Doctor Shopping Behavior and the Diversion of Opioid Analgesics: 2008-2012 A report prepared for The Executive Office of The President Office of National Drug Control Policy Ron Simeone Simeone Associates, Inc. November 5, 2014
Forward
Forward: General Considerations There are many ways to define prescription drug diversion. We define prescription drug diversion in terms of doctor shopping behavior. Doctor shopping occurs whenever prescriptions for the same class of drug overlap and involve several doctors and several pharmacies. Sometimes these overlapping prescriptions are obtained for personal (but nonmedical) use. Sometimes they are obtained by individuals who are sponsored by a drug dealer who accumulates the product for subsequent distribution on the black market. 3
Forward: General Considerations Our focus is on doctor shopping behavior involving prescription opioids that occurred during the period beginning January 1, 2008 and ending December 31, 2012. We include 21 opioids (generic and brand-name products) in our analysis. All of our findings are expressed in terms of morphine-equivalence. We consider 36 operational definitions ( contingencies ) for doctor shopping. These contingencies indicate the number of different doctors and the number of different pharmacies that have to be involved before we conclude that a prescription has been diverted. Trends in diversion are examined over time for the nation as a whole and by area (state and zip-three). 4
Sample Characteristics
Sample Characteristics: Base Year and Five Year Stability Samples The ability to detect doctor shopping behavior is a function of sample coverage. The more pharmacies included in the sample the greater the probability of detection. The ability to study change over time requires that pharmacies remain in play over the entire 2008-2012 period. It is not possible to maximize coverage and sample retention at the same time. Most pharmacies do not report consistently for five consecutive years. We solve this problem by working with two groups of pharmacies: a base year sample that maximizes detection capability and a five year stability sample that allows change to be measured over time. 6
Sample Characteristics: Base Year The base year sample includes 35,311 pharmacies that reported on at least 95% of their claims for all of calendar year 2012. This sample represents approximately 30% of all pharmacies in the United States. At the state level these pharmacies are associated with 60,732,837 unweighted prescriptions and with 265,644,177 weighted prescriptions. At the zip-three level these pharmacies are associated with 60,732,837 unweighted prescriptions and with 264,778,101 weighted prescriptions. 7
Sample Characteristics: Five Year Stability The five year stability sample includes 8,954 pharmacies that reported on at least 95% of their claims over the entire 2008-2012 period. This sample represents approximately 12% of all pharmacies in the United States. At the state level these pharmacies are associated with an average of 35,589,553 unweighted prescriptions per year and with an average of 257,483,435 weighted prescriptions per year. At the zip-three level these pharmacies are associated with an average of 35,589,553 unweighted prescriptions per year and with an average of 252,528,537 weighted prescriptions per year. 8
Defining Diversion
Defining Diversion: 36 Contingencies for Doctor Shopping There is no gold standard that would allow us to determine with certainty that a prescription is associated with diversion. Only a case investigation would suffice for this purpose. Instead diversion events are identified by the statistical improbability of their occurrence. This assessment is made by examining the frequency with which a given prescription overlaps with other prescriptions for the same class of drug and the numbers of different doctors and pharmacies that are involved in such an event. Between 1 and 6 doctors and 1 and 6 pharmacies can be involved in a diversion event. This gives us 6 X 6 = 36 possible contingencies. 10
Doctors Defining Diversion: 36 Contingencies for Doctor Shopping Upper and lower bounds for diversion are defined by identifying natural breaks in the probability of occurrence that are found when we examine the 36 alternative contingencies that might be associated with doctor shopping. The base year sample is used for this purpose. Pharmacies 1 2 3 4 5 6 1 35.2414% 5.4271% 0.4088% 0.0394% 0.0074% 0.0029% 2 10.5582% 5.3758% 0.5291% 0.0528% 0.0080% 0.0025% 3 0.5936% 0.5751% 0.4361% 0.0698% 0.0109% 0.0027% 4 0.0256% 0.0384% 0.0532% 0.0735% 0.0161% 0.0041% 5 0.0012% 0.0024% 0.0046% 0.0107% 0.0199% 0.0057% 6 0.0002% 0.0003% 0.0005% 0.0014% 0.0037% 0.0097% 11
