In my work with Médecins Sans Frontières I constantly face dilemmas when trying to decide whether a patient has tuberculosis or not. In the countries where we work, diagnosis for tuberculosis still relies on the use of sputum microscopy, a test developed over 100 years ago, that will detect only 45-65% of cases when performed well. In children or people living with HIV the performance of the test is even worse.
Tuberculosis is the most common cause of death of people living with HIV in resource limited settings. Patients co-infected with HIV and tuberculosis more commonly have extra-pulmonary tuberculosis or smear negative pulmonary tuberculosis or frequently are too ill to produce a good sputum sample. In children it is often not possible to get a good sample for sputum microscopy, and diagnosis relies on the use of various score charts that have not been validated in HIV infected or malnourished children – precisely the groups where we find we most need accurate and rapid diagnosis.
There are of course better diagnostic tests for tuberculosis including culture and genetic tests. However, culture can take weeks to provide an answer and requires highly trained staff, expensive laboratory and equipment and importantly an ongoing quality control system to ensure the validity of results of culture and drug sensitivity testing. Even with improvements in culture techniques such as liquid culture methods diagnosis takes on average 9 days for a positive result and still takes several weeks for a negative result (which excludes the logistics and time delays of getting a sample from the patient to a proficient laboratory and getting the result back to where the patient is.
I am currently working in Myanmar and saw 3 patients in one morning where current World Health Organization diagnostic algorithms are inadequate. If a rapid point of care test existed it would have made a huge difference to the care of these patients: a child with malnutrition and chronic cough but unable to cough up sputum, a patient with HIV and low CD4 count who clinically has been coughing for some weeks but the standard microscopy tests and chest X-ray were negative, and a patient with HIV who has nearly completed TB treatment and has clinically improved, but one out of 3 of the sputum microscopy tests used to monitor treatment has shown small amounts of acid fast bacilli which may represent failure of treatment with drug resistance, re-infection, colonisation with atypical mycobacterium or laboratory error, but according to protocol means he must start TB treatment again. A rapid point of care test within these primary care clinics could have made such a difference. As it is we are left relying on sputum microscopy which misses more patients than it diagnoses, following algorithms for children that are not validated, and sending tests for sputum culture which take several weeks to months to get results (if you can send to a reliable lab which is not always the case in many countries).
If we allow ourselves to day dream for a few minutes about the ideal test to diagnose tuberculosis in resource limited settings what would such a test look like? I would like a test that gives a result on the same day. A test that can actually help diagnose or exclude tuberculosis in the majority of patients I see, that means in patients with HIV and in children including malnourished children. A test that diagnoses patients regardless of whether they have pulmonary or extra-pulmonary tuberculosis. And a test that is cheap and can be performed reliably without the need for electricity.
I think that patients expect more from us than the current tests. To hear patients themselves talk about how the lack of adequate TB testing has affected them follow this link.
Philipp du Cros works in the Manson Unit of Médecins Sans Frontières UK, providing assistance to our programmes treating tuberculosis (TB).