ABSTRACT
OBJECTIVE: To test the accuracy of clinical symptoms and signs for anaemia in pregnant women, as assessed by nurse-midwives, in two locations in Northern Tanzania. METHODS: One location was at 1000 m above sea level, the other at 1800 m. Midwives performed examinations and conducted structured interviews to detect severe anaemia at the first antenatal care visit before haemoglobin (hb) results were revealed; 369 and 535 women of all parities were examined in consecutive order at the two locations. Severe anaemia was defined as hb <75 g/l in the first and <80 g/l in the second (higher) location, based on altitude effect on hb distributions. RESULTS: Hb distribution differed substantially between the two locations, with much higher hb levels among those living at 1800 m. Sensitivities for detection of severe anaemia based on individual signs (pallor, conjunctiva, etc.) were 0.85, but only 0.33 to 0.44 for those living at lower and for those at higher altitudes, respectively. Conversely, specificities were around 0.90 at higher and 0.55 at lower altitudes, respectively. Symptoms (headache, dizziness, palpitations, etc.) were too common among those without anaemia to be useful as distinguishing features. Changing the definition of severe anaemia to higher cut-off hb values did not materially alter the results. CONCLUSION: Validity of non-invasive tests to detect severe anaemia in pregnant women varies by locality. In a high-altitude area detection rate was low (sensitivity around 40%). In lower-lying areas detection rate was high, at the cost of low specificity (around 45% false positive tests). Symptoms like headache, dizziness and fatigue were too common to discriminate those with severe anaemia.