Doctor shopping reveals geographical variations in opioid abuse

Pain Physician 2013; 16:89-100 • ISSN 1533-3159 Epidemiologic Assessment Doctor Shopping Reveals Geographical Variations in Opioid Abuse Sandra Nordmann, MSc1, Vincent Pradel, MD, PhD2, Maryse Lapeyre-Mestre, MD, PhD3, Elisabeth Frauger, PharmD, PhD1, Vanessa Pauly, PhD2, Xavier Thirion, MD, PhD2, Michel Mallaret, PhD4, Emilie Jouanjus, PharmaD3, and Joëlle Micallef, MD, PhD1 From: 1Centre d’Evaluation et d’Information de la PharmacodépendanceAddictovigilance PACA-Corse, hôpital Timone, Pharmacologie clinique, Institut des Neurosciences de la Timone, Faculté de médecine, Aix Marseille Université UMR 7289 CNRS, Marseille, France; 2Centre d’Evaluation et d’Information de la PharmacodépendanceAddictovigilance PACA-Corse, Centre Associé, hôpital Sainte Marguerite, Laboratoire de Santé Publique, Faculté de médecine, EA 3279, Marseille, France; 3Centre d’Evaluation et d’Information de la PharmacodépendanceAddictovigilance Midi-Pyrénées, Service de Pharmacologie Clinique, Hôpitaux de Toulouse, UMR INSERM 1027, Unité de Pharmacoépidémiologie, Université de Toulouse, France; 4Centre d’Evaluation et d’Information de la PharmacodépendanceAddictovigilance, Centre hospitalouniversitaire de Grenoble, France Address Correspondence: Joëlle Micallef, MD, PhD Centre d’Evaluation et d’Information de la PharmacodépendanceAddictovigilance PACA-Corse Pharmacologie Clinique Hôpital Timone 264, rue Saint Pierre, 13385 Marseille Cedex 5 France E-mail: joelle.micallef@ap-hm.fr Background: Prescription opioid abuse is not homogeneous due to varying patterns of use and different geographic preferences. Because doctor shopping is one of the main sources of diversion, it has previously been used to estimate drug abuse. Objectives: The aim of this study was to describe and compare opioid abuse in 2008 using doctor shopping to estimate abuse in 3 French regions. Setting: Data for this study came from the General Health Insurance (GHI) reimbursement database, which covers 77% of the French population. All individuals living in ProvenceAlpes-Côte d’Azur-Corse (PACA), Rhône-Alpes (RA), or Midi-Pyrénées (MP) that received at least one reimbursement for oral opioids from the GHI in 2008 were included. Methods: Oral opioids under study were opioids for mild to moderate pain (dextropropoxyphene, codeine, tramadol, dihydrocodeine), opoids for moderately severe to severe pain (oral morphine, oxycodone, buprenorphine painkiller, hydromorphone), and opioid maintenance treatments (buprenorphine maintenance, methadone). For a given opioid, the Doctor Shopping Quantity (DSQ) is the quantity obtained by overlapping prescriptions from several prescribers. It is used to estimate the magnitude of abuse. The Doctor Shopping Indicator (DSI) is the DSQ divided by the total dispensed quantity. It is used to estimate the abuse corrected for use. Results: The total DSQ for opioids in PACA (213.3 DDD/1,000 inhabitants) was twofold superior to that in RA (115.1 DDD/1,000) and in MP (106.2 DDD/1,000). The DSQ of opioids for mild to moderate pain was 75.5DDD/1000 (DSI=1.1%), 19.7DDD/1,000 (DSI=5.0%) for opioids for moderately severe to severe pain, and 55.3DDD/1,000 (DSI=6.2%) for opioid maintenance treatments. Emergent signals of abuse have been observed at a regional level for oxycodone in MP and dihydrocodeine in RA and MP. Limitations: The main limitation of this study is that the GHI reimbursement database provides information about dispensed and reimbursed prescription drugs, and not necessarily the actual quantity used. Disclaimer: There was no external funding in the preparation of this manuscript. Conflict of interest: None. Conclusion: These results confirm important variations in the 3 French regions despite them being geographically close. Besides, they highlight different rates of opioid abuse between opioids for mild to moderate pain, opioids for moderately severe to severe pain, and opioid maintenance treatments, as well as differences within these groups. Manuscript received: 06-13-2012 Revised manuscript received: 08-13-2012 Accepted for publication: 09-11-2012 Key words: Prescription drug abuse, Opioid abuse, Prescription opioid analgesics, opioids for mild to moderate pain , Opioids for moderately severe to severe pain, Opioid maintenance treatments, Prescription drug database, Doctor shopping Free full manuscript: www.painphysicianjournal.com Pain Physician 2013; 16:89-100 www.painphysicianjournal.com