Defining Diversion: Percent Prescriptions by Contingency Doctors 30% 25% 20% Pharmacies 1 2 3 4 5 6 1 35.2414% 5.4271% 0.4088% 0.0394% 0.0074% 0.0029% 2 10.5582% 5.3758% 0.5291% 0.0528% 0.0080% 0.0025% 3 0.5936% 0.5751% 0.4361% 0.0698% 0.0109% 0.0027% 4 0.0256% 0.0384% 0.0532% 0.0735% 0.0161% 0.0041% 5 0.0012% 0.0024% 0.0046% 0.0107% 0.0199% 0.0057% 6 0.0002% 0.0003% 0.0005% 0.0014% 0.0037% 0.0097% 15% 10% 5% 0% 12
Defining Diversion: 36 Contingencies for Doctor Shopping Doctors Although there is no gold standard that might be used to validate our upper and lower bounds there are other data associated with each contingency that can be used to assess the internal consistency of our approach. One of these is the percent prescriptions paid for in cash. Pharmacies 1 2 3 4 5 6 1 7.1224% 8.3357% 10.1394% 16.1205% 25.3897% 33.8797% 2 5.7227% 14.5310% 16.6602% 19.5302% 23.9745% 32.3696% 3 6.5073% 14.1181% 28.6430% 28.7404% 30.1163% 35.4826% 4 8.1378% 16.0988% 28.5177% 38.2046% 34.3256% 34.6054% 5 9.3825% 19.9974% 25.6874% 36.1667% 41.6399% 37.0712% 6 9.0058% 19.0510% 25.6900% 35.9401% 40.0167% 42.9791% 13
Defining Diversion: Percent Cash Payments by Contingency Doctors Pharmacies 1 2 3 4 5 6 1 7.1224% 8.3357% 10.1394% 16.1205% 25.3897% 33.8797% 2 5.7227% 14.5310% 16.6602% 19.5302% 23.9745% 32.3696% 3 6.5073% 14.1181% 28.6430% 28.7404% 30.1163% 35.4826% 4 8.1378% 16.0988% 28.5177% 38.2046% 34.3256% 34.6054% 5 9.3825% 19.9974% 25.6874% 36.1667% 41.6399% 37.0712% 6 9.0058% 19.0510% 25.6900% 35.9401% 40.0167% 42.9791% 40% 35% 30% 25% 20% 15% 10% 5% 0% 14
Change Over Time
Change Over Time: Percent Prescriptions Diverted Upper Bound Lower Bound Upper Bound Lower Bound Prescriptions (%) Prescriptions (%) Prescriptions (n) Prescriptions (n) 2008 1.7512 0.1927 4,295,445 472,782 2009 1.5368 0.1368 3,882,450 345,561 2010 1.3530 0.1071 3,532,179 279,637 2011 1.2795 0.0945 3,362,412 248,451 2012 1.2685 0.0834 3,369,660 221,665 3.0000 2.5000 2.0000 1.5000 1.0000 0.5000 0.0000 2008 2009 2010 2011 2012 Upper Bound Lower Bound 16
Change Over Time: Percent Milligrams Diverted Upper Bound Lower Bound Upper Bound Lower Bound Milligrams (%) Milligrams (%) Milligrams (n) Milligrams (n) 2008 2.9492 0.3825 6,551,225,405 849,599,352 2009 2.7431 0.2865 6,093,491,236 636,360,172 2010 2.4804 0.2299 5,509,900,609 510,609,476 2011 2.3393 0.2000 5,196,533,648 444,249,460 2012 2.1928 0.1568 4,871,138,710 348,246,842 3.0000 2.5000 2.0000 1.5000 1.0000 0.5000 0.0000 2008 2009 2010 2011 2012 Upper Bound Lower Bound 17
Change Over Time and Across Space
Change Over Time and Across Space: Percent Prescriptions Diverted by State (2008) 19
Change Over Time and Across Space: Percent Prescriptions Diverted by State (2012) 20
Change Over Time and Across Space: Percent Prescriptions Diverted by Zip-Three (2008) 21
Change Over Time and Across Space: Percent Prescriptions Diverted by Zip-Three (2012) 22
Change Over Time and Across Space: Percent Milligrams Diverted by State (2008) 23
Change Over Time and Across Space: Percent Milligrams Diverted by State (2012) 24
Change Over Time and Across Space: Percent Milligrams Diverted by Zip-Three (2008) 25
Change Over Time and Across Space: Percent Milligrams Diverted by Zip-Three (2012) 26
Implications
Implications: Trends in Context There appears to have been a constant and geographically widespread reduction in doctor shopping behavior over time. This may be related to the concerted efforts of government to reduce the scope and magnitude of the problem. Although nonmedical use has declined modestly in the general population, admissions to drug treatment with prescription opioids as the primary drug of abuse have increased markedly. This may be indicative of behavior on the part of a committed core of nonmedical prescription opioid users; or conversely, the result of reduced availability of such drugs on the blackmarket. 28
Future Directions
Future Directions: Definitions of Diversion by Extension Diversion within other classes of drugs in a manner consistent with the National Survey on Drug Use and Health (Pain Relievers, Stimulants, Sedatives, and Tranquilizers). Diversion of benzodiazepines (using equivalence factors similar to those applied to opioids). Diversion of combinations of drugs across classes (opioids and benzodiazepines). 30
Future Directions: More Comprehensive Definitions of Diversion Doctors Patients Pharmacies 31