Why a real blind spot in pandemic preparedness is undiagnosed patients
As an infectious disease fellow attending the Infectious Diseases Society of America’s 2008 meeting, I was struck by a point Arturo Casadevall made in his Kass Lecture—one that has stayed with me ever since. He described a fundamental asymmetry in infectious diseases: a gradual move away from specific microbiologic diagnoses.
He contrasted today’s syndrome-based approach—“community-acquired pneumonia”—with an earlier era in which clinicians not only diagnosed pneumococcal pneumonia, but identified the specific serotype to guide targeted serum therapy.
Today, armed with the blunt but powerful tool of broad-spectrum antibiotics, that level of diagnostic specificity has largely faded.
That idea maps closely to what I’ve called biological dark matter—the vast universe of microbes that cause human disease—but are never identified.
Most discussions of this problem focus outward: environmental sampling, wastewater surveillance, metagenomics of soil, animals, and air.
But there’s a more immediate—and more consequential—form of biological dark matter that gets far less attention:
The patients in our hospitals who never receive a specific diagnosis.
The Diagnostic Blind Spot
Every day, clinicians treat patients with:
Pneumonia of unclear etiology
Sepsis without a defined pathogen
Encephalitis with negative standard workups
Meningitis where cultures never yield an answer
These aren’t rare edge cases. They are routine.
We assign syndromic labels—“community-acquired pneumonia,” “viral syndrome,” “sepsis of unknown origin”—and move on. Treatment is empiric. Outcomes vary. A patient survives or they don’t. The chart closes without resolution. And the system moves on.
Many unexplained deaths have infectious origins that are never recognized during life.
From a clinical standpoint, that’s often acceptable. From a biosecurity standpoint, it’s a profound vulnerability.
Because every one of those cases is a data point we are not fully capturing.
And those data points are exactly where new threats first appear.
The Cancer Contrast
What makes this even more striking is that medicine already knows how to behave differently.
In oncology, diagnostic ambiguity is treated as unacceptable. Tumors are routinely characterized with extraordinary precision including histology, immunophenotyping, molecular markers, and genomic sequencing.
A lung cancer is not just “lung cancer”—it is EGFR-mutant, ALK-rearranged, PD-L1 high, or something else entirely. Therapy hinges on that specificity.
In infectious diseases, by contrast, we often accept far less resolution—even in critically ill patients. In many ICUs, roughly half of septic shock cases lack a microbiologic diagnosis, despite the presence of life-threatening illness.
We tolerate a level of diagnostic imprecision in infection that would be unthinkable in cancer.
That gap is not technological. It is structural.
Why we don’t look harder (even when we could)
It’s not because the tools don’t exist.
Advanced diagnostics—metagenomic sequencing, plasma microbial cell-free DNA assays like the Karius test—can identify pathogens that conventional testing misses.
But they are not used routinely. The reason is economic architecture.
Under Medicare’s Diagnosis-Related Group (DRG) system, hospitals receive fixed payments based on diagnosis. That creates a simple calculus:
Additional tests increase cost
High-end diagnostics are expensive
Identifying the exact pathogen often does not change reimbursement—or even management
This logic extends beyond advanced tools. Even multiplex respiratory panels are often scrutinized by administrators and insurers, sometimes requiring approval before they can be ordered.
So, the usual scenario is to end diagnostic investigation once the treatment plan is clear enough.
The system optimizes for efficiency, not discovery. And in doing so, it systematically suppresses diagnostic curiosity at the bedside.
The preparedness paradox
We invest heavily in surveillance systems designed to detect emerging threats:
Wastewater monitoring
Environmental sequencing
Wildlife pathogen discovery
These approaches are valuable—but noisy.
Most detected organisms:
Are not human pathogens
Will never cause disease
Or lack context for interpretation
This creates a paradox:
We are getting better at detecting what exists—but not necessarily what matters.
Meanwhile, the most relevant signal—unexplained human illness—remains under-characterized. These illnesses may represent the first foray of a pathogen into humans—and because they cause disease, they are not noise or benign viral chatter. Recall that the 2009 H1N1 influenza pandemic was first detected in two children who happened to seek care at facilities participating in influenza typing studies—not routine clinical settings in which the diagnosis would have been “viral syndrome” or influenza A.
A Different Surveillance Anchor: Syndromes without Answers
If biosecurity is about detecting threats that impact humans, then the most valuable dataset is not environmental—it is clinical.
Specifically:
Undiagnosed pneumonia
Culture-negative sepsis
Encephalitis of unknown origin
Atypical respiratory failure clusters
These are real-time manifestations of disease.
They already represent:
Pathogens we failed to detect
Known organisms behaving in unexpected ways
Or genuinely novel threats
Focusing here does something environmental surveillance cannot:
It applies a filter of human pathogenicity by default.
A Pandemic Preparedness Constraint
Pandemic preparedness will always be constrained if:
We tolerate large fractions of serious infections going undiagnosed
We prioritize environmental detection over clinical attribution
We fail to integrate advanced diagnostics where uncertainty is highest
There are companies trying to push surveillance closer to the point of care by turning hospitals themselves into biosensors. Using genomic analysis of wastewater—and increasingly air and environmental sampling—they aim to detect pathogens circulating within a facility before they are clinically recognized, linking those signals back to patient cases and potential reservoirs.
Wastewater may tell us what’s circulating.
Astute clinicians may sound the alarm when they uncover a zebra cloaked in horse’s hair.
But undiagnosed patients tell us what’s dangerous before we know to look.
Any biothreat radar that does not encompass the biological dark matter within our health care facilities will remain fundamentally incomplete.
And until we align incentives, diagnostics, and surveillance around that reality, a significant portion of the biological world that matters most will remain—both scientifically and operationally—unknown.