[PPT] - Analytical Models Using PDMP Data Svetla Slavova, PhD Huong Luu, PowerPoint Presentation

SLIDE 1

Data-Driven Multidisciplinary Approaches to Reduce Prescription Drug Abuse in Kentucky

Analytical Models Using PDMP Data

Svetla Slavova, PhD Huong Luu, MD, MPH 2017 BJA Grantee National Meeting March 30-31, 2017

SLIDE 2

Study 1: Trends and Disparities of Opioid Prescribing in Kentucky (preliminary results)

SLIDE 3

Historically High Opioid Prescribing in Kentucky

SLIDE 4

McDonald DC, Carlson K, Izrael D. Geographic variation in opioid prescribing in the U.S. J Pain. 2012;13(10):988-96. doi:10.1016/j.jpain.2012.07.007.

SLIDE 5

Rate of Patients with Opioid Prescriptions by County, Kentucky, 2012

SLIDE 6

Slavova, S., Bunn, T., & Gao, W. (2015). Drug Overdose Hospitalizations in Kentucky, 2000-2013. KIPRC. Retrieved from http://www.mc.uky.edu/kiprc/projects/ddmarpdak/pdf/Drug_Overdose_Hospitalizations_2000-2013.pdf

SLIDE 7

Table 1: Characteristics of drug intoxication deaths, adult Kentucky residents, 2013-2014a. Demographic All Drugs By Involved opioid (Rate per 100,000) Characteristics

No. (%)

Rate per 100,000 Heroin- related N=415 Fentanyl- related N=151 Pharmaceutical

pioids

excluding morphine and fentanyl N=634 All 1,971 (100) 29 6.1 2.2 9.3 Gender Males 1,209 (61) 37 8.8 3.0 10.4 Females 762 (39) 22 3.2 1.5 8.2 Race Black 69 (4) 13 3.1c * 2.8 c White 1,885 (96) 31 6.4 2.5 10.0 Other 17 (<1) 14 * * * Age groups 18-24 131 (7) 15 7.4 1.3c 2.3 25-34 403 (20) 36 13.1 3.6 9.8 35-44 518 (26) 46 8.4 3.5 15.1 45-54 540 (27) 44 6.3 3.0 15.3 55-64 313 (16) 27 2.6 1.7 10.4 65+ 66 (3) 5 * * 2.0 Region Appalachian counties 626 (32) 34 2.0 2.4 16.4 Non-Appalachian counties 1,345 (68) 27 7.6 2.2 6.7 Marital status Divorced 662 (34) Single 643 (33) Married 528 (27) Widowed 95 (5) Not classified 43 (2) Place of death ED/outpatient 348 (18) Inpatient 190 (10) Residence 1,078 (55) Other 355 (18) Manner of deaths Unintentional 1,750 (89) Suicide/Self-harm 105 (5) Undetermined 116 (6)

a Data were provisional at the time of the analysis. b The reported rates could be different from rates reported elsewhere because the study population was limited to

adult Kentucky residents (i.e., 18 years of age or older) who died in Kentucky from drug intoxications/poisonings.

c Rates were marked as unreliable when they were based on less than 20 deaths

* Rates based on less than 10 counts were suppressed according to the state data reporting policy.

Slavova S, Bunn TL, LaMantia S, Corey T, Ingram V., Linking death certificates, post-mortem toxicology, and prescription history data for better identification of populations at increased risk for drug intoxication deaths, Pharmaceutical Medicine, 2017, DOI: 10.1007/s40290-017-0185-7

SLIDE 8

Research Questions:

1) Are there any positive changes in opioid prescribing in Kentucky in recent years?

Decline in the number/rate of patients with opioid prescriptions?
Decline in the number/rate of patients with high daily dose of morphine

milligram equivalent? 2) Are there significant differences in opioid prescribing for Appalachian vs. Non-Appalachian Kentucky residents?

