Abstract

BACKGROUND: Anaemia is a major cause of morbidity and mortality in low-income countries. Primary health-care workers in resource-poor settings usually diagnose anaemia clinically, but this is inaccurate. The WHO Haemoglobin Colour Scale (HCS) is a simple, cheap quantitative method to assess haemoglobin concentration outside of the laboratory. We systematically reviewed the literature to assess the accuracy of the HCS in primary care to diagnose anaemia, and compared this with clinical assessment. METHODS: We searched the electronic databases including MEDLINE, EMBASE, SCOPUS, Web of Science, Cochrane library, CINAHL plus, Popline, Reproductive Health Library, and Google Scholar and regional databases up to Nov 14, 2014, "haemoglobin colour scale" in alternative spellings published in any language. Two reviewers independently screened studies, extracted data, and assessed quality using the QUADAS-2 instrument. Statistical analyses were carried out in STATA using the bivariate model. FINDINGS: Of 141 records and abstracts screened, 14 studies were included. The pooled sensitivity of the HCS to diagnose anaemia was 80% (95% CI 68-88) compared with 52% for clinical assessment ([95% CI 36-67]; p=0·008). Specificity was similar between the HCS (80% [95% CI 59-91]) and clinical assessment (75% [56-88]; p=0·8250). For severe anaemia, diagnostic accuracy was again higher overall for the HCS (p<0·0001); sensitivity was 57% (36-76) for the HCS and 45% (95% CI 12-83) for clinical assessment, but specificity was 99·6% (95% CI 95-99·9) versus 92% (62-99). Combining clinical assessment and the HCS could result in higher sensitivity (anaemia: 91% [95% CI 81-96]); severe anaemia 83% (33-98), but at the expense of specificity (anaemia: 59% [35-79]; severe anaemia 90% [40-99]). Individual studies were highly heterogeneous but pooled results did not differ substantially in a series of sensitivity analyses for indicators of study robustness. INTERPRETATION: In so-called real-life primary health-care conditions, HCS can significantly reduce misdiagnosis of anaemia compared with clinical assessment alone. Future research is required to optimise training, and assess clinical outcomes and cost-effectiveness. FUNDING: None.