Reproducibility and validity of portable haemoglobinometer for the diagnosis of anaemia in children under the age of 5 years

• Published in: Journal of nutritional science (Cambridge) Vol. 9; p. e3
• Main Authors: da Silva Pereira, Alessandra, de Castro, Inês Rugani Ribeiro, Bezerra, Flávia Fioruci, Nogueira Neto, José Firmino, da Silva, Ana Carolina Feldenheimer
• Format: Journal Article
• Language: English
• Published: England Cambridge University Press 01-01-2020
• Subjects: Accuracy; Age; Anaemia; Anemia; Automation; Children; Children & youth; Guardians; Infectious diseases; Medical diagnosis; Public health; Reproducibility; Sample size; Studies; Test methods; Variables; Brazil; Rio de Janeiro Brazil; Accuracy; PABAK, prevalence-adjusted bias-adjusted κ; Anaemia; Reproducibility; NPV, negative predictive value; PPV, positive predictive value; Children; ICC, intraclass correlation coefficient
• ISSN: 2048-6790; 2048-6790
• DOI: 10.1017/jns.2019.43

Portable haemoglobinometers have been used in order to estimate the prevalence of anaemia in diverse settings. However, few studies have been conducted to evaluate their performance in children of different age groups in distinct epidemiological contexts. To evaluate the reproducibility and reliability of a portable haemoglobinometer for the diagnosis of anaemia in children <5 years Hb was measured in the venous blood of 351 children <5 years by an automated system (standard method) and in three capillary blood samples, using a portable haemoglobinometer (HemoCue
; test method). The reproducibility of the device and of the test method was evaluated using the intraclass correlation coefficient (ICC) (Hb in its continuous form), κ and prevalence-adjusted bias-adjusted κ (PABAK) (categorised variable: anaemia: yes/no). For test method validation, Bland-Altman analyses were performed and sensitivity, specificity, accuracy rate, positive predictive value (PPV) and negative predictive values (NPV) were calculated. The haemoglobinometer presented good device reproducibility (ICC = 0·79) and reasonable method reproducibility (puncture, collection and reading) (ICC = 0·71). Superficial and fair agreement (κ) and good agreement (PABAK) were observed among the diagnoses obtained through the test method. The prevalence of anaemia was 19·1 and 19·7 % using the standard and the test method, respectively, with no statistically significant differences. The test method presented higher specificity (87·7 %) and NPV (88·3 %) than sensitivity (50·7 %) and PPV (49·3 %), and intermediary accuracy rate (57·8 %). HemoCue
showed good device reproducibility and reasonable method reproducibility, as well as good performance in estimating the prevalence of anaemia. Nevertheless, it showed a fair reliability and low individual diagnostic accuracy.