Page 111

proefschrit Van Zwam inhoud met kaft en stellingen.indd

Follow up of coiled intracranial aneurysms 107 If only one follow-up per patient was assessed in the 14 patients who had follow-up at both six and 24 months, the results did not differ substantially from those also containing the data of the second study (Table 3). Table 3. Test characteristics of CEMRA in the follow-up of coiled aneurysms for two different dichotomizations in the three-point Roy and Raymond classification, only one follow-up assessment per patient. Adequate – Inadequate Complete – Incomplete N=74 (1 + 2) - (3)* (1) - (2+3)* Sens (%) 78.9 (60.6 – 97.3) 95.5 (89.4 – 100) Spec (%) 52.7 (39.5 – 65.9) 53.3 (35.5 – 71.2) NPV (%) 87.9 (76.8 – 99.0) 88.9 (74.4 – 100) PPV (%) 36.6 (21.9 – 51.3) 75.0 (63.7 – 86.3) Linear weighted Kappa for interobserver agreement in comparing occlusion status at follow-up with initial occlusion status at DSA after treatment was only 0.40 (95% CI: 0.18 – 0.54). Sensitivity and specificity for new or increased recanalization compared with initial occlusion status were 80.6 (95% CI: 67.6 – 93.5) and 64.7% (95% CI: 51.6 – 77.8) respectively for the first observer and 77.8 (95% CI: 64.2 – 91.4) and 64.7% (95% CI: 51.6 – 77.8) for the second observer. PPV was 61.7 (95% CI: 47.8 – 75.6) and 60.9 (95% CI: 46.8 – 75.0) for the first and second observer respectively and NPV 82.5 (95% CI: 70.7 – 94.3) and 80.5 (95% CI: 86.4 – 92.6) respectively. Discussion Most test characteristics dealing with occlusion status in this study show lower values than those previously published. In the two meta-analyses of Kwee and Kwee13 and Weng et al14 pooled sensitivities and specificities were calculated out of 16 studies in both cases, partly overlapping. Pooled sensitivities for detecting any residual flow (incomplete occlusion, R&R class 2 and 3) were 83.3% and 90% respectively, and specificities 90.6% and 95% respectively. However, the studies in these meta-analyses were very heterogeneous because different analyzing methodologies were used. A more recent multi-center study as well as some smaller single center studies used the same methodology as we did.16-19,24 Kappa values are comparable with those in our study, ranging from 0.49 to 0.57 for studies employing CEMRA, and from 0.55 to 0.64 for those in which TOF-MRA was used. The NPV for inadequate occlusion (R&R class 3) is also in the same range: NPV 92% in two studies evaluating CEMRA17,18 and NPV ranging from 85% to 94% in the studies evaluating TOF-MRA.17-19,24 For detecting any degree of recanalization (R&R class 2 and 3) the NPV in the present study is even better than in the other studies (72 and 81% in the CEMRA studies and 68 – 87% in the TOF MRA studies). Sensitivity in our study for detecting both inadequate and incomplete occlusion was 82%, and this is higher (though not significantly so) than in the above-mentioned studies, where sensitivity for detecting


proefschrit Van Zwam inhoud met kaft en stellingen.indd
To see the actual publication please follow the link above