Early detection of infants at high risk of cerebral palsy in a low-middle income country

Abstract

Objective: Many children living in low-middle income countries (LMIC) with cerebral palsy (CP) do not receive diagnosis until school age, delaying access to early intervention. An early detection program in India aimed to determine: (i) proportion of infants determined 'high risk' and/or confirmed diagnosis of CP, (ii) predictive validity and reproducibility of the tools, and (iii) risk factors associated with 'high risk'/ CP diagnosis. Design: Prospective cross-sectional study of a geographical cohort. Method: Eight hundred and thirty-four infants were referred with birth complications aged 12-40 weeks corrected age (CA) (n = 419 male, mean CA 22.3 weeks SD = 10.2) were identified in Kolkata urban slums and two surrounding rural districts. The infants were screened using General Movements (GMs) (12-17 weeks CA), and Hammersmith Infant Neurological Examination (HINE) (18-40 weeks CA). Trained assessors administered and videoed assessments; which were scored by certified GMs scorers masked to clinical history. 'High risk status' was determined by absent/ abnormal fidgety on single GMs video, and published cut-scores for HINE. Birth risk factors were extracted from medical records or parent report. CP diagnosis and Gross Motor Function Classification System (GMFCS) was diagnosed from video by an Australian physician at 18 months CA. Analysis included: (i) sensitivity/ specificity, (ii) Cohen's kappas, percentage agreement (inter-rater reproducibility), and (iii) logistic regression (risk factors). Results: Eight hundred and seven assessments were completed on 785 infants (95.4% rateable): (i) 51.8% were assessed on GMs and 48.2% on HINE, resulting in 165 (22.0%; 95% CI 19.2, 25.1) infants at high risk of CP. 94/139 infants at 18 months had CP diagnosis, (ii) sensitivity of screening was excellent (87.8-97.2%), however specificity was poor (38.1-60.0%), with HINE performing better than GMs (∼10%), (iii) inter-rater reproducibility was strong, screening overall 92.7% agreement, kappa = 0.79, p < 0.001, and (iv) higher odds of 'high risk' outcome for infants with seizures (OR = 8.5 [5.0, 14.5]; p < 0.001) and birth asphyxia (OR = 2.7 [1.9, 3.8]; p < 0.001), but lower odds with low birth weight (OR = 0.7 [0.5, 1.0]; p = 0.05) or NICU admission (OR = 0.6 [0.4, 0.8]; p = 0.002). Poor recall of post menstrual age and completion of only one GMs video may have influenced specificity. Conclusion: Early detection of infants with risk factors is effective for reducing age of identification of those at risk of CP in a LMIC. The GMs and HINE were feasible and efficient tools for use in this context, although overdetected infants at high risk (low specificity). Improved efficiency of screening may be achieved by targeting infants with birth asphyxia and seizures; and undertaking two GMs videos.L6381698252022-06-10 <br />