A Community Health Center Buyback Program to Reduce the Supply of Opioids to Secondary Users

We seek to provide proof of concept for an opioid buy-back program that could reduce the flow of opioids from patients to secondary users and the black market. To study the effectiveness of the intervention, we will randomly assign patients to receive information about the buyback program when the prescription is filled including details on the financial incentive and reminder text messages. We will assess the direct impacts of the program based on the number of patients that take up the buyback and the total quantity of opioid medications that are returned. We will also explore how the effect varies across different groups of patients (e.g. age range, gender, English proficiency, underlying reason for prescription) as well as whether there are any program spillovers to patients that not served by the community health center.

Implementing Partner: A community health care center in Massachusetts with five in-house pharmacy locations in the same city will implement the program. Serving a culturally diverse population, the community health center is responsive to emerging community needs and has developed programs and services designed to improve access to care as well as improve the quality of life for patients.

Target Population: The intervention will target patients prescribed an opioid for an acute diagnosis such as surgery, injury, or acute pain at all five of the community health center’s in-house pharmacy sites. Acute users will be identified as having an opioid prescription for a supply of 14 days or less.

Treatment Arms: There will be two main treatment arms that will be implemented across different pharmacy locations. At the main location, patients will be randomly assigned to receiving either no information or information on how to return unused medication plus the financial incentive. At the other four locations, patients will be randomly assigned to receiving either no information or information on how to return unused medication (e.g., no financial incentive). One of the larger sites in the second treatment arm, will also receive a text notification at the end of the month to remind them about the disposal program when their prescription is likely to be completed.

Randomization: To minimize contamination, patients will be randomized into the treatment and control groups within sites but across calendar days. The randomization will occur ex-ante and each week the staff will be provided with a schedule to indicate which days are designated for patient assignment to the intervention group and instructed to apply the intervention on only those days that week. Note that patients receiving an opioid prescription at locations other than the main site will only be informed on weekdays since those locations are not open on the weekends.

IRB Approval: The research team has obtained IRB approval from Northeastern University. All data provided by the pharmacy will be de-identified with all personally identifiable information (PII) such as first and last name, date of birth, and Rx number and date removed before the data is transmitted to the research team.

2019-05-06

2020-05-05

Our primary outcome of interest is whether individuals with an opioid prescription for acute who receive the intervention use are more likely to return unused medication to the pharmacy compared to those who do not receive the intervention. This is measured relative to the passive observation of the disposal kiosk (control), which is the baseline outcome for prior medication disposal programs. We will also explore heterogeneity of outcomes across different groups according to patient characteristics that predict opioid addiction such as age, gender, and diagnosis (e.g., injury versus surgery). We will also test which program features (e.g., text reminders, financial incentive) help to increase the increase take-up rate by comparing impacts across treatment arms. Finally, we will measure the amount of spillover to other patients and community residents who do not directly receive the intervention.

Secondary outcomes that we will measure include the type and estimated quantity of medication returned across the treatment and control groups based on the visual verification conducted by the pharmacists. The type of medication returned can reveal which opioids are more likely to be returned. The amount of medication returned can reveal the degree to which the program may have diverted consumption of opioids from either a primary or secondary user. We will compare the amount returned to the amount prescribed by type of medication in several ways. First, based on the visual verification and estimation conducted by the pharmacists we will compare the types and amount of medication returned by the treatment and control groups. Second, in cases where it is possible to match the Rx or patient name of those returning medication to those who were prescribed medication, we will directly compare the amount returned to the amount prescribed. Third, by comparing the total amount disposed of (e.g., pounds) based on reports from the disposal vendor to the total amount prescribed as recorded by the pharmacy, we will assess the aggregate amount returned versus prescribed across different treatment arms.

