06
Keeping Watch: Public Health Surveillance
How health departments spot problems before they explode into outbreaks
The word surveillance comes from the French sur ("over") and veiller ("to watch"). In epidemiology, it means the continued, systematic watchfulness over a health problem through the ongoing collection, analysis, and sharing of data. Surveillance is often summed up as "information for action." A health department that never looked at its own case data would have no way to know an outbreak was starting until it was already out of control.
The 5-Step Surveillance Process
This isn't a one-time checklist; step 5 always loops back into step 1, keeping the watch going continuously.
How tests word it: many tests list the five steps as data collection, data analysis, data interpretation, data dissemination, and link to action. Same process, just named by what happens to the data.
Four Types of Surveillance
Passive Surveillance
Health-care providers report cases to the health department on their own, following standard rules. Simple and cheap, but reporting can be incomplete.
ExampleA doctor mails in a form every time they diagnose a notifiable disease.
Active Surveillance
The health department reaches out and asks providers for case reports, instead of waiting. More complete, but more time and resources.
ExampleStaff call every hospital in the county each week during an outbreak.
Sentinel Surveillance
A pre-arranged network of "sentinel" providers (clinics, hospitals, or labs) agrees to consistently report specific conditions.
ExampleA network of doctors nationwide reports every flu-like illness they see each week.
Syndromic Surveillance
Instead of confirmed diagnoses, watchers track symptoms or proxies (like ER visits or pharmacy sales) to catch problems earlier, useful when timeliness matters most.
ExampleTracking a spike in over-the-counter cold medicine sales before flu is even diagnosed.
Trade-offs to know: passive surveillance is the cheapest but the least complete. Active surveillance finds more cases and is more timely, but costs the most staff time. Syndromic surveillance gives the earliest warning, but it's the least specific (many illnesses share the same symptoms).
Wastewater (environmental) surveillance: testing a community's sewage for a pathogen's genetic material. It tracks a whole town at once, including people who never get tested, and often shows a rise days before case reports do. It was used widely for COVID-19 and polio.
How a report travels: a doctor or lab reports a notifiable disease to the local (county) health department, which reports to the state health department, which notifies CDC (through the National Notifiable Diseases Surveillance System). States are legally required to collect reports; sending them on to CDC is voluntary.
🕐 Why Syndromic Surveillance Can Save a Full Week
Imagine someone is unknowingly exposed to an aerosolized biological agent on Day 0. Here's how the case actually gets discovered using only traditional (passive) reporting:
- Day 2: Feels feverish, buys medicine at a pharmacy.
- Day 3: Develops a cough, calls their doctor's office.
- Day 4: Sees a physician, is diagnosed with "the flu."
- Day 5: Feels much worse, calls 9-1-1, goes to the ER, but is sent home.
- Day 6: Admitted to the hospital with pneumonia.
- Day 7: A radiologist finally spots a telltale sign on a chest X-ray; the health department is notified that same day.
It took a full week for the health department to learn about this one exposure, but a syndromic system tracking pharmacy sales, 9-1-1 calls, or ER visits could have flagged an unusual pattern days sooner.
What Makes a Surveillance System "Good"?
Not every surveillance system is built the same way, and no system can maximize every quality at once. There are always trade-offs. Here are the attributes epidemiologists check when evaluating one:
Timeliness
Data are available fast enough for officials to actually act on them.
Sensitivity
The system catches most of the real cases that are actually out there.
Specificity
The system doesn't flag lots of people who don't really have the problem.
Simplicity
Easy to operate: simple case definitions, easy-to-get data.
Flexibility
Can adapt to new health problems or changing needs without a costly overhaul.
Representativeness
Accurately reflects the real pattern of disease by person, place, and time.
Acceptability
People and organizations are willing to actually participate and report.
Stability
Reliable and available: the system doesn't crash or lose data.
Data Quality
The recorded data are complete and correct (few blank or wrong fields).
Predictive Value Positive
Of the cases the system reports, the share that really have the condition.
CDC's official list has nine: simplicity, flexibility, data quality, acceptability, sensitivity, predictive value positive, representativeness, timeliness and stability. (Specificity, above, is a useful extra that isn't on CDC's list.) An evaluation also asks what the system is for, what it costs to run, and whether it's actually useful.
Cases can "rise" without a real increase. Before calling a jump in reported cases an outbreak, rule out: a bigger population (compare rates, not counts); more testing, or a new, more sensitive test; a new or changed case definition; a new screening program finding cases that were always there; and better reporting or news coverage making people see a doctor. This is also why a jump in prevalence can come from screening alone: it finds mild, long-lasting cases that would never have been diagnosed.
A trade-off in action: Making a system more sensitive (catching more real cases) often means it also flags more false alarms, which lowers its predictive value and can waste public health resources chasing down cases that turn out not to be real.
✓ Check Yourself
✓ Complete
Q10Instead of waiting for reports, a health department calls every hospital in the county each week to ask for updated case counts. What type of surveillance is this?
Active surveillance: the health department is initiating contact, rather than relying on providers to report on their own (passive surveillance).
Q11Why does timeliness matter so much in a surveillance system?
The whole point of surveillance is "information for action": if data arrive too late, public health officials can't act quickly enough to stop a problem from spreading further.