Early detection of infants with neurodevelopmental concerns indicative of cerebral palsy in a lower middle-income country (India).

Aim: To determine reproducibility and diagnostic accuracy of screening tools for neuromotor concerns indicative of cerebral palsy (CP) at 18 months corrected age by using the General Movements Assessment (GMA) and/or Hammersmith Infant Neurological Examination (HINE) in West Bengal, India.

Method: This prospective substudy tested psychometrics of screening tools nested within an overarching randomized control trial. A total of 785 infants with birth/infant-detectable risk factors, aged 12 to 40 weeks corrected age (n = 422 male, mean corrected age 22.6 weeks, SD = 10.2), were recruited. Infants were screened for 'high-risk CP' using the GMA (absent/abnormal fidgety, 12-17 weeks corrected age) and/or HINE (3 months < 56, 6 months < 59, 9 months < 62, 18-40 weeks corrected age). 'Neuromotor concerns indicative of CP' were classified at 18 months corrected age by a physician from a videoed neurological examination and semi-structured movement protocol. We analysed the results (1) using Gwet's AC1 and (2) for sensitivity and specificity.

Results: Interrater reproducibility was strong (Gwet's AC1 = 0.89, p < 0.001). A total of 165 out of 749 assessments were screened as 'high-risk CP' (22.0%; 95% confidence interval 19.2-25.1). The screening programme (GMA/HINE) was 80.1% accurate (GMA [only] sensitivity = 87.8%, specificity = 44.4%; HINE [only] sensitivity = 94.0%, specificity = 60.0%).

Interpretation: The GMA and/or HINE are reliable and accurate tools for screening high-risk populations in India, and may be useful in other low- and middle-income countries to identify infants with neuromotor concerns indicative of CP who could be triaged to early intervention.