We seek to provide proof of concept for an opioid buy-back program that could reduce the flow of opioids from patients to secondary users and the black market. To study the effectiveness of the intervention, we will randomly assign patients to receive information about the buyback program when the prescription is filled including details on the financial incentive and reminder text messages. We will assess the direct impacts of the program based on the number of patients that take up the buyback and the total quantity of opioid medications that are returned. We will also explore how the effect varies across different groups of patients (e.g. age range, gender, English proficiency, underlying reason for prescription) as well as whether there are any program spillovers to patients that not served by the community health center.

Implementing Partner: A community health care center in Massachusetts with five in-house pharmacy locations in the same city will implement the program. Serving a culturally diverse population, the community health center is responsive to emerging community needs and has developed programs and services designed to improve access to care as well as improve the quality of life for patients.

Target Population: The intervention will target patients prescribed an opioid for an acute diagnosis such as surgery, injury, or acute pain at all five of the community health center’s in-house pharmacy sites. Acute users will be identified as having an opioid prescription for a supply of 14 days or less.

Treatment Arms: There will be two main treatment arms that will be implemented across different pharmacy locations. At the main location, patients will be randomly assigned to receiving either no information or information on how to return unused medication plus the financial incentive. At the other four locations, patients will be randomly assigned to receiving either no information or information on how to return unused medication (e.g., no financial incentive). One of the larger sites in the second treatment arm, will also receive a text notification at the end of the month to remind them about the disposal program when their prescription is likely to be completed.

Randomization: To minimize contamination, patients will be randomized into the treatment and control groups within sites but across calendar days. The randomization will occur ex-ante and each week the staff will be provided with a schedule to indicate which days are designated for patient assignment to the intervention group and instructed to apply the intervention on only those days that week. Note that patients receiving an opioid prescription at locations other than the main site will only be informed on weekdays since those locations are not open on the weekends.

IRB Approval: The research team has obtained IRB approval from Northeastern University. All data provided by the pharmacy will be de-identified with all personally identifiable information (PII) such as first and last name, date of birth, and Rx number and date removed before the data is transmitted to the research team.

Intervention Protocol
Here we describe in detail the steps taken at each point in the intervention including an overview of the notification, return, and data matching process.

Identification of Treatment Day: Pharmacists will check the posted schedule for the week to see which days are treatment days. Notification cards will be made available in plain sight for pharmacists on treatment days as an additional visual reminder. A pop-up will be placed in the pharmacy calendar to alert pharmacists when they first log in to their computer that it is a treatment day.

Identification of Treatment Group: Patients receiving an opioid prescription for 14 or fewer days will be assigned to the treatment group based on the day that their prescription is filled by the pharmacist. Such patients will be identified when the medication is placed in the queue for pick-up, and a notification postcard will be stapled to the bag along with a “See Pharmacist” stamp.

Initial Notification: Notification of the incentive buy-back program will take place during pick-up as a pharmacist consultation. The pharmacist will verbally explain the program using a pre-developed script and provide the patient with the notification postcard as a reminder. Both the script and the written notification postcard will be developed in collaboration with pharmacy staff to ensure the appropriate literacy level, language (e.g. English and Spanish), and cultural considerations. We will also consult a behavioralist specializing in patient interventions to ensure that the script and the notification card appropriately educate the patient about the potential for addiction and the need to return any unused medication. The notification card will include the patient name and Rx number and will instruct the patient to return any unused medication in the original bottle.

Text Reminder: At one of the larger information-only locations, the pharmacy staff will send one text reminder at the end of the month to remind individuals in the treatment group to return any unused medication.

Receipt and Verification: When medication is returned to the pharmacy, a pharmacist will visually verify that the medication is an opioid and visually estimate the number of pills. Under no circumstances will the pharmacist ever touch the medication. An image chart will be posted in the pharmacy verification area (“the bench”) to help the pharmacist visually verify which medication is being returned. The pharmacist will record the patient’s name, the date, the Rx number (if possible), the type of medication returned, the estimated amount, and how the patient heard about the program (e.g., notified by pharmacist or word-of-mouth).

