JAMA Network Open April 2019 - Pubmed Link
Accompanying Commentary by friend of the show John Bailitz
1. In the largest prospective study yet, ocular ultrasound is highly sensitive for retinal detachment but only moderately specific.
2. It is specific for vitreous hemorrhage but otherwise not that accurate for vitreous hemorrhage or vitreous detachment.
3. Ocular ultrasound should not be used as an attempt to replace your ophthalmologist.
The eyeball was made for ultrasound: close to the surface AND mostly fluid!? Great combination for sound waves. This application has been used for many things: retinal detachment, lens dislocation, vitreous detachment and hemorrhage, optic nerve sheath diameter. Of those, the posterior ocular pathologies are probably the most common indication in the emergency department. The accuracy of this modality for retinal detachment, posterior vitreous detachment, and vitreous hemorrhage has been tested before but the populations were small and the results are all over the place. This group did a nice job of performing a robust study to address this issue.
Check out The Evidence Atlas - Orbital for a review of prior evidence on this topic.
How accurate is point-of-care ultrasound for diagnosing retinal detachment (RD), vitreous hemorrhage (VH), and vitreous detachment (VD)?
Emergency department patients at 4 sites. 2 academic, 2 country hospitals. All sites had EM residents, ophthalmology residency, and emergency ultrasonography fellowship.
Inclusion:
Exclusion:
No ophthalmologic consultation
younger than 18 years old
Non-English or non-Spanish speaking
Declined enrollment
Ocular trauma or suspicion for globe rupture
Incomplete data collection
Multicenter, prospective, observational, convenience sample
Patients enrolled between 8am and 12am by research assistants
POCUS performed by treating practitioner PRIOR to consultation with ophthalmologist
Ophthalmologist was blinded to POCUS results
More than one diagnosis could be reported
POCUS diagnoses were compared to final diagnosis by ophthalmologist after examination
Calculated would need sample size of 225 patients if there was at least 80% sensitivity
75 practitioners (70 physicians, 5 physician assistants) including:
EM attendings
EM residents
Supervised physician assistants
Variable POCUS experience and training
All practitioners received 30 minute lecture and 30 minutes hands-on scanning on healthy volunteers
Linear probe, ophthalmologic settings
Gel placed directly on eyelid (no barrier)
Both transverse and sagittal views obtained
Included static (operator fanning through eye) and kinetic (patient moving eye left and right).
The Pathology (authors' definitions):
Retinal Detachment: bright echogenic membrane tethered to optic disc but separated from choroid
Posterior vitreous detachment: detached, thin, mobile membrane at interface between vitreous and retina
Vitreous hemorrhage: fluid collection of variable echogenicity in posterior chamber that rotated with kinetic examination
Learn how to do Ocular Ultrasound from 5 Minute Sono!
Check out what ocular pathology looks like at the POCUS Atlas!
N = 225 (27 exclusions, mostly due to patients declining)
40% women
Mean age 51
Prevalence of Pathology: 36%
20.8% had RD
24% had VH
15.1% had VD
Primary Outcome
Retinal Detachment
Sens 96.9% (CI 80.6-99.6)
Spec 88.1% (CI 81.8-92.4)
Vitreous Hemorrhage
Sens 81.9% (CI 63-92.4)
Spec 82.3% (CI 75.4-87.5)
Vitreous Detachment
Sens 42.5% (CI 24.7-62.4)
Spec 96% (CI 91.2-98.2)
Convenience sample, no interrater reliability
Although multi-center, 173/225 (76.8%) were from one center, with 34, 14, and 4 from the other centers. Somewhat high prevalence of disease - 36% had one of the three pathologies.
Lower specificity for retinal detachment than prior studies. Authors state they were not aware of any other study differentiating RD, VH, or VD but we covered one from 2016 that showed high sensitivity and low specificity as well. A recent metaanalysis shows 94.2% sensitivity and 96.3% specificity, although lower accuracy if only looking at ED studies. The lower specificity seen in the current study could be based on the protocol, the machines, the training, or maybe its just that a larger study was able to give us a better answer.
Now we need some patient centered outcomes. Is time to diagnosis faster? Can we get them to OR faster? Less conversion to mac-off detachments?
Should we not use POCUS for VD and VH? These results are similar to other studies that say you can’t rule it out with POCUS. Maybe you could rule it in, especially VD with a fairly high specificity. Seems it would be wise to be cautious with these other diagnoses. This study didn’t look at retinal tears which can also be tricky to pick up.
What about macula-on or macula-off distinction for RD? We have no idea how accurate ultrasound is for this. There is one case report talking about it but most studies don’t attempt to differentiate. Theoretically you can extrapolate where the macula is based on the optic nerve sheath, but is that good enough?? That might be a cool study to do.
1. In the largest prospective study yet, ocular ultrasound is highly sensitive for retinal detachment but only moderately specific.
2. It is specific for vitreous hemorrhage but otherwise not that accurate for vitreous hemorrhage or vitreous detachment.
3. Ocular ultrasound should not be used as an attempt to replace your ophthalmologist.
Expert Reviewer for this Post
Reviewer's Comments
Confidence intervals aren’t very tight. Would also be helpful to know the test characteristics for each based on what level of provider and/or what level of PoC US experience providers had. Sensitivity for RD not THAT low, but lower than I would have expected. Would have helped to know whether the lower sensitivity for RD was because it was mistaken for something else (FP secondary to RH vs VD) or were no echoes seen in the post chamber. I would think, if performed correctly with high gain and pt moving their eyes (kinetic exam), if no echoes seen posteriorly, can r/o RD (as well as RH and VD). If lower sens b/c RD confused for RH or VD (ie, operator states “no RD” thinking RD was VD or RH, and this counts as a false neg even though the echo in the posterior chamber was picked up, it’s more a matter of that operator’s experience differentiating RD from other echoes as opposed to the ultrasound not picking up the detachment.