Opioid Abuse
Conditions
Keywords
opioid use disorder, randomized control trial, behavioral economics, nudges
Brief summary
The opioid epidemic is the largest man-made public health crisis the United States has faced. The objective of Trial 2 of the Application of Economics & Social psychology to improve Opioid Prescribing Safety (AESOPS-2) study, is to discourage unnecessary opioid prescribing by increasing the salience of negative patient outcomes associated with opioid use.
Detailed description
In AESOPS-2, a multi-site study, random assignment determines if prescribers to persons who suffer an opioid overdose (fatal or nonfatal) learn of this event (intervention) or practice usual-care (control). The AESOPS-2 trial will take place in 3 diverse health systems in the U.S. - Northwestern Medicine, AltaMed Health Services, and The Children's Clinic. At Northwestern Medicine, clinicians in the intervention group receive a letter notifying them of their patient's fatal or nonfatal ED overdose. At AltaMed Health Services, and The Children's Clinic, clinicians in the intervention group receive a letter notifying them of their patient's nonfatal ED overdose. The primary outcome is the change in clinician weekly milligram morphine equivalent (MME) dose prescribed in 6-month periods before and after receiving the letter. The secondary outcome is the change in the proportion of patients prescribed at least 50 daily MME. Group differences in these outcomes will be compared using an intent-to-treat difference-in-differences framework with a mixed-effects regression model to estimate clinician MME weekly dose. The AESOPS-2 trial will provide new knowledge about whether increasing prescribers' awareness of patients' opioid-related overdoses leads to a reduction in opioid prescribing. Additionally, this trial may better inform how to reduce opioid use disorder and opioid overdoses by lowering unnecessary population exposure to these drugs.
Interventions
BEHAVIORALOverdose Notification
We will identify overdoses from state vital records and insurance claims data linked to emergency departments. We will use electronic health record data to identify prescriptions of scheduled drug to patients who experienced a non-fatal or fatal overdose within the health system. If randomized to the overdose notification group, physicians who prescribed the controlled substances to the deceased or surviving patient in the year prior to their overdose will be informed of the overdose via letter. The letters will alert prescribers to the patient's opioid-related overdose, recommend the use of the state-level PDMP, and list evidence-based interventions to lower opioid-related overdoses. The letters will increase the salience and availability of opioid-related harms, which may cause clinicians to be more wary of a future overdose when prescribing opioids, benzodiazepines, muscle relaxants, or sedative-hypnotics.
Sponsors
Northwestern University
AltaMed Health Services Corporation
National Institute on Aging (NIA)
University of Southern California
Study design
Allocation
RANDOMIZED
Intervention model
PARALLEL
Primary purpose
HEALTH_SERVICES_RESEARCH
Masking
SINGLE (Caregiver)
Eligibility
Sex/Gender
ALL
Age
18 Years to No maximum
Healthy volunteers
No
Inclusion criteria
* 1\) The clinician prescribed a qualifying scheduled drug to a patient in the 12 months prior to their non-fatal or fatal overdose 2) the patient is 18 years old or older at the time of the overdose, 3) the provider practices within a health system enrolled in the study, and 4) the overdose occurs during the 12-month observation period. Qualifying prescriptions include those for one of the following scheduled drugs: opioids, benzodiazepines, muscle relaxants or sedative-hypnotics.
Exclusion criteria
* Prescriptions to patients in hospice or with active cancer
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Change in Daily Average Number of 5 MME Pill Counts | 13 months (6 months pre-intervention, 30-day washout period, 6 months post-intervention) | The primary outcome is the change in daily average number of 5 morphine milligram equivalent (MME) pill counts ordered by clinicians during the last week of baseline (week 26) and the last week of the intervention period (week 53) letter intervention. |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| Change in Clinician-level, Pre-to-post Mean Proportion of High Dose (=> 50 MME) Patient Visits | 13 months (6 months pre-intervention, 30-day washout period, 6 months post-intervention) | The change in the clinician-level, pre-to-post mean proportion of high dose (=\> 50 MME) patient visits by study arm. The numerator is whether a patient visit was =\> 50 MME, and the denominator is the number of opioids. We quantified this measure using logistic regression. Higher dose prescriptions are not recommended and are associated with the potential for greater patient harm. |
Countries
United States
Participant flow
Recruitment details
Prescriber did or did not receive the letter based on the randomization assignment (intervention or control) of the overdose victim. Clinician prescribing data was collected during the study timeframe.
Pre-assignment details
Patients who experienced an opioid overdose were randomized to either the letter intervention (fatal or non fatal overdose) or control; however, patients were not considered enrolled in the study. We analyze how the prescribing behavior of clinicians who had prescribed opioids to these patients changed pre-to-post intervention.
