Review: Serum ferritin radioimmunoassay is the most accurate test for determining iron-deficiency anemia
ACP J Club. 1992 July-Aug;117:23. doi:10.7326/ACPJC-1992-117-1-023
Guyatt GH, Oxman AD, Ali M, et al. Laboratory diagnosis of iron-deficiency anemia: an overview. J Gen Intern Med. 1992 Mar-Apr;7:145-53.
To determine, using meta-analysis, the diagnostic utility of mean cell volume, transferrin saturation, serum ferritin, red cell protoporphyrin, red cell volume distribution, and red cell ferritin in diagnosing iron deficiency.
MEDLINE was searched from 1966 through January 1990, using 2 systematic search strategies that included the words iron, anemia, bone marrow, and erythrocyte. All relevant articles were obtained and references were reviewed.
The following criteria were used to select studies: At least 10% of the population studied were > 18 years old with low hemoglobin levels (< 130 g/L for men, < 110 g/L for women); the diagnostic intervention was quantified; and the outcomes were presented comparing test results with findings on histologic examination of aspirated bone marrow for at least 50% of an identifiable patient group. Articles were assessed by duplicate review.
Data were extracted in duplicate on the number of patients in all age, gender, and disease categories; the process of patient selection; whether the patients were anemic by the reviewers' definition; the proportion of patients who had had bone marrow aspiration; and the results of the investigations.
55 studies were judged to be relevant. The area under the receiver operating characteristic curve for serum ferritin radioimmunoassay was 0.95 (95% CI 0.94 to 0.96), substantially better than any other test. Serum ferritin was as powerful in diagnosing iron deficiency in patients with chronic inflammatory disease as in those without the disease, but compared with results from patients without inflammatory disease, the probability of iron deficiency was slightly higher for each value of ferritin. 5 studies assessed the combined value of tests, and no consistent increase in information was found when other tests were added to the serum ferritin.
Serum ferritin radioimmunoassay is the most powerful test for the diagnosis of iron-deficiency anemia. It can be applied to a complete range of patients but must be interpreted differently for patients with inflammatory conditions. Little additional information is gained from other noninvasive tests.
Source of funding: In part, Ontario Ministry of Health.
Address for article reprint: Dr. G.H. Guyatt, Department of Clinical Epidemiology and Biostatistics, McMaster University Health Sciences Centre, Room 2C12, 1200 Main Street West, Hamilton, Ontario L8N 3Z5, Canada.
Guyatt and colleagues have taken a sophisticated approach in using meta-analysis to evaluate tests for iron deficiency. Their results support several conclusions. First, ferritin is the most useful test. Second, ferritin can and should be interpreted quantitatively—values below 15 mg/L confirm iron deficiency, but low-normal ferritin values (between 15 and 45 mg/L) often indicate iron deficiency as well. Third, ferritin can provide useful information in patients with inflammatory conditions, if interpreted appropriately—a ferritin level of 50 mg/L increases the probability of iron deficiency in a patient with inflammatory disease, but not in the general population. Finally, ferritin measurement can replace routinely measuring iron, iron-binding capacity, and transferrin saturation.
How to proceed when ferritin levels are indeterminate (50 to 100 mg/L) remains open to question and depends on pretest suspicion. Mean cell volume and red cell volume distribution do not appear to add much information, although extreme values of mean cell volume (< 70 fL) indicate either iron deficiency or thalassemia (1). We should probably be more willing to use bone-marrow aspiration to assess occult blood loss when the diagnosis is uncertain.
Meta-analysis has some important limitations when the results of different studies vary substantially. The best guide to interpreting a value in an individual patient (e.g., a ferritin value of 100 mg/L in a patient with rheumatoid arthritis) may be a few reports from comparable populations, rather than summary estimates derived from combining all studies. If studies would report separate test results for patients with characteristics that affect the level, future meta-analyses could provide estimates of the value of a given test result tailored to specific clinical features.
David Atkins, MD
University of WashingtonSeattle, Washington, USA