Web only rx16 pdmp-tues_330_1_kreiner_2ringwalt-schiro
Download as PPTX, PDF2 likes755 views
This document discusses three projects in North Carolina aimed at reducing harm from prescription drug abuse: 1) Identifying prescribers who prescribe very high levels of controlled substances using PDMP data and algorithms, in partnership with state agencies and the medical board. 2) Identifying prescribers with multiple patients who died from opioid-related overdoses. 3) Providing immediate feedback to prescribers on high-risk patients through electronic health records integrated with PDMP data, in partnership with health systems. The goals are to develop valid methods to identify problematic prescribing patterns using multiple data sources and address technical and policy hurdles to information sharing.
Web only rx16 pdmp-tues_330_1_kreiner_2ringwalt-schiro
- 1. Risk Behaviors, Morbidity and Mortality Presenters: • Peter Kreiner, PhD, Senior Scientist, Brandeis University • Christopher Ringwalt, DrPH, MSW, Senior Scientist, Injury Prevention Center, University of North Carolina - Chapel Hill • Sharon Schiro, PhD, Associate Professor/Data Scientist, University of North Carolina - Chapel Hill PDMP Track Moderator: John J. Dreyzehner, MD, MPH, FACOEM, Commissioner, Tennessee Department of Health, and Member, Rx and Heroin Summit National Advisory Board
- 2. Disclosures • Peter Kreiner, PhD; Christopher Ringwalt, DrPH, MSW; and Sharon Schiro, PhD, have disclosed no relevant, real, or apparent personal or professional financial relationships with proprietary entities that produce healthcare goods and services. • John J. Dreyzehner, MD, MPH, FACOEM – Ownership interest: Starfish Health (spouse)
- 3. Disclosures • All planners/managers hereby state that they or their spouse/life partner do not have any financial relationships or relationships to products or devices with any commercial interest related to the content of this activity of any amount during the past 12 months. • The following planners/managers have the following to disclose: – John J. Dreyzehner, MD, MPH, FACOEM – Ownership interest: Starfish Health (spouse) – Robert DuPont – Employment: Bensinger, DuPont & Associates-Prescription Drug Research Center
- 4. Learning Objectives 1. Identify indicators of risk behaviors by prescribers. 2. Demonstrate the prescriber risk behavior indicators used by the nation’s Prescriber Behavior Surveillance System. 3. Explain how North Carolina shares its PDMP data with multiple agencies to reduce morbidity and mortality related to Rx drug abuse. 4. Provide accurate and appropriate counsel as part of the treatment team.
- 5. A Validation Study of Prescriber Risk Measures from the Prescription Behavior Surveillance System PDMP Track: Risk Behaviors, Morbidity and Mortality March 29, 2016 Peter Kreiner, Ph.D. PDMP Center of Excellence, Brandeis University
- 6. Disclosure Statement • Peter Kreiner, PhD, has disclosed no relevant, real or apparent personal or professional financial relationships with proprietary entities that produce health care goods and services.
- 7. Learning Objectives • Identify indicators of risk behaviors by prescribers. • Demonstrate the prescriber risk indicators used by the Prescription Behavior Surveillance System. • Describe a validation study of these indicators in one state, using actions taken by the state Medical Board. • Provide accurate and appropriate counsel as part of the treatment team.
- 8. Overview • High-level summary of Prescription Behavior Surveillance System (PBSS) project • Validation study – Analytic strategy – Methods – Results – Conclusions/Limitations – Next steps
- 9. The Prescription Behavior Surveillance System (PBSS) A longitudinal, multi-state database of de-identified PDMP data, to serve as: 1. An early warning public health surveillance tool 2. An evaluation tool, in relation to state and local laws, policies and initiatives Info available at: http://www.pdmpexcellence.org/content/ prescription-behavior-surveillance-system-0 Publication: Paulozzi LJ, Strickler GK, Kreiner PW, Coris CM. Controlled substance prescribing patterns – Prescription Behavior Surveillance System, eight states, 2013. MMWR Surveillance Summaries. Oct. 16, 2015; 64(9): 1-14.
