New WHO algorithm to prevent TB deaths in very ill people with HIV

This article is more than 7 years old. Click here for more recent articles on this topic

The World Health Organization (WHO) has released a new diagnostic algorithm in order to reduce the likelihood of ‘seriously ill’ people with HIV dying of undiagnosed and untreated tuberculosis (TB).

“Despite being preventable and curable TB is still the leading cause of morbidity and mortality, even in the era of antiretroviral therapy (ART) going to scale. In fact, in 2014, it accounted for nearly a third of HIV-related deaths,” said Annabel Baddeley, of WHO’s Global TB Programme, who moderated a session at the 21st International AIDS Conference (AIDS 2016) in Durban, South Africa dedicated to discussing the need, evidence base and application of the algorithm and other strategies to reduce mortality from HIV-associated TB.

The persistent dilemma of TB in people living with HIV

Although people living with HIV are being started on treatment earlier – data from some large cohorts show that the median CD4 cell count at treatment initiation has doubled – the proportion of people who present for care with very advanced disease has remained relatively unchanged and “mortality is largely, in most parts of the world, due to the same problems,” said Dr Nathan Ford of the WHO, who presented more data on the enduring burden of TB morbidity and mortality among people living with HIV.

A systematic review carried out by Ford and colleagues found that AIDS-defining illnesses continue to be the leading cause of hospitalisation in people living with HIV, and around one in five of those people are hospitalised due to TB.



When using a diagnostic test, the probability that a person who does have a medical condition will receive the correct test result (i.e. positive). 


Antibiotics, also known as antibacterials, are medications that destroy or slow down the growth of bacteria. They are used to treat diseases caused by bacteria.


Material coughed up from the lungs, which can be examined to help with diagnosis and management of respiratory diseases.


A doctor, nurse or other healthcare professional who is active in looking after patients.


A test used to measure something.

Importantly, 30% of people living with HIV who were hospitalised were only diagnosed with HIV at the time of hospitalisation – underscoring the challenge of reaching people living with HIV earlier in the course of infection. The median CD4 cell count at diagnosis was 131. Another systematic review which looked at post-mortem data on the cause of death, found that in resource-limited settings, TB accounts for approximately 40% of facility-based HIV/AIDS-related adult deaths (43.2% in sub-Saharan Africa and 63.2% in Southeast Asia) (Gupta 2015). 

But the ‘late presenters’ are not the only ones at risk of being hospitalised with advanced HIV disease and TB.  A South African study of medical admissions found that more than a third (35.7%) of the people living with HIV admitted to a public sector district hospital in Cape Town had not yet started ART, 45% reported being currently on ART, while another 19.3% reported that they had initiated therapy but, for one reason or another, had interrupted care. Again, the most frequent primary diagnosis was TB (33.5%). Outcomes among the cohort were poor – within 90 days of follow-up, 29.9% required readmission and 13.3% had died. TB was the most common cause of death (in 37.2%) (Meintjes 2015).

Case fatality rates among people being treated for TB are more than three times greater if they are HIV-positive than if they are HIV-negative – and even higher if they are not put on ART as soon as possible after starting TB treatment.

But according to Dr Ford, it often does not happen.

“It's indicative of the need in the TB community and in the TB/HIV community to do a much better job at ascertaining gaps in the cascade of care, and where we need to focus our attention in the continuum from diagnosis to long-term treatment success,” he said. “There are major challenges in diagnosing HIV among those who are TB-positive, and major challenges in putting HIV/TB co-infected individuals on ART despite many, many years of of recommending immediate ART in all patients who are TB co-infected.”

Progress since earlier diagnostic algorithms

WHO has long recognised the need to respond more effectively to prevent deaths from undiagnosed TB in HIV-positive people. In effect, the new algorithm is descended from the algorithm for diagnosis of smear-negative TB in HIV-prevalent settings, only updated in light of the latest technology and data on presumptive treatment – and targeted to the most seriously ill people with HIV at greatest risk of dying of TB.

