 to 1.40, SD = 0.245. There was an underestimation of ABPI when performed by GPs, on average of -0.13 (95% CI: -0.19 to -0.06, p = 0.0003).
The Pearson correlation coefficient between them is r = 0.516 (p < 0.001). Using an ABPI value of 0.9 as the threshold for peripheral vascular occlusive disease, moderate agreement was seen between GP and VL diagnoses (Î = 0.416 ± 0.245). Using 0.9 as the threshold diagnostic ABPI value, 36 of 47 (76.6%) patients were diseased according to GP-performed measurements, and 25 of 47 (53.2%) were diseased according to VL performed measurements. Sensitivity for GP-performed measurements was 96%, specificity was 45.5%, positive predictive value was 66.7%, negative predictive value was 90.9%, and accuracy was 72.34%.
Using the Bland and Altman method, plotting the difference in paired, observed values against the mean of paired, observed values for assessing agreement, a systematic underestimation of the ABPI was associated with the GP assessment.

Discussion
This study demonstrates that GPs perform ABPI measurements with only a moderate degree of agreement with results obtained from experienced VL technicians. Comparing means and the analysis of the Bland and Altman plot demonstrates a systematic underestimation of the ABPI associated with GP performance of measurements. The very high sensitivity and accuracy (96% and 72.34%) of GP-performed assessments suggest that the ABPI is a useful screening test for peripheral vascular disease (PVD) in general practice. Although an angiogram is the “gold standard” test for PVD, since most patients were managed conservatively and did not go on to have angiograms, this comparison was not possible.

Kaiser et al³ report that the intraobserver variability of experienced practitioners was 7.3% compared to 12% for less-experienced practitioners. The authors suggest using repeated measurement or more experienced practitioners as a means of reducing variability. In general practice, there is a wide spectrum of personnel with variations in training and experience performing the test. A study from the UK showed that the variation in technique for measuring the brachial systolic pressure resulted in significantly different measurements.⁴ This may account for the wide variation of measurements obtained.

A study in primary healthcare investigating the screening of hypertensive patients over the age of 60 years diagnosed undetected PAOD in 7%.⁵ The authors also found that up to 53% of those identified by screening were not on the best medical treatment for PAOD. A study of patients with lower limb ulceration from Portugal reports ABPI measurement prevalence as low as 8%. An Irish audit reports that of all patients being treated for lower limb ulceration, only 59.9% had had ABPI measurements performed.⁷ Overall it may be concluded that the ABPI is an underused investigation in primary healthcare.

Conclusion
We suggest that the ABPI should be more widely used in targeted case findings for PAOD in at-risk groups. Also, the ABPI should be used to stratify the severity of PAOD and, hence, streamline the referral process to secondary care. Community-based training for ABPI measurement as well as repeated measurements may increase the accuracy of community-based GP or nurse-performed ABPI measurements.