- 10. PBSS Continued • Began in FY2012 with support from CDC and FDA, administered through BJA • Guided by Oversight Committee: – Federal partners: CDC, FDA, BJA, SAMHSA, ASPE – State partners to date: CA, DE, FL, ID, KY, LA, ME, OH, TX, VA, WA, WV – Additional state partners in process – No release of data or findings without Oversight Committee approval
- 11. PBSS Measures • Prescribing measures – Rates of opioid, benzodiazepine, stimulant prescriptions • By quarter and year, by drug class, sex, and age group • By quarter and year, by major opioid, benzodiazepine, and stimulant drug category • Patient risk indicators – Average daily dosage of opioids (MMEs) – Days of overlapping prescriptions – Multiple provider episode rates • By drug class, age group, and drug category
- 12. PBSS Measures Continued • Prescriber risk indicators – Prescriber percentile ranking, based on daily prescribing volume • By quarter, year, and drug class – Average daily dosage for opioid patients (MMEs) – Median distance in miles, patient to prescriber – Percentage of patients with MPE – Percentage of prescriptions by payment type – Percentage of patients prescribed LA/ER opioids who were opioid-naïve • Pharmacy risk indicators – Analogous to prescriber risk indicators
- 13. Prescriber Risk Indicators • Prescribing volume by prescriber decile: Proportion of total prescriptions accounted for by prescriber 10% groupings • Average daily opioid dosage (MMEs) by prescriber decile (volume) • Distance patients travel to prescriber and proportion of prescriber practice who meet MPE threshold
- 14. 0 10 20 30 40 50 60 70 1st 2nd 3rd 4th 5th 6th 7th 8th 9th 10th California, 2012: Proportion of Total Opioid, Stimulant, and Benzodiazepine Prescriptions Written by Prescriber Deciles Percent of total opioid prescriptions Percent of total stimulant prescriptions Percent of total benzodiazepine prescriptions
- 15. 0 10 20 30 40 50 60 70 80 1st 2nd 3rd 4th 5th 6th 7th 8th 9th 10th AveragedailydosageofpatientsinMMEs Prescriber decile based on volume of opioid prescriptions California 2012: Average Daily Dosage of Patients by Prescriber Decile Based on Volume of Opioid Prescriptions
- 16. 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0 20 40 60 80 100 120 140 PercentageofprescriberpatientswithMPE Average distance patients travel in miles California 2012: Prescriber Deciles Based on Average Distance Patients Travel, Compared with Percentage of Prescriber Patients with an MPE Percentage of patients with MPE Log. (Percentage of patients with MPE)
- 17. Validation Studies Purpose: 1. Examine frequency of prescribers highest on prescriber risk indicators having actions taken against them (Medical Board, DEA, law enforcement) – Vs. prescribers lower on these indicators 2. Develop predictive models of actions taken to estimate relative risk of different prescriber behaviors
- 18. Validation Studies: Analytic Strategy • Prescriber outcomes – Identify prescribers against whom actions have been taken • By the state Medical Board/Board of Osteopathic Medicine • By the DEA • By other law enforcement – Categorize types of offense and types of action taken – Examine/take into account prescriber license type and physician specialty
- 19. Validation Studies: Analytic Strategy • Logistic regressions of actions taken (2010-2014) • Predictor variables – Prescriber risk indicators • In 2010; vs. action(s) taken in 2010-2014 • Examine factor structure of prescriber indicators • Trajectory analysis: identify different groups/patterns over time – Measure of prescribing complexity? • Pattern of drugs prescribed, in relation to peers • Control variables – Prescriber sex, specialty (categorical variable)
- 20. Medical Specialties Used in the Analysis: Maine ProvidersMedical Specialty Frequency Share Share (exc. missing) Anesthesiology 26 0.3% 0.5% Dermatology 10 0.1% 0.2% Ear, nose, & throat (ENT) 33 0.4% 0.6% Emergency medicine 263 3.2% 4.8% GP/FM/DO a 845 10.4% 15.4% Internal Medicine 446 5.5% 8.1% Obstetrics and Gynecology (OB/GYN) 135 1.7% 2.5% Oncology b 65 0.8% 1.2% Ophthalmology 48 0.6% 0.9% Orthopedics c 132 1.6% 2.4% Pain Medicine d 32 0.4% 0.6% Pediatricians 198 2.4% 3.6% Physical Medicine and Rehabilitation 52 0.6% 0.9% Podiatrist 54 0.7% 1.0% Psychiatry & Neurology 330 4.1% 6.0% Radiology 35 0.4% 0.6% Surgery 198 2.4% 3.6% Dentist 489 6.0% 8.9% Veterinarian 195 2.4% 3.6% Othere 474 5.8% 8.6% Missing f 2621 32.4% Physician Assistants g 536 6.6% 9.8% Non-physician prescriber h 886 10.9% 16.2%
