Most organizations bury this. We would rather it were one page an agency can read in five minutes, because every item here is something you would otherwise find out later, in a worse setting.

Claims we do not make

• That positive identification prevents a large share of diversion. There is no US epidemiological estimate of the share of diversion attributable to prescription forgery or false-identity dispensing. It is not separated out in the national survey, it is bundled into an unmeasured residual in the pooled peer-reviewed analysis, and the Government Accountability Office names forgery as a mechanism without assigning it a magnitude. See the diversion evidence.
• That urine drug testing mitigates risk. The AHRQ 2020 review found that no study evaluated it. See urine drug testing.
• That risk assessment instruments reduce overdose, addiction, abuse or misuse. Same AHRQ finding, same sentence. This covers opioid management plans, patient education, monitoring instruments, more frequent monitoring intervals, pill counts and abuse-deterrent formulations as well.
• That pharmacogenomic testing improves pain outcomes. The randomized evidence is against it. See pharmacogenomics.
• That prescription drug monitoring programs reduce overdose death. Contested, with documented substitution effects. See PDMPs.
• That the 2016 CDC guideline caused the rise in overdose deaths. Population trends cannot establish that, in either direction. See step four.

Figures we found and deliberately did not publish

• “DEA quotas have fallen 65% for oxycodone and 73% for hydrocodone since 2013.” Widely repeated. We could not verify it in any Federal Register order or DEA document, and it appears only in advocacy and trade press. What is documented — DEA’s own record of shortage complaints in its 2025 quota order — is stronger anyway.
• “98 of the top 100 opioid prescribers in the country practiced in Florida.” The published figure is narrower: 98 of the 100 US physicians who dispensed the highest quantities of oxycodone directly from their offices. We use the accurate version. See step two.
• “Chronic pain incidence is 60 in 1,000.” The published estimate is 52.4 per 1,000 person-years; 60 is the upper bound of the confidence interval. Our own earlier material carried the wrong figure. See step one.
• “Patients prescribed opioids are 8 times more likely to start injection drug use.” A figure in our own founding material. The study it appears to derive from reports a 3.2-fold higher risk of hepatitis C seroconversion — a proxy for injection initiation — in a cohort of 382,478 people, not 8-fold injection initiation in 59,804. We do not publish it, and the paper’s own authors state their findings “should not be used to justify abrupt discontinuation of long-term therapy.”
• Pharmacy refusal rates for analgesic opioid prescriptions. Figures exist for buprenorphine — 62% of surveyed North Carolina community pharmacists had refused a buprenorphine prescription — and it would be wrong to extrapolate them to pain patients. We do not.
• Any dollar figure. IntellaRx does not publish pricing, and does not publish cost-of-illness figures on program pages where they would read as one.

Terminology we use carefully

• “Pain refugee.” Journalistic, not clinical. It returns no results in the medical literature. We use it in quotation marks with attribution, or say what we mean.
• Fentanyl and carfentanil potency ratios. The published estimates — fentanyl 50 to 100 times morphine, carfentanil roughly 100 times fentanyl or 10,000 times morphine — are hedged by the CDC and DEA sources that state them, because analog potency has not been evaluated in humans. We reproduce the hedge and we do not blend figures from different sources into a single number.
• “Deaths of despair.” The mortality trends are not disputed; the despair mechanism is. Estimates of the economic contribution range from near zero to about a third. See social determinants.