A-404 Determinants of Analytical Performance in Centrifugation–Based Hematocrit: Effects of Calibration and Internal Quality Control Frequency Across Hospital and Primary Point-of-Care Sites

Abstract

Manual centrifugation–based hematocrit (Hct) remains widely used as point-of-care testing (POCT) in hospital laboratories and subdistrict health promotion hospitals (SDHHs), particularly in decentralized settings. However, evidence is limited on whether quality-related factors influence satisfactory performance when evaluated using both precision (z-score) and trueness (bias%). This study aimed to determine whether calibration status, reading types, participation in external quality assessment (EQA), and internal quality control (IQC) frequency were associated with satisfactory performance in both settings. A multicenter analysis included six reference blood materials (68-Hct001–006) tested by hospital laboratories and SDHH POCT sites. Z-scores were calculated using group-specific robust means determined by Algorithm A (ISO 13528). Bias% was calculated against target values derived from the mean results of three reference laboratories in accordance with ISO Guide 35. All blood materials were produced by an ISO 17034–certified reference material producer. Satisfactory performance was defined as |z|=2 and bias thresholds of =6% and the stricter =4%. Associations were evaluated using Chi-square analysis. Hospital laboratories demonstrated higher satisfactory performance than SDHH sites across all samples, with the largest differences observed when accuracy was assessed using bias% =4%. Calibration was significantly associated with improved accuracy in both settings (?² = 12.4, p < 0.001). Increased IQC frequency was associated with improved precision in hospital laboratories (|z|=2; ?² = 4.9, p = 0.027) but showed no significant association in SDHH sites. Performance classification differed between z-score– and bias%-based evaluations, with larger between-setting differences observed when the bias =4% criterion was applied compared with z-score–based evaluation. Calibration and IQC frequency were associated with satisfactory performance. However, their effects varied depending on the performance metric and assigned value approach. Use of target values with a strict bias criterion (=4%) resulted in different performance classification compared with z-scores based on group-specific robust means, underscoring the importance of assigned value definition when interpreting performance outcomes, particularly for SDHH POCT sites.