- 21. Types of Maine Medical Board Actions Action Brief Description Revocation of License Physician’s license is terminated; individual can no longer practice medicine within the state or territory. Suspension of License Physician may not practice medicine for a specified period of time, perhaps due to disciplinary investigation or until other state board requirements are fulfilled. Probation of License Physician’s license is monitored by a state board for a specified period of time. Probation of License with Condition(s) Physician must fulfill certain conditions to avoid further sanction by the state board. Examples: Subject to urine test/ substance abuse surveillance; require to complete training/ course Restricted or Conditional License Physician's ability to practice medicine is limited (e.g., loss of prescribing privileges). Denial of License or Denial of License Renewal Physician’s application for a medical license or renewal of a current license is denied. Fine (monetary) A monetary penalty against a physician. Reprimand/ Warning Physician is issued a warning or letter of concern. Voluntary Surrender or Withdrawal of License Physician voluntarily surrenders or withdraws medical license, sometimes during the course of a disciplinary investigation. Amendment or Removal of Adverse Action Amendment or Removal of adverse action Reinstatement of License Reinstatement of License
- 22. Board Actions/Decisions and Individuals Disciplined Maine, 2010 - 2014 Total 2010 2011 2012 2013 2014 n % n % n % n % n % n % Number of Board Actions/Decisions (an individual can receive more than one action a year) 199 100 36 18 34 17 41 21 42 21 46 23 Number of unique individuals disciplined: Unique individuals sanctioned by year (with replacement across years) 164 100 30 18 29 18 37 23 35 21 33 20 N individuals placed on probation or probation with conditions 27 100 5 19 7 26 7 26 6 22 2 7 N individuals with a license suspension 18 100 3 17 4 22 5 28 2 11 4 22 N individuals whose license was revoked or voluntarily surrendered license 24 100 6 25 4 17 7 29 4 17 3 13 N individuals whose license application was denied or denied renewal of license 12 100 2 17 4 33 2 17 1 8 3 25 Unique individuals sanctioned 2010-2014 (without replacement) 119 100 30 25 25 21 23 19 23 19 18 15
- 23. Share of prescribers in ME PDMP dataset with criminal records (2002- 2011) and medical board disciplinary actions taken against them (2010- 2014), by specialty
- 24. Prescriber Indicators: Factor Structure Factor 1: # of patients # opioid prescriptions # benzodiazepine prescriptions # prescriptions written daily # opioid prescriptions written daily Factor 2: Average daily MME % of patients with average daily MME > 100 Factor 3: # stimulant prescriptions (-) % of prescriptions paid for by cash Factor 4: % of patients meeting MPE threshold Average distance travelled by patients
- 25. Conclusions • Being in the top 1% of Maine prescribers on number of patients issued a CS prescription, prescriptions per day, opioid prescriptions per day, or average daily dosage of opioids prescribed (MMEs) is associated with being subject to a Medical Board action for inappropriate prescribing and with prescription restrictions • Being in the top 1% for MMEs per day is associated with being subject to a severe action by the Board • Being in the top 2% of prescribers for average distance traveled by their patients is also associated with being subject to a severe action by the Board
- 26. Study Limitations • Findings may not be generalizable to other states: rural state; relatively small number of Board actions taken – We have validation studies in process with Ohio and Washington • Board actions reactive rather than proactive • Unknown lead time before Board takes action – We have follow-up study of Board actions to address this question; also effects of Board actions and Board decision- making process
