Geographic Variation of Inappropriate Prescription Opioid Use in - PowerPoint PPT Presentation

Geographic Variation of Inappropriate Prescription Opioid Use in Medicare W. Jenny Lo-Ciganic, MSPharm, MS, PhD University of Arizona, College of Pharmacy June 24, 2017 1

Research Team  University of Arizona • Drs. Jenny Lo-Ciganic, Kent Kwoh, Daniel Malone, Sandipan Bhattacharjee, Jeannie Lee, Melanie Bell, Ms. Lili Zhou and Mr. Westra Jordan  University of Pittsburgh • Drs. Walid Gellad & Julie Donohue  University of Wisconsin • Dr. Anne Roubal  Pharmacy Quality Alliance • Dr. Lisa Hines  ESRI Inc. • Mr. Jeremiah Lindemann 2

Funding & Disclosure  Dr. Lo-Ciganic is supported by a University of Arizona Health Sciences Career Development Award  Dr. Gellad is supported in part by VA HSR&D Merit Award I01 HX001765-01  Dr. Kwoh has received grant funding from Abbvie and EMD Serono 3

Overdose Deaths Involving Opioids, US 2000-2015 Any Opioid Deaths Per 100,000 population Commonly prescribed opioids (natural & semi- synthetic opioids and methadone) Heroin Other synthetic opioids (fentanyl, tramadol) Source: CDC/NCHS, National Vital Statistics System, Mortality. CDC WONDER, Atlanta, GA: US Department 4 of Health and Human Services, CDC; 2016. https://wonder.cdc.gov/.

Chronic Pain and Opioid Use in Medicare  Chronic pain conditions: 60%  Having ≥1 opioid prescription among Part D enrollees: 35%  Having polypharmacy with benzodiazepines: 25%

Gaps in Research and Science  Current evidence focuses on opioid utilization • Medicaid fee-for-service • Medicare disabled enrollees • Medicare Part D opioid drug mapping tool  Little is known about geographic variations of potentially inappropriate prescription opioid use 1. Morden NE et al. Med Care 2014;52:852-859; 2. Zerzan JT et al. Med Care 2006;44:1005-1010

Objective To examine geographic variation of potentially inappropriate prescription opioid use among non-cancer Medicare beneficiaries

Methods

Study Design, Data Source, & Cohort  Cross-sectional study  5% random sample of Medicare beneficiaries from 2011-2013  Exclusion criteria: • Not continuously enrolled in Part A, B, and D for 12 months • Medicare Advantage enrollees • Non-US residents • End-stage renal disease (ESRD) patients • Patients in hospice or with cancer • Had <2 opioid prescription fill • Had 2 prescription opioid fills but on the same day • Had <15 cumulative days of supplied opioids

Sample Size Flow Chart in 2013 Total beneficiaries (5% sample; N= 2,972,192) Excluded those who were (1) in Medicare Advantage program and not continuously enrolled in Part A, B, and D for 12 months (n=1,184,530), (2) non-US residents (n=27,343), (3) hospice patients (n=17,919), (4) ESRD patients (n=9,816); (5) cancers (except non-melanoma; n=338,191) Of these, non-hospice and non-cancer patients were continuously enrolled and had complete health claims history ( n= 1,394,393 ) Excluded patients who had (1) no opioid prescriptions (n=902,952), (2) only 1 opioid fill (n=159,584), (3) only filled 2 opioid prescriptions on the same day (n=1,337), and (4) only had days supplied of opioids <15 Of these, patients had ≥2 opioid filled on ≥2 cumulative days in 2013 (n=37,879). separate days, and ≥15 cumulative days of supply (n= 292,641) Of these, disabled Of these, non-disabled beneficiaries beneficiaries (n=135,252; 46.2%) (n=157,389; 53.8%)

Main Independent Variable  Dartmouth Atlas of Healthcare Hospital Referral Regions (HRR) • 3,436 health service areas were assigned to 306 HRR regions where the greatest proportion of major procedures were performed • Each HRR has at least one city where major cardiovascular surgery and neurosurgery are performed 1. http://www.dartmouthatlas.org/data/region/; 2. Donohue JM et al. N Eng J Med 2012;366: 530-538

Primary Outcomes of Interest: Pharmacy Quality Alliance (PQA) Measures  Among disabled Medicare opioid users (≥ 2 prescriptions with total days supply ≥15 days ) by HRR each year, % of beneficiaries with • High-dose : daily dosage >120 morphine milligram equivalent (MME) for ≥90 consecutive days • Multiple providers : opioid prescriptions from ≥4 prescribers and ≥4 pharmacies • Concurrent benzodiazepine use for ≥30 cumulative days

Covariates Sociodemographics Health status factors Regional/access-to- care factors  Age  Prescription  Rural vs urban Hierarchical  Sex Clinical Conditions geographic location  Race/ethnicity (RxHCC)  No. hospitals with  Low-income  Musculoskeletal pain management subsidy status disorders programs  Dual Medicaid  Depression  No. hospitals with eligibility physical therapy  Other serious programs mental illness

Statistical Analyses  Multivariable logistic regression at the individual level • Categorical regional HRR indicator • Adjust for sociodemographics, health status, regional/access-to-care covariates  Using marginal effect models to obtain the predicted probabilities of inappropriate prescription opioid use measures in each HRR  SAS 9.4, Stata 14.0, ArcGIS 10.4.1

