Stata 12 for windows

STATA12 for Windows was used for the statistical analysis. Data are presented as the means ± s.e.m. Two-sided tests were used throughout the analysis, and the differences were considered statistically significant at P < 0.05. Pairwise (univariate) comparisons were performed using Student’s t test or the Mann-Whitney U test as appropriate.

STATA 12 for Windows (Stata Corp, College Station, TX) was used for analysis. Prevalence of cognitive impairment using MMSE, MoCA and LCT and their respective 95% confidence intervals (95% CI) were estimated. Mean and standard deviation (SD) or median and interquartile range (IQR) were utilized to describe the distribution of quantitative variables. To allow for comparisons between tests, we used non-standardized (raw scores) and standardized (z-score) values of MMSE, MoCA and LCT. This standardization technique rescaled raw scores into a new variable with a mean of 0 and a standard deviation of 1. Pearson correlation coefficients (r) were calculated to measure the strength of the linear relationship between two tests. As previous studies comparing two quantitative methods using correlation have been criticized, Bland-Altman plots, calculated using the difference between the methods (X − Y) against the average of them (X + Y) / 2 [37] (link), were used. This simple parametric approach allows us to assess error and bias, spot outliers and detect trends.
Finally, using each of the score's cutoff values for cognitive impairment, Kappa statistics (κ) were calculated.

Summary values were expressed as counts, mean ± s.d., or medians as appropriate. Several PSA values were reported as <1.5 ng/mL. PSA values were therefore censored at 1.5 ng/mL and these values were included. Summary measure for PSA was expressed as medians and inference done by nonparametric means. Data were analyzed using Stata 12 for Windows (College Station, USA).

Data management and rate calculation of END were estimated by STATA 12 for Windows (Stata, College Station, TX, USA). Since meta‐analysis of rate generally has a significant heterogeneity, random effects model was chosen. A chi‐square test was used to examine heterogeneity between the included studies, and a p < 0.10 was considered statistically significant (Higgins, Thompson, Deeks, & Altman, 2003). To explore sources of heterogeneity, subgroup analyses were performed by continent (Asia, Europe or North America), OTT (≤120.0, 120.1–179.9 or 180.0–270.0 min), NIHSS at admission (≤4.0, 5.0–15.0, or 16.0–20.0), thrombolytic drug dose (0.6, 0.7, 0.8, 0.9 or 1.1 mg/kg), and study quality (high quality, moderate quality, or low quality). To evaluate the influence of each study on the pooled rate of END, sensitivity analysis was conducted. Publication bias was examined by using Begg's correlation and Egger's regression (Begg & Mazumdar, 1994; Egger, Davey, Schneider, & Minder, 1997). A p < 0.05 was considered to be statistically significant.