Anaemia Indices and Thalassaemia Discrimination Calculator

Evidence & Clinical Pearls

1. The One Clue Behind Every Index

All eight indices on this page are variations on a single observation. In iron deficiency the marrow cannot make haemoglobin, so it makes fewer red cells and they are small: the count falls with the volume. In beta-thalassaemia trait the marrow makes plenty of red cells and each one is small: the count is preserved or high while the volume is low. Every formula here is a way of asking how small the cells are relative to how many there are.

That is why the red cell count is the most informative single number on the report, and why a microcytic anaemia with a red cell count above about 5.0 × 10¹²/L should make you think of a trait before you think of iron. The Mentzer index is simply that thought written as a division.

2. The Eight Indices This Tool Computes

Note that Matos and Carvalho runs the opposite way to the other seven: a high value favours the trait. It is also the only one that uses the mean corpuscular haemoglobin concentration and the only one whose original paper was read directly during the building of this tool.

3. Two Indices This Tool Deliberately Does Not Compute

The England and Fraser index is MCV minus RBC minus five times the haemoglobin, minus a constant, with a value below zero favouring the trait. Published secondary tables give that constant as 3.4 in some places and 8.4 in others. Those are not the same index: on a borderline count they give opposite answers. The 1973 Lancet paper could not be read directly to settle it, and there is a 1973 follow-up letter from the same authors that may be the origin of the divergence.

The Ricerca index is RDW divided by the red cell count, and the published cut-off appears as 3.3 in some sources and 4.4 in others, with a locally derived Indian value of 3.8 in a third.

Both are omitted rather than shipped with a guessed constant. A calculator that prints a confident number from a constant nobody can pin down is worse than one that says it does not know. Worse in the way that is hardest to catch, too: the output looks exactly like the seven that are right.

4. The Cut-offs Travel Badly, and They Travel Worst to Here

The largest analysis of these indices, 99 studies and 135,409 results, found their performance was markedly worse in South-East Asian populations than in European ones, with a pooled diagnostic odds ratio of 9.3 against 53.1. That is not a small drift. It means a threshold validated in Milan or Manchester should not be applied to a report from an Indian laboratory without knowing what it does here.

The most directly relevant work was done at Manipal, on the coast of Karnataka. Kumar and colleagues ran the classical indices against HPLC-confirmed cases and found that every published cut-off performed better once it was re-derived locally, and that the RDW index came out best of them:

The result panel prints both columns wherever a local value exists. Where the two disagree on a given patient, that disagreement is the honest state of the evidence and not a defect in the tool. A separate North Indian series from Ambala found Shine and Lal to have a specificity of 3.3 per cent, calling almost every iron-deficient patient a thalassaemia trait, which is the strongest argument available for never acting on one index alone.

5. The Patient Who Has Both

This is the failure mode that matters most in India and the one no index handles. In a Delhi series of 463 people with beta-thalassaemia trait, 27.2 per cent also had iron deficiency, and their haemoglobin, mean corpuscular volume and mean corpuscular haemoglobin were all significantly lower than in trait carriers who were iron replete. The added iron deficiency drives the volume down further, which pushes the Mentzer index further below 13, so the tool becomes more confident of the trait at exactly the moment the patient also needs iron.

The reassuring half of that same study is that haemoglobin A₂ stayed elevated in all but one of the iron-deficient carriers, so HPLC still found the trait. The evidence on whether iron deficiency masks a trait is genuinely mixed: a Turkish paediatric series found that 45 per cent of iron-deficient carriers fell below a 3.5 per cent A₂ threshold, while a Sicilian series found no significant difference. Either way, the safe practice is the same: if HPLC is borderline and the patient is iron deficient, correct the iron and repeat the HPLC.

6. What Actually Settles It

Haemoglobin HPLC. The Indian national guideline on haemoglobinopathies sets haemoglobin A₂ at 2.3 to 3.5 per cent as normal, 3.5 to 3.9 per cent as equivocal and requiring further evaluation, and 4.0 to 8.0 per cent as diagnostic of beta-thalassaemia trait. The equivocal band is not a formality. A Mumbai study from the national haematology institute found a beta-globin mutation in 73 per cent of people whose A₂ sat in the borderline zone. Send a borderline result for molecular testing rather than filing it as normal.

For iron, WHO sets the ferritin threshold at below 15 µg/L in anyone aged 5 and over, and below 70 µg/L where infection or inflammation is present. Ferritin is an acute phase protein. A value of 40 in a patient with an infection is entirely compatible with iron deficiency, and WHO recommends measuring C-reactive protein and alpha-1-acid glycoprotein alongside it wherever inflammation is common, which in an Indian ward is most of the time.

