Diagnosing TB in HIV

FASH Graphic

Point-of-care ultrasound predictors for the diagnosis of tuberculosis in HIV-positive patients presenting to an emergency center

J Acquir Immune Defic Syndr Jan 2020 - Pubmed Link


Take Home Points

1. The FASH exam was not sensitive enough to rule out TB in this population.

2. Intraabdominal lymphadenopathy, ascites, and pericardial effusion are predictive of TB in HIV positive patients.

3. More research is needed to evaluate this in different populations and with less expert users.


Background

Most people who know ultrasound are familiar with the FAST exam, but have you heard of the FASH?! Focused Assessment with Sonography for HIV/TB (FASH) is a protocol designed to find signs of HIV-associated tuberculosis using point-of-care ultrasound. This has been described before in areas of high HIV prevalence, but the accuracy of the exam has not been well established. Most of the prior studies have been retrospective, small, or lacked robust reference standards. The article cites a prior systematic review showing pooled sensitivity of 63% and specificity of 68%, but very low quality evidence. Using ultrasound can be helpful because many of these patients might not have a typical presentation of TB and may not have positive cultures. These authors perform a study to validate this protocol and examine just how good it is at diagnosing TB in these patients.


Questions

What is the diagnostic accuracy of individual POCUS features of the FASH exam?

What components of the FASH are independent predictors of HIV-associated tuberculosis?


Population

Cross-sectional diagnostic study performed in single center in Cape Town, South Africa

HIV prevalence at this hospital is 27%, TB rate is 917 per 100,000


Inclusion:

  • Adults ≥18 years

  • HIV positive (by history or a rapid test)

  • Any symptoms of TB (cough, fever, night sweats, weight loss)


Exclusion:

  • Anti-TB treatment within past 3 months

  • Pregnancy

  • Presented to emergency center >24 hours prior to screening (meaning they missed them because of their limited window of screening)

  • Meningitis syndrome or new focal neurologic sign

  • Primary trauma, gynecological, psychiatric condition


Design

Prospective cross-sectional study

Consecutive patients presenting to the emergency center were screened Monday to Thursday for over 1 year


Directly after consenting, POCUS was performed

Physician performing POCUS was not blinded to clinical information, but was blinded to reference standard

They divided the FASH into categories (see Scan section below). FASH-basic was positive if there was pericardial effusion, ascites, or pleural effusion. FASH-plus was considered positive if there were upper abdominal lymph nodes ≥15 mm in diameter, focal splenic lesions, or focal liver lesions. FASH-combined was positive if any of either of those was positive.

They measured what they found and used a few different cut offs to test which would work better:

  • Pericardial effusions - any size and ≥ 5 mm

  • Intraabdominal lymph nodes - any size, ≥5 mm, ≥10mm, ≥15mm)


Reference standard was M. tuberculosis Xpert MTB/RIF (this is a rapid sputum or urine TB and Rifampin resistance test) or culture

Sub-optimal POCUS images counted as negative.


Power analysis calculated they would need 400 HIV patients to get about 100 TB cases


Who did the ultrasounds?

Single emergency physician - first author of this paper


The Scan

Curvilinear probe (but also could use phased array if you want, especially for thoracic components)

Curvi


The authors describe two versions of the FASH exam

FASH-Basic

  • Pericardial effusion?

  • Pleural effusion?

  • Ascites?

FASH-Plus

  • Upper abdominal lymph nodes (assessed peri-portal, para-aortic, splenic, mesenteric)

  • Focal hypoechoic splenic lesions (any size or number)

  • Focal hypoechoic liver lesions (any size or number)

FASH-Basic + FASH-Plus = FASH Protocol (or FASH-Combined)


Great introduction to the FASH video from the Academy of Emergency Ultrasound


Images of FASH Pathology from GrepMed


Results

556 screened, many excluded due to HIV negative, refusal, or no reference standard done


N = 414

  • 41.5% had confirmed TB

  • Median age 36

  • 52.8% had prior TB at some point

  • 47.1% on antiretroviral therapy currently

  • Median weight 54 kg, BMI 20

  • Mean HR 122

  • Median CD4 count 86

  • 7.2% in-hospital mortality


Primary Outcome - Diagnostic Accuracy of Individual POCUS Findings

FASH-Combined (≥1 finding)

