Prescription Behavior Surveillance Using PDMP Data Dagan Wright, PhD, MSPH (Oregon Health Authority) Denise Penone, PhD (New York City Department of Health) Special thanks and acknowledgement to Len Paulozzi who could not attend as all contributors

Outline of the PDMP Talk What is PMP or PDMP? Why so important? What are general characteristics and data elements? What are questions that can be answered? Examples of data Examples of outreach and evaluation

What is PMP or PDMP? Tool utilized for reducing prescription drug misuse and diversion Drug Epidemic Warning System Drug Diversion & Fraud Investigative Tool Public Health Surveillance tool to collect, monitor, and analyze dispensing data Avoidance of Drug Interactions Patient Care Tool Identification & Prevention of “Doctor Shopping”* Data now can used to support states’ efforts in education, research, quality assurance (better healthcare), enforcement and abuse prevention Not meant to infringe on the legitimate prescribing of controlled substances *Doctor Shopping: Practice of obtaining multiple controlled substance prescriptions from multiple doctors Source: http://www.pmpalliance.org/content/prescription-monitoring-frequently-asked-questions-faq

Why so Important?

Opioid analgesic overdose deaths increased 65% Opioid analgesic overdose deaths, NYC, 2005-2011 2011 had the highest / Source: New York City Office of the Chief Medical Examiner & New York City Department of Health and Mental Hygiene 2005-2011 6

Oregon Drug Related Trends Counts and rates/100,00

As I mentioned, we are developing SAS code to identify the drugs involved using the death certificate data. We did a quick and dirty run using Oregon data with our program to find the drugs listed on the death certificates in Oregon. For anyone familiar with ICD-10 coded data, you will notice that we cannot identify some of these drugs using the ICD-10 coded data as they are including in some of the “other specified” categories. Such as “other specified synthetic narcotics” and “other opioids”. This graph shows the data for 2010 & 2011. You can see it is following as we expected. The heroin & cocaine up, methadone, methadmephetemine down a very little. Heroin & cocaine have their own ICD-10 code, but Methamphetamine is with T43.6 Psychostimulants with abuse potential It is possible to look specifically at fentanyl which will be important as there is perhaps some new non-pharmetical fentanyl. An CDC epi-aid showed 16 deaths in RI involving these drugs.

Methadone Death Rates Parallel Methadone Sales

More Drug Overdose Deaths than Motor Vehicle Crash Deaths Year Source: Oregon Vital Records

Oregon Hospitalization Rate/10,000 residents

What are General Characteristics and Data Elements?

PDMP: General Characteristics Typically require monthly or bi-weekly reporting Some States require weekly reporting i.e., Florida, Oregon Oklahoma, requires reporting at time of sale Reactive vs. Proactive Reactive: Generate solicited reports only in response to a specific inquiry Proactive: Generate unsolicited reports whenever suspicious or potentially at risk to the patient behavior is detected Drug Schedules Monitored by states: 24 collect Schedules II -V 17 collect Schedules II –IV 1 collect Schedule II only 2 collect Schedules II & III Source: http://www.simeoneassociates.com/simeone3.pdf Reactive: made by a prescriber, dispenser, or other party with appropriate authority Proactive: thus seen as a law enforcement tool Majority of states monitor classes 2 through 5

PDMP: Information Collected Patient identification Name & Address DOB & Gender Prescriber Information & Dispenser Information DEA number Drug Information National Drug Code (NDC) Info: Name Type Strength Manufacturer Quantity & date dispensed Source: http://www.pmpalliance.org/

PDMP Attributes As a Surveillance System Simplicity: single data source, few data elements, drug code (NDC) is complicated Flexibility: limited fields Data quality: insurance and system error checks Acceptability: mandatory See: Lee et al, eds., Principles and Practice of Public Health Surveillance, 3rd edition, 2010.

PDMP Attributes As a Surveillance System Sensitivity: high, required by law Predictive value positive: metrics untested Representativeness: population-based Timeliness: days to weeks Stability: in most cases operating for years Cost: support for many is inadequate for most PDMPs Other sources Oregon uses a provider licensing fee to support the PDMP See: Lee et al, eds., Principles and Practice of Public Health Surveillance, 3rd edition, 2010.