Results

Potentially Inappropriate Prescription Opioid Use in Disabled Medicare, 2011-2013 35 % of disabled beneficiaries having High-dose 32.9 30 Multiple providers inappropriate opioid use 25 Concurrent benzodiazepine use* 20 15 10 8.9 8.8 8.6 5 5.2 5.1 4.6 0 2011 (N=114,696) 2012 (N=124,929) 2013 (N=135,252) Calendar Year (Number of Disabled Beneficiaries) *Centers for Medicare & Medicaid Services (CMS) began coverage for benzodiazepines in 2013

Potentially Inappropriate Prescription Opioid Use in Non-disabled Medicare, 2011-2013 18 % of disabled beneficiaries having High-dose 16.98 16 14 Multiple providers inappropriate opioid use 12 Concurrent benzodiazepine use* 10 8 6 4 1.27 1.28 1.25 2 0.77 0.76 0.81 0 2011 (N=132,654) 2012 (N=141,632) 2013 (N=157,389) Calendar Year (Number of Disabled Beneficiaries) *Centers for Medicare & Medicaid Services (CMS) began coverage for benzodiazepines in 2013

Selected Characteristics, Disabled Beneficiaries, 2013 High Multiple Concurrent All cohort dose providers benzodiazepine N 135,252 11,691 6,149 44,458 Age ≥65, % 28.9 12.2 9.6 23.3 Female, % 59.2 50.6 61.2 64.8 Race/ethnicity, % White 75.0 85.8 68.8 83.6 Black 19.1 9.9 25.8 12.0 Hispanic/others 5.9 3.3 5.5 4.4 Low-income subsidy, % 66.8 70.4 83.8 70.0 Dual eligibility, % 57.6 57.0 75.6 60.3 Musculoskeletal disorders, % 61.3 68.1 73.8 65.4 Depression, % 26.4 29.5 41.8 36.6 Other mental illness, % 9.0 7.9 19.4 13.8

High-dose, 2011-2013 2011 2012 2013

High dose, Top 10 HRRs 2012 Adjusted 2013 Adjusted 2011 Adjusted rates, % rates, % rates, % Sun City, AZ 16.6 Sarasota, FL 21.0 Sarasota, FL 22.6 Sarasota, FL 16.2 Ridgewood, NJ 18.2 Pueblo, CO 18.9 Pueblo, CO 16.9 Lawton, OK 15.5 Fort Lauderdale, FL 18.9 Pueblo, CO 15.5 Sun City, AZ 16.8 Ridgewood, NJ 16.9 New Brunswick, NJ 14.8 Fort Lauderdale, FL 16.8 Hudson, FL 16.9 Hudson, FL 15.9 Clearwater, FL 14.5 Clearwater, FL 16.8 Napa, CA 14.3 New Brunswick, NJ 15.7 New Brunswick, NJ 16.3 Fort Lauderdale, FL 14.0 Salisbury, MD 14.8 Morristown, NJ 16.3 Clearwater, FL 14.5 Mesa, AZ 13.7 Salisbury, MD 16.1 Knoxville, TN 13.6 Paterson, NJ 15.6 Medford, OR 14.5

Multiple Providers, 2011-2013 2011 2012 2013 2013

Multiple Providers, Top 10 HRRs 2013 Adjusted 2011 Adjusted 2012 Adjusted rates, % rates, % rates, % Arlington, VA 11.2 Arlington, VA 10.7 Minot, ND 10.8 Lake Charles, LA 9.6 Sun City, AZ 9.7 St. Cloud, MN 7.7 Takoma Park, MD 8.6 Provo, UT 9.0 Pueblo, CO 7.3 Anchorage, AK 7.2 Chattanooga, TN 8.2 Paterson, NJ 8.9 St. Paul, MN 6.9 Metairie, LA 8.2 Arlington, VA 8.3 Phoenix, AZ 6.8 Sun City, AZ 7.9 Washington, DC 7.4 Dubuque, IA 7.1 Norfolk, VA 7.3 Provo, UT 6.8 Lake Charles, LA 6.4 Temple, TX 6.9 Fort Wayne, IN 7.0 San Jose, CA 6.3 Fort Myers, FL 6.8 Honolulu, HI 7.0 Jacksonville, FL 6.1 Honolulu, HI 6.5 Newport News, VA 6.9

Concurrent Benzodiazepine Use, 2013

Concurrent Benzodiazepine Use, Top 10 HRRs 2013 Adjusted rates, % Slidell, LA 49.8 Miami, FL 49.5 Clearwater, FL 47.3 Detroit, MI 46.1 Paterson, NJ 46.0 Panama City, FL 45.5 Dearborn, MI 44.8 Hudson, FL 44.6 Lake Charles, LA 44.0 Spartanburg, SC 43.7

Discussion  Substantial HRR-level variation exists  Concurrent opioid and benzodiazepine use was common  Implications: Potential target interventions • Verification of appropriate diagnoses and treatment monitoring • Consultation with mental health or pain specialists • Prior authorization of long-term use o Pharmacy lock-in program

Limitations  Lack of socio-behavioral and other information (e.g., reason for use)  Uncertainty if the medications were actually taken by the patients  Substance use disorders claims were redacted by CMS  Limited generalizability to other populations

Thank you for your attention! Jenny Lo-Ciganic MSPharm, MS, PhD, lociganic@pharmacy.arizona.edu

28