Upper level of the spina bifida defect: how good are we?

To assess the accuracy of obstetric sonography in determining the upper level of myelomeningocele lesions.

This was a retrospective study of 171 consecutive cases of spina bifida repaired in utero. The upper level of the lesion as determined by obstetric sonography was assigned by community physicians prior to referral in the second trimester and by the authors at Vanderbilt University Medical Center during preoperative evaluation. One hundred and eleven cases had levels established by plane-film X-ray or magnetic resonance imaging after delivery and this was regarded as the gold standard.

Of the 171 community examinations, only 29% identified a specific upper level of the lesion; our corresponding examinations specified the lesion level in all cases. Of the 111 cases that had upper levels of the lesion established by post-delivery imaging, corresponding levels were available for comparison from 35 of the community examinations and from 111 of the examinations performed at Vanderbilt. All three assigned levels were available for comparison in 35 cases. In 26% of cases, community-assigned levels agreed exactly with post-delivery levels, while 66% agreed within one level and 80% agreed within two levels. In 38% of cases, levels assigned at Vanderbilt agreed exactly with post-delivery levels, while 78% agreed within one level and 96% agreed within two levels. Upper levels of the lesion assigned at Vanderbilt were significantly more accurate overall compared with those assigned by community physicians (signed rank test [paired comparison], P = 0.048). However, comparison of lesion levels assigned at Vanderbilt in the first 50 vs. the last 61 cases revealed a significant learning effect (Fisher's exact test, P = 0.03). When comparison of lesion levels assigned by community physicians was restricted to the first 50 cases at Vanderbilt, accuracy was similar (n = 13; t-test, P = 0.16; rank sum test, 0.31).

Community physicians were successful in assigning the upper level of the spina bifida lesion only 29% of the time. When successful, the accuracy of these determinations was similar to that of the authors at Vanderbilt. A significant learning effect was demonstrated by improved accuracy over time at Vanderbilt. A concerted continuing medical education effort is indicated to improve the imaging skills of physicians in the accurate diagnosis of the severity of spina bifida in fetuses.