Public-health clinic streams include body-temperature readings that can be impossible. What should quality controls do?

Public-health clinic streams can emit body-temperature readings that are physically impossible. Quality controls should range-check temperatures and quarantine values outside clinically plausible bounds before analytics or alerting consume them. Range validation protects outbreak dashboards and triage models from sensor faults, unit errors, and transcription spikes without discarding the entire clinic visit when other fields remain valid.

Standardizing temperature scales so Fahrenheit readings convert before analytics fixes a consistency problem, not a plausibility one—a reading can convert perfectly and still be impossible, so standardization never catches or quarantines an out-of-range value. Deduplicating body-temperature readings so repeated transmissions collapse before analytics fixes duplication, not plausibility—a lone impossible value with no duplicate still flows through, so deduplication never catches or quarantines an out-of-range value. Profiling temperature distributions so drift triggers recalibration before analytics catches slow, gradual bias over many readings; it is not built to quarantine one impossible reading the moment it arrives, so an out-of-range value can still reach analytics before any drift is visible.

Exam caveat: Range checks are complementary to type checks. Quarantine or correct at the reading grain; do not "fix" outliers by blending them into other facilities. Define bounds with clinical SMEs (and unit—C vs F).

Operational check: Inject a reading such as an impossible 200° value, confirm quarantine, and verify gold clinic metrics exclude it. Pass normal temperatures and confirm they flow. Ensure text-label conversion is not used to bypass numeric range rules on the production stream.