Evaluation of the accuracy of two point-of-care haemoglobin meters used in sub-Saharan African population: a cross-sectional study.

BACKGROUND: Point-of-care haemoglobin meters are attractive solutions to improve timely diagnosis of anaemia in resource-limited settings. However, concerns regarding the accuracy of these meters may affect their adoption. The accuracy of two hand-held point-of-care haemoglobin meters was evaluated against reference full blood count analyser. METHODS: This was a hospital-based cross-sectional study conducted at the Douala General hospital, Cameroon. Two handheld haemoglobin meters were assessed: Urit12® (URIT Medical Electronics Co.,Ltd. Guangxi, China) and MissionHb®(ACON Laboratories, Inc., San Diego, USA); against a reference standard CELL-DYN RUBY® (ABBOTT DIAGNOSTICS, Illinois, USA). The Pearson's correlation and Bland-Altman agreement were used to assess the technical accuracy of the meters. Clinical accuracy was evaluated using total error allowable and area under the Receiver Operating Curve. Finally, their agreement with the reference in diagnosing anaemia was assessed using the kappa statistic. RESULTS: A total of 228 participants were included in the study. The mean haemoglobin values of both haemoglobin meters (MissionHb®: 11.6 ± 2.5 g/dl; Urit12®: 10.9 ± 2.7 g/dl) were significantly higher than the reference value (10.5 ± 2.5 g/dl), p < 0.001 for both meters. Both haemoglobin meters had good correlation with the reference analyser (r = 0.89 and r = 0.90 for Urit12® and MissionHb® respectively) and good agreement on the Bland-Altman plots. However, the MissionHb® meter did not meet the clinical accuracy requirements (p < 0.001). Even though both meters were excellent at identifying the presence of anemia (MissionHb® - AUC = 0.9161, Urit 12® - AUC = 0.9009), they, however, both had weak agreement with the reference analyser in diagnosing the severity of anaemia (K = 0.39 for MissionHb®, p < 0.001 and K = 0.54 for Urit12®, p < 0.001). CONCLUSION: Although both devices showed technical accuracy with a positive correlation with the reference analyser and were able to accurately diagnose the presence of anemia, both meters however, had sub-optimal agreement with the reference analyser in diagnosing the degree of severity of anaemia among our participants.