Data analysis
Group differences were assessed by independent t-test,Mann-Whitney U test ,or x^2 - test,whichever was appropriate. Analysis of covariance(ANCOVA) was used to assess the mean differences of each outcomeat three and six months follow-up.Generalized estimating equation(GEE) to account for the correlated data over six months was used to estimate the average treatment effects. These analyses adjusted for age ,dilation of each outcome. A P