Scholarlink

<p>Iron deficiency anemia (IDA) and beta thalassemia trait (BTT) are the leading causes of microcytic hypochromic anemia, requiring fundamentally different management. Differentiation remains challenging, and commonly used red cell discrimination indices show inconsistent diagnostic performance. We developed and internally validated a hemogram-based multivariable predictive model using routinely available complete blood count (CBC) parameters and compared its performance with twelve conventional discrimination indices. In this prospective cross-sectional study at a tertiary care center in Pune, India, 224 adults with confirmed IDA (<i>n</i> = 114) or BTT (<i>n</i> = 110) were enrolled. Hemoglobin A2 (HbA2) by high-performance liquid chromatography (HPLC) and serum ferritin were measured in all patients. Predictor selection used a random forest classifier. A logistic regression model incorporating total red blood cell count (TRBC), hemoglobin (Hb), mean corpuscular hemoglobin concentration (MCHC), and mean corpuscular volume (MCV) with restricted cubic splines (RCS) was developed. Performance was assessed using discrimination (C-index) and calibration with bootstrap validation (1,000 iterations), and compared with twelve discrimination indices.he model demonstrated an optimism-corrected C-index of 0.984 and a mean absolute calibration error of 0.014. Among conventional indices, TRBC count alone achieved the highest accuracy (89.73%; Youden’s J 0.791). The model substantially outperformed all indices, including the Mentzer Index (C-index 0.890) and the Green &amp; King Index (0.874). A hemogram-based multivariable model substantially improves diagnostic accuracy compared with traditional indices and represents a scalable decision-support tool for resource-constrained settings. External validation is warranted.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Hemogram-Based Predictive Model for Differentiating Iron Deficiency Anaemia from Beta Thalassemia Trait: Development, Internal Validation, and Comparison with Conventional Discrimination Indices

  • 期刊论文

  • 发表日期：

    2026年7月20日

查看全文
  • Santosh Kumar Singh,
  • Rahil Arora,
  • Suman Kumar Pramanik,
  • Abhishek Kumar,
  • Sandeep Kumar,
  • Gurpreet Kaur,
  • Deepanjan Dey,
  • Anjali Gautam,
  • Ashutosh Kumar,
  • Vani Singh,
  • Venkatesan Somasundaram,
  • Uday Yanamandra

摘要

Iron deficiency anemia (IDA) and beta thalassemia trait (BTT) are the leading causes of microcytic hypochromic anemia, requiring fundamentally different management. Differentiation remains challenging, and commonly used red cell discrimination indices show inconsistent diagnostic performance. We developed and internally validated a hemogram-based multivariable predictive model using routinely available complete blood count (CBC) parameters and compared its performance with twelve conventional discrimination indices. In this prospective cross-sectional study at a tertiary care center in Pune, India, 224 adults with confirmed IDA (n = 114) or BTT (n = 110) were enrolled. Hemoglobin A2 (HbA2) by high-performance liquid chromatography (HPLC) and serum ferritin were measured in all patients. Predictor selection used a random forest classifier. A logistic regression model incorporating total red blood cell count (TRBC), hemoglobin (Hb), mean corpuscular hemoglobin concentration (MCHC), and mean corpuscular volume (MCV) with restricted cubic splines (RCS) was developed. Performance was assessed using discrimination (C-index) and calibration with bootstrap validation (1,000 iterations), and compared with twelve discrimination indices.he model demonstrated an optimism-corrected C-index of 0.984 and a mean absolute calibration error of 0.014. Among conventional indices, TRBC count alone achieved the highest accuracy (89.73%; Youden’s J 0.791). The model substantially outperformed all indices, including the Mentzer Index (C-index 0.890) and the Green & King Index (0.874). A hemogram-based multivariable model substantially improves diagnostic accuracy compared with traditional indices and represents a scalable decision-support tool for resource-constrained settings. External validation is warranted.