Participants by arm
| Arm | Count |
|---|---|
| Non Fatal Control Group Physicians in the non fatal control group will receive no notification of their patient's nonfatal overdose. | 19 |
| Non Fatal Overdose Notification Group Non Fatal Overdose Notification: We will identify overdoses from insurance claims data linked to emergency departments. We will use electronic health record data to identify prescriptions of scheduled drug to patients who experienced a non-fatal overdose within the health system. If randomized to the overdose notification group, physicians who prescribed the controlled substances to the deceased or surviving patient in the year prior to their overdose will be informed of the overdose via letter. The letters will alert prescribers to the patient's opioid-related overdose, recommend the use of the state-level PDMP, and list evidence-based interventions to lower opioid-related overdoses. The letters will increase the salience and availability of opioid-related harms, which may cause clinicians to be more wary of a future overdose when prescribing opioids, benzodiazepines, muscle relaxants, or sedative-hypnotics. | 20 |
| Fatal Control Group Physicians in the fatal control group will receive no notification of their patient's fatal overdose. | 11 |
| Fatal Overdose Notification Group Fatal Overdose Notification: We will identify overdoses from state vital records. We will use electronic health record data to identify prescriptions of scheduled drug to patients who experienced a fatal overdose within the health system. If randomized to the overdose notification group, physicians who prescribed the controlled substances to the deceased or surviving patient in the year prior to their overdose will be informed of the overdose via letter. The letters will alert prescribers to the patient's opioid-related overdose, recommend the use of the state-level PDMP, and list evidence-based interventions to lower opioid-related overdoses. The letters will increase the salience and availability of opioid-related harms, which may cause clinicians to be more wary of a future overdose when prescribing opioids, benzodiazepines, muscle relaxants, or sedative-hypnotics. | 11 |
| Total | 61 |
Withdrawals & dropouts
| Period | Reason | FG000 | FG001 | FG002 | FG003 |
|---|---|---|---|---|---|
| Overall Study | Lost to Follow-up | 0 | 1 | 0 | 0 |
Baseline characteristics
| Characteristic | Non Fatal Overdose Notification Group | Fatal Control Group | Fatal Overdose Notification Group | Total | Non Fatal Control Group |
|---|---|---|---|---|---|
| Age, Customized Age | 0 Participants | 0 Participants | 0 Participants | 0 Participants | 0 Participants |
| Race and Ethnicity Not Collected | — | — | — | 0 Participants | — |
| Region of Enrollment United States | 20 Participants | 11 Participants | 11 Participants | 61 Participants | 19 Participants |
| Sex/Gender, Customized Female | 11 Participants | 5 Participants | 2 Participants | 27 Participants | 9 Participants |
| Sex/Gender, Customized Male | 9 Participants | 6 Participants | 9 Participants | 32 Participants | 8 Participants |
| Sex/Gender, Customized Missing | 0 Participants | 0 Participants | 0 Participants | 2 Participants | 2 Participants |
Adverse events
| Event type | EG000 affected / at risk | EG001 affected / at risk | EG002 affected / at risk | EG003 affected / at risk |
|---|---|---|---|---|
| deaths Total, all-cause mortality | 0 / 861 | 0 / 973 | 0 / 524 | 0 / 801 |
| other Total, other adverse events | 0 / 861 | 0 / 973 | 0 / 524 | 0 / 801 |
| serious Total, serious adverse events | 1 / 861 | 1 / 973 | 1 / 524 | 0 / 801 |
Outcome results
Primary
Change in Daily Average Number of 5 MME Pill Counts
The primary outcome is the change in daily average number of 5 morphine milligram equivalent (MME) pill counts ordered by clinicians during the last week of baseline (week 26) and the last week of the intervention period (week 53) letter intervention.
Time frame: 13 months (6 months pre-intervention, 30-day washout period, 6 months post-intervention)
Population: Participating clinicians from Northwestern Medicine and AltaMed who prescribed an opioid during the study period
| Arm | Measure | Value (MEAN) |
|---|---|---|
| Non Fatal Control Group | Change in Daily Average Number of 5 MME Pill Counts | -3.73 Difference in estimated mean pill counts |
| Non Fatal Overdose Notification Group | Change in Daily Average Number of 5 MME Pill Counts | 3.89 Difference in estimated mean pill counts |
| Fatal Control Group | Change in Daily Average Number of 5 MME Pill Counts | 9.8 Difference in estimated mean pill counts |
| Fatal Overdose Notification Group | Change in Daily Average Number of 5 MME Pill Counts | -4.85 Difference in estimated mean pill counts |
Comparison: Null hypothesis is the difference in difference in estimated mean pill counts from the last week of baseline (week 26) to the last week of the intervention period (week 53) between control and non-fatal is 0.p-value: 095% CI: [2.4, 13.59]Zero-inflated Poisson mixed regression
Comparison: Null hypothesis is the difference in difference in estimated means from the study start (week 0) to the study end (week 23) between control and fatal is 0.p-value: 095% CI: [-25.14, -6.97]Zero-inflated Poisson mixed regression
Secondary
Change in Clinician-level, Pre-to-post Mean Proportion of High Dose (=> 50 MME) Patient Visits
The change in the clinician-level, pre-to-post mean proportion of high dose (=\> 50 MME) patient visits by study arm. The numerator is whether a patient visit was =\> 50 MME, and the denominator is the number of opioids. We quantified this measure using logistic regression. Higher dose prescriptions are not recommended and are associated with the potential for greater patient harm.
Time frame: 13 months (6 months pre-intervention, 30-day washout period, 6 months post-intervention)
| Arm | Measure | Value (MEAN) |
|---|---|---|
| Non Fatal Control Group | Change in Clinician-level, Pre-to-post Mean Proportion of High Dose (=> 50 MME) Patient Visits | 0.004 proportion of high dose patient visits |
| Non Fatal Overdose Notification Group | Change in Clinician-level, Pre-to-post Mean Proportion of High Dose (=> 50 MME) Patient Visits | 0.014 proportion of high dose patient visits |
| Fatal Control Group | Change in Clinician-level, Pre-to-post Mean Proportion of High Dose (=> 50 MME) Patient Visits | 0.01 proportion of high dose patient visits |
| Fatal Overdose Notification Group | Change in Clinician-level, Pre-to-post Mean Proportion of High Dose (=> 50 MME) Patient Visits | -0.09 proportion of high dose patient visits |
p-value: 0.21995% CI: [-0.43, 1.86]Regression, Logistic
p-value: 0.22995% CI: [-1.9, 0.46]Regression, Logistic