SLIDE 9

Method: Repeated Measures Regression

Negative Binomial Generalized Estimating Equations (GEE) Modeling

Variables Data Sources Count Outcome Variables (unit of analysis: county-quarter) (1) Number of patients with opioid prescriptions* (Offset=Log of population) KASPER/Office of Inspector General (OIG), 2012 – 2015 (2) Number of patients receiving more than three days with 100 MME or more* (Offset=Log of population) KASPER/OIG, 2012 - 2015 (3) Number of patients receiving more than three days with 100 MME or more* (Offset= Log of patients with opioid prescriptions*) KASPER/OIG, 2012 - 2015 Covariates: (1) Region: Appalachian vs. Non-Appalachian (2) Rate of ED visits due to injury excluding poisoning per 1,000 population Outpatient Claims Data, Office

f Health Policy, 2012 - 2015

(3) Age-adjusted rate of invasive cancer incidence per 100,000 population (average rate over previous two years ) Kentucky Cancer Registry, 2010 - 2014 (4) Rate of buprenorphine prescriptions (per 1,000 population) KASPER/OIG, 2012 - 2015 (5) Percent of persons ages 55+ with opioid prescriptions* KASPER/OIG, 2012 - 2015

* Excluding prescriptions of buprenorphine or buprenorphine/naloxone combination

SLIDE 10

GEE model: Number of patients with opioid prescriptions Offset: Log of population

RRQ1-12=1.25 (1.16 -1.36)

RRQ1-12=1.25 (1.16 -1.36) RRQ4-15=1.23 (1.14 -1.32)

SLIDE 11

SLIDE 12

SLIDE 13

GEE model: Number of patients with high daily MME dose Offset: Log of population

RRQ1-12=1.35 (1.20 -1.52) RRQ4-15=1.09 (0.96 -1.23)

SLIDE 14

SLIDE 15

SLIDE 16

GEE model: Number of patients with high daily MME dose Offset: Log of patients with opioid prescriptions

RRQ1-12=1.10 (1.002 -1.21) RRQ4-15=0.92 (0.83 -1.01)

SLIDE 17

SLIDE 18

SLIDE 19

http://www.chfs.ky.gov/os/oig/kaspertrendreports

SLIDE 20

SLIDE 21

SLIDE 22

SLIDE 23

SLIDE 24

If the ACME is 100 or greater, a

warning symbol will appear along with a note that increased clinical vigilance may be appropriate.

This warning threshold was

established by consensus of the KASPER Advisory Council members based on a recommendation from the Kentucky Injury Prevention and Research Center.

For patients with an ACME of 100 or

greater, the last page of the report will also include information and links to additional resources about naloxone prescribing and dispensing to help in situations where a provider believes the patient may be at risk of an overdose.

KASPER Tips: Morphine Equivalent Dose and Naloxone Information on KASPER Reports, http://www.chfs.ky.gov/os/oig/KASPERtips

The Kentucky Board of Medical Licensure advises that when a patient’s MED level reaches the 100 threshold,

prescribers are expected to increase safeguards (such as increased monitoring and the use of naloxone) and that

ngoing treatment be supported by increased documentation of clinical reasoning.

SLIDE 25

Study 2: Transition from non-medical prescription opioid use to heroin among Kentucky residents who died from heroin-related intoxication deaths

SLIDE 26

Data sources:

Linked death certificates, post-mortem toxicology, and PDMP data for 2013-2014

Kentucky overdose decedents; Results:

Kentucky heroin intoxication decedents had substantial exposures to legally

acquired prescription opioids in the years before their death;

91% of decedents had PDMP records for opioid analgesics (excluding

buprenorphine) during the 5 years before death;

77% had opioid prescriptions filled within two-years before death;
20% had a history of continuous opioid use (COU) during the two years before

death (COU defined as at least 90 days’ supply in a six-month period, and with gaps between prescriptions of no more than 30 days);

41% did not meet the criteria for COU but had a history of high-dose opioid use

(≥50 MME daily).

SLIDE 27

Study 3: Ecological analysis of fentanyl overdose deaths

SLIDE 28