Assume positive intent

About patients, about prescribers, about pharmacists, and about the agencies enforcing the rules. Almost every actor in this history was doing what their information and incentives told them to. A program that starts by identifying a villain will design instruments to catch them, and those instruments will mostly catch other people. See counting without context.

Break the cycle of shame

The single most consistent thing patients in this population report is being judged — and the consequence is that they stop telling clinicians the truth. Every clinical component of this program depends on honest answers. Shame is therefore not only cruel here; it is the mechanism by which the program would fail.

Feedback and education rather than optics

It is possible to build something that produces reassuring reports and changes nothing. The test we try to apply: does this component change a decision somebody makes? If nothing changes when a measurement changes, the measurement should not be collected. That rule is applied explicitly on remote patient monitoring and in data governance.

Publish the weak evidence

Where the evidence for a component is thin, absent or contested, it goes on the site — at the same prominence as the rest. What is not established is a page, not a footnote, and it includes figures from our own founding material that we corrected.

Never make treatment conditional on the program

Declining biometric enrollment, a site visit, monitoring or screening does not remove anyone from care and does not reduce anyone’s treatment. A consent that cannot be refused is not a consent, and a program that punishes refusal has become the thing it was built to prevent.

Never tell an agency what its own legal status is

Whether HIPAA, 42 CFR Part 2, state confidentiality law, corrections rules or public records law apply to a given arrangement is a legal determination about that organization. It varies, it is consequential, and it is not ours to make. We describe what the program does in enough detail for counsel to decide. See data governance.

Identify outliers instead of marginalizing them

A patient whose dose does not match the arithmetic is a person to be understood, not a number to be corrected. This is the whole content of metabolic determination, and it generalizes.