37th ICPE - International Society for Pharmacoepidemiology Conference Presentations

User Collection Public

This collection contains work presented at the 37th Annual ICPE - International Society for Pharmacoepidemiology Conference from collaborators at the Opioid Data Lab.

Parent Collection:

Oral/Poster Presentations, Oral Presentations, Poster Presentations

Methodological Advances in Evaluating Abuse Deterrent Opioid Analgesics

Works (7)

1. Four Competing Definitions of Morphine Equivalence Insidiously Inhibit Evidence Synthesis

Title Tesim:: Four Competing Definitions of Morphine Equivalence Insidiously Inhibit Evidence Synthesis
Creator:: Chidgey, Brooke, Delcher, Chris, Dasgupta, Nabarun, Bae, Jungjun, Wang, Yanning, Cooper, Toska, and Kinlaw, Alan
Date of publication:: August 23, 2021
Abstract Tesim:: Background: Morphine-standardized analgesic doses are used in clinical practice and research to account for molecular potency. 90 milligrams of morphine equivalents (MME) per day are considered a “high dose” risk threshold in guidelines and laws. Theoretically total MME (numerator) is divided by days supply (denominator) to calculate daily MME. Although ubiquitously cited the “CDC definition” of daily MME lacks a clearly defined denominator. Objectives: To assess the sensitivity of denominator-dependency on “high dose” classification across 4 competing definitions of daily MME. Methods: To identify definitional variants, we examined text descriptions from 27 studies cited in the 2016 CDC Pain Management Guideline, literature, and electronic prescribing tools; our review yielded 4 unique definitions. Using Prescription Drug Monitoring Programs data (Jul-Sep 2018), we conducted a population-based cohort study of 3,916,461 patients who received outpatient opioid analgesics in California (CA) and Florida (FL). The binary outcome was whether patients were deemed “high dose” (>90 MME/day), which we compared across the 4 definitions. We calculated I^2 for heterogeneity attributable solely to definition, and stratified by extended/immediate release. Results: In a simple example, a 30-day supply of 2 partially overlapping oxycodone prescriptions have daily MME varying across 4 definitions: 76mg (total days supply); 93mg (on-therapy days); 31mg (fixed observation window); or 105mg (maximum daily dose). Among 9,436,640 prescriptions, 42% overlapped, which led denominator definitions to impact daily MME values. Across definitions, average daily MME varied 3-fold [range: 17-52 (CA) and 23-65 mg (FL)]. Across the definitions, prevalence of “high dose” individuals ranged 5.9%-14.2% (FL) and 3.5%-10.3% (CA). Definitional variation would impact a hypothetical surveillance study trying to establish how much more “high dose” prescribing was present in FL than CA: from 34% to 79% more. Meta-analyses revealed strong heterogeneity (I^2 range 86%-99%) based solely on MME definition, impacting immediate- and extended-release differentially. In the sensitivity analysis, including unit interval 90.0-to-90.9 in the boundary increased the “high dose” population fraction by 15.4%. Conclusions: Previously unrecognized definitional variation confounds surveillance metrics and constrains real-world applicability of observational studies. While 90 MME may have cautionary mnemonic benefits, without harmonization of calculation, its utility is limited. Comparison between studies using daily MME requires explicit attention to definitional variation for validity.
Affiliation Label Tesim:: University of North Carolina at Chapel Hill and Injury Prevention Research Center
Type:: http://purl.org/dc/dcmitype/Text
DOI:: https://doi.org/10.17615/rr05-c370
Language Label:: English
ORCID:
Other Affiliation:: , University of Kentucky, and University of Florida
Person:: Chidgey, Brooke, Delcher, Chris, Dasgupta, Nabarun, Bae, Jungjun, Wang, Yanning, Cooper, Toska, and Kinlaw, Alan
Rights Statement Label:: In Copyright