- 27. Next Steps • Examine Ringwalt measures of co-prescribing of benzodiazepines and > 100 MME opioids, and overlapping CS prescriptions • Explore measure interactions (e.g., top 1% of prescribers on volume and top X% on risk indicator) • Examine prescribing profiles based on prescriber – drug networks, in relation to specialty
- 28. Maine Prescribers Above 90th Percentile: Prescriber – Drug Network with Medical Board and Criminal Justice Actions
- 29. Contact Information Peter Kreiner, Ph.D. Principal Investigator PDMP Center of Excellence Brandeis University 781-736-3945 pkreiner@brandeis.edu www.pdmpexcellence.org
- 30. Reducing morbidity and mortality from prescription drug abuse through partnerships with NC’s PDMP Sharon Schiro, Chris Ringwalt, Rachel Seymour, Scott Proescholdbell, Joseph Hsu, David Henderson, Alex Asbun
- 31. Disclosure statement • Sharon Schiro, PhD, has disclosed no relevant, real or apparent personal or professional financial relationships with proprietary entities that produce health care goods and services. • Chris Ringwalt, DrPH, has disclosed no relevant, real or apparent personal or professional financial relationships with proprietary entities that produce health care goods and services.
- 32. Learning objectives • Explain how North Carolina shares its PDMP data with multiple agencies to reduce morbidity and mortality related to Rx drug abuse. • List other sources of data that augment the PDMP for the purpose of addressing problematic prescribing • Identify methodologies utilized to overcome technical and policy-related hurdles. • Provide accurate and appropriate counsel as part of the treatment team.
- 33. Introduction • Three projects • Identifying prescribers who prescribe very high levels of controlled substances • Identifying prescribers with multiple patients with opioid-related death • Immediate feedback to prescribers on potentially high risk patients through health care system electronic prescription records and PDMP data • Partners • Injury andViolence Prevention Branch, NC Division of Public Health • Injury Prevention Research Center (IPRC), University of North Carolina-Chapel Hill • Department of Surgery, University of North Carolina-Chapel Hill • Carolinas Healthcare Department of Orthopedics • NC Medical Board (NCMB) • Drug Control Unit, Division of Mental Health, Developmental Disabilities and Substance Abuse Services (DMH/DD/SAS)
- 34. Identifying prescribers who prescribe very high levels of controlled substances
- 35. Identifying prescribers who prescribe very high levels of controlled substances: Partners • Partnership between DMH/DD/SAS, UNC IPRC, and UNC Department of Surgery to develop algorithms • Partnership with NC Medical Board for possible investigation of identified prescribers
- 36. Identifying prescribers who prescribe very high levels of controlled substances: Methodology • Project goal: Develop and validate algorithms using PDMP (NC CSRS) data to identify prescribers with unusual prescribing patterns. • Algorithms evaluated: • Rates of prescriptions for daily dose > 100 MME • High average daily MME* • High total MME per prescription • Prescription rates for opioids, benzodiazepines, stimulants* • Rates of co-prescribed opioids (> 100 MME) and benzodiazepines* • Temporally overlapping prescriptions* • Long travel distance by patient to prescriber or pharmacy • Multiple prescribers for controlled substances • Multiple pharmacies for controlled substances * Algorithms most closely associated with deaths from overdose
- 37. Identifying prescribers who prescribe very high levels of controlled substances: Methodology • NC Medical Board receives lists of prescribers who are in: • Top 1% of those prescribing 100 milligrams of morphine equivalents (“MME”) per patient per day • OR • Falls within the top 1% of those prescribing 100 MME’s per patient per day in combination with any benzodiazepine and who are within the top 1% of all controlled substance prescribers by volume
