Estimation of the Prevalence of Delayed Dispensing Among Opioid Prescriptions From US Surgeons and Dentists

Key Points Question How often are opioid prescriptions from US surgeons and dentists dispensed more than 30 days after writing (delayed dispensing)? Findings In this cross-sectional analysis of a national pharmacy database representing 63% of US prescriptions, 194 452 opioid prescriptions (0.9%) from surgeons and dentists in 2019 were dispensed more than 30 days after writing. The prevalence of delayed dispensing decreased after a Minnesota law was enacted to preclude dispensing of opioid prescriptions more than 30 days after writing. Meaning These findings raise concerns that opioids prescribed by surgeons and dentists may sometimes be used for reasons or during a time frame other than that intended by the prescriber.


eAppendix 2. Search Strategy for State Laws
Laws regulating maximum period between writing and dispensing of controlled substance prescriptions During winter 2021, we used NABPLAW Online to compile all state laws regulating the maximum period between writing and dispensing of controlled substance prescriptions that were effective at the time of the search. We started with NABPLAW Online as opposed to WestLaw Edge because the former only contained pharmacy laws, making the relevant laws easier to identify. We used the following search terms in NABPLAW for each of the 50 states and the District of Columbia (the example below is for Alabama): • "controlled substance" "day" "AL" • "controlled substance" "month" "AL" • "controlled substance" "year" "AL" • "Schedule II" "day" "AL"; "Schedule II" "month" "AL"; "Schedule II" "year" "AL" • "Schedule III" "day" "AL"; "Schedule III" "month" "AL"; "Schedule III" "year" "AL" • "Schedule IV" "day" "AL"; "Schedule IV" "month" "AL"; "Schedule IV" "year" "AL" • "Schedule V" "day" "AL"; "Schedule V" "month" "AL"; "Schedule V" "year" "AL" Of 180 state laws identified, we excluded laws that only applied to partial refills, laws that only applied to patients in long-term care facilities, laws that only apply to disciplinary measures for violations of regulations surrounding controlled substance prescribing regulations, and laws that were not applicable to the pharmacy setting. The 96 remaining laws all pertained to the maximum period between writing and dispensing of controlled substance prescriptions.
Although NABPLAW included the text of the current 96 laws along with the dates of amendments, it did not contain the text of the amendments. Therefore, we used Thomas Reuters Westlaw Edge to compile the text and effective dates of all amendments. By comparing the text of the law effective as of the time of the search and any amendments, we determined which laws were effective as of December 31, 2019 and whether states had passed laws regulating the maximum period between writing and dispensing of controlled substance prescriptions between 2014-2019 (for the difference-in-differences analysis).
We conducted two additional checks to ensure that we had identified all relevant laws. First, we used the same search terms as above in WestLaw Edge and obtained the same set of laws. Second, we used more general search terms to search for relevant laws for each state in WestLaw Edge: "controlled substance", "Schedule II", "Schedule III", "Schedule IV", "Schedule V." This strategy yielded the same set of laws as in our original search of NABPLAW.
Laws regulating maximum period between writing and dispensing of non-controlled substance prescriptions When states did not have specific laws regulating the maximum period between writing and dispensing of controlled substances in DEA Schedule III-IV, we set the maximum period to 6 months, the maximum allowed under the Controlled Substance Act.
When states did not have specific laws regulating the maximum period between writing and dispensing of controlled substances in DEA Schedule II or V, we set the maximum period to the period allowed for non-controlled substance prescriptions by the state. To identify state laws regulating this period for noncontrolled substance prescriptions, we used WestLaw Edge and additionally contacted state pharmacy associations. a These prescriptions met all study inclusion and exclusion criteria (e.g., they were not refills of prior prescriptions) except that they were written by any prescriber, as opposed to surgeons and dentists specifically.

eAppendix 6. Sensitivity Analyses for Difference-in-Differences Analyses
For prescription i dispensed in state s during month t, the model specification in the main analysis was: E(Yist) = β1*States + β2*Montht + β3 Minnesotaist*Postist Where Yist is an indicator that equals 1 if the prescription was dispensed more than 30 days after writing and 0 otherwise, Minnesota is an indicator that equals 1 if the prescription was dispensed to a patient residing in Minnesota and 0 if dispensed to a patient in a control state, and Post is an indicator that equals 1 if the prescription was dispensed on or after July 1, 2019 (the date on which Minnesota's law was enacted) and 0 otherwise. State is a vector of state fixed effects (accounting for time-invariant confounders at the state level) and month is a vector of yearmonth fixed effects (accounting for confounders associated with specific time periods). Models employed standard errors clustered at the state level.
Analyses excluded nine states that restricted the maximum period between writing and dispensing of prescriptions for Schedule II, III, or IV drugs to 30 days or less. Analyses also excluded Louisiana, New Mexico, and Florida, the last of which prohibited dispensing of prescriptions for Schedule II-III drugs more than 14 days after the date of a surgical procedure. The remaining 38 states were control states.
We conducted several sensitivity analyses, all of which yielded similar difference-in-differences estimates as in the main analysis (-0.22 percentage points, 95% CI: -0.32, -0.13).
• Controlled for patient age, patient sex, DEA Schedule, whether prescription was written by a surgeon or dentist, and prescription method of payment: -0.23 (-0.33, -0.14). • Used standard errors clustered at the patient level instead of state level: -0.22 (-0.25, -0.20).
• Limited to data from 2017-2019: -0.22 (-0.32, -0.13) • Excluded data from July 2019 as a wash-out period: -0.23 (-0.33, -0.14) We also constructed an alternative control group comprised of the 9 states that restricted the maximum period between writing and dispensing of prescriptions for Schedule II, III, or IV drugs to 30 days or less throughout the study period. Using the same model specification as in the primary analysis, the difference-in-differences estimate was -0.34 (-0.38, -0.29).