2. Off-Label Day-of-Surgery Gabapentinoids and Prolonged Opioid Use: A Retrospective Cohort Study of Medicare Patients using Electronic Health Records

Title Tesim:: Off-Label Day-of-Surgery Gabapentinoids and Prolonged Opioid Use: A Retrospective Cohort Study of Medicare Patients using Electronic Health Records
Creator:: Pate, Virginia, Jonsson-Funk, Michele, Hudgens, Michael, Young, Jessica, Chidgey, Brooke, Dasgupta, Nabarun, and Sturmer, Til
Date of publication:: August 23, 2021
Abstract Tesim:: Background: Gabapentinoids are increasingly incorporated into multimodal analgesia protocols for surgery. However, these drugs are not approved by the US FDA for use in managing surgical pain , and little is known regarding the effects of this off-label use on prolonged post-surgical opioid use following surgery. Objectives: We estimate the association between preoperative day-of-surgery gabapentinoid administration on the risk of prolonged postsurgical opioid use. Methods: We identified older adults (65+ years) undergoing surgery using electronic health records (EHR) from a large integrated healthcare system in the southeast (2016-2019) . Exposure to preoperative gabapentinoids was measured using inpatient medication administration records on the day of surgery, and the outcome of prolonged opioid use was measured using outpatient medication order s. We used stabilized inverse probability of treatment weights with 1% asymmetric trimming to adjust for patient demographics, baseline health indicators, preoperative pain score, and surgical details. We used log-binomial regression to estimate risk ratios and 95% confidence intervals (Cl) of prolonged opioid use comparing patients who received preoperative gabapentinoids to those who did not. Results: Overall, 17,435 surgical patients met inclusion criteria, of whom 17.2% received preoperative gabapentinoids. The overall observed 90-day risk of prolonged opioid use following surgery was 1.20% (95% Cl: 1.05, 1.38) patients. The adjusted risk ratio of prolonged opioid use comparing patients who received a dose of preoperative gabapentinoids on the day of surgery to those who did not was 1.67 (95% Cl: 0.85,3.23). Conclusions: Using data from a large integrated health system, we did not find evidence that preoperative gabapentinoids reduced the risk of prolonged opioid use in patients undergoing a broad range of surgeries. Off-label use of these medications to manage surgical pain should be carefully evaluated.
Affiliation Label Tesim:: Gillings School of Global Public Health, University of North Carolina at Chapel Hill, and Injury Prevention Research Center
Type:: http://purl.org/dc/dcmitype/Text
DOI:: https://doi.org/10.17615/y3xd-0f27
Language Label:: English
ORCID:
Other Affiliation:
Person:: Pate, Virginia, Jonsson-Funk, Michele, Hudgens, Michael, Young, Jessica, Chidgey, Brooke, Dasgupta, Nabarun, and Sturmer, Til
Rights Statement Label:: In Copyright