- 38. Identifying prescribers who prescribe very high levels of controlled substances: Next steps • Identify additional data elements to be added to metrics to improve sensitivity of metrics and reduce false positives. Concern is time required for investigations and impact of investigations on legitimate prescribers. • Work with NCMB Advisory Group (2 Board members and 3 consultants) to develop objective criteria to decide whether or not to open an investigation
- 39. Identifying prescribers with more than one patient with an opioid-related death
- 40. Identifying prescribers with more than one patient with an opioid-related death: Partners • Partnership between NC Division of Mental Health, Developmental Disabilities and Substance Abuse Services and NC Injury andViolence Prevention Branch of the Division of Public Health to develop algorithms • Partnership with NC Medical Board for possible investigation of identified prescribers
- 41. Identifying prescribers with more than one patient with an opioid-related death: Methods • Identified deaths attributed to unintentional or undetermined controlled substance-related poisoning usingVital Records Death Certificate file. • Linked to CSRS data to identify those who prescribed the decedent a controlled substance prescription within 60 days prior to death. • Linked to CSRS andVital Records to determine whether the prescriber(s) had any other patient deaths due to substance-related poising during the prior 12 months • Prescribers with two or more controlled-substance related deaths in the past 12 months referred to NCMB
- 42. Identifying prescribers with more than one patient with an opioid-related death: Results • Validation results: • 465 opioid overdose-related deaths identified in 2012 • 651 prescribers prescribed opioids to these patients within 30 days of their death • Match to prescribers who fall in top 1% of all CS prescribers and 100+ MME prescribers • 30-46% of prescribers who fell in top 1% of metrics also had prescribed an opioid to a patient within 30 days of their death • Publication:The use of a prescription drug monitoring program to develop algorithms to identify providers with unusual prescribing practices for controlled substances. Journal of Primary Prevention October 2015.
- 43. Identifying prescribers with more than one patient with an opioid-related death: Next steps • Identify additional data elements to be added to metrics to improve sensitivity of metrics and reduce false positives. Concern is time required for investigations and impact of investigations on legitimate prescribers • Working with NCMB Advisory Group (2 Board members and 3 consultants) to develop objective criteria to decide whether or not to open an investigation
- 44. Immediate feedback on high risk patients through healthcare system prescription records and PDMP data
- 45. Immediate feedback on high risk patients through healthcare system prescription records and PDMP data: Partners • Partnership between NC Division of Mental Health, Developmental Disabilities and Substance Abuse Services and Carolinas Healthcare System (CHS).
- 46. Immediate feedback on high risk patients through healthcare system prescription records and PDMP data: Methods • Integration of decision support in to electronic medical record (EMR) using 5 criteria to trigger an alert. • Existing prescription with >50% prescription duration remaining • 2+ visits to ED with on-site administration of controlled substance • 3+ controlled substance prescriptions in last 30 days • Positive screen for blood alcohol, cocaine, or marijuana • Previous presentation to ED for overdose
- 47. Immediate feedback on high risk patients through healthcare system prescription records and PDMP data: Results • Two silent phases used to tune alert triggers to have acceptable number of relevant alerts at point of care. • Alerts generated for 5.9% of prescribing encounters • Majority of prescribing encounters in outpatient setting. • Inpatient discharges and ED/Urgent Care encounters have disproportionately high rates of opioid and benzodiazepine prescriptions.