The reasons why TB can be difficult to diagnose in people with advanced HIV disease have changed little, even with current tests:

  • People living with HIV are often unable to provide sufficient and high quality sputum specimens for lab tests.
  • Even when they can produce sputum, the amount of TB bacilli in the specimen may be very low; also people with HIV have high rates of extrapulmonary TB.

The algorithm

The new algorithm, which is published in WHO’s 2016 version of the Consolidated Guidelines on the Use of Antiretroviral Drugs for Treating and Preventing HIV Infection, is specifically designed to provide earlier access to TB treatment to the most seriously ill people with HIV or of unknown HIV status in resource-limited settings – particularly peripheral settings where access to TB diagnostics is limited.

  1. If a person with HIV or unknown HIV status visits a peripheral health facility – such as a primary health clinic or outpost – and is suspected of having TB and has danger signs, they should be immediately referred to a higher level medical facility if possible.
  2. Often referral is not possible, and in those cases, the algorithm recommends performing an Xpert MTB/RIF test and a lateral flow LAM test (according to the guidelines mentioned above). Meanwhile, the individual should be given intravenous antibiotics to treat any possible bacterial infections, and treatment for Pneumocystis pneumonia should also be considered. If available, a chest x-ray should be performed to help guide the clinician’s decision.

  3. If any of the TB tests come back positive, TB treatment should be initiated, but if the TB tests are negative or not available, then:

  4. If there is improvement of symptoms after 3-5 days, TB is unlikely, and the care provider should look for other HIV-related diseases, consider starting ART as soon as possible, provide isoniazid preventive therapy and cotrimoxazole preventive therapy and complete the course of parenteral antibiotics.

But if there is no improvement or symptoms worsen after three to five days, presumptive TB treatment should be started, as well as ART and cotrimoxazole preventive therapy. The clinician should perform further investigations for TB (such as culture) and other diseases and complete the course of parenteral antibiotics.

Potential and limitations of new diagnostics: Xpert MTB/RIF and LAM

Yohhei Hamada of WHO’s Global TB Programme, who presented the new algorithm, noted that there are some tools (or approaches) that can help to expedite TB diagnosis and treatment.

The Xpert MTB/RIF is much more sensitive than smear microscopy (manually looking for bacilli in a treated sputum sample under a microscope), detecting up  to 79% of pulmonary TB in people living with HIV (a future generation of the test, ‘the Ultra’ promises even higher sensitivity). In addition, the test is automated and can generate a result in less than two hours.

WHO recommends that Xpert MTB/RIF, “should be used as the initial diagnostic test in people living with HIV to diagnose pulmonary TB and TB meningitis,” said Hamada. It may also be used rather than conventional tests (culture) for testing other extrapulmonary specimens (from lymph nodes and other tissues).

However, even though the cost of the Xpert MTB/RIF test cartridges has been brought down, the equipment that runs the test is expensive, severely limiting access to the tests in many resource-limited settings. Furthermore, in most countries where the test has been scaled up, such as South Africa, it is primarily performed at district or regional laboratories – meaning that specimens must usually be transported from the primary health care clinic to the lab, and the results then sent back to the clinic – which must then track down the patient.

Another tool that can speed up TB diagnosis in some of the most ill individuals, the LAM assay, is a urine-based test that can routinely be used at point-of-care, generating results in 25 minutes. Unfortunately, the sensitivity of the test is limited in people who aren’t in the hospital or who have CD4 cells higher than 200. However, WHO reviewed the available literature and found that in five studies, the pooled sensitivity of the LAM assay for TB in people with CD4 cell counts of 100 or less was 56% and 49% in those with CD4 counts of 200 or less, while in six studies, the pooled sensitivity for TB in hospital inpatients was 54%.

Consequently, WHO recommends that the lateral flow LAM test should be only used to assist in the diagnosis of active TB in HIV-positive adults and children (both in- and out-patients), presumed to have TB, who:

  • Have a CD4 cell count of 100 or below, or
  • Are seriously ill (see definition above), regardless of CD4 cell count
  • The LAM should not be used as a screening test for TB, however.

