A nurse developed fever, muscle aches and fatigue on 24 April at the Evangelical Medical Centre in Bunia, Ituri Province. She was tested for Ebola. The result came back negative.
Four weeks later, she was dead. And by then, the virus she carried had spread to at least 8 confirmed cases across three health zones, sparking a continent-wide emergency.
The delay was not due to incompetence or lack of equipment. It was due to a machine specifically designed to detect Ebola, but programmed only to recognise one species of it.
This Ebola diagnostic failure represents a testing bottleneck that went unrecognised until genomic sequencing confirmed the truth. It has exposed a critical vulnerability in Africa’s outbreak detection infrastructure.
Health authorities now acknowledge that the Ebola diagnostic failure evident here stems from standard diagnostic tools across the continent being built around a single dominant threat, leaving rarer pathogens invisible until they have already spread.
The machine behind the Ebola diagnostic failure
Bunia Regional Health Centre uses GeneXpert, an automated diagnostic platform originally developed to detect tuberculosis. The system was adapted for rapid Ebola testing and has become standard equipment in dozens of African health facilities.
GeneXpert works well. It is fast, reliable, and can return results within hours rather than days. But the Bunia version only detects Ebola Zaire, the most common Ebola species and the cause of almost all previous outbreaks in the Democratic Republic of Congo. This limitation created the Ebola diagnostic failure that delayed outbreak detection.
It does not detect Bundibugyo.
When samples from the Evangelical Medical Centre nurse were tested at Bunia on 25 April, the machine read them as negative. A second batch of samples, collected between 28 April and 1 May from the Aru health zone, also tested negative at Bunia. Those same samples, retested at the National Institute of Biomedical Research (INRB) in Kinshasa with equipment designed to detect multiple Ebola species, also came back negative initially.
But when samples from Rwampara health zone, collected between 3 May and 7 May, arrived at Bunia, GeneXpert again reported negative. When those identical samples reached INRB in Kinshasa for further testing, they tested positive for Ebola Bundibugyo.
Four weeks had passed since the nurse’s initial symptoms.
Placide Mbala, head of epidemiology and global health at INRB, explained the testing gap to science journalists this week. “The regional health center of Bunia uses a diagnostic machine called Genexpert that only recognises Ebola Zaire,” Mbala said. “Part of the reason early tests didn’t detect Ebola Bundibugyo was this limitation in diagnostic capacity.”
Why the delay matters for containment
Those four weeks were critical. They allowed the nurse’s contacts to scatter across multiple health zones. Mining workers who may have crossed into Uganda departed undetected. Healthcare staff in Bunia who treated other suspected cases did not know they needed heightened infection control precautions.
By the time confirmation came, the World Health Organisation noted, the outbreak suggested “a potentially much larger outbreak than what is currently being detected and reported.”
The WHO also recorded a pattern that confirmed transmission had already deepened: at least four deaths among healthcare workers across multiple facilities in Ituri, all displaying symptoms consistent with viral haemorrhagic fever. This healthcare-associated transmission indicated the virus had moved beyond initial community contacts and was being amplified inside hospitals where staff were unaware they were handling a Bundibugyo case.
The delay, the WHO stated, reflected “low clinical index of suspicion among healthcare providers” in the region. But that suspicion could only form if providers knew what they were looking for. They did not.
A testing architecture built for one threat
Bunia is not an outlier. Across eastern DRC and neighbouring countries, most diagnostic facilities use GeneXpert systems specifically configured for Ebola Zaire screening. This configuration was rational when Zaire was responsible for nearly every outbreak. The 2018 to 2020 North Kivu and Ituri outbreak involved Zaire. Previous major outbreaks involved Zaire.
But it created a blind spot for rarer species, leaving the continent vulnerable to the Ebola diagnostic failure now unfolding. Bundibugyo has only appeared twice before in the 21st century: a 2007 outbreak in Uganda that killed 42 people, and a 2012 outbreak in the DRC that killed 29. Because those events were so infrequent, investment in Bundibugyo-specific diagnostics remained minimal.
“The difficulties in detecting [Bundibugyo] show why a rapid test that can identify Bundibugyo in the field instead of a specialized lab is urgently needed,” Mbala told journalists. Such a test does not currently exist as a field-deployable tool.
Currently, if a GeneXpert system in the field returns a negative result for Ebola Zaire, samples must be packaged, transported, and retested at a national reference laboratory with broader diagnostic capacity. That process takes days. In this outbreak, it took weeks because samples were not prioritised for secondary testing until the outbreak had already spread.
The sequencing that came too late
INRB scientists eventually performed polymerase chain reaction testing (a more sophisticated genetic technique) which confirmed Bundibugyo virus. But this confirmation came after four weeks of false negatives had allowed the outbreak to advance unchecked. This delay exemplified the broader Ebola diagnostic failure facing the region.
The WHO’s Disease Outbreak News statement, released after sequencing confirmed Bundibugyo, now advises that “early tests didn’t detect Ebola Bundibugyo” due to diagnostic limitations. The agency has recommended that affected countries establish “decentralised laboratory capacity for testing of Bundibugyo virus” and strengthen “enhanced community surveillance, particularly focused on community deaths.”
But those recommendations acknowledge a problem that already exists: the testing infrastructure was not built for this scenario.
A continental pattern, not a Bunia anomaly
Health authorities across East Africa are now conducting urgent audits of their GeneXpert configurations. Preliminary findings suggest that most systems in DRC, Uganda, South Sudan, and Kenya are similarly limited to Zaire detection.
This diagnostic architecture developed logically over two decades. Zaire caused the overwhelming majority of Ebola cases. Governments and donors invested in tools that addressed the most common threat. But that efficiency created brittleness, and when a pathogen outside the expected set emerged, the system could not see it until it had already spread widely.
The Bundibugyo outbreak offers a hard lesson in the dangers of such Ebola diagnostic failure: diagnostic readiness requires capacity for the unexpected, not just the predictable. INRB in Kinshasa has the equipment to detect multiple Ebola species. But Kinshasa is over 1,000 kilometres from Ituri, and a single nurse in a remote health centre cannot access that expertise in time to change the course of an emerging outbreak.
“A field-deployable rapid test for Bundibugyo would have caught this in days,” one INRB scientist involved in the sequencing told AfricanQuarters on condition of anonymity. “Instead we caught it in weeks, after the virus had already moved.”
That gap between what the continent has and what it needs is now defining the entire response to this outbreak. Until the Ebola diagnostic failure is addressed at the infrastructure level, Africa remains vulnerable to the same delays when the next rare pathogen emerges.
