The opioid analysis published in MMWR on March 17, 2017 reported 1,294,247 eligible patients, including 33,548 who continued therapy for at least a year: a raw share of 2.6 percent. It also reported an estimated one-year continuation probability of 6.0 percent. The two numbers should not be presented as rival calculations of the same simple fraction. [mm6610a1]
University of Arkansas researchers used a Kaplan-Meier time-to-event method, allowing for observation ending before discontinuation. Follow-up ended with loss of enrollment, the study's end or discontinuation, defined as at least 180 opioid-free days. The selected US adults were opioid-naive and cancer-free. These design features explain why the reported estimated probability serves a different purpose from dividing the observed count by the full sample. [mm6610a1]
Continuation also was not a diagnosis of addiction or proof that an extra prescription caused long-term use. The records lacked pain intensity and duration, and pain's cause was not considered in the analysis. An initial episode could span multiple prescriptions separated by no more than 30 days. Calling episode duration the length of one prescription would change what the researchers measured. [mm6610a1]
The linked 2016 CDC guideline contributed historical recommendations, not another replication of these statistics. It was voluntary guidance for adult primary-care chronic pain outside active cancer treatment, palliative care and end-of-life care. Reading the two together separates observation from clinical guidance: neither the sample share nor the estimated probability is a personal forecast, and this retrospective supplies no current prescribing advice. [mm6610a1][rr6501e1]
Source documents
- Characteristics of Initial Prescription Episodes and Likelihood of Long-Term Opioid Use — United States, 2006–2015https://www.cdc.gov/mmwr/volumes/66/wr/mm6610a1.htm
- CDC Guideline for Prescribing Opioids for Chronic Pain — United States, 2016https://www.cdc.gov/mmwr/volumes/65/rr/rr6501e1.htm