- 48. Immediate feedback on high risk patients through healthcare system prescription records and PDMP data: Next steps • Assess burden on prescribers. • Evaluation of impact on prescribing patterns. • Integration of PDMP data in to EMR and alerts
- 49. Navigating policy hurdles: Lessons learned • Access to PDMP (CSRS) data: sensitive data requires high level of security and confidentiality. Striking the balance between legal protection, legal authority, and clinical interest. • Investigating prescribers is a big deal, so need to ensure that metrics identify only prescribers truly warranting investigation. • Need to balance vigorous regulatory efforts with potential chilling effect on other prescribers and the right for adequate pain control for many patients
- 50. Navigating technical hurdles: Lessons learned • Storage and processing power requirements are significant for analysis of large databases. • Extensive cleaning of PDMP (CSRS) data required. • Clear, complete data dictionaries needed for all datasets to avoid errors due to assumptions on how data are collected and defined.
- 51. Importance of the partnerships • Utilization of research, evaluation, and data science skill sets from an academic institution (UNC), a large healthcare system (CHS), the NC Division of Public Health, and the NC Division of Mental Health/Developmental Disabilities/Substance Abuse Services • Integration by research partners of topic-specific knowledge and experience from DMH/DD/SAS and NCMB staff, as well as CHS prescribers
- 52. Importance of the partnerships, continued • Translation of research into practice through investigative efforts of NCMB • Because this is a multi-disciplinary, multi-jurisdictional problem, increased coordination and communication are essential. NC continues to identify ways in which key stakeholders can collaborate and share knowledge to address common problems.
- 53. Questions? • Contact information: • Sharon Schiro: Sharon_Schiro@med.unc.edu • Chris Ringwalt: cringwal@email.unc.edu • Alex Asbun: Alex.Asbun@dhhs.nc.gov • David Henderson: David.Henderson@NCMedBoard.org • Joseph Hsu: Joseph.Hsu@CarolinasHealthcare.org • Scott Proescholdbell: Scott.Proescholdbell@dhhs.nc.gov • Rachel Seymour: Rachel.Seymour@CarolinasHealthcare.org
- 54. Risk Behaviors, Morbidity and Mortality Presenters: • Peter Kreiner, PhD, Senior Scientist, Brandeis University • Christopher Ringwalt, DrPH, MSW, Senior Scientist, Injury Prevention Center, University of North Carolina - Chapel Hill • Sharon Schiro, PhD, Associate Professor/Data Scientist, University of North Carolina - Chapel Hill PDMP Track Moderator: John J. Dreyzehner, MD, MPH, FACOEM, Commissioner, Tennessee Department of Health, and Member, Rx and Heroin Summit National Advisory Board
Editor's Notes
- #15: This distribution – approximately 80% of total prescriptions accounted for by the top 20% of prescribers, is quite consistent across states.
- #16: Note the association between average volume of opioid prescriptions prescribed and average daily dosage of patients.
- #17: This graph shows an association between the average distance patients travel to prescribers (aggregated into prescriber deciles based on distance patients travel) and the percentage of the prescribers’ patients who meet the multiple provider episode criteria (MPE: 5 or more prescribers and 5 or more pharmacies in a 3-month period). The association appears to be logarithmic.
- #30: For questions or more information.
- #37: Algorithms evaluated: Rates of prescriptions for daily dose > 100 MME High average daily MME (what the patient takes each day) High total MME per prescription (what’s in the bottle – daily dose x days) Prescription rates for opioids, benzodiazepines, stimulants Rates of co-prescribed opioids (> 100 MME) and benzodiazepines Temporally overlapping prescriptions Long travel distance by patient to prescriber or pharmacy Multiple prescribers for controlled substances (doctor shopping) Multiple pharmacies for controlled substances (pharmacy shopping)
- #39: Initial screen, then other processes to maximize sensitivity and minimize false positives. Burden of opening investigations – could damage the career of the prescriber, even if the prescriber is prescribing appropriately.
- #42: May need to consider # of prescriptions written by prescriber, since 2 deaths
- #50: Legal protection for patients (HIPAA) Legal interest – investigation by SBI
- #52: Notes – multiple disciplinary. Inclusive vs working in siloes.
- #53: Notes – multiple disciplinary. Inclusive vs working in siloes.
- #54: Notes – multiple disciplinary. Inclusive vs working in siloes.