Table 2.
Subgroup analyses of the associations between RDW and OS in cancer.
| Stratified analyses | No. of patients | No. of studies | Model | Pooled HR (95%CI) | P value | PD value | Heterogeneity | |
|---|---|---|---|---|---|---|---|---|
| I2 | PH value | |||||||
| Tumor type | <0.001 | |||||||
| Hematologic malignancies | 1979 | 10 | fixed | 2.046 (1.623-2.580) | <0.001 | 21.2% | 0.248 | |
| MM | 748 | 5 | fixed | 1.692 (1.256-2.281) | 0.001 | 18.8% | 0.295 | |
| DLBCL | 881 | 2 | fixed | 3.178 (1.853-5.450) | <0.001 | 0.0% | 0.793 | |
| UGI cancer | 3805 | 6 | random | 1.091 (0.925-1.286) | 0.303 | 73.4% | 0.001 | |
| HCC | 1510 | 5 | random | 1.430 (1.232-1.660) | <0.001 | 79.9% | <0.001 | |
| NSCLC | 2304 | 4 | random | 1.440 (1.103-1.880) | 0.007 | 57.2% | 0.053 | |
| Breast cancer | 2627 | 3 | random | 2.092 (0.833-5.255) | 0.116 | 80.3% | 0.006 | |
| Colorectal carcinoma | 843 | 3 | fixed | 1.932 (1.397-2.673) | <0.001 | 0.0% | 0.521 | |
| Gliomas | 287 | 2 | fixed | 1.466 (1.129-1.904) | <0.001 | 23.9% | 0.252 | |
| UTUC | 420 | 1* | fixed | 2.172 (1.599-2.949) | <0.001 | 3.5% | 0.309 | |
| Stage | <0.001 | |||||||
| Mix stage | 16786 | 33 | random | 1.494 (1.372-1.626) | <0.001 | 80.5% | <0.001 | |
| Early stage | 1545 | 5 | fixed | 1.690 (1.180-2.422) | 0.004 | 41.0% | 0.148 | |
| Advanced Stage | 1416 | 6 | random | 1.717 (1.235-2.386) | 0.001 | 57.7% | 0.038 | |
| Age | <0.001 | |||||||
| ≤60 | 7979 | 19 | random | 1.590 (1.321-1.914) | <0.001 | 82.6% | <0.001 | |
| >60 | 7992 | 22 | random | 1.515 (1.351-1.699) | <0.001 | 75.7% | <0.001 | |
| Gender distribution | <0.001 | |||||||
| Female dominant | 5059 | 9 | random | 1.401 (1.153-1.703) | 0.001 | 74.9% | 0.001 | |
| Balanced | 6418 | 21 | random | 1.696 (1.441-1.997) | <0.001 | 74.8% | <0.001 | |
| Male dominant | 5325 | 14 | random | 1.413 (1.232-1.620) | <0.001 | 81.6% | <0.001 | |
| Country | <0.001 | |||||||
| Eastern | 10608 | 28 | random | 1.716 (1.458-2.020) | <0.001 | 79.8% | <0.001 | |
| Western | 8180 | 17 | random | 1.316 (1.203-1.439) | <0.001 | 80.9% | <0.001 | |
| Cut-off value | <0.001 | |||||||
| >15% | 3356 | 6 | random | 1.608 (1.107-2.335) | 0.013 | 89.5% | <0.001 | |
| >14% and ≤ 15% | 7911 | 21 | random | 1.510 (1.351-1.688) | <0.001 | 79.2% | <0.001 | |
| >13% and ≤ 14% | 3409 | 11 | random | 1.869 (1.493-2.340) | <0.001 | 57.5% | 0.004 | |
| ≤13% | 1982 | 5 | fixed | 1.534 (1.262-1.865) | <0.001 | 0.0% | 0.655 | |
| Definition of cut-off value | <0.001 | |||||||
| ROC curve analysis | 6276 | 22 | fixed | 1.569 (1.434-1.718) | <0.001 | 42.6% | 0.015 | |
| Upper limit | 3558 | 11 | random | 1.504 (1.296-1.746) | <0.001 | 70.8% | 0.000 | |
| Median | 2357 | 3 | random | 1.400 (0.961-2.040) | 0.080 | 62.4% | 0.046 | |
| 4th quartile | 2757 | 3 | random | 1.647 (1.430-1.897) | <0.001 | 0.0% | 0.645 | |
| Arbitrary# | 922 | 5 | random | 1.682 (1.073-2.638) | 0.023 | 63.2% | 0.028 | |
| HR calculation‡ | <0.001 | |||||||
| Multivariate | 13572 | 28 | random | 1.477 (1.342-1.626) | <0.001 | 83.9% | <0.001 | |
| Univariate | 4275 | 17 | fixed | 1.525 (1.380-1.686) | <0.001 | 8.5% | 0.355 | |
Abbreviations: MM = Multiple Myeloma; DLBCL = Diffuse large B-cell lymphoma; UGI cancer = upper gastrointestinal tract (UGI) cancers (including esophagus cancer, gastric cancer, and small intestine cancer); HCC = hepatocellular carcinoma; NSCLC = non-small cell lung cancer; UTUC = upper tract urothelial carcinoma; OS = overall survival; HR = hazard ratio; CI = confidence interval; PD = P for subgroup difference; PH = P for heterogeneity.
*: Cheng et al 2015 separately evaluated the survival outcome in two cohorts, which were derivation cohort and validation cohort.
#: Definition of cut-offs value of RDW was based on previous study.
‡: HRs were extracted from multivariate cox proportional hazards models, univariate cox proportional hazards models or survival curve analysis.