A meta-analysis of variability in conjunctival microvascular hemorheology metrics

Conjunctival hemorheology has been used analytically to assess qualities of blood flow associated with various forms of cardiovascular disorders including diabetes mellitus, stroke, and sickle cell disease. Although conjunctival axial red blood cell velocity (V_ax) has been demonstrated in varying disease states, benchmark measures of V_ax are not well-defined. Due to various methodologic differences in assessment of V_ax, interstudy consistency of hemorheological metrics is susceptible to both systematic and random error. Our study examines interstudy heterogeneity of V_ax as measured in the conjunctival microvasculature of healthy subjects and assesses the overall perturbation of V_ax based on disease state. Furthermore, our study aims to establish a potential range of normative V_ax by comparing inter-study measurements in healthy patients. The most widely employed analytic approach to assess V_ax was space-time analysis (n = 30). Using a meta-analytic approach, the prediction interval for V_ax in healthy subjects among 20 studies ranged from 0.32-2.60 mm/s with a combined effect size of 0.52 ± 0.03 (CI: 0.46-0.59) mm/s. Inter-study comparison of V_ax in healthy patients showed a high degree of variability (I²: 98.96%), due to studies with low measurement precision and/or dissimilar analytic methodology. Neither age nor diameter was a clinically significant moderator of V_ax measurements in healthy patients. The combined effect size, defined as the composite Hedge's g of studies comparing healthy and disease state mean V_ax, was 0.21 ± 0.13. High heterogeneity (I²: 80.48%) was observed in studies analyzing the difference between mean V_ax in healthy and disease state patients. This heterogeneity was also observed when the difference in mean V_ax between healthy and disease state patients was assessed in subgroups based on disease condition (I²: vascular disease 33%, sickle cell disease 62.22%, other 83.43%). Age was found to be a significant moderator (p = 0.048, β = -0.40) of Hedge's g while diameter was not. No significant publication bias was observed in studies presenting healthy patient V_ax or in studies comparing V_ax between healthy and disease state patients. In summary, although homogeneity can be seen in healthy group V_ax measurements, a high degree of statistical heterogeneity is found in V_ax assessment comparing healthy and disease conditions that is not fully explained by methodologic variability.