Introduction Iron deficiency anaemia (IDA) is common. It is of particular importance because about 10% of subjects with IDA over the age of 50 will have an underlying gastro-intestinal (GI) malignancy, often in the absence of other clinical pointers to the diagnosis. IDA is therefore an accepted indication for examination of the GI tract, generally through bidirectional endoscopy.

Investigation of IDA is labour-intensive however, and most examinations will not reveal significant pathology. The aim of this study was to determine whether simple and objective clinical variables can identify sub-groups of subjects with IDA who are at clinically useful extremes of risk for underlying malignancy – arbitrarily defined as < 1% for low risk and > 20% for high risk.

Methods A retrospective study of 720 subjects referred to a single IDA clinic between 2004 and 2012. All had confirmed iron deficiency, minor or no localising symptoms, and subsequent GI tract investigation. Recorded information included age, sex, haemoglobin concentration (Hb), mean cell volume (MCV), iron studies, and final diagnosis.

Results A total of 68 (9.4%) of the study population had a GI malignancy. In the model generated by logistic regression analysis, age (> 70 v < 71), sex (M v F) and Hb quartile were all predictive of the probability of underlying malignancy. The effects of these variables were cumulative.

Percentage of cases of GI malignancy in each subgroup (cases/number in subgroup):

The prevalence of GI malignancy ranged from 0.0% in younger females with mild anaemia, to over 25% in older males with more severe anaemia. By the pre-defined criteria, the model identified sub-populations of 84 (11% of the total) at extreme low risk, and 117 (16%) at extreme high risk.

Conclusion The results confirm previous work identifying age, sex and haemoglobin concentration as variables predictive of underlying malignancy in IDA. Furthermore, the findings suggest that over a quarter of subjects with IDA can be predicted to be of extremely low or high risk on the basis of these simple and objective clinical criteria. This may be of clinical relevance for patient counselling, prioritisation of investigations and allocation of resources. Work is ongoing to validate risk prediction in a prospective study, and to refine the model by inclusion of additional variables.

Disclosure of Interest None Declared.

Reference

• James MW, Chen CM, Goddard WP, Scott BB, Goddard AF. Risk factors for gastrointestinal malignancy in patients with iron deficiency anaemia. Eur J Gastroenterol Hepatol 2005; 17(11):1197–203.