7. The Reticulocyte Production Index

A reticulocyte percentage is a proportion of a red cell count that is itself abnormal, so in anaemia it flatters the marrow twice over. Two corrections fix it. The first divides out the anaemia: the corrected count is the percentage multiplied by the patient's haematocrit over a normal 45. The second divides out early release: as anaemia deepens the marrow pushes out reticulocytes earlier, so they survive longer in the blood and are counted for longer. That second correction is the maturation factor.

The maturation factor table is not agreed, and this tool names which one it uses. Three mutually incompatible versions circulate. The one used here is the peer-reviewed version: 1.0 at a haematocrit of 40 and above, 1.5 from 30 to below 40, 2.0 from 20 to below 30, and 2.5 below 20. A widely reproduced alternative shifts every band down by about four points, which changes the reported index by up to half at a haematocrit in the thirties. Below 2 the marrow is not responding. Above 3 it is responding appropriately, which points at haemolysis or blood loss rather than a production failure. Between 2 and 3 is genuinely indeterminate, although one widely used reference collapses that zone and calls 2 adequate.

The absolute reticulocyte count sidesteps the whole argument and many haematologists prefer it. It is the percentage multiplied by the red cell count, and this tool prints it alongside.

8. Thresholds, Altitude and the Indian Argument About Them

WHO issued a new guideline on haemoglobin cut-offs on 5 March 2024. Almost everything was retained: the only threshold that moved was for children aged 6 to 23 months, from 110 to 105 g/L. Adult thresholds remain 12.0 g/dL for non-pregnant women and 13.0 g/dL for men, and pregnancy is now stated by trimester: 11.0, 10.5 and 11.0 g/dL.

Two details in that document are routinely got wrong. The altitude adjustment starts at 500 metres, not at 2500. A resident of Shimla, Gangtok or Ooty needs the correction, and it reaches 1.1 g/dL by 1500 metres. And the adult rows are labelled 15 to 65 years, with no row above 65: applying the adult cut-off to an 80 year old is an extrapolation the guideline does not make.

The Indian argument. Sachdev and colleagues took venous samples from healthy Indian children and adolescents in the Comprehensive National Nutrition Survey, and derived thresholds 1 to 2 g/dL below the WHO values. On their own cut-offs, anaemia prevalence fell from 30.0 to 10.8 per cent. That work is contested, partly because most of the original sample was excluded from the healthy reference group. India has not adopted a different cut-off. What did change is the measurement: anaemia was dropped entirely from the National Family Health Survey 6, whose fact sheets were released on 29 May 2026 with no haemoglobin indicator at all, and the estimate has moved to the ICMR Diet and Biomarkers Survey. The most recent national figures therefore remain those of NFHS-5: 57.0 per cent of women aged 15 to 49, 25.0 per cent of men, 67.1 per cent of children aged 6 to 59 months. This tool uses the WHO thresholds, and says so.

9. Screening, and Why Any of This Matters Beyond One Patient

The average beta-thalassaemia carrier frequency in India is 3 to 4 per cent, around 42 million carriers, and 10,000 to 15,000 children with thalassaemia major are born each year. A systematic review of 69 studies put the general population figure at 3.74 per cent and tribal groups at 4.6 per cent. Carrier rates are higher in Sindhi, Punjabi, Gujarati and Bengali communities and in several tribal groups. The National Sickle Cell Anaemia Elimination Mission, launched on 1 July 2023, is screening 7 crore people aged 0 to 40 across 278 districts in 17 states, with an elimination target of 2047. By July 2025 it had screened 6 crore and identified 16.7 lakh carriers. That mission is sickle cell specific, and thalassaemia prevention still sits under the 2016 national haemoglobinopathies guidelines and state programmes rather than a comparable national mission.

The practical consequence is that finding a trait is never only about the patient in front of you. It is an indication to test the partner before a pregnancy, and to counsel the family, and that conversation is worth more than the index that started it.

10. What This Tool Deliberately Does Not Do

It does not diagnose. It does not see alpha-thalassaemia trait, in which the same indices behave unpredictably and haemoglobin A₂ is normal, and it does not see haemoglobin E trait, which matters in the North East and in Bengal. It does not compute a probability, because the published indices give a classification and not a likelihood, and manufacturing one from their agreement would be inventing precision. It does not interpret a count taken during or within three months of a transfusion. And it does not stage or manage the anaemia it finds: what it produces is a direction for the next two investigations.