Skip to main content
. 2015 Jul 15;57(2):161–170. doi: 10.1111/jcpp.12447

Table 2.

Classification statistics for DAWBA

DAWBA result Sensitivity Specificity PPV NPV Correctly classified overall (%)
Computer generated probability a
Positive if using this cut‐off ‘Very low (.1%)’ 1.00 .00 .41 40.6
‘Low (.5%)’ .94 .79 .75 .95 84.8
‘Low (3%)’ .88 .85 .80 .91 86.2
‘Moderate (20%)’ .69 .91 .84 .81 81.9
‘50/50%’ .38 .97 .89 .69 72.8
‘High (80%)’ .20 .99 .96 .64 67.0
DAWBA clinical rater a
Positive if rating of Any ASD .86 .87 .82 .90 86.6
Autism/Asperger's .54 .98 .95 .76 80.5
DAWBA clinical rater ± ADOS b
Positive if rating of Any ASD .86 .80 .83 .83 83.0
AND ADOS +ve .74 .95 .94 .77 84.0
OR ADOS +ve .98 .69 .78 .97 84.4
Autism/Asperger's .54 .97 .95 .66 74.5
AND ADOS +ve .49 .99 .98 .63 72.6
OR ADOS +ve .92 .82 .85 .90 87.3

Classification statistics for DAWBA probability bands (upper panel), for diagnosis assigned by clinical rater (middle panel), and for diagnosis by clinical rater in combination with ADOS classification (met ASD criteria according to revised algorithm; see online supplement). PPV, positive predictive value, indicates the proportion of children at and above each cut‐off who have BERD of ASD; NPV, negative predictive value, indicates the proportion of children below each cut‐off without a BERD of ASD.

a

Comparison group includes unaffected cotwins and low‐risk twins.

b

Comparison group is unaffected cotwins only.