3. Distinguishing Death from Disenrollment in Claims Data Using a Readily Implemented Machine Learning Algorithm

Title Tesim:: Distinguishing Death from Disenrollment in Claims Data Using a Readily Implemented Machine Learning Algorithm
Creator:: Shiv, Shalu, Gibson, Teresa, Dasgupta, Nabarun, Irwin, Debra, Cooper, Toska, Yoon, Frank, Young, Jessica, Bloemers, Sarah, and Pack, Kenneth
Date of publication:: August 23, 2021
Abstract Tesim:: Background: The inability to identify dates of death in insurance claims data is a major limitation to retrospective claims based research. If not an outcome, death is a competing risk and poses a threat to validity when treated as non-informative right censoring. Objectives: We aim to develop a user-friendly public algorithm to predict death within the year of disenrollment using an administrative claims database. Methods: We identified adults (18+ years) with at least 2 years of continuous enrollment prior to disenrollment between 01/2007 and 01/2018. Leveraging unique linkages in addition to data that are typically unavailable in the publicly licensed data, we ascertained date of death from the Social Security Death Index, inpatient discharge status, and death indicators in the administrative data. Models including candidate predictors for age, sex, Census region, month of disenrollment, year of disenrollment, chronic condition indicators (components of the Elixhauser score), and prior healthcare utilization were estimated using used elastic net regression tuned by 5-fold cross-validation and final models evaluated in an independent testing set. Weighted analysis adjusts for rare outcome (i.e., class imbalance). Sensitivity, specificity, and ROC associated with various thresholds of predicted probability to classify death at disenrollment were calculated. Results: Overall, we identified 13,360,460 beneficiaries who disenrolled during the study period, with 5% of patients who died within the year of disenrollment. The strongest predictors of death were age at disenrollment, diagnosis of metastatic cancer in the year prior to death, and type of care received (e.g., inpatient stay, hospice care). Using a prediction threshold of 30%, the algorithm classified death at disenrollment with a sensitivity of 0.684 and specificity of 0.985 (ROC=0.97. At the same prediction threshold, the weighted algorithm classified death with a sensitivity of .947 and a specificity of 0.898 (ROC=.973). Conclusions: Our algorithm uses publicly defined chronic conditions and utilization patterns that are easy to implement in claims data and predicts death at disenrollment with high specificity and varying sensitivity depending on the chosen prediction threshold. Users can easily implement the algorithm and can choose the prediction threshold (balancing sensitivity and specificity) to meet the needs of the specific study at hand.
Affiliation Label Tesim:: Injury Prevention Research Center and University of North Carolina at Chapel Hill
Type:: http://purl.org/dc/dcmitype/Text
DOI:: https://doi.org/10.17615/42zz-g634
Language Label:: English
ORCID:
Other Affiliation:: IBM Watson Health and
Person:: Shiv, Shalu, Gibson, Teresa, Dasgupta, Nabarun, Irwin, Debra, Cooper, Toska, Yoon, Frank, Young, Jessica, Bloemers, Sarah, and Pack, Kenneth
Rights Statement Label:: In Copyright

4. Analytical Enhancements in Kentucky’s Drug Overdose Mortality Surveillance: Rapid Monitoring of Trends and Decedents’ Recent Controlled Substance Prescription History

