484 CLINICAL DISAGREEMENT IN THE ASSESSMENT OF BPD IN A CLINICAL TRIAL

Disagreement between clinicians in interpreting clinical signs is ubiquitous in medicine. In the interpretation of endpoints in clinical trials, this can be crucial. A randomised clinical trial to test the effectiveness of Vitamin E in LBW babies in the prevention of BPD required the assessment of BPD to be as objective as possible. Three radiologists agreed to interpret the X-rays, independently and blindly. One was a pediatric radiologist; one, a general radiologist, regularly reviews neonatal films; and one was interested in pulmonary disease. The criteria used were those of Northway. X-rays, taken at days 1, 10, 21, 28 and 42 were coded and read. The results of the first batch of 268 films indicated that there was complete agreement in 28% (usually the normal films); agreement between 2 of the 3 in 36% and no agreement in 35%. Possible sources of disagreement included the quality of the films and the comorbidity of the babies as well as reviewing out of context. The decoded films were placed back in order. Agreement was reached on 35% of babies; two agreed on 17% and all 3 agreed on a further 13% but not on the severity of the disease. 35% still had questionable conclusions. Strategies to resolve these differences included adding the complete series of baby films; checking autopsy reports if insufficient films were available and bringing the 3 radiologists together with a neonatologist to resolve interpretative differences.