Chronological order: cause event occurs before effect event

Correlation: the effect event changes as the cause event changes

Causal variability: changes in causal events are caused by changes in causal events

Temporal order and correlation are directly observable, and variability can be demonstrated indirectly by excluding the possibility of other factors.

Definition: The theoretical framework used in modern medicine to distinguish different causes and explain their relationship with disease, their relationship with each other, and their mechanisms of action.

use

The ultimate goal: to discover new causes or grasp the main causes to develop a more comprehensive and effective disease prevention strategy.

Definition: A disease often has multiple independent or interrelated causal chains. Different causal chains of the same disease are connected and intertwined with each other to form a more complex and complete causal network. The overall network structure of connections from etiology to onset is called etiology network

Removing any one factor in a cause chain can completely cut off the entire cause chain, thereby preventing the disease from occurring through this cause chain.

Different causal chains may have different effects on the occurrence of the disease. Effective prevention should cut off the main causal chains.

Different causal chains may independently affect the occurrence of the disease, and cutting off multiple causal chains at the same time will definitely prevent more cases.

Method of seeking common ground: Examine different situations where a certain phenomenon occurs. If all the different situations have different conditions except for one condition, then this same condition may be the cause of a certain phenomenon under study.

Difference method: Compare the occasions when a certain phenomenon occurs and the occasions when it does not occur. If the two occasions are the same except for one difference, then this difference is the cause of the phenomenon.

How to find common differences and similarities: If there is only one common factor in various situations where a certain phenomenon under investigation occurs, and there is no such common factor in various situations where the phenomenon under investigation does not occur, then this common factor is the cause of the phenomenon under investigation (two Seeking common ground multiple times and seeking differences multiple times)

Co-variation method: When there is a variation or change in a certain phenomenon, there is a corresponding variation or change in another phenomenon, and no matter what the variation and change of the latter are, then there may be a causal relationship between the two. When a dose-response relationship exists, a causal relationship is more likely to exist

Residue method: If a certain composite phenomenon is determined to be caused by a certain composite cause, subtract the parts that are confirmed to have causal connections, then the remaining parts must also have causal connections.

Mill's methods of seeking agreement, seeking differences, and seeking consensus are only applicable to sufficient and necessary causes or extremely strong causal relationships, and cannot be directly used to study the causes of chronic non-communicable diseases → Change "all" to "many" or "majority"

Randomized controlled trials use random grouping to completely eliminate confounding and are the most reliable method to verify causal relationships in the population. However, it can only be used to evaluate the effects of treatments and interventions, and cannot be used to directly study the causes of diseases.

Cohort studies have a causal relationship and are the most reliable method to verify the cause of disease in the population. They are also often used to study adverse effects of treatment.

Epidemiological research is based on the principle of Muller's law, collecting evidence of the three conditions of the cause in the population, and using this to infer the existence of causal relationships in medicine.

Three levels of inference: 1. Inference based on a specific study; 2. Inference based on all similar studies; 3. Inference based on all relevant evidence

The validity of the inferences depends on the type of study design, the methodological quality of the study, and the sample size, with study quality being more important than sample size.

Authenticity and research quality

Factors that determine research quality

Methods for evaluating research quality

The Nine Criteria of Hill's Criteria

The existence of correlation and the time characteristics of correlation are necessary and specific conditions for judging causality, and are the basic conditions for judging causality.

Among non-specific criteria, consistency of results is most important

Except for the first 3 items, the lack of any one or all 7 items is not enough to deny the existence of causal relationship.

Even if the relationship between two events meets all 10 conditions, it is not 100% certain that it is a causal relationship.

Definition: A brand-new literature synthesis method that applies certain standardized methods to comprehensively and systematically retrieve all published or unpublished original studies on a specific issue, and uses clinical epidemiology to rigorously evaluate the literature. Principles and methods were used to screen out original studies that met the inclusion criteria, extract relevant data, and perform integrated analysis to draw more scientific and reliable comprehensive conclusions.

Strengthened the systematic collection of original research and quantitative inferences from results

For any controversy about efficacy, SR/Meta analysis is the most solid and reliable evidence