Opioid medications from other pharmacies will also be accepted to determine spillovers to other patients in the community who are not served by the community health center. Other medications brought to the pharmacy that are not opioids will be directed to a collection site in the community. If a patient insists a medication is an opioid that the pharmacist cannot verify, then the medication will be treated as if it were an opioid but the pharmacist will note that it cannot be verified as such.

Disposal of medication into the kiosk by the patient: Once the medication has been visually verified by the pharmacist, the patient will dispose of the medication in a kiosk in compliance with DEA and Massachusetts Board of Pharmacy regulations. In accordance with DEA regulations, the disposal kiosk will be placed in the pharmacy waiting area under video surveillance.

Disposal of medication from the kiosk: A DEA-approved vendor who will install the kiosks in accordance with DEA regulations and handle the disposal of the medication. The units will be shipped to each location with the bolts to have a maintenance team install them to the floor. When disposing of the medication from the kiosk, the unit will be opened with two members of the pharmacy per DEA regulations and a security guard at all times. The two pharmacy members would open the unit together, immediately pull the self-adhesive tape and seal the liner.. The box will remain in the kiosk and will be locked until the common carrier picks it up. After locking the kiosk, the common carrier will be contacted immediately to schedule a pick up within 24 hours. Once the carrier drops off the filled liner to the vendor, they take possession of the liner. The vendor will weigh the box containing the liner with at minimum two employees present and scan the tracking label to upload the weights into the reporting site. The box is then moved into the secure cage area waiting for the DEA representatives’ arrival for transport with minimum of two vender employees. From there the box is transported to the DEA approved controlled clean energy burn site. At the site, the tracking number is scanned, and the liner is weighed again. The liner is then burned in the presence of both the DEA officials and vendor representatives. A destruction manifest is then signed and uploaded to the web portal.

Financial Incentive (main location only): The pharmacist will dispense the gift card and the individual returning the medication will print and sign their name to acknowledge that they have received the gift card. We propose a $10 gift card for any amount of medication returned as sufficient to compensate individuals for their time but not so large as to encourage adverse behavior.

Notification Tracking: For internal purposes only, pharmacy staff will run a report of all opioid prescriptions for 14 days or less that will be placed in an excel spreadsheet each month. The spreadsheet will contain one row for each prescription that provides information on patient first and last name, date or birth, Rx number and date, type of medication and amount prescribed, and other patient characteristics (e.g., gender, English proficiency). At a later date, a pharmacy researcher will merge in data on diagnosis associated with the prescription (e.g. surgery, injury). All data will be de-identified by removing personally identifying information (e.g., name, date of birth, Rx number and date) before sending to the Northeastern University research team.

Return Tracking: Pharmacy staff will use a paper system to track which patients have returned medication. This information will include the patient’s first and last name, date of birth, and Rx number (if available from the notification card or prescription bottle). The pharmacist will also record the type and estimated quantity of opioids collected for disposal when visually verifying the medication as well as the date the medication was returned. Pharmacy staff will also record how the patient heard about the program to determine whether the patient was given the information directly or if they heard it through another source (e.g., word of mouth). This data will then be merged into the notification data using Rx number or first and last name. All data will be de-identified by removing personally identifying information (e.g., name, date of birth, Rx number and date) before sending to the Northeastern University research team.

Data Matching: The notification and return data will be matched to determine if the patient who returned medication was in the treatment group, the control group, or outside GLFHC. Data will be matched by Rx number or the combination of patient name and/or date of birth. This will allow us to measure any spillovers to those who were not initially informed as well as the amount returned relative to the amount prescribed.

