Examples of bad clinical data

Realistic examples showing why clinical data must be checked before it is used in analytics or automation.

Impossible date of birth

BLOCKED

Use blocked

Raw data

{
  "first_name": "Giulia",
  "last_name": "Bianchi",
  "date_of_birth": "2099-01-01",
  "email": "giulia@@mail.com"
}

Why this is a problem

The date of birth is in the future and the email format is invalid. This data cannot be trusted.

System outcome

The system blocks this record and prevents it from being used downstream.

ATTENTION

Review suggested

Raw data

{
  "first_name": "Marco",
  "last_name": "Rossi",
  "date_of_birth": "1987-03-12",
  "smoker": "maybe"
}

Why this is a problem

Smoking status is present but unclear. This may affect clinical decisions.

System outcome

The data can be used, but the system flags it for human review.

Safe to use

Raw data

{
  "first_name": "Laura",
  "last_name": "Verdi",
  "date_of_birth": "1990-08-07",
  "email": "laura.verdi@email.com",
  "smoker": "no"
}

Why this is a problem

All required fields are present and values are plausible.

System outcome

The system allows this data to be used without restrictions.