There are a large number of effective clinical cases of traditional Chinese medicine（ TCM） treatment for primary liver cancer

but the relevant clinical data has the characteristics of large amount data

high heterogeneity and strong dispersibility

and it is difficult to form high-level clinical evidence. Therefore

efficacy evaluation on TCM treatment for primary liver cancer requires an innovative research model. Real-world study（ RWS） is a research method that selects interventions based on the patient’s actual condition and willingness and conducts long-term evaluation of their efficacy. It focuses on the matching between characteristics of clinical practice and the holistic view of TCM. It is suitable for TCM efficacy evaluation in the treatment of primary liver cancer. However

there are many confounding factors and biases in the clinical data based on TCM individualized treatment of primary liver cancer in real environment. At the same time

long-term multiple treatment of liver cancer is easy to cause data loss

affecting the scientific and authenticity of RWS results. In view of the above problems

the confounding factors and biases of primary liver cancer data were analyzed by propensity score method and instrumental variable method. The multi-point follow-up was used to solve the integrity problem of RWS data

and provide ideas for medical efficacy evaluation in primary liver cancer.