Interobserver variation in interpretation of arteriography and management of severe lower leg arterial disease.

OBJECTIVE: arteriography is the reference standard for the assessment of the lower leg arteries in patients with severe lower limb ischaemia. Interobserver variation in arteriography interpretation may cause disparities with non-invasive imaging modalities. We determined interobserver variation in lower leg artery assessment with intra-arterial digital subtraction angiography (IaDSA) and subsequent patient management. MATERIALS: iaDSA studies of patients evaluated for severe claudication (n =5) or critical ischaemia ( n =43). METHODS: arteriograms were independently judged by four observers. The popliteal and tibial arteries were graded as fully patent, severely diseased, occluded or non-diagnostic. The dorsalis pedis, common and deep plantar artery were graded as directly, indirectly or not filling the pedal arch or non-diagnostic. Agreement on grading arteries was expressed as kappa-values. Treatment plans (conservative, PTA, surgery, amputation, non-diagnostic) proposed by each observer based on clinical information and iaDSA were compared. RESULTS: the rate of non-diagnostic judgements ranged from 1% in the popliteal to 22% in the pedal arteries. Overall agreement was good for grading the popliteal arteries (kappa=0.64), moderate for the tibial (kappa=0.47--0.54) and fair for the pedal arteries (kappa=0.39). Agreement was good to excellent for grading occluded or fully patent popliteal and tibial artery segments, and fair to moderate for grading severe disease. In 57% of cases at least 3 observers proposed identical treatment, which indicates fair overall agreement (kappa=0.33). CONCLUSION: interobserver agreement on iaDSA is good to determine occluded or fully patent popliteal or tibial arteries, but not for severe disease. This should be taken into account when other diagnostic modalities are compared with iaDSA. Evaluation of diagnostic modalities as concordance in treatment plans is flawed by interindividual variation.