| Objective:The purpose of this study was to explore whether and how rapid eye movement sleep(REM)sleep behavior disorder(RBD),depression and anxiety mediate impulse control disorders(ICDs)in patients with dopaminergic replacement therapy(DRT)and Parkinson’s disease(PD)using longitudinal cohort data.Methods:Subjects were selected from the Parkinson’s Progression Marker Initiative(PPMI).After excluding missing data,268,223,218,238,and 219 patients with PD diagnosed 12,24,36,48,and 60 months earlier,respectively,were included.We used the Questionnaire for Impulsive-Compulsive Disorders,the RBD Screening Questionnaire,the Geriatric Depression Scale,and the State-Trait-Anxiety Inventory to assess,respectively ICD,RBD,depression and anxiety.A causal mediation model based on Bayesian Monte Carlo algorithm proposed by Imai et al.and a causal mediation model embedding the counterfactual structure into a natural effects model proposed by Steen et al.were used to construct a causal mediation analysis model with depression,RBD,and anxiety as mediating variables,DRT as exposure variable,and ICD and its subtype development as outcome variable to infer depression,anxiety,and RBD mediated potential causal pathways from DRT to ICD and its subtypes.The confounders in this study were corrected using logistic regression and multiple linear regression models.Results:The results of causal mediation model based on Bayesian Monte Carlo integration algorithm show that DRT is associated with the increased risk of PD incidence rate.ICD is mediated by depression,RBD and anxiety,and the average causal mesomeric effect accounts for 8.0%,16.4% and 12.7% of the total effect respectively.The study on specific subtypes of ICD found that when depression was the mediator,the total effect of DRT on ICD eating was significant.The average causal mesomeric effect of depression was 0.4%,accounting for 7.2% of the total effect,and the direct effect of DRT was 5.8%,accounting for 92.8% of the total effect.For impulse buying,the total effect of DRT on ICD buying is significant.The average causal mesomeric effect effect of depression is 0.2%,accounting for 5.3% of the total effect.With RBD as the mediating variable,the direct effect of DRT on ICD eating is 5.6%,and the average causal mesomeric effect of RBD is 4.7%,accounting for 16.4% of the total effect.For ICD hobbyism,the direct effect of DRT on ICD hobbyism is 4.8%,and the average causal mesomeric effect effect of RBD is 0.7%,accounting for 13.7% of the total effect.With anxiety as the mediating variable,the direct effect of DRT on ICD eating was 5.8%,and the average causal mesomeric effect of anxiety was 0.6%,accounting for 10.3% of the total effect.The direct effect of DRT on impulsive hobbies was 4.9%,and the average causal mesomeric effect effect of anxiety was 0.4%,accounting for 7.5% of the total effect.The single causal mediation model based on the estimation model of natural effects shows that when depression is used as the mediating variable,the total effect of DRT on ICD is 0.663,of which the natural indirect effect of depression is 0.060,the OR value is1.602(95% CI: 1.341,2.458),the mesomeric effect accounts for 9.05%,and the natural direct effect is 0.603.The total effect of DRT on ICD-eating was 1.038,the natural indirect effect of depression was 0.079,the OR value was 1.082(95% CI: 1.047,2.130),and the natural indirect effect was 0.079.With RBD as the intermediary variable,the total effect of DRT on ICD is 0.717,in which the natural indirect effect of RBD is 0.097,the OR value is1.101(95% CI: 1.030,2.162),the proportion of mesomeric effect is 13.52%,and the natural direct effect is 0.620.Taking ICD eating as the dependent variable,the total effect of DRT on ICD eating is 1.051,in which the natural indirect effect of RBD is 0.164,the OR value is 1.178(95% CI: 1.063,2.263),the proportion of mesomeric effect is 15.60%,and the natural direct effect is 0.887.With anxiety as the mediating variable,the total effect of DRT on ICD was 0.667,of which the natural indirect effect of anxiety was 0.085,the OR value was 1.089(95% CI: 1.021,2.146),the proportion of mesomeric effect was12.74%,and the natural direct effect was 0.582.For ICD eating,the total effect is 1.049,of which the natural indirect effect of anxiety is 0.104,the OR value is 1.109(95% CI:1.027,2.180),the proportion of mesomeric effect is 9.91%,and the natural direct effect is0.946.The multiple causal mediation model shows that when the mediating variables are depression and RBD,the total effect of DRT on ICD is 0.675,in which the combined natural indirect effect of depression and RBD is 0.139,the OR value is 1.149(95% CI:1.051,2.226),the proportion of mesomeric effect is 20.59%,and the natural direct effect is0.563.The total effect of DRT on ICD-eating is 1.035,with a natural indirect effect of0.198,an OR value of 1.219(95% CI: 1.064,2.331),and a natural direct effect of 0.838.When the mediating variables are RBD and anxiety,the total effect of DRT on ICD is0.677,the natural indirect effect is 0.158,and the OR value is 1.171(95% CI: 1.067,2.249,the proportion of mesomeric effect is 23.34%,and the natural direct effect is 0.518.The total effect of DRT on ICD eating is 1.035,the natural indirect effect is 0.104,and the OR value is 1.109(95% CI: 1.083,2.337),the proportion of mesomeric effect is 20.29%,and the natural direct effect is 0.825.Conclusions :1.The results of the single causal mediation model showed that the independent causal mediation effects of depression,anxiety,and RBD based on the Bayesian Monte Carlo integration algorithm hold,indicating that PD patients treated with DRT due to the appearance of depressive symptoms may lead to the emergence of ICD and its subtypes(including ICD-eating and ICD-buying),PD patients treated with DRT due to the The presence of RBD increases the risk of developing ICD and its subtypes(including ICD-eating and ICD-hobbyism),and PD patients with anxiety increase the incidence of ICD and its subtypes(including ICD-eating and ICD-hobbyism)when using DRT.In the single causal mediator model based on the natural effects estimation model,depression,RBD,and anxiety also affect DRT and ICD as independent mediators,but the difference is that when PD patients are depressed,treatment with DRT only affects the occurrence of the subtype of ICD-eating,and similarly,when PD patients have RBD and anxiety,the results remain consistent with depression.2.The results of the multiple causal mediator model: based on the natural effects estimation,if PD patients with depression and RBD or anxiety and RBD,DRT treatment will affect the occurrence of ICD and its subtype(ICD-eating).In conclusion,the results suggest that for PD patients and their clinical investigators,in addition to focusing on changes in RBD,depression and anxiety associated with high dopaminergic status,focusing on the link between RBD,depression and anxiety can help PD patients to better prevent ICDs.3.The results of model application: the conclusions of the causal mediator model based on the Bayesian Monte Carlo integration algorithm and the causal mediator model of the natural effects estimation algorithm are basically the same,the former can estimate the mediator effects present in the pathway more extensively than the latter,but the latter can provide indicators such as OR values and 95% confidence intervals that the former cannot provide.The latter can also provide joint causal mediation effects that are not available in the former when there is a link between mediating variables,which suggests that a Bayesian Monte Carlo integration algorithm is recommended for more extensive estimation of possible mediating effects.In contrast,when there are continuous exposures or covariates in the model,or when there are multiple mediating variables that are linked to each other,a causal mediation model based on natural effects estimation can provide us with joint mediating effect estimates. |
PDF Full Text Request