When people hear “disease surveillance,” they usually picture something abstract — dashboards, case counts, maybe a map with dots on it. What they don’t picture is the actual work: a small team of epidemiologists and data analysts combing through case reports every week, looking for patterns that don’t announce themselves.
That’s what Cluster Detection and Response (CDR) is. It’s a core part of how public health departments track HIV transmission, and it’s one of the more quietly demanding disciplines in epidemiology — because the signal you’re looking for is almost never obvious until it is.
The work behind the map
A “cluster” in HIV surveillance is a group of cases that share enough genetic or epidemiologic similarity to suggest they’re connected — a chain of transmission moving through a community faster than average. Spotting one early can mean the difference between a contained outbreak and a much larger one.
Finding it requires layering several types of evidence: molecular data from viral sequencing, case interview data, timing, and geography. None of these are decisive alone. A city the size of Houston — with a metro population well over two million — generates enough case data that the real skill isn’t collecting it, it’s knowing which threads are worth pulling.
That’s where a lot of surveillance programs stumble. It’s easy to build a system that flags statistical anomalies. It’s much harder to build one that a team can actually act on every week, under real staffing and time constraints, without drowning in false positives or missing the case that matters.
Why the “boring” parts matter most
The unglamorous truth is that most of what makes cluster detection effective isn’t the analysis — it’s the operational scaffolding around it. Standard operating procedures for how a suspected cluster gets escalated. Clear ownership of who investigates what. A weekly case conference where findings actually get discussed, not just reported. Quality assurance protocols that catch data entry errors before they become false alarms.
None of that shows up in a case study headline. All of it determines whether a real cluster gets caught in week two or week twelve.
This is also where the relationship between local health departments and state or federal partners matters. A city-level team might spot a pattern, but confirming it — and mobilizing the right response — often depends on tight coordination with state health departments and the CDC’s Division of HIV Prevention. Getting that coordination right, consistently, is as much a program management problem as an epidemiologic one.
What I’ve taken from this work
Leading a surveillance team taught me that good epidemiology and good operations aren’t separate skills — they’re the same skill applied at different altitudes. You can have brilliant analysts and still miss clusters if the escalation process is unclear. You can have a clean SOP and still miss them if nobody’s empowered to question the data.
The programs that work well are the ones where the science and the process reinforce each other: rigorous enough to trust the signal, disciplined enough to act on it fast.
That’s the part of public health leadership I find most underrated — and most necessary. Anyone can build a dashboard. Building a team and a system that consistently turns data into timely, correct action is a different, harder job. It’s also, I’d argue, the job that actually protects people.