An mpox result at the bedside in under 40 minutes

Design

prospective diagnostic accuracy study with nested qualitative usability evaluation, against laboratory qPCR

Population

300 consecutively enrolled people of any age meeting the WHO suspected-case definition for mpox with an active cutaneous lesion, at six facilities in Kampala and Wakiso, Uganda

Primary outcome

sensitivity and specificity of the point-of-care platform for monkeypox virus and orthopoxvirus against qPCR

Effect

overall agreement 98.7% (95% CI 96.6 to 99.5); MPXV sensitivity 98.5% (95.6 to 99.5), specificity 96.2% (90.5 to 98.5); results in under 40 min

Six health facilities in Kampala and Wakiso enrolled 300 consecutive people of any age who met the WHO suspected-case definition for mpox and had at least one swabbable lesion. Each lesion swab was tested at the point of care on Dragonfly, a sample-to-result molecular platform with dual targets for orthopoxvirus and monkeypox virus, and then by confirmatory laboratory qPCR. Of the 300, 196 (65%) were positive by qPCR, at a median cycle threshold of 21.1.

Agreement with qPCR was 98.7% (95% CI 96.6% to 99.5%), with results available in under 40 minutes. For monkeypox virus, sensitivity was 98.5% (95.6% to 99.5%) and specificity 96.2% (90.5% to 98.5%); for orthopoxvirus, sensitivity 100% (98.1% to 100%) and specificity 96.2% (90.6% to 98.5%). Front-line staff in focus groups rated it highly, largely for removing the sample-transport step, while flagging training and supply chain as the conditions for routine use.

The reason this matters is turnaround, not accuracy - centralised qPCR is already accurate, but a result that arrives days later cannot drive isolation or contact tracing. A diagnostic accuracy study is not an outbreak-control study, though, and the evaluation was done in a high-prevalence population where 65% of suspected cases were positive. Specificity of 96.2% behaves very differently when the suspected-case rate is low: in a setting where few suspects are truly positive, one in twenty-five negatives called positive becomes the dominant error.

• Judge a point-of-care platform on turnaround and on who can run it, not only on sensitivity.
• Recalculate the predictive values for your own setting before deploying - 96.2% specificity is not the same test in a low-prevalence clinic.
• Confirm positives by laboratory qPCR where the result changes public health action or clade assignment.
• Plan the consumable supply chain and operator training before the platform arrives; users named both as the limiting factors.
• Performance across clades and in key populations was not established here - do not assume it transfers.

The statistics, in plain English

Sensitivity and specificity do not change with prevalence, but what a result means to the clinician does. At the 65% positivity seen here, a positive is almost certainly true. Drop the prevalence to 10% and the same 96.2% specificity produces roughly one false positive for every three true ones. The confidence intervals also tell you how many negatives there were to test specificity on: an interval of 90.5% to 98.5% is wide because only about a hundred participants were qPCR-negative. Median cycle threshold of 21.1 means most positives carried a high viral load; performance at the low-load end is where a rapid platform usually loses ground, and that group was small here.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.