Transmission of Data to Northeastern University: Pharmacy staff will remove any personally identifiable information (PII) from the excel spreadsheet that is shared with the researchers including first and last name, date of birth, Rx number, and date filled. To assess the impact of the program on different types of patients, patient characteristics will be provided at the group level (e.g., males age 25-34 years old) as long as there are at least five individuals in that group to prevent identification. To assess any seasonal variation as well as the spread of the program, the month that the prescription was filled and the month that it was returned will be provided. The spreadsheet will then be password protected and sent via email with the password sent separately. Data will be shared on a monthly basis with the researchers to monitor data robustness and identify emerging trends.

Pharmacy Staff Training: Training materials will be developed and used to educate and monitor pharmacy staff regarding the notification of patients, no handling of returned opioid medication, proper disposal or medication, and secure tracking of information. The training module will be delivered online, including a short quiz on the material, and an online verification when the pharmacists completes the training by the pharmacist that they have completed the training. Pharmacists will be required to take the training twice during the intervention period at six-month intervals to ensure compliance with the research protocol and DEA regulations. A one-page description of the return protocol will be posted in the pharmacy bench area to remind pharmacists of the steps for returning medication. Similarly, a one-page description of the notification protocol will be posted in the pharmacy bench area to remind pharmacists of script for notifying patients about the program.

Follow-up: At the six-month mark, the research team will conduct three focus groups with patients to uncover any unforeseen barriers or unintended consequences that could inform future programming.The three groups be comprised of individuals who were informed and returned medication, individuals who were informed and did not return medication, and individuals who were not informed but returned medication anyway.

Randomization: To minimize contamination, patients will be randomized into the treatment and control groups within sites but across calendar days. The randomization will occur ex-ante and each week the staff will be provided with a schedule to indicate which days are designated for patient assignment to the intervention group and instructed to apply the intervention on only those days that week. Note that patients receiving an opioid prescription at locations other than the main site will only be informed on weekdays since those locations are not open on the weekends.

Randomization will be at the individual level within clusters.

Yes

5 pharmacy locations

Based on a sample of prescriptions over a three-month period, the community health center projects that they fill about 2,400 opioid prescriptions for acute users over a 12-month period across their five locations. The largest location, fills half of this amount with the remaining half divided between their next largest location, and the other three locations.

We make several assumptions to arrive at an estimated take-up rate for each of the treatment and control groups. First, even with recent restrictions on prescribing opioids for acute users, most experts agree that patients still end up with excess medication. As such, we assume that all acute users will have at least some minimal amount (e.g., one pill) that could be returned. Second, we assume the potential take-up rate will be similar to that found in a 2014 survey of beliefs and behaviors regarding unused and expired medication among community pharmacy patients. The survey found that 15 percent had utilized a take-back disposal location in their community, although 77 percent of patients were willing to return unused medication when asked about it (Kozak et al. 2015). As such, we use a range of take-up rates across each group: 10-15 percent for the control group, 30-40 percent for the information-only treatment group, and 50-75 percent for the incentivized treatment group. Finally, we assume spillovers to individuals outside the pharmacy would be minimal with one patient per week per location.

A back-of-the-envelope power calculation using the projected sample size and anticipated program impacts based on estimates from the literature indicates that the study is sufficiently powered. Assuming a power threshold of 80 percent, we show the MDE needed given our projected sample sizes across the two arms. Given a control mean of 15 percent, an MDE of 6.5 percentage points would be required to demonstrate a significant program impact at the 5 percent level for the largest site which has 1,200 patients. For the smaller sites with only 600 patients, the MDE is 9.0 percentage points. Given that prior surveys indicate upwards of 77 percent of individuals would be willing to return medication even without a financial incentive, achieving a treatment mean of 21.5 to 24.0 percent seems likely. In addition, we estimate the total sample size required to achieve a policy relevant MDE that corresponds to doubling the percent of individuals returning medication relative to the control group. The required sample size ranges from 236 to 394, about one-half to two-thirds the sample size projected at our smallest site. We believe effect sizes of this magnitude would be of interest to state policymakers deciding whether to expand the program.