Abstract:
Reference intervals(RIs) are vital for interpreting clinical laboratory
results and guiding patient management. As RIs vary with demographic, genetic, and
environmental factors, population-specific establishment is necessary for clinical
accuracy. Refining hematological RIs is complex, requiring careful data selection,
statistical analysis, and validation, typically achieved via the direct method. Indirect
methods offer a cost-effective alternative, but their applicability in the Jaffna
population remains unclear.
Objectives: This study aimed to establish sex-specific RIs for Full Blood Count (FBC)
parameters in the Jaffna population using the indirect refineR algorithm.
Methodology: A retrospective cross-sectional study was conducted from July 2023 to July
2024 using routine FBC data from the Laboratory Information Systems of Base Hospitals in
Thellipalai and Point Pedro. Data were curated to exclude pathological values and stratified
by sex. RIs were estimated using refineR, an open-source R package that uses inverse
modelling to identify non-pathological distributions within mixed datasets. Established RIs
were compared with direct-method RIs using the bias ratio, with values ≥0.375 indicating
significant differences.
Results: During the study period, 43,638 FBC records were collected, of which 15,844
(6,537 male and 9,307 female) non-pathological data met the criteria for RI calculation.
Sex-specific RIs for 13 FBC parameters were established for the Jaffna population, with
95% inclusion: WBC (3.61–10.7 vs 4.06–10.8 ×10⁹/L), neutrophils (1.53–7.44 vs 1.46
7.46 ×10⁹/L), lymphocytes (0.97–3.67 vs 1.00–3.78 ×10⁹/L), monocytes (0.19–0.72 vs
0.15–0.69 ×10⁹/L), eosinophils (0.16–0.63 vs 0.03–0.60 ×10⁹/L), basophils (0–0.05 vs
0–0.05 ×10⁹/L), RBC (3.94–5.64 vs 3.46–5.01 ×10¹²/L), hemoglobin (11.3–16.7 vs 9.78
14.2 g/dL), hematocrit (34.8–47.8 vs 28.9–42.0%), platelets (130–399 vs 146–408
×10⁹/L), MCV (75.9–97.2 vs 74.2–95.3 fL), MCH (27.0–32.2 vs 24.2–32.6 pg), and
MCHC (32.3–35.7 vs 31.9–35.3 g/dL) for males and females, respectively. Furthermore,
bias ratio comparison with the direct method showed comparable results in 8 parameters
for males and 9 for females.
Conclusions: Sex-specific RIs for 13 FBC parameters were established for the selected
Jaffna population using refineR, with most parameters showing good agreement with
direct-method RIs, supporting its utility for accurate clinical interpretation.
