Introduction Point-of-care CD4 testing attracts global interest particularly in resource-constrained settings where it is needed the most. We evaluated the diagnostic performance (DP) of the most commonly used POC CD4 test in field settings, the PimaTM, as compared to flow cytometry at CD4 threshold of 350 cells/µl.
Method A systematic literature search and data extractions were performed electronically. Meta-analysis was conducted applying a bivariate multi-level random-effects modelling approach to provide pooled estimates of sensitivity and specificity, and positive and negative likelihood ratios (±LRs) of the PimaTM. In producing estimates, the model accounts for correlation between test sensitivity and specificity and between-study heterogeneity in test performance.
A covariate extended model was also explored to test for difference in DP between capillary and venous blood. Diagnostic statistics and sensitivity analyses were used to examine impacts of outlier bias. User-written STATA programs, MIDAS and Generalised Latent and Linear Mixed Modelling (GLLAMM) were used to undertake statistical analyses.
Results The search identified 13 studies with data currently available for meta-analysis (6 capillary, 7 venous). Pooled sensitivity and specificity of PimaTM were 0.92 (95% CI: 0.88–0.95) and 0.87 (95% CI: 0.85–0.88) respectively with ±LRs indicating strong DP (+LR: 7.0, 95% CI: 6.1–7.9; -LR: 0.09, 95% CI: 0.06–0.13). The extended model showed some difference in DP by blood sample type (venous vs. capillary): sensitivity (0.94 vs 0.89), specificity (0.86 vs 0.87); however, these differences were jointly marginally non-significant (Wald c2(2) = 4.77, p = 0.09).
Conclusion Our study is the first to present pooled data on in-field test performance of the PimaTM. The PimaTM was found to have strong DP in field settings. The difference found in DP by blood sample type, although not statistically significant, may have significant clinical implications which warrant further analysis once more data are available. The recommendation on use of one blood sample type (venous) over the other (capillary) could hinder the scalability of the test.
Disclosure of interest statement Funding for this study was provided by the National Health and Medical Research Council (NHMRC) and Monash University, Australia.