Table 5.

Results of (random-effects) meta-analysis of cross-sectional studies reporting mean raw test scores

Test	k	$\sum n$	Est. (SE)	95% CI	z-value (P)	${\widehat{\tau }}^2$ (SE)	95% CI ${\widehat{\tau }}^2$	Q-value (P)	$I^2$ -statistic
MMSE	2	2008	−3.513 (0.417)	(−4.330;−2.695)	−8.425 (<.001)*	0.132 (0.952)	(0;>100)	1.244 (.265)	19.6
Coding/substitution	4	890	−0.256 (0.548)	(−1.330;0.817)	−0.468 (.640)	1.082 (0.980)	(0.263;17.144)	35.802 (<.001)*	93.4
TMT A	4	538	−0.227 (0.287)	(−0.789;0.335)	−0.791 (.429)	0.270 (0.268)	(0.054;3.953)	23.659 (<.001)*	88.0
Digit span forward	2	402	−0.410 (0.100)	(−0.607;−0.214)	−4.087 (<.001)*	0 (0.030)	(0; 10.317)	0.062 (.803)	0
Semantic fluency	8	2511	−0.628 (0.223)	(−1.066;−0.190)	−2.809 (.005)*	0.345 (0.213)	(0.123;1.711)	70.037 (<.001)*	91.5
Phonemic fluency	3	938	−0.388 (0.551)	(−1.469;0.692)	−0.705 (.481)	0.822 (0.913)	(0.173;33.907)	35.186 (<.001)*	92.9
Stroop speed interference task	5	642	−0.763 (0.261)	(−1.275;−0.251)	−2.922 (.003)*	0.268 (0.240)	(0.051;2.779)	19.993 (<.001)*	83.9
TMT B	4	538	−0.521 (0.223)	(−0.958;−0.084)	−2.335 (.020)*	0.145 (0.161)	(0.016;2.226)	13.638 (.003)	79.5
Finger tapping dominant hand	2	625	−0.650 (0.425)	(−1.483;0.183)	−1.530 (.126)	0.156 (0.511)	(0;>100)	1.761 (.184)	43.2
Finger tapping non-dominant hand	2	625	−0.424 (0.395)	(−1.197;0.350)	−1.074 (.283)	0.107 (0.440)	(0;>100)	1.523 (.217)	34.3
Digit span backward	3	426	−0.583 (0.099)	(−0.778;−0.388)	−5.873 (<.001)*	0 (0.030)	(0;6.613)	2.370 (.306)	0
Verbal memory delayed recall	8	1410	−0.056 (0.203)	(−0.455;0.342)	−0.277 (.782)	0.258 (0.175)	(0.072;1.321)	34.857 (<.001)*	87.1
Verbal memory delayed recognition	2	513	0.251 (0.561)	(−0.848;1.351)	0.448 (.654)	0.585 (0.893)	(0.079;>100)	13.601 (<.001)*	92.6
Verbal memory immediate	8	1410	−0.172 (0.220)	(−0.603;0.259)	−0.782 (.435)	0.312 (0.205)	(0.097;1.471)	44.979 (<.001)*	88.9

Note. k = number of included studies, $\sum n$ = total number of included patients in a meta-analysis, Est. (SE) = Average effect size estimate and standard error, CI = confidence interval, z-value (P) = z-value and two-tailed P-value to test the null-hypothesis of no effect. ${\widehat{\tau }}^2$ (SE) = estimated between-study variance in true effect size using the restricted maximum likelihood estimator and corresponding standard error, 95% CI ${\widehat{\tau }}^2$ = 95% confidence interval of the between-study variance obtained with the Q-profile method (Viechtbauer, 2007),⁵³Q-value (P) = Q-statistic and P-value to test the null-hypothesis of no between-study variance. $I^2$-statistic = percentage of variance that can be attributed to between-study variance. * indicates a P-value <.05. Tests with moderate or high effect sizes are indicated in bold.