TABLE 2.

Six criteria for risk of bias within and across studies.

Source	Steps taken to avoid selection bias	Blinding of participants (1), personnel (2) or outcome assessors (3)	Control groups or tools (*)	More than one clinician involved in evaluations	Inter-rater agreement achieved	Validity (a), internal item consistency (b) and/or reliability testing (c)	Possible bias identified in the selected study
Dammeyer, 2012	✓	Unclear	Unclear	✓	Unclear	Unclear	Small sample size. Comparisons made between the difficulties common between the syndromes.
Peltzer-Karpf, 2012	Unclear	Unclear	✓	Unclear	Unclear	Unclear	Unclear
Funnell & Wilding, 2011	Single case study	Unclear	✓	✓	Unclear	Unclear	Unclear
Absoud et al., 2011	✓	✓ (3)	✓ (*)	✓	X consensus rating used	✓ (a) ✓ (b) ✓ (c)	Referral pattern used. Small sample size. Variation in participant group size.
Parr et al., 2010	✓	✓ (3)	✓	✓	Unclear	Unclear	Retrospective nature of the study as the true rate of ASD in the sample may be higher because ASD knowledge has developed since 1977. The impact of individual differences in environmental experience and input was not assessed.
Hoevenaars-van den Boom et al., 2009	✓	✓ (3)	✓	✓	✓	✓ (b)	Difficulty diagnosing ASD. Small sample size. Diverse aetiologies of VI in participants. Adjustment for behaviour was limited due to the standardisation of the assessment. Psychometric properties of the assessment procedures required further testing.
Peltokorpi & Huttunen, 2008	Unclear	✓ (3)	✓ (*)	✓	✓	Unclear	Modifications were required for both analysis methods due to the participants’ VI. Short duration of the sample of behaviour for each child may only reveal some features of communication. Testing with multiple partners in different environments is required.
Rattray & Zeedyk, 2005	✓	✓ (3)	✓	✓	Unclear	✓ (c)	Unclear
Ashkenazy et al., 2005	✓	Unclear	✓	✓	Unclear	Unclear	Small sample size. Some children in the control group may have developed difficulties later, after the upper age limit of the study.