模板:剂量-反应方法学部分描述

经常有小伙伴发邮件或者微信问我剂量-反应部分的方法学如何描述。这个部分的描述确实是一个加分项，描述得专业档次立马就上来了。以下是我们给大家准备的3个模板（请忽略可能存在的语法问题），使用的时候自行做点修改哈。

 

装逼版1

A dose-response relationship was established by treating ABC as independent continuous variable and regress it against the dependent variable XYZ. Considering the dose-response relationship could likely be a non-linear pattern, we used the restricted cubic spline function to fit the dose-response curve [1]. This was done by setting three fixed knots at the 5, 50, and 95 quartiles of the of ABC distribution (This was done by setting three random knots on the quartiles of the ABC distribution) [2]. The Wald test was used to test the non-linearity with the null hypothesis as the coefficients of the non-linear terms equals to zero. The robust-error meta-regression (REMR) model was used to synthesis the dose-response relationship across available studies [3]. This is a one-stage procedure that treat each study as a cluster and weight the effects of each study by its inverse variance while employs the robust-variance to address the potential correlation of the within-study effects. The advantage of the one-stage procedure is that it only requires 2 (including the reference) or more categories for each study, rather than 3 or more in the classical two-stage procedure [4]. In addition, the robust-variance has been proved to be a good solution in case of the occurrence of outliers [3].

参考文献

1. Durrleman S, Simon R. Flexible regression models with cubic splines. Stat Med 1989:8(5):551–561

2. Xu C, Liu Y, Jia PL, et al. The methodological quality of dose-response meta-analyses needed substantial improvement: a cross-sectional survey and proposed recommendations. J Clin Epidemiol. 2019; 107:1-11.

3. Xu C, Doi SAR. The robust error meta-regression method for dose-response meta-analysis. Int J Evid Based Healthc. 2018; 16(3):138-144.

4. Xu C, Doi SAR. Dose-response meta-analysis (Book Chapter). Springer, Singapore, 2020.

 

装逼版2

We attempts to establish a dose-response relationship between ABC and XYZ through the robust-error meta-regression (REMR) model [1]. This is a one-stage method that considering all included studies as a whole while treat each study as a cluster in order to validate the fitting of dose-response curve. There are generally three possible dose-response gradients in epidemiological studies: linear,non-linear, and categorical dose-response relationship [2]. The latter one is not as precise as the former two [3]. A linear or non-linear relationship could be established when there is sufficient dose-specific points, say, 10 points for linear, and 15 points for the non-linear relationship [4]. In order to considering both linearity and non-linearity in one model, we used the restricted cubic spline (RCS) function to fit the dose-response trend. Three fixed knots (5, 50, 95)/random knots was used in the RCS function [4]. These three knots will divided the data into four pieces and the dose-response curve was fitted within each piece and further smoothed at the knots. The first and last pieces will be restricted as linear while for the 2^nd and 3^rd pieces a cubic spline was fitted. When the cubic and quadratic terms in the function equals to zero, RCS automatically degrade into a simple linear function. The Wald test was used to test the probability that the cubic and quadratic terms equals to zero [5].

参考文献

1. Xu C, Doi SAR. The robust error meta-regression method for dose-response meta-analysis. Int J Evid Based Healthc. 2018; 16(3):138-144.

2. SteenlandK, Deddens JA. A practical guide to dose-response analyses and risk assessment in occupational epidemiology. Epidemiology. 2004; 15(1):63-70.

3. XuC, Doi SAR. Dose-response meta-analysis (Book Chapter). Springer, Singapore, 2020.

4. DurrlemanS, Simon R. Flexible regression models with cubic splines. Stat Med 1989:8(5):551–561

5. XuC, Liu Y, Jia PL, et al. The methodological quality of dose-response meta-analyses needed substantial improvement: a cross-sectional survey and proposed recommendations. J Clin Epidemiol. 2019; 107:1-11.

 

简洁版

In order to establish an “average” dose-response relationship between ABC and XYZ based on the data of all available studies, the robust-error meta-regression (REMR) model was employed in this study [1]. The robust variance was used in the REMR model to address the potential correlations among the with-study effects as these effects share the same reference within the study [2]. A non-linear curve was fitted by the restricted cubic spline in the regression [3]. The Wald test was used to test the non-linearity by assuming the coefficients of the non-linear terms as zero.

参考文献

1. Xu C, Doi SAR. The robust error meta-regression method for dose-response meta-analysis. Int J Evid Based Healthc. 2018; 16(3):138-144.

2. Xu C, Liu Y, Jia PL, et al. The methodological quality of dose-response meta-analyses needed substantial improvement: a cross-sectional survey and proposed recommendations. J Clin Epidemiol. 2019; 107:1-11.

3. Durrleman S, Simon R. Flexible regression models with cubic splines. Stat Med 1989:8(5):551–561