Classification of mediation analysis methodology in clinical research - Pubrica
                     CLASSIFICATION OF  
MEDIATION 
ANALYSIS  
METHODOLOGY IN  
CLINICAL RESEARCH
An Academic presentation by
Dr. Nancy Agens, Head, Technical Operations, 
Pubrica  Group: www.pubrica.com
Email: [email protected]
Today's Discussion
Outline
In Brief
What is Mediation Analysis
Classification of Mediation Analysis 
Methodology  Conclusion
Future Scopes
In Brief
Mediation Analysis as a method to connect the causal-effect relation in  
being used popularly in Medical writing for clinical research, specially since 
 the last decade. The type of Mediation Analysis model is dependent on 
the  needs of the study and each has its own set of advantages and  
disadvantages. Since there is significant heterogeneity in the manner it is  
used, there is need for methodological uniformity.
What is  Mediation analysis (MA) is a form of statistical analysis and is  
Mediation very commonly used in epidemiology, psychology, sociology, and 
 medicine by Scientific Medical Writing Companies.
 Analysis?
It attempts to predict the relation between the antecedent  
variable and the outcome variable by Clinical Research Services.
In other words, Medical Writing in Clinical Research use MAto try 
 and find methodology by which outcome is related to exposure.
Thus, it is useful in both interventional and observational studies 
 while scientific research paper writing or Medical Writing for  
Clinical Trials.
Research paper writing help use mediation model as a causal model where 
 mediator is presumed to cause the effect, while the vice versa is not true.
Statistics is used to evaluate whether the presumption of the MA is correct 
or not.
Scientific writing help use this information while research paper writing and it is 
 useful to find the association with causation.
This helps understand the disease aetiology better and this information is used 
 to improve or alter existing treatment and management modality.
Classificatio 1. TRADITIONAL APPROACHES SINGLE MEASURE: -
n  of 
Mediation  In this method, separate regression models are used for  
Analysis  the antecedent and the outcome variable in 
clinical medical  research. E.g.
Methodology
Baron and Kenny’s framework is commonly used by
Medical Research Companies.
X is the causal variable and Y is the outcome in the  
absence of any mediation as per the below representation. 
 Path C is the total effect.
Contd.
.
Mediating variable M in the below figure is the process or intervention variable.
Path C´ in this figure is the direct effect. When the path from X to Y is reduced but 
 above zero, there is partial mediation, while when C´ becomes zero, there is  
complete mediation.
There are four steps in the Baron and Kenny’s model in establishing mediation.
Contd.
.
Step 1: - To show that the antecedent variable correlates with the outcome variable.
Step 2: - To show that the antecedent variable correlates with the mediator.
Step 3: - To show that the mediator affects the outcome variable.
Step 4: - To establish that the mediator completely mediates the causal- effect relationship. 
 Total effect = direct effect + indirect effect OR
C = C´ + ab
Complete mediation is proved when all the four steps are met, while partial mediation is 
 indicated if Step 4 is not met.
Contd.
.
Contd.
.
2. COUNTERFACTUAL APPROACHES SINGLE MEDIATOR: -
Lewis’s 1973 counterfactual analysis is based on the understanding that, if an  
antecedent event is the complete causal factor for the effect, then, there would not be 
 the effect if there is no antecedent event.
Medical Writing CRO use different regression models for the antecedent and the  
outcome. Causal contrast is highlighted between the exposed and the non-exposed.
In the causal-effect model, when there are multiple mediators that need to be 
 evaluated, one can consider one mediator at a time.
This is possible only when mediators do not affect each other.
Contd.
.
However, if we need to evaluate several mediators simultaneously or in cases where 
 mediators affect each other, we need to use either parallel or serial MA.
Contd.
.
Mediation with single mediator M, A is exposure, Y is the outcome and C is 
 the confounder.
Contd.
.
Mediation with two mediators M(1), M(2), A is exposure, Y is outcome and C is 
 the confounder.
Contd.
.
Mediators with unknown common cause, where mediators do not affect each other.
Contd.
.
3. P ARALLEL MEDIATORS:
Medical Writing for Clinical research uses parallel multiple mediators that need to 
 be evaluated at the same time.
Regression model is used for all mediators and separate models are used for 
 different mediators.
4. S ERIAL OR SEQUENTIAL MEDIATION: -
Serial mediation is used when mediators act one after the other for the outcome or 
 the chain reaction.
Contd.
.

Conclusion
MA is a statistical technique which is being used
increasingly by Research Paper Writing Services and 
Medical  Writing in Clinical Trials.
There is no single approach to MA, researchers may or may 
 not have a prior hypothesis.
Scientific research writing help may focus on direct or 
indirect
antecedent causal evidence or association-based studies.
Each model of MA has its own advantages and disadvantages 
 and the model chosen depends on the type of study.
Future 
 MA is based on strong assumptions that need to be 
Scope  discussed to avoid misinterpretations.
s Planning MA in advance is preferred as it helps in adjusting
in the methodologies to account for appropriate sample size, 
 key confounders, and measurement errors in mediators.
There is considerable heterogeneity in the planning, conduct 
 and reporting of MA, necessitating the need to have uniform  
and consensus-based guidelines on the methodology.
Contact 
Us
UNITED KINGDOM
+44-1143520021
INDIA
+91-9884350006
EMAIL
[email protected]