3.3 Tumor response according to RECIST criteria
Fifty-seven studies used RECIST criteria to assess TRR. A meta-analysis of CR and TRR was performed. RR ≥ 1 (IV model, fixed, 95% confidence interval), it is beneficial for the test group. Based on the different dosage forms of medicine, they were divided into 6 groups for meta-analysis: total (57 studies); capsule group (7 studies); granule group (2 studies); decoction group (22 studies); TCM syndrome group (5 studies); injection group (21 studies).
Total group. In 57 studies (n = 4266, Table 1), ORR improved significantly in the experimental arm (RR 1.39, 95% CI [1.30, 1.50]); P=0.621 > 0.05, I2 = 0%), indicating low heterogeneity, the fixed effect model was used for calculating OR value of combined effect size (Figure 2).
Capsule group. Seven studies were included in the capsule group (n = 547, Table 1). TRR improved significantly (RR 1.30 [1.05-1.63], I2 = 0%). And the TRR funnel plot is symmetric. Two studies were included in the granule group (n = 73, Table 1). There may be effective  improvements for TRR (RR 0.81 [0.40-1.67], I2 = 0%). TRR funnel plot is slightly asymmetric, maybe the sample size is too small.
Decoction group. Twenty-two studies were included in the decoction group (n = 1629, Table 1). Significant improvement in TRR (RR 1.34 [1.19-1.50], I2 = 0%). The TRR funnel plot is obviously symmetrical.
TCM syndrome group. Five studies were included in the TCM syndrome group (n = 372, Table 1). There was also a significant improvement in TRR (RR 1.51 [1.17-1.93], I2 = 0%). The TRR funnel plot is symmetric.
Injection group. Twenty-one studies were included in the injection group (n = 1645, Table 1). TRR improved significantly (RR 1.48 [1.32-1.66], I2 = 0%). The TRR funnel plot is symmetric.
To put it briefly, the curative effect of adjuvant treatment of lung cancer with traditional Chinese medicine was observed in the order: TCM syndrome group > injection group > decoction group > capsule group > granule group.