Anaemia: serum ferritin is the best test for diagnosing or excluding iron deficiency anaemia.

The study

• Patients: older than 18 years, with suspected iron deficiency anaemia
• Intervention: laboratory investigations compared with bone marrow aspiration
• Outcome: diagnosis


Articles found in most languages (except Chinese and Japanese) using MEDLINE, 1966 to1990 (search terms: [iron or iron (tw)] and [anemia/diagnosis or bone marrow/analysis or bone marrow/metabolism] or [iron (tw) or anemia or anemia (tw)] and [erythrocytes/analysis or erythrocytes/pathology or erythrocyte count] ) and bibliographies were also searched, authors contacted for unpublished studies or other relevant articles

Selection criteria: as above
Appraisal criteria: articles were selected by two independent reviewers using set criteria (detailed in text)
Articles excluded if: not stated

55 studies were found
Significant heterogeneity between studies (especially for ferritin- noted to depend on population being studied).

The evidence




diagnostic test	iron deficiency anaemia	no iron deficiency anaemia	LR (95% CI)	post-test probability
serum ferritin (µg/l) < or = 15	474	20	52 (42 to 62)	%
serum ferritin 15 <25	117	29	8.8 (7.2 to 10)	%
serum ferritin 25 <35	58	50	2.5 (2.1 to 3.0)	%
serum ferritin 35 <45	36	43	1.8 (1.5 to 2.2)	%
serum ferritin 45 <100	76	398	0.54 (0.48 to 0.60)	%
serum ferritin =100	48	1320	0.08 (0.07 to 0.09)	%
MCV (µm³) =70	58	7	12 (6.1 to 19)	%
MCV 70 <75	31	14	3.3 (2.0 to 4.7)	%
MCV 75 <80	26	39	1.0 (0.69 to 1.3)	%
MCV 80 <85	43	71	0.91 (0.71 to 1.1)	%
MCV 85 <90	32	63	0.76 (0.56 to 0.96)	%
MCV =90	24	128	0.29 (0.21 to 0.37)	%
red cell volume distribution =21	13	8	2.7 (1.3 to 4.1)	%
red cell volume distribution 17 <21	35	33	1.8 (1.4 to 2.2)	%
red cell volume distribution	25	50	0.84 (0.63 to 1.1)	%
red cell volume distribution =15	29	80	0.61 (0.48 to 0.74)	%
transferrin saturation =5%	72	11	10 (6.4 to 15)	%
transferrin saturation 5% <10%	70	44	2.5 (2.0 to 3.1)	%
10% <20%	90	178	0.81 (0.70 to 0.92)	%
transferrin saturation 20% <30%	36	111	0.52 (0.41 to 0.63)	%
transferrin saturation 30% <50%	22	82	0.43 (0.31 to 0.55)	%
transferrin saturation =50%	4	44	0.15 (0.06 to 0.24)	%
red cell protoporphyrin (µg/dl) =350	14	3	8.3 (2.6 to 14)	%
red cell protoporphyrin 250 <350	17	5	6.1 (2.8 to 9.3)	%
red cell protoporphyrin 150 <250	26	23	2.0 (1.4 to 2.6)	%
red cell protoporphyrin 50 >150	42	132	0.56 (0.48 to 0.64)	%
red cell protoporphyrin =50	1	15	0.12 (0.00 to 0.25)	%
total



• Likelihood ratios calculated as continuous values and taken from the article.

Comments

1. As MCV is independent of the serum ferritin, serial likelihood ratios can be used, though the MCV will not add much when it is in the 70-90 range. It is very important to understand your own clinical population for estimating the prior probability. In areas with a higher prevalence of haemoglobinopathy (esp, thalassaemia), the MCV will be less useful as an initial screening test to help decide who should have a serum ferritin measured.
2. Ferritin was found to be less helpful in patients with liver or inflammatory disease.



Citation

1. Guyatt GH, Oxman AD, Ali M, et al: Laboratory diagnosis of iron-deficiency anemia: an overview. [published erratum appears in JGen Intern Med 1992 Jul-Aug;7(4):423]. Journal of General Internal Medicine 1992; 7: 145-153
Contributor: Chris Ball and Clare Wotton, July 2000
Reviewer:

_{Clinical Question.}

Patient	suspected iron-deficient anaemia
Intervention or Exposure	laboratory investigations
Comparison	bone marrow aspiration
Outcome	diagnosis