Mistake 1: Case Definitions That Cannot Be Applied in the Field
A case definition is only useful if a health worker with five minutes per patient can apply it consistently. Most are written for epidemiologists reviewing records, not community health extension workers triaging patients. The result is systematic misclassification at the point of first contact.
The fix: Translate every case definition into a maximum three-question algorithm applicable in under 60 seconds. Field-test the algorithm with actual health workers before deployment.
Mistake 2: Surveillance Data That Reports Outputs, Not Signals
Output-focused surveillance asks: "How many cases were reported this week?" Signal-focused surveillance asks: "Is the pattern of cases this week a departure from the expected baseline that requires investigation?" Most DHIS2 implementations are configured to report outputs without calculating whether that count exceeds the alert threshold for that disease in that location at that time of year.
The fix: Build alert thresholds into the surveillance system explicitly either fixed counts (three or more suspected cholera cases in a single facility in seven days) or statistical thresholds (two standard deviations above the seasonal mean). Configure the system to flag threshold crossings automatically. See DHIS2 Dashboard Best Practices for how to make these thresholds visible in real time.
Mistake 3: Zero Reporting That Is Not Enforced
A facility that does not report provides no information which looks identical to zero cases in the aggregate. A state dashboard showing low case counts may reflect genuine low transmission or 40% missing reports. I have seen programmes declare disease burden reduction when what actually happened was that reporting completeness declined.
The fix: Enforce zero reporting at facility level. Every facility should submit a weekly report even if it records zero for every indicator. Build completeness tracking into the surveillance dashboard and treat below-85% completeness as an alert condition requiring the same response as a disease threshold crossing.
Mistake 4: Investigation Capacity That Does Not Match Alert Volume
A surveillance system that detects outbreak signals faster than the health system can respond does not accelerate outbreak control it accelerates fatigue. When alert volume exceeds investigation capacity, triage happens informally; eventually alerts are acknowledged but not acted on.
The fix: Map investigation capacity before deploying or reconfiguring a surveillance system. Set alert thresholds that generate a volume the system can actually respond to. A system with 95% alert specificity generating two actionable alerts per week is more useful than one generating twenty that overwhelm the team.
Mistake 5: Laboratory Linkages That Break the Signal Chain
Specimens are collected at investigation. Transport takes days. Results take additional days or weeks. By the time confirmation arrives, the contact tracing window has closed and the data has aged out of operational relevance. The practical consequence: most confirmed cases in outbreak settings are confirmed after the response is already underway or over.
The fix: Separate the signal chain from the confirmation chain. Design the response protocol so that a cluster of epidemiologically linked suspected cases with consistent clinical presentation triggers an investigation response without waiting for laboratory confirmation. Laboratory confirmation then validates or revises the response; it does not initiate it.
Mistake 6: Surveillance Data That Is Not Used in Response Planning
Decisions about where to deploy resources and which districts to prioritise are being made on field team intuition and political considerations rather than surveillance data. The dashboard is running in another browser tab. No one opens it.
The fix: Bring the surveillance dashboard into the decision-making process institutionally. Every programme review meeting should open with five minutes of dashboard review. Response planning documents should be required to cite the specific surveillance data justifying prioritisation decisions. This is a governance change, not a technical one.
Mistake 7: No After-Action Review Mechanism
Surveillance systems that are not reviewed after outbreaks cannot improve. Most programmes conduct AARs when major outbreaks conclude; almost none conduct them when the surveillance system failed to detect an outbreak at all precisely the scenario where the most learning is available.
The fix: Build after-action review into the surveillance calendar as a scheduled, recurring activity. For every outbreak or cluster that triggered an investigation, document the full timeline: first case, alert triggered, investigation completed, response initiated. Make the latency between each step visible. Then systematically work to reduce it.
The Common Thread
Every mistake on this list reflects a version of the same problem: a surveillance system designed around data collection rather than decision support. The question a surveillance system should answer is not "how much data did we collect?" It is "how quickly did we move from the first case to a coordinated response?" For a systems-level approach to surveillance design, see Integrated Disease Surveillance and Response: A Systems Perspective.