Most of the studies reporting outcome of ECMO patients are based on developmental assessment at one year using the Bayley Scales of Infant Development. There is evidence from other groups of patients in long term follow-up programs that neurodevelopmental outcome beyond one year of age cannot be accurately predicted from the one year assessment. In patients enrolled in our ECMO follow-up program, we assessed the predictive value of Bayley Scales of Infant Development at one year for late (2 to 4 years) severe neurodevelopmental delay (IQ ≤ 70). Between November 1987 and June 1993, 151 patients were treated with ECMO; 101 of the 119 survivors (85%) were followed and 66 patients had cognitive testing at 1 and 2 to 4 years. Patients followed did not differ from the overall ECMO population in terms of pre, intra and post ECMO variables. Mental developmental index (MDI) and psychomotor developmental index (PDI) score correlation coefficients with the Stanford Binet or Differential Ability Scales scores at 2 to 4 years were 0.48 and 0.28, respectively (both <p 0.05). Sensitivity, specificity, positive and negative predictive values of Bayley Scales using a score of ≤70 (1SD) or ≤85 (2SD) as cut off values are shown in thetable: In the ECMO population neurodevelopmental assessment using the Bayley Scales of Infant Development at one year of age allowed the identification of only about 50% of the patients that later had severe neurodevelopmental delay. The sensitivity did not improve when the threshold was increased to 85. We conclude that the Bayley Scales of Infant Development at 1 year cannot be used to identify the group of patients at risk for developmental delay later in infancy or childhood.

Table 1
Full size table