Guide · Surveillance

Disease cluster detection: how to see an outbreak in days, not in the weekly report.

A cluster is usually visible in the data well before anyone acts on it. The delay comes from weekly reporting, separate systems and nobody owning the signal.

By · Updated

Cluster detection means spotting when cases of an illness in a small area rise above what is normal for that place and season, early enough for a response to matter.

Why outbreaks are seen late

How to detect clusters earlier

  1. Bring cases, lab results and facility reports together by district and block.
  2. Build a baseline for each block for the season, from past years.
  3. Compare current cases with that baseline every day, not every week.
  4. Join helpline calls by area: a spike in fever calls is an early signal too.
  5. Raise each cluster with its evidence: the numbers, the area and the trend.
  6. Assign every cluster an owner and a next action, and track it to closure.

Pitfalls to avoid

PitfallWhat to do instead
Alerting on raw countsCompare with the seasonal baseline for that block
Too many alertsRank by risk and show the evidence with each one
Alerts by emailRoute to a named owner with a time to act
Hiding stale dataShow source and timestamp on every number

Thresholds, case definitions and response protocols belong to your surveillance programme; confirm them with your epidemiologists.

How HealthWatch does it

HealthWatch is a public-health command centre for health secretaries, mission directors and district health officers. It pulls surveillance and HMIS data, lab results and facility reports per district; refreshes ICU, oxygen and ventilator availability every 15 minutes; and joins helpline cases and ambulance response to the same districts. Cases are compared with the seasonal baseline by block, clusters are raised with evidence, and every exception is ranked and assigned to a district or hospital owner. Each district gets a morning brief on WhatsApp with decisions that close in one tap, and every exception keeps its owner, time and closure for the weekly review. Every number keeps its source and timestamp, so stale or estimated data is never hidden. It runs on web and WhatsApp, on managed cloud, in your own VPC or on-premises.

Clinical and programme decisions stay with your health officials and their protocols; the command centre puts the evidence and the owner in front of them sooner.

Questions

How are disease clusters detected?

By comparing current cases in a small area, such as a block, with the normal level for that place and season, and raising a cluster when cases rise above that baseline.

Why use a baseline rather than case counts?

Some areas always have more cases in certain seasons. A baseline shows what is unusual for that block, not just what is large.

Can helpline calls help detect outbreaks?

Yes. A rise in calls about similar symptoms from one area can be an early signal, especially when joined to surveillance data for the same district.

See it on your own data. HealthWatch — Outbreaks and ICU load, one screen. Book a 30-minute working session with an engineer.

General guidance, current as of the date above. Figures and examples are illustrative unless a source is linked.