Accuracy of various iron parameters in the prediction of iron deficiency in an acute care hospital.

INTRODUCTION: Iron parameters like serum ferritin and iron saturation are routinely used in diagnosing iron deficiency. However, these tests are influenced by many factors. We aimed to review the accuracy of iron parameters among inpatients in an acute care hospital.
MATERIALS AND METHODS: From October 1997 to April 2002, bone marrow aspirate samples from patients on whom concurrent iron studies had been done were analysed. Accuracy of the various iron parameters was analysed using receiver operating characteristic curves.
RESULTS: Among 92 bone marrow aspirate samples, 58, 86 and 83 had a concurrent serum ferritin, serum iron and percentage iron saturation done respectively. Serum ferritin is the best marker for predicting the presence of iron deficiency. This is followed by percentage iron saturation and lastly by serum iron. At the most optimal, a serum ferritin of <60 ng/mL has a positive likelihood ratio of 24.35, sensitivity of 69.6%, specificity of 97.1% and positive predictive value of 94.1%. An iron saturation of <7% has a positive likelihood ratio of 21.62, sensitivity of 44.1%, specificity of 98.0% and positive predictive value of 93.8%. Seven out of 50 samples, which had concurrent measurements of iron saturation and serum ferritin, were discordant in indicating the presence of iron deficiency. In 6 cases, the iron saturation was the spurious result.
CONCLUSION: A serum ferritin of <60 ng/mL or a percentage iron saturation of <7% is highly predictive of iron deficiency among inpatients in an acute care hospital.