TABLE 2.

Summary of grading of recommendations assessment, development and evaluation (GRADE) analysis in randomized trials

Certainty Assessment					Certainty
No. of Studies	Risk of Bias	Inconsistency	Indirectness	Imprecision
Negative symptom domain (follow up: range 8 to 24 wk; assessed with PANSS negative score)
3	Seriousa	Seriousb	Not serious	Not serious	++○○ Low
Cognition symptom domain (follow up: range 12 to 24 wk; assessed with various scales)
2c	Not serious	Seriousd	Not serious	Seriouse	++○○ Low
Positive symptom domain (follow up: range 8 to 24 wk; assessed with PANSS positive score)
3	Seriousf	Not serious	Not serious	Not serious	+++○ Moderate
Treatment attrition for any reason (follow up: range 8 to 24 wk; assessed with all-cause discontinuation)
3	Not serious	Seriousg	Not serious	Not serious	+++○ Moderate

PANSS = Positive and Negative Syndrome Scale.

^aSelective reporting and incomplete outcomes data in study by Sepehrmanesh et al23; did not include prespecified between-group differences in mean PANSS score changes at trial end point. Instead, included findings with F statistic from ANOVA. No SD reported with mean PANSS score, yielding inability to complete t test analysis through independent investigator using summarized data.

^bAlthough consistency of findings were reported among all 3 trials,21-23 the statistics provided by Sepehrmanesh et al23 did not show a between-group difference as stated in the text. So evidence is lacking to state that all 3 studies were consistent in finding.

^cPublications by Berk et al21 and Sepehrmanesh et al23 evaluated for cognitive symptom improvement.

^dBerk et al21 found no differences in subgroup of patients analyzed, whereas Sepehrmanesh et al23 found benefits with N-acetylcysteine across all cognitive function assessments.

^eCannot assess magnitude of effect given published data from Sepehrmanesh et al23 and unable to assess sample size in each group within Berk et al21 study, which was likely not powered to detect a difference.

^fSelective reporting and incomplete outcomes data in study by Sepehrmanesh et al23; did not include prespecified between-group differences in mean PANSS score changes at trial end point. Instead, included findings with F statistic from ANOVA. No SD reported with mean PANSS score, yielding inability to complete t test analysis through independent investigator using summarized data.

^gLarge variability in reported drug reactions and discontinuations between studies.