Testing two screening instruments for autism spectrum disorder in UK community child health services

AimThe aim of this study was to test the accuracy of two screening instruments in UK Community health services: Modified Checklist for Autism in Toddlers (M-CHAT) and Social Communication Questionnaire (SCQ) for autism spectrum disorder (ASD). A two-stage screening and in-depth assessment procedure, combined with sampling stratification and statistical weighting, allowed the accuracy of the screens to be estimated in the entire population of referred children. MethodThe study included all referrals of children aged 18 to 48months to community paediatric and speech and language therapy services in two London districts over a 12-month period between September 2004 and September 2005. Parents of 808 children were approached; screen data were obtained on 543 children (67.2%). A stratified subsample of 120 children received an in-depth assessment for ASD as defined by the International Statistical Classification of Diseases and Related Health Problems, 10th edition. Community clinician judgement of likely ASD was available for 98 out of the 120 children. ResultsThe sensitivity and specificity were 64% (95% confidence intervals; range 51-80%) and 75% (63-85%) for the SCQ, and 82% (72-92%) and 50% (33-64%) for M-CHAT. There was no evidence that the area under the curve differed between the two screening instruments. There was also no evidence that clinician judgement of likely ASD differed from either of the screening tests. The screening tests did not perform well to confirm preliminary clinical judgement to refer (in series), nor as an alternative indicator for referral (in parallel). InterpretationWhile screening tests may provide useful information, their accuracy is moderate. Screening information in isolation should not be used to make referral decisions regarding specialized ASD assessment.