Table 2.
All subgroups |
HIV serostatus subgroups |
||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Q Test† |
HIV+ | HIV- | Meta-regression ROR‡ (95% CI) for HIV+ / HIV- | P‡ | |||||||
Biomarker | No. | OR (95% CI) | I2* (95% CI) | Q-statistic | P | OR (95% CI) | I2 (95% CI) | OR (95% CI) | I2 (95% CI) | ||
IL-6 | 10 | 1.22 (0.97 to 1.54) | 80 (65 to 89) | 45.61 | <.001 | 2.07 (1.19 to 3.60) | 82 (44 to 94) | 1.01 (0.97 to 1.06) | 0 (0 to 69) | 1.96 (1.53 to 2.50) | <.001 |
IL-10 | 8 | 1.24 (0.93 to 1.63) | 82 (65 to 90) | 38.43 | <.001 | 1.20 (0.64 to 2.24) | —§ | 1.25 (0.91 to 1.72) | 84 (69 to 92) | 0.96 (0.33 to 2.83) | .943 |
TNF-α | 9 | 1.18 (1.04 to 1.34) | 63 (23 to 82) | 21.46 | .035 | 1.79 (1.35 to 2.37) | 0 (– to –) | 1.12 (1.02 to 1.23) | 44 (0 to 76) | 1.58 (1.15 to 2.18) | .005 |
CXCL13 | 5 | 1.47 (1.03 to 2.08) | 89 (78 to 95) | 37.00 | <.001 | 2.56 (1.32 to 4.96) | — | 1.35 (0.95 to 1.92) | 91 (79 to 96) | 1.89 (0.69 to 5.23) | .218 |
sCD23 | 4 | 1.57 (1.21 to 2.05) | 90 (77 to 96) | 29.24 | <.001 | 1.59 (1.23 to 2.06) | 0 (– to –) | 1.58 (0.93 to 2.66) | 0 (0 to 69) | 1.00 (0.54 to 1.87) | .996 |
sCD27 | 7 | 2.18 (1.20 to 3.98) | 92 (87 to 96) | 79.67 | <.001 | 4.93 (3.00 to 8.08) | 0 (– to –) | 1.61 (0.89 to 2.93) | 84 (69 to 92) | 3.35 (1.05 to 10.71) | .041 |
sCD30 | 9 | 1.65 (1.22 to 2.22) | 90 (84 to 94) | 83.01 | <.001 | 3.69 (2.40 to 5.69) | 11 (– to –) | 1.40 (1.11 to 1.76) | 44 (0 to 76) | 2.55 (1.38 to 4.73) | .003 |
Higgins’ I2 statistic measuring the proportion of the observed variance between studies relative to the total variance of a set of studies. CI = confidence interval; OR = odds ratio; ROR= ratio of odds ratios.
Q test assessing the degree to which study effect sizes are concordant.
The ratio of odd-ratios compares the odds-ratio for the HIV+ subgroup with that of the HIV- subgroup (HIV+/HIV-). The corresponding P values are computed from a test of the null hypothesis of no difference between the serostatus groups.
“—” and “–” denote Higgins’ I2 statistics and confidence intervals that were not calculated because of inadequate sample size, n = 1 and n = 2, respectively.