Policy & Priorities: Rethinking University Research with State Data Grand Rounds: NIH HCS Collaboratory and PCORnet | 6/29/18 Aaron McKethan, PhD | @a_mckethan Chief Data Officer, NC Department of HHS | aaron.mckethan@dhhs.nc.gov Assistant Professor of Population Health Sciences, Duke School of Medicine Senior Policy Fellow, Duke-Margolis Center for Health Policy | aaron.mckethan@duke.edu
2
Three-part punchline 1. States need help developing analytic priorities 2. Start with the simplest available research methods 3. Consider policy implications from the beginning 3
Analytic priorities 4
Analytic priorities 5
Analytic priorities “How many OB - GYNs billed at least one claim in Harnett County in 2017?” 6
Analytic priorities “How many OB - “What was the GYNs billed at fiscal impact of least one claim in shifting to Harnett County in Medicaid managed 2017?” care?” 7
Analytic priorities “How many OB - “What was the GYNs billed at fiscal impact of The opportunity space for policy-oriented least one claim in shifting to health services researchers Harnett County in Medicaid managed 2017?” care?” 8
Analytic priorities 9
Analytic priorities 1. What do we already know? …and where is policy not aligned with available evidence? • 10
Analytic priorities 1. What do we already know? …and where is policy not aligned with available evidence? • 2. What do we not know? …and how valuable would it be to know? • 11
Analytic priorities 1. What do we already know? …and where is policy not aligned with available evidence? • 2. What do we not know? …and how valuable would it be to know? • 3. What are the highest-priority questions? …that can be answered with available data? • … that can inform specific policy actions in the near term? • 12
Example: Prescription Opioids 13
https://files.nc.gov/ncdhhs/documents/NC%20Opioid%20Action%20Plan%20Metrics_April%202018%20V2.pdf 14
Number of opioid prescription claims and percentage of all Medicaid prescription claims that are opioids 1,200,000 8% 7% 1,000,000 6% 800,000 5% 600,000 4% 3% 400,000 2% 200,000 1% 0 0% 2013 2014 2015 2016 2017 Number of claims Percentage of all prescription claims 15
Average morphine milligram equivalents (MME) per day and average days's supply per prescription 58 17.4 18 17.1 16.8 16.5 16.3 17 16 56 55.6 15 14 53.7 53.6 13 54 12 11 MME Days 51.6 52 10 9 8 50 7 48.6 6 48 5 4 3 46 2 1 44 0 2013 2014 2015 2016 2017 Average Morphine Milligram Equivalent (MME) per day Average days' supply per prescription 16
Average MME/day per prescription in 2017 by county, with state average comparison 100 90 80 70 MME 60 50 40 30 20 10 0 4,000+ 1,000,000+ smallest population 100 counties largest population 17
Number and percentage of Medicaid beneficiaries 18 to 64 years old with concurrent use of prescription opioids and benzodiazepines, 2013-2017 30 30% 25 25% 20 20% Thousands 15 15% 10 10% 5 5% 0 0% 2013 2014 2015 2016 2017 Number of beneficiaries Percentage of beneficiaries Measure specification is from Pharmacy Quality Alliance: “Concurrent Use of Opioids and Benzodiazepines” 18
County variation in rate of concurrent opioids and benzodiazepines, 2017 DRAFT 600 Gaston Those in Denominator with 30+ Days Overlapping Rural Mecklenburg 500 Urban Cumberland Rockingham 400 Benzodiazepine Rxs Robeson Wake 300 200 Durham 100 Scotland 0 0 500 1000 1500 2000 2500 3000 3500 Beneficiaries 18-64 Who Received 2+ Opioid Prescriptions for a Total of 15+ Days Measure specification is from Pharmacy Quality Alliance: “Concurrent Use of Opioids and Benzodiazepines” 19
Number and percentage of concurrent opioid and benzodiazepine users 18-64 with a fill of naloxone in the previous 24 months, 2013-2017 1,400 10% 9% 1,200 8% 1,000 7% 6% 800 5% 600 4% 3% 400 2% 200 1% 0 0% 2013 2014 2015 2016 2017 Number of beneficiaries Percentage of beneficiaries Measure specification for denominator is from Pharmacy Quality Alliance: “Concurrent Use of Opioids and Benzodiazepines” 20
NC Opioid Symposium: Developing an Analytic Agenda • What are the most important ‘known unknowns’? • >70 experts (including government officials) • Medicaid claims and controlled substances data 21
What else do we not know re: opioid prescribing and use? McKethan A., Powell E., Patel A., Daniels C., Campbell H., Marshall S., & Proescholdbell S. 22
NC Opioid Symposium - Examples • “Does proactively informing prescribers on where they fall on opioid prescribing metrics change prescribing behavior ?” • “What has the effect of the STOP Act been on prescribing behaviors, opioid action plan metrics, and other outcomes ?” • “Is geographic clustering of harm reduction strategies associated with reduced negative outcomes ?” • “What is the current rate of referral from the hospital (E.D., inpatient) to treatment ?” • “What are the predictors of success in treatment in OBOTs? What are the best metrics to define treatment success (retention, relapse, etc.)?” • “What is the best set of outcomes and metrics that can be used across treatment studies ?” And 100+ more 23
DHHS Data Lab • Data sharing and research agreements with: 24
State-University Partnership Learning Network (SUPLN) Multi-State Medicaid OUD Project Principal Investigator: Julie Donohue, PhD (Pitt) Selected Draft Measures o Initiation and engagement of alcohol and other drug dependence treatment o Continuity of pharmacotherapy for opioid use disorder o Follow-up after Emergency Department visit for alcohol and other drug abuse or dependence States Kentucky • North Carolina Virginia • • Ohio • Maryland • • West Virginia Michigan Pennsylvania • • • Wisconsin http://www.academyhealth.org/SUPLN 25
Opportunities for PCORnet https://pcornet.org/ 26
Three-part punchline 1. States need help developing analytic priorities 2. Start with the simplest available research methods 3. Consider policy implications from the beginning 27
24% - Social Circumstances
30
31
32
How can we use these data products? 1. Better front-end technology 2. Benchmarking and business processes at county level 3. Measurement and support for health plans 4. Measurement and support for medical home providers 5. Collaboration with community-based organizations 6. Other 33
Three-part punchline 1. States need help developing analytic priorities 2. Start with the simplest available research methods 3. Consider policy implications from the beginning 34
Policy Implications Introduction Background Background Data & Methods Methods Methods Results Results Results Discussion Discussion Discussion Policy Implications Conclusion Conclusions 35
Policy Implications Paraphrase: “Thus , policy makers could further encourage these trends by continuing to invest in education and training.” 36
Three-part punchline 1. States need help developing analytic priorities 2. Start with the simplest available research methods 3. Consider policy implications from the beginning 37
Recommend
More recommend