Title Tesim:: Analytical Enhancements in Kentucky’s Drug Overdose Mortality Surveillance: Rapid Monitoring of Trends and Decedents’ Recent Controlled Substance Prescription History
Creator:: McAninch, Jana, Slavova, Svetla, Brancato, Candace, Oh, GYeon, Secora, Alex, Dasgupta, Nabarun, Collins, Candice, and Freeman, Patricia
Date of publication:: August 23, 2021
Abstract Tesim:: Background: Timely drug overdose mortality surveillance is key to informing public health action to reduce overdose-related fatalities. States are increasingly using linked data sources to enhance surveillance activities, yet approaches to their effective utilization and analyses are needed. Objectives: The objective of this study is to describe the development and utilization of analytical tools for rapid, ongoing monitoring of drug overdose trends in Kentucky (KY) and decedents’ exposure to prescribed controlled substances (CS). Methods: KY established a monthly process of linking all-cause death certificate (DC) with prescription drug monitoring program (DC-PDMP) data to enhance mortality surveillance. Using provisional 2018-2020 DC-PDMP data we developed scheduled quarterly analytical reports. Drug overdose deaths are identified based on underlying cause of death (ICD-10 X40-X44, X60-X64, X85, or Y10-Y14); involved drugs/drug classes are identified from multiple cause of death codes (T36 – T50). Common contributing substances are identified from DC cause of death section text fields. Drugs listed on DCs are compared with decedents’ past 90 days CS prescriptions. Results: KY resident drug overdose deaths accounted for 2.8% of all-cause mortality, but among age group 26-40 years, 28.6% of all-cause deaths were due to drug overdose. Drug overdose decedents were disproportionally male (65.4% vs. 51.8% among all-cause deaths). From 2018 to 2020, the number of drug overdose deaths increased 42%. Deaths involving synthetic opioids and psychostimulants increased (56.2%vs 71.7% and 27.3% vs 35.1%, respectively) and deaths involving heroin (10.4% vs 6.0%), benzodiazepines (24.1% vs 15.3%), cocaine (9.6% vs 8.4%) and natural/semi-synthetic opioids (22.7% vs 21.3%) declined. The five substances most frequently listed in the DC in 2020 were fentanyl, methamphetamine, 4-ANPP, gabapentin, and acetyl fentanyl. Sixty-three percent of deaths involving natural/semi- synthetic opioids and 76% of cases involving benzodiazepines had no dispensed prescriptions for those drug classes in the previous 90 days, suggesting possible diversion. A historically high level of drug overdose deaths was observed in the first months of COVID-19 pandemic, with April-June 2020 overdose deaths (n=557), 80%higher than the same period in 2019. Conclusions: The analytical enhancement of KY’s drug overdose surveillance supports rapid assessment to inform public health action and provides a rich dataset for pharmacoepidemiologic studies.
Affiliation Label Tesim:: Injury Prevention Research Center
Type:: http://purl.org/dc/dcmitype/Text
DOI:: https://doi.org/10.17615/d4mt-5a51
Language Label:: English
ORCID:
Other Affiliation:: U.S. Food and Drug Administration, University of Kentucky, and
Person:: McAninch, Jana, Slavova, Svetla, Brancato, Candace, Oh, GYeon, Secora, Alex, Dasgupta, Nabarun, Collins, Candice, and Freeman, Patricia
Rights Statement Label:: In Copyright

5. Characterizing Patients using Abuse-deterrent Formulations of Extended-release Opioid Analgesics

Title Tesim:: Characterizing Patients using Abuse-deterrent Formulations of Extended-release Opioid Analgesics
Creator:: DiPrete, Bethany, Slavova, Svetla, Dasgupta, Nabarun, Ranapurwala, Shabbar, Oh, GYeon, Slade, Emily, Delcher, Chris, and Moga, Daniela
Date of publication:: August 23, 2021
Abstract Tesim:: Background: Abuse-deterrent formulations (ADFs) of extended-release (ER) opioids are manufactured to address opioid abuse. However, little is known about characteristics of patients who initiate ADF opioids, which is important to identify appropriate comparators to address confounding by indication. Objectives: To describe demographics and medical characteristics of patients prescribed ADF and non-ADF ER opioids in two sources of commercial claims. Methods: Using IBM Marketscan commercial claims (Data A) and a large private insurance provider in North Carolina [USA] (Data B) (both 2009-2018), we conducted a retrospective cohort study to examine patterns of ADF opioid use compared to non-ADF ER opioid use. Patients who initiated ADF and non-ADF ER opioids (18-64 years-old) were selected using both a traditional new user design (no opioid claims during the washout period, defined as six-months prior to ER opioid initiation) and a prevalent new user design (allowed non-ER opioid claims during the washout period and excluded the patients with no six-months eligibility prior to the first immediate-release (IR) opioid claim). Patient characteristics including demographics, medications (gabapentin, benzodiazepine, antidepressants, IR opioids), pain-related symptoms, and cancer were measured during the washout period for patients with ADF and non-ADF ER opioids. Results: Among eligible ER opioid initiators in Data A (N=330,728) and B (N=20,992), 31% and 34% initiated with ADF opioids, respectively. Among these patients, demographics were as follows (Data A and B): age [mean (SD)] = 49.4 (11.8) and 48.4 (11.8); male sex = 51.2% and 55.4%. Among patients with non-ADF ER opioids, demographics were as follows (Data A and Data B): age [mean (SD)] = 49.2 (11.4) and 47.8 (11.3); male sex = 45.8% and 50.4%. About 50% and 62% of patients with ADF opioids initiated with IR opioids, whereas 29%and 34% of patients with non-ADF ER opioids initiated with IR opioids in Data A and B, respectively. In both data sources, the prevalence of several types of pain was higher among patients with ADF opioids than in non-ADF ER group, including acute pain (Data A: 54.5% vs. 40.3%; Data B: 56.7% vs. 41.5%), arthritis pain (35.7% vs. 20.1%; 36.4% vs. 22.7%), and chronic pain (84.8% vs. 76.3%; 89.5% vs. 85.3%). The prevalence of use of medications and cancer was higher in patients with non-ADF ER opioids than in patients with ADF opioids in both data sources. Conclusions: Both data sources revealed differences in characteristics between patients with ADF and non-ADF ER opioids. The implications for research design include identifying appropriate comparator groups when examining ADF opioid use related outcomes.
Affiliation Label Tesim:: Injury Prevention Research Center
Type:: http://purl.org/dc/dcmitype/Text
DOI:: https://doi.org/10.17615/dqb8-at59
Language Label:: English
ORCID:
Other Affiliation:: and University of Kentucky
Person:: DiPrete, Bethany, Slavova, Svetla, Dasgupta, Nabarun, Ranapurwala, Shabbar, Oh, GYeon, Slade, Emily, Delcher, Chris, and Moga, Daniela
Rights Statement Label:: In Copyright

