Design: prospective diagnostic accuracy study with mixed-methods usability evaluation, six facilities in Uganda
Population: 300 consecutively enrolled patients of any age meeting the WHO suspected-case definition with an active cutaneous lesion; 196 qPCR-positive
Primary outcome: sensitivity and specificity for monkeypox virus and orthopoxvirus against centralised qPCR
Effect: MPXV sensitivity 98.5% (95% CI 95.6-99.5), specificity 96.2% (90.5-98.5); OPXV sensitivity 100% (98.1-100); overall agreement 98.7%; results in under 40 minutes

Three hundred patients meeting the WHO suspected-case definition for mpox, with at least one swabbable cutaneous lesion, were enrolled consecutively at six health facilities in Kampala and Wakiso. Lesion swabs were tested on Dragonfly, a sample-to-result molecular platform detecting orthopoxvirus and monkeypox virus, then confirmed by centralised quantitative PCR. Of the 300, 196 were qPCR-positive for monkeypox virus.

Sensitivity for monkeypox virus was 98.5% and specificity 96.2%, with overall agreement of 98.7% and results available in under 40 minutes. For orthopoxvirus the sensitivity was 100%. Front-line users rated it acceptable, specifically because it removed the sample-transport step.

The reason to report a diagnostic evaluation on this desk is that mpox control is limited by turnaround time rather than by test accuracy. Centralised qPCR is accurate and slow, and a patient who has gone home before the result arrives is an isolation decision that was never made. Compressing that to under 40 minutes at the point of care changes what a clinic can do on the day.

The users' own caveats are the practical ones: training and supply chain. A platform that works in an evaluation with dedicated support behaves differently in a clinic that runs out of cartridges.

Where suspected mpox presents, the isolation decision is time-critical — record when the sample was taken and when the result is expected
Treat a negative point-of-care result in a high-suspicion case as provisional pending confirmation
Ask about cartridge supply and operator training before planning a service around a point-of-care platform
Swab an active cutaneous lesion; this evaluation enrolled only patients who had one

Don't overread it

This was a single-country evaluation in a high-prevalence outbreak setting; performance across clades and in low-prevalence populations was not assessed.

The statistics, in plain English

A specificity of 96.2% with a 95% confidence interval of 90.5 to 98.5 was measured against 104 qPCR-negative patients, which is a small denominator — that is why the interval is wide, and it means roughly 4 in 100 uninfected patients would test positive. In a setting where 65% of suspected cases were genuinely positive, as here, that matters little; in a low-prevalence setting the same specificity would generate a much higher proportion of false positives among those testing positive. Sensitivity of 98.5% was measured across a broad range of viral loads, with a median cycle threshold of 21.1 indicating mostly high viral loads.