Sensitivity 73% (CI 65-79)

Specificity 54% (CI 47-60)


Secondary Outcomes

✱This seems like similar results to primary outcome, but here the authors are a) analyzing the protocols instead of each individual findings and b) they are attempting a statistical external validation


External Validation of Protocols

FASH-combined had sensitivity of 71% and specificity of 57%

C-statistic for FASH basic was 0.609, FASH-plus 0.598, and FASH combined 0.630

  • C-statistic is also called concordance, it is similar to the area under the ROC curve


Independent POCUS Predictors

Intra-abdominal lymph nodes - (adjusted diagnostic odds ratio) OR 3.7 (CI 2-6.7)

Ascites - OR 3 (CI 1.5-5.7)

Pericardial effusion - OR 1.9 (CI 1.2-3.0)


Any two findings had 91% specificity (CI 86-94), LR+ of 3.7

Having ≥3 features was 93% specific, ≥4 was 99% specific.


Other Findings - Subgroup Analysis

In patients with CD4 ≤100/mm3, the FASH sensitivity increased to 82% and specificity was similar at 55%.

Even in patients with CD4 ≤100, having ≥1 positive feature only gave LR- 0.3 and having ≥4 features gave LR+ 7.3.


Limitations

Single center, single skilled operator

Somewhat convenience sample, sampling bias. Only patients on Monday to Thursday.

Sub-optimal images were counted as negative. This could increase false negatives. This could make the sensitivity appear lower than reality.

Reference standards. There were two standards which is a little tricky because it makes the results a little harder to interpret. The rapid test MTB/RIF seems less accurate than the traditional cultures and according to the CDC, should be confirmed by cultures. Additionally, there were several (63) patients diagnosed with TB who did not actually have microbiologically confirmed TB.

Patients with a low pre-test probability for TB were included (these patients were essentially being screened for TB) which could decrease the accuracy.


Discussion

In a study that has multiple components it is interesting to note that ascites, lymph nodes, and pericardial effusion were independent predictors of TB. However, in this patient population almost half of the patients had the disease. In an area of lower incidence these findings would unlikely be from other conditions. These authors suggest that finding any two findings on the FASH would be a reasonable test to justify initiating treatment. I think this obviously depends on your population and a risk-benefit consideration. A +LR of 3.7 and a specificity of 91% is not great, but perhaps if you are getting a confirmatory test, this would allow you to rapidly initiate treatment in the meantime.

Another recent article on the FASH exam asked a slightly different question. They routinely used FASH as one data point to help decide if they should start treatment. They found that having a positive FASH exam did make providers more likely to initiate treatment. This goes along with the findings of this study. People are already using this in the field and it seems to work okay.


Take Home Points

1. The FASH exam was not sensitive enough to rule out TB in this population.

2. Intraabdominal lymphadenopathy, ascites, and pericardial effusion are predictive of TB in HIV positive patients.

3. More research is needed to evaluate this in different populations and with less expert users.


More Great FOAMed on this Topic

Ultrasound Podcast - FASH Podcasts


Our score

3 Probes


Expert Reviewer for this Post

Panebianco

Nova Panebianco, MD, MPH @Novaleda

EM Ultrasound Division Director, Ultrasound Fellowship Director, Associate Professor, University of Pennsylvania, Department of Emergency Medicine


Reviewer's Comments

While this exam may not be relevant to most North American settings, in populations where HIV and TB are prevalent, the presence of +FASH findings in addition to high clinical suspicion, may be used to expedite care. While the FASH protocol has many parts and includes imaging above and below the diaphragm, the presence of intra-abdominal lymphadenopathy, ascites, and/or pericardial effusion were predictive of TB in HIV positive patients. The lower the CD4 count, the more sensitive the FASH exam becomes.


Cite this post as

Michael Prats. Diagnosing TB in HIV. Ultrasound G.E.L. Podcast Blog. Published on April 27, 2020. Accessed on May 28, 2020. Available at https://www.ultrasoundgel.org/90.
Published on 04/27/20 05:00 AM
comments (0)