Objective To establish a multilevel statistical model

quantitatively evaluate therapeutic effect and intervention level of acupuncture and moxibustion treating primary depression and study the similarities and differences between above results and the results of Meta analysis. Methods The researchers retrieved database to collect literatures acupuncture and moxibustion treating primary depression. First of all

the researchers constructed linear model with incidence rate of fitting logarithmic event that evaluated therapeutic effect of acupuncture and moxibustion. They used software MLwi N 2. 26 to fit results for establishing combined model with different interventions as indicator variable. The researchers established the model to evaluate incidence to curative effect of different intervention with acupuncture and moxibustion. Then they used Rev Man 5. 3 software to make Meta analysis to above data and compared the results between the two models. Results Eventually 13 studies were brought into this research. Results of multilevel statistical model showed acupuncture and moxibustion or the two combined with medicine can better improve clinical control rate of patients with primary depression compared with only medicine [RR = 1. 54

95% CI（ 1. 02

2. 32）; RR = 2. 14

95% CI（ 1. 63

2. 81） ]. Results of Meta analysis showed acupuncture and moxibustion combined with medicine can better improve clinical control rate and total effective rate of patients with primary depression compared with only medicine [RR = 2. 24

95% CI（ 1. 68

2. 98）; RR = 1. 12

95% CI（ 1. 05

1. 19） ]. Results of data model with interventions as indicator variable showed that there were no statistically significant difference in clinical control rate

effective rate and total effective rate between only acupuncture and moxibustion and acupuncture and moxibustion combined with medicine（ P > 0. 05）. Collecting results of multilevel statistical model

clinical control rate that intervened with acupuncture and moxibustion was 27. 97%- 28. 14%

however clinical effective rate was30. 36%- 35. 79% and clinical total effective rate was 90. 85%- 91. 82%. Clinical control rate that intervened with acupuncture and moxibustion combined with medicine was 36. 42%- 36. 64%

however clinical effective rate was38. 52%- 38. 84% and clinical total effective rate was 90. 91%- 93. 58%. Conclusion The method of multilevel statistical model may evaluate therapeutic effect of acupuncture and moxibustion and intervention level. It has sufficient evidence to support quality therapeutic effect of acupuncture and moxibustion

and is more accurate than results of Meta analysis.