You don't need to be signed in to read BMJ Blogs, but you can register here to receive updates about other BMJ products and services via our site.


Primary Care Corner with Geoffrey Modest MD: Urine-Based Rapid TB Test

12 Apr, 16 | by EBM

By Dr. Geoffrey Modest

The lancet just reported a study looking at a rapid, low-cost urine test to guide TB treatment in HIV-positive individuals in areas with high TB prevalence (see Lancet 2016; 387: 1187).


  • TB is the leading cause of death in people with HIV, accounting for 360K deaths in 2013; post-mortem analysis in resource-limited countries find that TB is the cause of death in about 40% of HIV-infected people, in 85% of cases it is disseminated, and half are undiagnosed at the time of death
  • Those with HIV have a higher case-fatality rate when co-infected with TB: increased disseminated extra-pulmonary TB, more severe TB as immunocompromise increases.
  • It is harder to diagnose TB in those with advanced immunosuppression, since they often have low bacillary loads in their bodily fluids, reducing the sensitivity of smears and cultures
  • ​There is a urine test detecting the lipoarabinomannan Ag (LAM), a glycolipid antigen of the M tuberculosis cell wall, which reflects hematogenously disseminated TB. It requires 60 ml of urine, takes 25 minutes to get a result, and costs $2.66/test. It has a specificity of approx 94% and a sensitivity of around 56% in a meta-analysis. but, compared to sputum-smear microscopy, it does identify patients with the most severe illness

Details of study:

  • 2659 patients (median age 37, 51% female, CD4 of 84, 48% on antiretroviral therapy (ART) at time of hospitalization, 73% on ART by 8-week follow-up) in 10 hospitals in Africa (South Africa, Tanzania, Zambia, Zimbabwe) were randomly assigned to urinary LAM testing, with 2528 ultimately in the modified intention-to-treat analysis
  • 8-week mortality was 578 (23%)
    • 261 (21%) in the LAM group and 317 (25%) in the no LAM group, an absolute reduction of 4% (1-7%) and relative risk reduction of 17% [RR 0.83 (0.73-0.96, p=0.012]
  • Overall sensitivity of LAM was 45.6%, specificity was 88.7%, positive likelihood ratio of 4.03, negative likelihood ratio of 0.61
  • But for those with CD4 ≤50, sensitivity of LAM was 63.7%, specificity was 83.2%, positive likelihood ratio of 3.80, negative likelihood ratio of 0.44; and 46% of the deaths happened in those with CD4 ≤50
  • The overall % of patients begun on antiTB therapy was much higher in the LAM group (55% on day 1 and up to 91% on day 8) vs those without LAM (40% on day 1 and up to 89% on day 8), p=0.024 for difference.
  • In the subgroup of patients with CD4 ≤​50, LAM reduced mortality by 29%
  • The attending clinicians delayed TB therapy in the no LAM group because they favored a different diagnosis (e.g. bacterial pneumonia), or they were waiting for the results of diagnostic investigations.

Of note, a recent concern has developed with the spread of MDR-TB (multiply drug-resistant) in Daru Island in New Guinea, with a commentary in Lancet Respiratory Medicine (see ). The spread of TB, especially resistant TB, is referred to as a “time bomb”. And, of rather concerning note, TB is not so common in Daru and most patients with MDR have never taken TB meds (i.e., they did not develop resistance by taking meds, but were infected with a very difficult-to-treat TB infection), and lots of people there got infected (200 people in this 6 km2 island with only 15,000 individuals, i.e., about 1% of the population). This scary outbreak “will almost certainly eclipse those of both the Ebola and the recent Zika virus outbreaks, deemed a global public health emergency, combined”, per the Lancet commentary.

So, I bring up this LAM study for several reasons:

  • It seems pretty likely to me that TB really is a ticking time-bomb, and reading this article on LAM is a reality check on this
  • TB tends to have the highest morbidity and mortality in resource-poor countries, which have to deal with an array of very expensive and urgent other issues (little things like famine, war, etc.). A couple of days ago the NY Times ran a very concerning article on Vietnam (see​ ), a country with a remarkably effective TB program (90% cure rate for uncomplicated TB, 75% in drug-resistant cases — with a global average is 50%), where improving outcomes further is getting more expensive (needing to do more outreach into rural areas, finding/treating those addicted to heroin…), but their hospital wards are packed (with more potential to lead to spread of MDR) and their “money is close to running out”
  • TB has the potential to develop highly virulent strains, and the above article on New Guinea is shocking at how virulent a multi-drug resistant TB can be (and MDR, of course, is associated with much more cost to control as well as an inherently higher mortality). And the risk of worldwide spread of a virulent and less-treatable TB is much increased as travel becomes easier and more frequent
  • We are continually dealing with outbreaks of newer, potentially devastating diseases (e.g., Zika, Chikungunya, Ebola,…) which divert attention and resources from staid old TB
  • And LAM really may be a huge benefit in resource-limited countries, where AIDS-related deaths are still very high (i.e., AIDS has not evolved into a chronic disease requiring taking one pill to control, as in the US and many resource-rich countries), and LAM is a cheap and easy test to identify and treat those with likely imminent death
  • But even in the US, TB infection in HIV has been increasing, with increasing numbers of observed vs expected cases (see​ ), noting that the increase in cases in US cities is both from reactivation of latent TB by HIV-related immunocompromise, but that 41% were due to recent transmission
  • Now, one might argue that there should just be empiric anti-TB therapy in all HIV-infected patients in TB-endemic areas, given the high prevalence and mortality of HIV-TB coinfection, at least in those with more severe immunocompromise. But, empiric treatment is more complicated, more expensive, and more toxic, especially in areas of MDR, which reinforces the potential benefit of LAM as a test in those co-infected with HIV and even only a remote possibility of hematogenous TB. And, with the very real possibility of TB spreading outside of the currently endemic areas (as now shown in New Guinea), and co-infecting those with HIV, LAM may become a reasonable test in more and more countries around the world.


Primary Care Corner with Geoffrey Modest MD: Latent TB Infections: Screening and Treatment and Probs with IGRAs

15 Mar, 16 | by EBM

By Dr. Geoffrey Modest

The USPSTF just circulated a draft recommendation for screening for latent TB infection (LTBI) — see​ . In brief, they are recommending that adults “who are at increased risk for TB” be screened, grade B recommendation (high certainty that net benefit is moderate).


  • The 2011-12 National Health and Nutrition Examination Survey estimates the prevalence of LTBI in the US at 5% (worldwide, about one-third of the world’s population is infected with TB, with 1.5 million deaths in 2014. in the US there were 555 deaths in 2013)
  • 30% of those exposed to Mycobacterium TB develop LTBI; of these, 5-10% progress to active TB disease
  • Those at increased risk for TB: born in or were former residents in countries with increased TB prevalence, or have lived in high-risk congregate settings (e.g., homeless shelters, correctional facilities)
  • In 2014, 66.5% of active TB infections were among foreign-born, with 13.4x the incidence of US-born
  • ​>50% of foreign-born with active TB were from: Mexico, the Philippines, Vietnam, India and China. The CDC also adds Haiti and Guatemala to the list of high risk.
  • In those >15 yo with active TB, 5.5% were homeless within past year, 2.2% were in long-term care facilities, 4.2% in correctional facilities at time of diagnosis
  • Estimates of those with LTBI by TST (Mantoux tuberculin skin test): 23.1-87.6% of prisoners and 18.6-79.8% of homeless
  • Those in health care settings may also be at higher risk of infection and may be an important site for TB surveillance
  • And, the subgroup who are much more likely to develop active TB from LTBI: immunosuppressed [including HIV, those on immunosuppressives (including those on TNF-a inhibitors, and those receiving organ transplant) and silicosis]. They did not comment on other common subgroups: those with renal failure, diabetes, on prednisone >=15mg/d for one month, and I would consider testing for LTBI in all of these groups and treating if positive.
  • Screening can be done by either TST or one of the IGRA assays. Both are supported by the CDC (for full updated CDC guidelines from 2010, see ). But see below blog from July 2014 for my concerns about the IGRA tests.
  • Benefits of detecting LTBI: adequate evidence that treating LTBI with one of recommended CDC regimens decreases progression to active TB
  • Harms: primary harm is hepatotoxicity
  • Screening intervals: unclear
  • Treatment: as per CDC guidelines

So, a few points:

  • I work in a community with a pretty high prevalence of TB exposure, BCG use, and also with several cases of active TB over the years (mostly in older people, perhaps from waning immune systems). One of our big issues is interpreting the TST in those with BCG exposure. The reality is that we do not have an exact method, since there is no gold standard for the diagnosis of LTBI. And I do not trust the new IGRA tests, though on paper they make a lot of sense especially in the subgroup who received BCG immunization (see the blog below for details of 2 rather disturbing studies, suggesting that the IGRA results are not stable and therefore not completely reliable). Unclear why they are unstable. perhaps different numbers of interferon-g-producing T cells in the periphery, perhaps reversion to normal from a positive reflects that the immune system cleared the TB infection, maybe technical errors in performing the assay, or the definition of a positive IGRA may be too lenient (see articles below)
  • I have been prescribing the CDC-recommended rifampin-based treatment in patients who are not concurrently on meds that interact badly with rifampin, a strong CYP3A inducer. The rifampin regimen is shorter (4 months, vs 6-9 months for INH) and has significantly less hepatotoxicity

For some reason, the blog from July 2014 did not make it onto the BMJ website, so I will reprint it below:

A recent article found the disturbing result that the Interferon-γ release assays (IGRAs) For tuberculosis seems to have many false positive results (see DOI: 10.1164/rccm.201305-0831OC). This is particularly disturbing since many institutions (including some of the most prestigious hospitals) rely heavily or exclusively on IGRAs [70 have adopted QuantiFERON-TBGold (QTF)]. I sent out a blog a couple of years ago questioning the validity of IGRAs at the Cleveland clinic (see below). The current study is more impressive and with larger numbers of health care workers. For background reference, MMWR did a full issue on PPD vs IGRAs in 2010 (see ). There was some concern voiced to me by a TB specialist member of the CDC committee involved in producing this paper at that time, stating that most of the committee members voicing approval for IGRAs (not including him) were receiving funding from producers of the tests, and that he felt that there was insufficient data to endorse IGRAs at that time. (Of course, one of the issues is that there is no good gold standard – the comparison is with PPDs, which can have false pos and neg results). The current study was at Stanford (which adopted serial QFT screening in 2008, with more than 10,000 employees).  Baseline at Stanford:

  • 611-bed hospital with mean of 14 cases of pulmonary TB/yr between 2006-2011
  • Surrounding community with estimated TB rate of 10.7-10.9 cases per 100K people
  • From 2008-2010, all employees were screened annually with QFT, independent of prior PPD status or if had received therapy for LTBI (latent TB infection)
  • Analyzed 9153 health care workers with at least 2 results over time from QFT, 72% female, man age 43


  • Overall 13.9% of QFT tests were positive
  • Of 1223 individuals initially with positive QFT,5% remained positive (“persistently positive”)
  • Of 8227 initially negative, 4.4% converted to positive (which was 11x the rate of prior conversions with PPDs!!)
  • Of those whose QFT reverted from positive to negative, 73.9% had titers between 0.35 and 1.0 IU/ml (0.35 being the cutoff value for positive), vs that titer in 62.4% going from negative to positive and 27.8% of persistently positive
  • Of the 361 workers who converted to positive, repeat testing was done in 60 days in 261 of them and 8% reverted to normal.
  • In 16 low-risk workers with conversions from negative to positive who were persistently positive after the repeat test, 12 (75%) reverted ultimately to negative
  • If change cutoff for positive increased to be 1.0 IU/ml, that would eliminate 67% of discordant repeats, but fail to pick up 33.7% of persistently positives
  • And one interesting analysis: if one were to assume that the PPD conversion rate at the institution (0.4%) were a true reflection of LTBI (and it has the longest track record with such repeat testing), that would translate to a QFT cutpoint of 5.3 IU/ml to get the same 0.4% conversion rate. In that case the persistent positive QFTs (who presumably have the highest likelihood of real LTBI) would be reduced by 68.6%

So… one would have thought (including me) that this more specific test (the IGRAs) would be great, esp for those of us working in communities with a high prevalence of prior BCG vaccination and a moderately high rate of TB. This article is pretty impressive in demonstrating the variability of QFT results. Not sure if this applies to other IGRAs (the MMWR considers the QFT and T-SPOT as pretty equivalent), or if QFT would really be more accurate with a higher cutoff value for positive (again, hard to assess directly since there is no great gold standard. and, would one then miss a lot of true positives??). Also would have been interesting to know what % of workers at Stanford with positive PPD had negative QFT, but this was not reported in this article. Bottom line: probably best to stick with PPDs as the test-of-choice

An additional blog was circulated in 2012, prior to the BMJ blogs, finding false positives in the Cleveland Clinic (see Chest 2012; 142: 55). In this retrospective chart review of health care workers from 2007-2010, these workers had received the QuantiFERON-TB Gold test. In very brief:

  • 7374 IGRAs were done on newly hired health care workers
  • 486 (6.6%) were positive at baseline 305 (4.1%) were indeterminate and 6,583 (89.3%) were negative
  • Serial IGRAs were done and detected 52 (2.8%) who converted to positive (mean value of 0.63 IU/ml, 29% with values >1). [Note that the baseline conversion rate for TST prior to IGRA testing was 0.09%, though there may have been people who did not return for TST readings]. for the converters:
  • 41% had a history of BCG immunization
  • None were part of documented TB outbreak
  • 43 of these 52 had further evaluation an 26 (61%) got INH therapy
  • 10 who did not get the INH (including one with the very high titer or 10 IU/mL) had a repeat IGRA 1-6 months later and 8 were negative!!
  • This article cites several studies and systematic reviews finding higher rats of reversions (to normal) and conversions (to positive) than TSTs.

EBM blog homepage

Evidence-Based Medicine blog

Analysis and discussion of developments in Evidence-Based Medicine Visit site

Creative Comms logo

Latest from Evidence-Based Medicine

Latest from EBM