The objectives were to assess the difference in serumretinol binding protein (RBP)levels among pre-school aged children (PSC; 6-59 months) and women of reproductive age (WRA; 15-49 years) during inflammation and/or malaria parasitemia and to investigate different adjustment algorithms to account for these effects.
Cross-sectional data from 8 surveys for PSC (n=8,803) and 4 for WRA (n=4,191) were analyzed individually and combined using meta-analysis. Several approaches were used to adjust RBP concentrations for inflammation and malaria: 1) exclude those with elevated APPs based on C-reactive protein (CRP) >5mg/L or a1-acid glycoprotein (AGP) >1 g/L; 2) apply arithmetic correction factors (CFs), and 3) use a regression correction (RC) approach.
Making no adjustments to RBP concentrations in the presences of inflammation and malaria resulted in an over-estimation of the prevalence of VAD in PSC (defined as <0.7 µmol/L) and vitamin A insufficiency (VAI) in WRA (<1.05µmol/L). Depending on the adjustment approach used, the prevalence of VAD decreased by 4 to 22 percentage points in PSC and VAI decreased by less than 6 percentage points in WRA in comparison to unadjusted prevalence. RBP was associated with malaria independently from CRP and AGP.
The use of CFs and RC adjustments are promising approaches to estimate the prevalence of VAD and VAI in regions with inflammation and malaria. The RC adjustment should be used when feasible, since it accounts for severity of inflammation and likely better reflects RBP values in the absence of inflammation.