The structural problem

Every count-based instrument in this field — a payer edit, a pharmacy corporate threshold, a prescriber-outlier report — operates on the same data: how many prescriptions, for how many days, at what converted dose, from which prescriber. That data is genuinely useful. It is also missing the two facts that determine whether any given number is appropriate: what is wrong with the patient, and what the clinician is trying to achieve.

Without those, a high number has at least four completely different explanations, and the instrument cannot tell them apart:

1. A clinician selling prescriptions.
2. A specialist to whom every complex, high-need patient in a region is referred — which is what a referral system is for.
3. A clinician managing cancer pain, palliative care or end-of-life care, populations that were never in scope for the chronic-pain guidance in the first place.
4. A clinician whose patients are pharmacokinetic outliers, needing doses that look wrong on a conversion table and are correct for them.

Only the first is the thing anyone wants to find. Numbers two through four are what a functioning system looks like. An instrument that flags all four identically has a false-positive problem whose absolute size scales with the population — which is why step one matters.

Base rates are why this is worse than it sounds

Consider the best available estimate of concentrated, potentially diversionary prescription acquisition. Analyzing 146.1 million opioid prescriptions dispensed in 2008 across pharmacies representing 76% of US retail volume, researchers identified an extreme outlier group of 0.7% of purchasers, averaging 32 prescriptions from 10 prescribers, accounting for 1.9% of prescriptions and about 4% of the weighed amount. The authors were careful about what that does and does not establish, and we quote them directly:

Very few of these patients can be classified with certainty as diverting drugs for nonmedical purposes.

McDonald & Carlson, PLOS ONE, 2013

When the behavior you are screening for is that rare, even a highly specific test produces mostly false positives in absolute terms. That is not a criticism of the analysts; it is arithmetic that applies to any screening instrument with a low base rate. The remedy is not a better threshold. It is a second source of information that resolves the ambiguity — which is exactly what clinical context is.

What the guideline authors themselves now say

This is not a fringe position. The 2022 CDC clinical practice guideline explicitly addressed how its predecessor had been used, listing among the misapplications: extension to populations it did not cover, such as cancer and palliative care patients; rapid tapers and abrupt discontinuation without collaboration with patients; rigid application of dosage thresholds; duration limits imposed by insurers and pharmacies; and patient dismissal and abandonment.

It also stated directly that the guideline “should not be applied as inflexible standards of care across patient populations by health care professionals; health systems; pharmacies; third-party payers; or state, local, or federal organizations or entities.” More on that on the prescribing guidelines page.

What we are not claiming

We are not claiming that any particular agency’s analytics are badly designed, that enforcement targeting is improper, or that any specific case was wrongly brought. We have no basis for any of those claims and do not make them. The argument is narrower and, we think, harder to dispute: an instrument that lacks diagnosis and treatment intent cannot supply them, and conclusions drawn from it inherit that gap. The fix is to add the missing information, not to stop looking.

That is the whole reason this program exists. See ProviderSynch for the mechanism and for agencies for how it is deployed.

Questions this raises

Are you saying prescription monitoring should stop?

No. Monitoring produces real and useful information, and we support it — the evidence for and against prescription drug monitoring programs is set out honestly on our PDMP page. The argument is that count data alone is an incomplete basis for a conclusion about a clinician or a patient, and that the missing half can be supplied. See what a prescribing limit actually does.

Why does a low base rate make screening harder rather than easier?

Because when the thing you are looking for is rare, most of the people a test flags will not have it, even if the test is accurate. With an outlier group of under 1%, a test with 95% specificity still produces several false positives for every true one. This is the same arithmetic that governs any rare-disease screening program. Worked through in base rates and outlier flags.