ABSTRACT
Objective:To characterize the distribution characteristics of choroidal thickness in healthy normal subjects and to define the diagnostic cut-off value for pachychoroid.Methods:A cross-sectional study design was carried out.Four hundred and forty-six eyes of 230 healthy subjects from the pachychoroid disease spectrum (PCD) cohort in Beijing Tongren Hospital from April 2018 to June 2021, were enrolled for the choroidal thickness distribution analysis.Three hundred and fourteen eyes of 274 patients with PCD including 149 eyes of 113 patients with central serous chorioretinopathy, 95 eyes of 81 patients with polypoid choroidal vasculopathy, 70 eyes of 60 patients with neovascular age-related macular degeneration, along with 382 eyes of 199 normal subjects matched for refractive error, age and gender with PCD were selected for likelihood ratio analysis.Routine eye examinations including the best corrected visual acuity, intraocular pressure, slit-lamp microscopy, dilated fundus examination and color fundus photography were performed in all subjects.Swept-source optical coherence tomography (SS-OCT) of 9 mm×9 mm scanning mode was used to measure the subfoveal choroidal thickness (SFCT) automatically in nine macular regions according to the Early Treatment Diabetic Retinopathy Study classification system using TOPCON Advanced Boundary Segmentation (TABS) software.Pearson linear correlation analysis and Spearman rank correlation analysis were adopted to evaluate the correlations between SFCT and age, diopter.Multiple linear regression was employed to analyze the factors affecting SFCT.After age and refractive error adjustment, the likelihood ratio test was used to determine the diagnostic cut-off value for pachychoroid.This study adhered to the Declaration of Helsinki.The study protocol was approved by an Ethics Committee of Beijing Tongren Hospital (No.TRECKY2016-054). Written informed consent was obtained from each subject prior to entering the cohort.Results:A negative correlation was found between SFCT and age in normal eyes ( r=-0.34, P<0.001), in both normal male and female subjects ( r=-0.43, P<0.001; r=-0.38; P<0.001). A weak positive correlation was found between SFCT and diopter ( rs=0.19, P<0.001). It was found that age and diopter were strongly correlated with SFCT (both at P<0.001). The cut-off values for pachychoroid in 20-39 years group, 40-59 years group, 60-79 years group and ≥80 years group were 320-330 μm, 330-340 μm, 250-275 μm and 200-225 μm, respectively.The percentages of eyes with pachychoroid in 20-39 years group, 40-59 years group and ≥60 years group were 14.71%(10/68), 24.48%(47/192) and 28.89%(55/184), respectively, showing statistically significant differences among them ( χ2=6.170, P=0.046; LR=6.579, P=0.037). The proportion of pachychoroid in ≥60 years group was significantly higher than that of 20-39 years group, showing a statistically significant difference ( χ2=5.982, P=0.014; LR=6.479, P=0.011). Conclusions:The distribution characteristics of pachychoroid vary in normal subjects over age.Age and diopter are the independent influencing factors of SFCT.
ABSTRACT
The purpose of this article was to introduce the goodness of fit test and its SAS implementation. The main contents included the following four aspects： ① Pearson΄s goodness of fit test； ② deviance or likelihood ratio goodness of fit test； ③ Hosmer-Lemeshow goodness of fit test； ④ goodness of fit test for the sparse data. In the aforementioned “fourth aspect”， there were six specific test approaches， namely “information matrix test” “information matrix diagonal test” “Osius-Rojek test” “unweighted residual sum of squares test” “Spiegelhalter test” and “Stukel test”. The paper implemented the four types of the goodness of fit tests mentioned above with the help of the SAS software through an example， explained the output results， and made statistical and professional conclusions.
ABSTRACT
The purpose of this article was to introduce the likelihood ratio test， six nonparametric tests， and the SAS implementation of the survival data. Based on the assumption that the survival data had the exponential distribution， the likelihood ratio test method was derived， the main difference between six nonparametric test methods was that they had different weight functions. Under the conditions of non-stratification and stratification， the seven survival data hypothesis testing methods mentioned above could be used， and their common point was that their test statistics all followed the χ2 distribution. Through two examples and by means of the SAS software， the article realized the various hypothesis tests for two or more groups of survival data， outputed and explained SAS calculation results， and made statistical and professional conclusions.
ABSTRACT
The purpose of this article was to introduce the likelihood ratio test and the SAS implementation. Specifically， three definitions of the likelihood ratio test statistics and six more commonly used likelihood ratio test statistics were introduced. The three definitions were called based on the spatial size of the parameter vector to construct the likelihood ratio test statistic， based on the two nested statistical models to construct the likelihood ratio test statistic and based on the full model and partial model to construct the likelihood ratio test statistic； the six commonly used likelihood ratio test statistics were the general likelihood ratio χ2 test statistic，the adjusted likelihood ratio χ2 test statistic，the profile likelihood ratio χ2 test statistic，the quasi-likelihood ratio χ2 test statistic，the pseudo-likelihood ratio χ2 test statistic，and the Rao-Scott likelihood ratio χ2 test statistic. In the paper， through two examples， and with the help of the SAS software， the likelihood ratio χ2 tests were realized， the SAS calculation results were output and explained， and statistical and professional conclusions were made.
ABSTRACT
In many studies about biomedical research factors influence on the outcome variable,it has no influence or has a positive effect within a certain range.Exceeding a certain threshold value,the size of the effect and/or orientation will change,which called threshold effect.Whether there are threshold effects in the analysis of factors (x) on the outcome variable (y),it can be observed through a smooth curve fitting to see whether there is a piecewise linear relationship.And then using segmented regression model,LRT test and Bootstrap resampling method to analyze the threshold effect.Empower Stats software developed by American X & Y Solutions Inc has a threshold effect analysis module.You can input the threshold value at a given threshold segmentation simulated data.You may not input the threshold,but determined the optimal threshold analog data by the software automatically,and calculated the threshold confidence intervals.