Pain Physician: January/February 2013; 16:89-100 O ver the past 10 years the therapeutic use of opioids has escalated as has their abuse and non-medical use (1). However the public health impact of non-medical use and abuse of prescription opioids is not homogeneous due to varying patterns of use and different geographic preferences (2-8). Evaluating opioid abuse at a regional level may facilitate the detection of an emergent medication abuse problem that is restricted to one area before it spreads to other areas. Such an approach may optimize the local intervention strategies due to a better knowledge of determinants involved in abuse and non-medical use such as population characteristics or product availability (3,9). In order to identify product availability, some studies have focused on the key diversion routes of prescription opioids and shown that the 2 main sources were friends or family and prescription or doctor shopping (10-13). Doctor shopping is when a patient consults several prescribers over the same period of time and thus obtains overlapping prescriptions (14-16). This behavior has been linked to substance abuse-related deaths in Australia (17) and in Ontario (18). Since the establishment of prescription drug monitoring programs in the US, this behavior can be identified (19-22). It has thus become a focus for clinical practice and authorities (23,24). Some years ago, a method that quantifies doctor shopping using the General Health Insurance (GHI) reimbursement database was developed to give the doctor shopping indicator (14,25,26). This quantitative assessment was used to estimate the magnitude of buprenorphine diversion (25) and to assess the impact of a national prescription drug monitoring program for buprenorphine (14). Recently, 2 other studies assessed the relative abuse potential of benzodiazepines in reallife settings using the doctor shopping indicator (26,27). Even if some studies concerning doctor shopping have been published (15,16,27,28), few involved geographic information which may give a better comprehension of doctor shopping behavior (15). Product availability is an important determinant of opioid misuse (4,9). The consumption of opioids has increased in France (29) as in other European countries (30,31) raising concerns about their misuse. Therefore this work focused on opioids including opioid analgesics and opiate maintenance treatments. In this context, we performed a study based on 3 regions in the south of France: Provence-Alpes-Côte d’Azur-Corse (PACA), Rhône-Alpes (RA), and MidiPyrénées (MP) which represented a total of 14 million people in 2008. 90 The main objective of this study was to describe and compare opioid use and abuse using doctor shopping to estimate the abuse over a one year period (2008) in 3 French regions. METHODS Settings Data for this study came from the GHI reimbursement database. The GHI is a public insurance system, which covers 77% of the French population. The remaining part of the French population is insured by other public insurance systems (32). It should be noted that in France, medication is dispensed in a pharmacy and then reimbursed by the GHI, either to the patient or directly to the pharmacist. Everyone covered by the GHI in 2008 in the PACA (4,054,669), RA (4,732,936), and MP (1,980,913) regions was included. In PACA, RA, and MP, there were respectively 47, 36, and 27 care centers dedicated to drug users. This study analyzed, for every insured inhabitant of these regions, all oral and sublingual forms of prescription opioids dispensed and sent for reimbursement between January 1, 2008 and December 31, 2008. Medications dispensed in hospitals were not included in the GHI reimbursement database. Included medications were oral opioids for mild to moderate pain (codeine combinations [N02AA59], dextropropoxyphene combinations [N02AC54], dihydrocodeine [N02AA08], tramadol as a single-ingredient drug [N02AX02] or combination [N02AX52]), oral opioids for moderately severe to severe pain (buprenorphine painkiller [N02AE01], hydromorphone [N02AA03], immediate and sustained release oral morphine and morphine syrup [N02AA01], immediate and sustained release oxycodone [N02AA05]), and oral opioid maintenance treatments (methadone syrup, methadone tablets [N07BC02] and buprenorphine used as maintenance treatment [N07BC01]). Fentanyl was not included because no oral form was available in 2008 in France. Five variables were extracted: the date