The Three Types of Health Audience Data and Why the Differences Matter

Most health marketers are missing this. There is not one type of health audience data. There are three. And legally, they do not follow the same rules.

Examples:

• Diagnosed with diabetes
• Had knee surgery
• Filled a prescription

These are concrete health facts. They are usually the most tightly regulated, particularly under HIPAA when handled by covered entities.

Examples:

• Might have diabetes
• Probably needs surgery
• Could be trying to get pregnant

Inferences are assumptions about health based on behavior, demographics, or other signals. They are not facts, but many state privacy laws treat them as sensitive data.

Examples:

• 78% likely to have condition X
• 92% likely to need procedure Y

Predictions are scored probabilities about future health status. Like inferences, they fall under the scope of many state privacy laws and can create compliance risk even when no medical records are involved.

HIPAA: Mostly covers health facts, and only for covered entities and their business associates.

State privacy laws: Frequently regulate inferences and predictions, even when HIPAA does not apply.

This is where many audience vendors get into trouble. Once you infer or predict a person's health condition, you may be creating sensitive data, even if you never touch a medical record.

It is not just what you target.

It is how your audience was built.

Ask the right questions: When building or buying health audience segments, probe into how the audience is constructed.

Look beyond labels: The terminology matters less than the method. Transparency into audience creation is essential for compliance and trust.