Epitrochlear lymph nodes as marker of HIV disease in subSaharan Africa

BMJ 1994; 309 doi: (Published 10 December 1994) Cite this as: BMJ 1994;309:1550
  1. A Malin,
  2. I Ternouth,
  3. S Sarbah
  1. Department of Clinical Pharmacology, University of Zimbabwe, Harare, Zimbabwe, lecturer. Parirenyatwa Hospital, Harare physician, house physician.
  1. Correspondence to: Dr A Malin, Department of Clinical Sciences, London School of Hygiene and Tropical Medicine, London WC1E 7HT.
  • Accepted 6 July 1994

The association of lymphadenopathy and HIV infection has been well described,1 but both the size and distribution of the enlarged lymph nodes have been poorly characterised. In Africa serological testing for HIV infection is both expensive and difficult to obtain, and health care workers often have to rely on clinical features to assess the likelihood of HIV disease. Knowledge of the rate of occurence of enlarged lymph nodes and variations in their size and distribution could help identify useful diagnostic markers. Generalised lymphadenopathy, although suggestive of HIV infection, does not exclude other diagnoses such as viral infections, tuberculosis, and haematological malignancy. Our impression, in a large central hospital in sub-Saharan Africa, was that palpable epitrochlear lymph nodes were a good marker of early HIV disease. Osler noted that such nodes were a prominent clinical feature of secondary syphilis,2 but nowadays the palpability of this node is not emphasised.3 4 5 We assessed the value of regional lymph node enlargement, by site and size, as a predictor of HIV disease.

Subjects, methods, and results

All sequential adult patients (>/= 12 years) admitted to an acute general medical ward at this hospital over three months were eligible for the study. Readmissions and patients from specialist clinics were excluded. We obtained informed consent (from parents when appropriate), and all those who entered were given counselling for HIV testing. HIV infection was diagnosed using a double enzyme linked immunosorbent assay (ELISA) technique and western blotting for equivocal cases. Lymph nodes were scored for palpability and size to the nearest 0.5 cm in each patient. Three regions were assessed: epitrochlear, submandibular, and axillary. The doctor scoring lymph node size was blinded to clinical data and results of HIV serological tests.

Of 260 eligible subjects, 259 agreed to take part (age range 12-88). There were 184 men (mean age 40) and 75 women (mean age 36). One hundred and forty six were positive for HIV (age range 17-76; mean age men 37, women 33), giving a seroprevalence of 56%. Positive and negative predictive values, sensitivities, specificities, and odds ratios are given in the table. The odds ratios were all significant by χ2 analysis (P<0.001).


Enlargement of axillary and submandibular lymph nodes by >/= 1 cm gave high positive predictive values (91% and 89% respectively) and specificities (95% and 96%), but negative predictive values and sensitivities were all low (<60%). Thus, the presence of these enlarged lymph nodes was a strong marker of HIV disease but their absence was unhelpful. Their use as markers of HIV disease is limited in that they were only occasionally present (24% of all subjects had enlarged axillary nodes and 12% enlarged submandibular nodes). When the criterion for enlargement was >/=0.5 cm the total number of those with palpable nodes increased, but positive prediction and specificity were poor.

Enlargement of epitrochlear nodes by >/= 0.5 cm gave a positive predictive value of 85% and the highest overall sensitivity (84%); lymph nodes were palpable in 47% of all subjects. This small enlargement of the epitrochlear nodes was less specific (81%) than more enlarged (>/=1 cm) axillary (95%) and submandibular (96%) nodes, but small epitrochlear nodes occurred twice as commonly as small axillary nodes and four times as commonly as small submandibular nodes. Larger epitrochler nodes (>/= 1 cm) improved specificity to 90%, but positive prediction and sensitivity fell dramatically. Thus, unlike for axillary and submandibular nodes, enlargement by >/= 0.5 cm was a much better predictor than enlargement by >/= 1 cm (odds ratio 23.4 (upsilon) 5.3). A combination of two different regions improved the positive predictive value to > 90% but with a noticeable fall in sensitivity.

Predictive values, sensitivities, and specificities for peripheral lymphadenopathy as marker of HIV infection by site and size of lymph nodes

Predictive values, sensitivities, and specificities for peripheral lymphadenopathy as marker of HIV infection by site and size of lymph nodes

View this table:

A small enlargement of epitrochlear nodes was a common and useful marker for HIV disease in an acute general medical ward in sub-Saharan Africa where the prevalence of HIV infection was 56%. Extreme caution must be taken not to extrapolate these predictive values beyond the setting in which they were derived, particularly where prevalences are lower. It remains to be seen whether enlarged epitrochlear nodes predict early HIV disease in other clinical settings. However, the clinical importance of epitrochlear nodes has clearly been forgotten and deserves more attention.

We thank Dr Max Marshall and Kathy Roberts for their critical advice and Mary Gibson for her research support.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
View Abstract