Model Act 2010 Revision Data Elements for PDMPs Prescription Number, Date issued by prescriber, Date filled, New or refill, Number of refills, State-issued serial number (optional) Drug NDC code for drug, Quantity dispensed, Days’ supply dispensed Comes from pmp model act 2010 revision document on Alliance website: http://www.pmpalliance.org/pdf/PMPModelActFinal20100628.pdf

Model Act 2010 Revision Data Elements for PDMPs Patient Identification number Name, Address, Date of birth, Sex Source of payment Name of person who receives prescription if other than patient Prescriber Dispenser Comes from pmp model act 2010 revision document on Alliance website: http://www.pmpalliance.org/pdf/PMPModelActFinal20100628.pdf

Descriptive Measures: Prescription Counts Specific compound, formulation Drug class Opioids, benzodiazepines, stimulants, etc. All extended-release formulations of opioids Class within a schedule, e.g., Schedule II opioids Daily dosage of an opioid prescription

Questions that can be Answered

Descriptive Measures: Denominators Person, e.g., rx per 1,000 people (most common) Patient, e.g., rx per 1,000 patients Prescriber, e.g., mean daily dose/prescriber Pharmacy, e.g., rx/pharmacy Time period is specified: e.g., in 2012, in past quarter

Descriptive Measures: “By” Variables Patient sex, age group Patient/prescriber/pharmacy by county or zip code Month, year (prescribed or dispensed) Prescriber specialty (requires linkage based on prescriber number) Source of payment (where collected) Patient type, e.g., opioid-naive

Risk Measures: Daily Dose for Opioids Converted to morphine milligram equivalents (MME) Usually categorized, e.g., High, e.g., >100 MME/day Going beyond specific dosing guidelines e.g., more than 30 mg of methadone per day for an opioid-naïve person Also quantified by measures of central tendency: mean, median , quartiles dose SAS coding to do MME conversions available from CDC

Examples of Data

Number of Patients Receiving Opioid Dosages > 100 MME/day, Tennessee, 2007‒2011 This graph shows the number of unique patients receiving a high dose of greater than 100 morphine milligram equivalents by year. The number of people who received greater than 100 morphine milligram equivalents on average per day for a year has increased from 2007 through 2011. Baumblatt J. Prescription Opioid Use and Opioid-Related Overdose Death TN, 2009–2010, CDC EIS Tuesday Morning Seminar, 1/8/2013

Opioid Prescriptions Filled by Staten Islanders Are More Frequently High Dose Schedule II opioids + hydrocodone, New York State Prescription Drug Monitoring Program

Number of people/1,000 residents receiving an opioid Oct 1, 2011 to March 31, 2012

Number of people/1,000 residents receiving an opioid and benzodiazepine Oct 1, 2011 to March 31, 2012

Number of people/10,000 residents using 4 or more prescribers and 4 or more pharmacies Oct 1, 2011 to March 31, 2012

Rates of Unintentional Poisoning Mirrors Rates of Dispensed Prescriptions Neighborhoods with Highest Rates of Opioid Prescriptions Also Have the Highest Rates of Overdose Deaths, 2008-2009 Of the five NYC neighborhoods with the highest rates of hydrocodone and/or oxycodone prescriptions filled,four were in Staten Island and overlapped with four of the five neighborhoods where the rate of unintentional opioid analgesic poisoning (overdose) deaths was highest during the years 2008-2009. Source: http://www.nyc.gov/html/doh/downloads/pdf/epi/epi-data-brief.pdf

Use of PMP Data by MA Dept. of Public Health “Shopping” as a portion of all prescriptions Overdoses in ED Data Slide provided courtesy of Peter Kreiner, PMP Center of Excellence at Brandeis. Doctor shopping, the questionable activity, was defined as 4+ prescriber s and 4+ pharmacies for CSII in six months.

Measures of “Shopping” or “Multiple Provider Episodes” Author (year) Drug No. of Prescribers Pharmacies Rx Overlap Time Period Hall (2008) Any CS 5+ NA 1 yr Peirce (2012) 4+ 6 mo Ohio DOH (2010) Opioid Avg of 5+ Over 3 yrs Gilson (2010, 2012) “Same medication” 2+ 30 d Katz (2010) Any CSII Cepeda (2012) 3+ 1+ day 18 mo BJA criteria CSII-IV 3 mo. No standard. Not like 5 or more drinks per day. Sensitivity and specificity not determined, but measures have been associated with abuse or overdose Shorter time periods might be desirable if looking for short-term impact of an intervention. More specific measures might be chosen in response to limited resources for followup.

