Dried Blood Spots in Neonatal Studies: A Computational Analysis for the Role of the Hematocrit Effect

Abstract

Dried blood spot (DBS) microsampling is extensively employed in newborn screening (NBS) and neonatal studies. However, the impact of variable neonatal hematocrit (Ht) values on the results can be a source of analytical error, and the use of fixed Ht for calibration (Ht_cal) is not representative of all neonatal subpopulations. A computational approach based on neonatal demographics was developed and implemented in R^® language to propose a strategy using correction factors to address the Ht effect in neonatal DBS partial-spot assays. A rational "tolerance level" was proposed for the Ht effect contribution to the total analytical error and a safe Ht range for neonatal samples, where the correction of concentrations can be omitted. Furthermore, an "alert zone" for a false positive or negative result in NBS was proposed, where the Ht effect has to be considered. Results point toward the use of Ht_cal values closely representative of populations under analysis and an acceptable level of percentage relative error can be attributed to the Ht effect, diminishing the probability of correction. Overall, the impact of the Ht effect on neonatal studies is important and future work may further investigate this parameter, correlated to other clinical variables potentially affecting results.

Keywords: Monte Carlo simulations; computational analysis; cut-off; dried blood spots; hematocrit effect; neonates; pharmacokinetic; screening.

Figure 1

Percentage relative error as a function of the individual neonates’ Ht values.

Figure 2

Percentage relative error versus the Ht value used for calibration (Ht_cal). Panel (A): Simulations for term neonates at post-natal day 3; Panel (B): Simulations for preterm neonates at post-natal day 3. Under each point, a number of 1,000,000 Monte Carlo studies were performed.

Figure 3

Graphical presentation of the probability of correction (p) in relation to the different neonate subpopulations and sampling times, assuming a fixed Ht_cal of 50% and an indicated “tolerable” percentage relative error of ±5%.

Figure 4

Graphical presentation of the Ht effect on the NBS results for specific markers (Val: valine; C0: free carnitine; C_exp: experimental (initial) concentration; C_theor: theoretical (corrected) concentration). C_exp was assumed to be 10% different of the respective cut-off limit. Ht_cal was set at 50%.