ASQ3 and/or the Bayley-III to support clinicians' decision making.

BACKGROUND: Appropriate tools are essential to support a clinician's decision to refer very preterm infants to developmental resources. Streamlining the use of developmental assessment or screening tools to make clinical decisions offers an alternative methodology to help to choose the most effective way to assess this very high-risk population.

OBJECTIVE: To examine the influence of the Ages and Stages Questionnaire-3rd edition (ASQ3) and the Bayley Scales of Infant Development-3rd edition (Bayley-III) scores within a clinically-based decision-making process.

METHODS: This retrospective cohort study includes children born at less than 29 weeks gestation who had completed both psychologist-administered Bayley-III and physician-observed ASQ3 assessments at 18 months corrected age. Theoretical referral decisions (TRDs) based on each assessment results were formulated, using cut-off scores between the lower first and second standard deviation values and below the lower second standard deviation values. TRDs to refer to developmental resources were evaluated in light of the multidisciplinary team's actual final integrated decisions (FID).

RESULTS: Complete data was available for 67 children. The ASQ3 and the Bayley-III had similar predictive value for the FID, with comparable kappa values. Comparisons of the physicians' and psychologists' TRDs with the FIDs demonstrated that the ASQ3 in conjunction with the medical and socio-familial findings predicted 93% of referral decisions.

CONCLUSION: Taking into consideration potential methodological biases, the results suggest that either ASQ3 or Bayley-III, along with socio-environmental, medical and neurological assessment, are sufficient to guide the majority of clinicians' decisions regarding referral for specialty services. This retrospective study suggests that the physician-supervised ASQ3 may be sufficient to assess children who had been extremely preterm infants for referral purposes. The findings need to be confirmed in a larger, well-designed prospective study to minimize and account for potential sources of bias.