Patient vs. Provider Metrics? Top 1% of prescribers based on number of prescriptions might account for 33% of the morphine equivalents (MME) in your state.(1) Top 1% of patients might account for 40% of MME.(2) 1. Swedlow 2011; 2. Edlund 2010

15% of prescribers write 82% of opioid analgesic prescriptions Prescriptions filled by NYC residents, 2010 15% 82% Percent -more prescription data from PDMP PH and PS go after this a little differently -X axis shows prescriber on left and rx on right. Y axis percent. -point: 15% of prescribers write 82% of opioid analgesic rx. -we can then use data to better understand high volume prescribers. Source: New York State Department of Health, Bureau of Narcotic Enforcement, Prescription Drug Monitoring Program, 2008-2010

Distribution of CS II-IV prescriptions to prescribers, Oregon, 1/12 to 9/12 % of Prescribers % of CS Prescriptions Oregon Health Authority. Prescription Drug Dispensing in Oregon, October 1, 2011 – March 31, 2012

Examples of Outreach and Evaluation

Patient vs. Provider Metrics? 100 patients in the PMP for every prescriber It takes roughly 100 times more effort to address the same fraction of problematic prescriptions. For interventions, provider case-finding is preferred based on efficiency.

1st Evaluation of Oregon PDMP soon followed by NIH study – survey use 65% say it is very helpful to monitor patients’ prescriptions for controlled substances 64% report it is very helpful to control “doctor shopping” 78% have spoken with patient about controlled substance use after using system 59% reduced or eliminated prescriptions for a patient after using system 49% contacted other providers or pharmacies Source: Oregon Prescription Drug Monitoring Program Evaluation

NYC Opioid Treatment Guidelines Avoid prescribing opioids for chronic non-cancer, non-end-of-life pain E.g. low back pain, arthritis, headache, fibromyalgia When opioids are warranted for acute pain, 3-day supply usually sufficient Avoid whenever possible prescribing opioids in patients taking benzodiazepines If dosing reaches 100 MED, reassess and reconsider other approaches to pain management Clinical experts in primary care, rheumatology, psychiatry, emergency medicine, and pain management, including Theodore Strange and Mark Jarrett from Staten Island University Hospital and Anne Marie Stilwell from Interventional Pain Management of Staten Island, helped us develop these guidelines.

References Cited Cepeda, M., D. Fife, et al. (2012). "Assessing opioid shopping behavior." Drug Safety. Edlund, M. J., B. C. Martin, et al. (2010). "Risks for opioid abuse and dependence among recipients of chronic opioid therapy: results from the TROUP study." Drug Alcohol Depend 112(1-2): 90-98. Forrester, M. B. (2011). "Ingestions of hydrocodone, carisoprodol, and alprazolam in combination reported to Texas poison centers." Journal of Addictive Diseases 30: 110-115. Hall, A. J., J. E. Logan, et al. (2008). "Patterns of abuse among unintentional pharmaceutical overdose fatalities." JAMA 300: 2613-2620. Katz, N., L. Panas, et al. (2010). "Usefulness of prescription monitoring programs for surveillance---analysis of Schedule II opioid prescription data in Massachusetts, 1996--2006." Pharmacoepidemiol Drug Safety 19: 115-123. Ohio Department of Health. (2010). "Epidemic of prescription drug overdoses in Ohio." Retrieved September 1, 2010, from http://www.healthyohioprogram.org/diseaseprevention/dpoison/drugdata.aspx. Peirce, G., M. Smith, et al. (2012). "Doctor and pharmacy shopping for controlled substances." Med Care. Swedlow, A., J. Ireland, et al. (2011). Prescribing patterns of schedule II opioids in California Workers' Compensation, California Workers' Compensation Institute. White, A. G., H. G. Birnbaum, et al. (2009). "Analytic models to identify patients at risk for prescription opioid abuse." Am J Manag Care 15(12): 897-906. Wilsey, B. L., S. M. Fishman, et al. (2010). "Profiling multiple provider prescribing of opioids, benzodiazepines, stimulants, and anorectics." Drug Alcohol Depend 112: 99-106.