Presumptive TB treatment

Finally, in certain cases, WHO recommends presumptively treating suspected TB, which it defines as the “initiation of TB treatment for people living with HIV in peripheral facilities, based exclusively on clinical suspicion [of TB in] ‘seriously ill patients’ with respiratory distress based on the judgment of the clinician.”

Several recently reported trials of empiric or presumptive TB treatment have failed to show benefit – however, Hamada stressed that the protocols did not meet WHO’s definition of presumptive treatment and, in particular, were not targeted to individuals meeting the algorithm’s definition of seriously ill, which includes any individual with HIV or of unknown HIV status presenting with one or more of the following danger signs:

  • Unable to walk unaided
  • A respiratory rate over 30/min
  • A fever of more than 39oC, and/or
  • A pulse rate of over 120/min.

The first of these studies was the REMEMBER trial which found no benefit providing empirical TB treatment over providing isoniazid preventive therapy (IPT) to people living with HIV with less than 50 CD4 cells. However, Hamada pointed out that the study only included participants in whom TB had been ruled out by extensive investigations and was therefore not suspected.

A more recent study, the TB Fast Track Trial, which was presented at CROI earlier this year, did evaluate empiric treatment in individuals with CD4 cell counts of 150 or below, who were suspected of having TB (a positive LAM test and/or low haemoglobin or body mass indexes). Indeed, those considered to have a high probability of TB were more likely to receive an eventual confirmed TB diagnosis (in 11.5%), but there was no significant benefit to empiric treatment. This may have been due to the fact, that in practice and contrary to expectations, empiric TB treatment was also associated with delayed ART initiation. A closer look at the data found that delayed ART was linked to poorer outcomes – one must remember that ART was delayed in the majority of those put on empiric TB treatment who did not in fact have TB). This underscores the need to not delay ART treatment in people provided with empiric TB treatment no matter what the protocol – and that it may be important to limit presumptive TB treatment to those at the highest risk of having TB.

Note, all the participants in the TB Fast Track Trial were ambulatory, and they were not required to have any of the other danger signs.

The failure of these particular empiric therapy strategies to improve survival does not mean other presumptive TB treatment will not work. In fact, in a recently published study, empiric TB treatment achieved a statistically significant 44% reduction in 8-week mortality among seriously ill HIV-positive inpatients with WHO’s ‘danger signs’ (Katagira 2016).


Clearly, access to better, more reliable diagnostics would lead to more expeditious TB treatment for seriously ill individuals with co-infection, but it is hoped that, until such tools become available, use of this algorithm could minimise delay in TB treatment initiation and prevent early mortality. In the meantime, programmes are urged to “adopt rapid TB diagnostic tools (i.e. Xpert MTB/RIF and LF-LAM) to increase earlier diagnoses of TB,” concluded Hamada. 


Ford N et al. TB as a cause of hospitalization and in-hospital mortality among people living with HIV worldwide: a systematic review and meta-analysis. J Int AIDS Soc 19(1):20714, 2016.

Gupta RK et al. Prevalence of tuberculosis in post-mortem studies of HIV-infected adults and children in resource-limited settings: a systematic review and meta-analysis. AIDS 29(15):1987-2002, 2015.

Hamada Y. Overview of the new algorithm to expedite diagnosis, to include elements such as the use of rapid diagnostics, LF-LAM and presumptive TB treatment. 21st International AIDS Conference, Durban, abstract THWS0303, 2016.

Download the presentation slides from the conference website.

Meintjes G et al. HIV-related medical admissions to a South African district hospital remain frequent despite effective antiretroviral therapy scale-up. Medicine 94(50):e2269, 2015.

Katagira W et al. Empiric TB treatment of severely ill patients with HIV and presumed pulmonary TB improves survival. J Acquir Immune Defic Syndr 72(3):297-303, 2016.