Table 2 Moderation of cortical metabolite effects by coefficient of variation (COV) as a quality metric.

	Subgroups	K	Pts	HC	Effect Size with (95% CI), or r² and IP	P value	Percent Diff Pt - HC	Heterogeneity I²% P value
Cortical tCho	COV logistic fit	16	234	244	r² = 0.61, IP (g) = −0.31	0.035	n/a	n/a
	COV logistic fit excl 1° outliers	14	193	209	r² = 0.96, IP (g) = −0.48	0.00086	n/a	n/a
	Lo vs Hi COV^a	14	193	209	Lower tCho with <COV	0.05	n/a	0.0, 8.4, 0.76
	COV ≤ 17%^a	8	100	116	−0.64 (−0.36 to −0.92)	<0.00001	−8.0%	0.0, 5.6, 0.58
	COV ≥ 19%^a	6	93	93	−0.24 (+0.05 to −0.53)	0.10	−5.5%	0.0, 2.7, 0.74
Cortical NAA	COV logistic fit	18	269	275	r² = 0.05, IP = n/a	0.37	n/a	n/a
	COV logistic fit excl 1° outliers	16	232	238	r² = 0.18, IP = n/a	0.27	n/a	n/a

^aLow and high COV datasets defined based on COV at the inflection point of the logistic curve. See Methods.
COV coefficient of variation of measured metabolite values, calculated as the unweighted mean of the COVs of patient and control participants for each dataset, IP empirical inflection point of logistic curve in effect size units; excl 1° outliers, after exclusion of datasets with an effect size that fell outside the 95% confidence limits for the primary meta-analysis of patient versus control differences, as shown in Table 1. Other abbreviations as in Table 1.
Rows with bolded values indicate statistically significant effects.

Quick links

Search