6. Matching Study Design to Prescribing Intention: The Prevalent New User Design in Opioid Research

Title Tesim:: Matching Study Design to Prescribing Intention: The Prevalent New User Design in Opioid Research
Creator:: Oh, GYeon, Pence, Brian, DiPrete, Bethany, Delcher, Chris, Moga, Daniela, Ranapurwala, Shabbar, Slavova, Svetla, Slade, Emily, and Dasgupta, Nabarun
Date of publication:: August 23, 2021
Abstract Tesim:: Background: In drug studies, research designs requiring no prior exposure to certain drug classes may restrict research on important populations. For example, currently marketed abuse-deterrent formulation (ADF) opioids are routinely used in patients with prior prescription opioid exposure. The traditional new user design excludes patients with prior exposure to prescription opioids, hence incident ADF users are not representative of the overall ADF user population. A prevalent new user design, wherein patients are prescribed similar treatments (or potential comparators) before starting the new treatment, likely better represents the intended ADF patient population. Objectives: To evaluate the appropriateness of new user versus prevalent new user design for estimating post-market effectiveness of ADFs and examine patterns of ADF initiation. Methods: We used pharmaceutical claims data from a large private insurer in North Carolina [USA] from 2009-2018. Included patients were new ADF users age 18-64 with 6 months of continuous enrollment prior to their first ADF claim. Incident users were identified as those with no prescription opioid claims in a 6-month washout period prior to ADF initiation. Prevalent new users were identified as those with non-ADF opioid claims during the 6 months before ADF initiation, so long as they also had a 6-month washout period of no opioid claims prior to first non-ADF opioid claim. We compared sample sizes by study design and described ADF utilization patterns. Results: We identified 8,841 eligible patients who initiated an ADF. Of these, 2,332 (26%) were classified as incident users, whereas 6,509 (74%) were prevalent new users and would be excluded in a traditional new user design. Most incident ADF users started with both an ADF and an immediate-release (IR) opioid concurrently (85%). Among prevalent new users, common ADF initiation patterns were: adding an ADF to an IR opioid regimen (43%), an immediate switch from IR opioids to an ADF (15%), and a delayed switch from IR opioids to an ADF (14%). Conclusions: Three-quarters of patients initiating ADFs had prior prescription opioid use and would be excluded in a traditional new user study design. A prevalent new user design would increase sample size and better capture clinically meaningful patients. These findings may apply to studies of other medications where prior exposure is a labeled prerequisite, such as higher dose ER opioids and second-line therapies. Future work will explore prevalent new user designs and consider nuances in ADF initiation such as immediate versus delayed switching by incorporating time-matching to address opioid tolerance.
Affiliation Label Tesim:: Gillings School of Global Public Health and Injury Prevention Research Center
Type:: http://purl.org/dc/dcmitype/Text
DOI:: https://doi.org/10.17615/4cbq-t195
Language Label:: English
ORCID:
Other Affiliation:: University of Kentucky and
Person:: Oh, GYeon, Pence, Brian, DiPrete, Bethany, Delcher, Chris, Moga, Daniela, Ranapurwala, Shabbar, Slavova, Svetla, Slade, Emily, and Dasgupta, Nabarun
Rights Statement Label:: In Copyright

