Lipid profile alterations in pulmonary tuberculosis patients with and without diabetes mellitus: a systematic review and meta-analysis

Figures

• 

  Fig. 1. PRISMA flow diagram of the study selection process

• 

  Fig. 2. Forest plots of mean differences in lipid concentrations (mg/dL) between TB-DM and TB-only patients across included studies (n = 9). Positive MD values indicate higher concentrations in the TB-DM group relative to the TB-only group. Pooled estimates were derived using random-effects models with between-study variance estimated by REML and HK-adjusted CIs. Six studies [13–16, 19, 20] required data transformation, including median-to-mean conversion, unit conversion, or weighted subgroup pooling

• 

  Fig. 3. Forest plots of mean differences in lipid concentrations (mg/dL) between TB-DM and DM-only patients across included studies (n = 6). Negative MD values indicate lower concentrations in the TB-DM group relative to the DM-only group. Pooled estimates were derived using random-effects models with between-study variance estimated by REML and HK-adjusted CIs. Four studies [14, 15, 19, 20] required data transformation, including median-to-mean conversion, unit conversion, or weighted subgroup pooling

• 

  Fig. 4. Sensitivity analysis based on studies not requiring median-to-mean transformation (n = 5) in the primary comparison (TB-DM vs. TB-only). Positive MD values indicate higher concentrations in the TB-DM group relative to the TB-only group. Pooled estimates were derived using random-effects models with between-study variance estimated by REML and HK-adjusted CIs. No studies required median-to-mean transformation

• 

  Fig. 5. Sensitivity analysis based on studies with low risk (n = 6) in the primary comparison (TB-DM vs. TB-only). Positive MD values indicate higher concentrations in the TB-DM group relative to the TB-only group. Pooled estimates were derived using random-effects models with between-study variance estimated by REML and HK-adjusted CIs. Of the six included studies, Dong et al. [16] required data transformation, including median-to-mean conversion, unit conversion, or weighted subgroup pooling. Gui et al. [19] required unit conversion or weighted subgroup pooling. No study required median-to-mean transformation

• 

  Fig. 6. Sensitivity analysis based on studies that excluded patients on lipid-lowering drugs (n = 6) in the primary comparison (TB-DM vs. TB-only). Positive MD values indicate higher concentrations in the TB-DM group relative to the TB-only group. Pooled estimates were derived using random-effects models with between-study variance estimated by REML and HK-adjusted CIs. Of the six included studies, three [13, 15, 16] required median-to-mean conversion, two [13, 16] required unit conversion and one [16] required weighted subgroup pooling. The remaining three [23–25] required no data transformation

• 

  Fig. 7. Leave-one-out sensitivity (n = 9) analysis among the primary comparison (TB-DM vs. TB-only). Pooled estimates were derived using random-effects models with between-study variance estimated by REML. Six studies [13–16, 19, 20] required data transformation, including median-to-mean conversion, unit conversion, or weighted subgroup pooling