7. Variation among Pharmacists in Calculating Days’ Supply

Title Tesim:: Variation among Pharmacists in Calculating Days’ Supply
Creator:: Freeman, Patricia, Brown, John, Dasgupta, Nabarun, Slavova, Svetla, and Nocera, Maryalice
Date of publication:: August 23, 2021
Abstract Tesim:: Background: Pharmacoepidemiologic studies utilize the days’ supply field in prescription (Rx) claims to classify exposures and outcomes; however, data on the accuracy of days’ supply are lacking. In the United States, days’ supply is calculated at the time of Rx filling. Pharmacists’ interpretation may influence days’ supply calculations leading to variability. Objectives: The objectives of this study are to assess consistency of pharmacists’ calculation of days’ supply for opioid Rxs and identify factors pharmacists consider when calculating days’ supply for a given Rx. Methods: A cross-sectional survey was designed to collect information from Kentucky-licensed pharmacists about opioid analgesic dispensing practices and days’ supply calculations. The survey was distributed electronically via email through the Kentucky Board of Pharmacy. Responses were collected anonymously over a 4-week period in December 2019. Inclusion criterion was experience dispensing opioid analgesics; other questions were voluntary. For days’ supply calculations, pharmacists saw 2 hypothetical opioid Rxs and calculated days’ supply. Additional questions examined factors influencing days’ supply and pharmacists’ perception of the importance of accuracy. Results: Of 753 responses, 643 were eligible. For hydrocodone tablets, 19.4% (n=347) of pharmacists assigned inaccurate days’ supply, ranging from 7 days to 90 days. Common mistakes were 40 and 45 days instead of the correct 30. For morphine solution, days’ supply was inaccurate in 10.0% (n=351), ranging from 2 to 90 days. Common mistakes were 10 and 15 days instead of the correct 7. Most reported considerations when assigning days’ supply were instructions from the prescriber on intended duration (75.5%), third-party payer restrictions (48.4%) and familiarity with practitioners prescribing habits (45.3%). Most pharmacists perceived that accurate days’ supply are important to determine when medication is due to be filled/refilled (89.8%) and to support safe medication use for patients (80.8%), although fewer perceive accuracy as important for calculating morphine milligram equivalents (61.8%) and assessment of doctor shopping in the state prescription drug monitoring program (63.7%). Conclusions: Between 10-20% of pharmacists inaccurately assigned days’ supply values to opioid analgesic Rx. While most believe accuracy in days’ supply is important to dispensing decisions, fewer consider it important for other activities such as surveillance monitoring. The estimated error rate from the survey will be used to inform the design of a validation study using stratification-sampled pharmacy records.
Affiliation Label Tesim:: Injury Prevention Research Center
Type:: http://purl.org/dc/dcmitype/Text
DOI:: https://doi.org/10.17615/awwd-d722
Language Label:: English
ORCID:
Other Affiliation:: University of Kentucky and
Person:: Freeman, Patricia, Brown, John, Dasgupta, Nabarun, Slavova, Svetla, and Nocera, Maryalice
Rights Statement Label:: In Copyright