of dispensing, the CIP code (drug box identification code, which is a French equivalent to the national drug code in the USA), the patient’s anonymous number, the prescriber’s anonymous number, and the quantity of reimbursed medication given as defined daily doses (DDD). The DDD is the assumed average maintenance dose per day for a drug used according to its main indication in adults; DDD are defined by the World Health Organization (WHO) Collaborating Centre for Drug Statistics www.painphysicianjournal.com Geographical Variations of Opioid Abuse Methodology, according to the ATC (Anatomical Therapeutic Chemical-code) classification index. One purpose of the ATC/DDD system is to allow comparison of drug consumption statistics at an international level. We used the 2010 version of this index (WHO, 2010) (33). Calculation of Doctor Shopping Quantity (DSQ) The principle of DSQ calculation is based on the number of overlaps of different prescribers’ prescriptions for a given patient. This is illustrated in the appendix with an example of a fictitious patient with 2 prescribers. A prescription period is defined for each prescriber/patient couple as the period between the first and the last observed dispensing. This prescription period is not necessarily continuous and may be interrupted. For instance the patient may consult another prescriber if the regular prescriber is on holiday. So when the interval between 2 consecutive dispensings is superior to a threshold, the prescription period is declared interrupted. This threshold is defined as the eightieth percentile of the observed intervals between 2 consecutive dispensings for all prescriber/patient couples. The threshold is calculated separately for each region and for each medication. In the doctor shopping method, it is assumed that within the quantity obtained by multiple prescribers during overlapping prescription periods, a certain proportion is medically legitimate. For instance, in the case of overlapping prescription periods from 3 different prescribers, it is assumed that one-third of the total quantity is medically legitimate and the remaining twothirds are obtained using doctor shopping. Therefore, the DSQ is computed for each patient using the formula: tor shopping quantities of all patients. It reflects the magnitude of abuse. It is given in DDD/1,000 inhabitants covered by the GHI per year (DDD/1000) to allow geographical comparison. The Doctor Shopping Indicator (DSI) is the DSQ divided by total dispensed quantity and reflects the abuse corrected for use. The DSI is considered clinically significant over 1% (27,34). Below this value, we consider that there is no signal of abuse. Separate analyses were conducted on each medication and each region. Results were computed using SPSS V13.0®. RESULTS Opioid User Population The number of individuals that received at least one dispensing of oral opioids reimbursed by the GHI in 2008 was 885,941 in PACA (21.8% of the insured population), 945,102 in RA (20.0% of the insured population), and 386,834 in MP (19.5% of the insured population). The male/female ratio was 0.43 in PACA, 0.45 in RA, and 0.44 in MP. The proportion of individuals under 30 years old was 19% in PACA, 18% in RA, and 19% in MP. The proportion of individuals over 60 year -old was 30% in PACA, 31% in RA, and 29% in MP. Thus, in the 3 regions studied, there was very little difference observed in the general profile of opioid users. Product-Specific Analysis Dispensed Quantity For the 3 regions taken together, opioids for mild to moderate pain represented 83.8% (n = 70, 388, 614 DDD) of the total dispensed quantity of opioids, opioids for moderately severe to severe pain represented 5.0% (n = 4, 120, 808 DDD) and opioid maintenance treatments represented 11.2% (n = 9, 536, 221 DDD). The total dispensed quantities in 2008 in PACA, RA, and MP are presented in Table 1. 2.2. Doctor Shopping Quantity where ni is the number of simultaneous prescription periods at the date of dispensing i and Qi the quantity dispensed. When there is no overlap between prescription periods of several prescribers for a patient (one or several prescribers with non overlapping prescriptions), ni=1 for all dispensings and therefore DSQ is null. For a population, the total DSQ is the sum of doc- www.painphysicianjournal.com The total DSQ for all oral opioids represented 150.5 DDD/1,000. Opioids for mild to moderate pain represented 50.2% (75.5 DDD/1,000) of the total DSQ for oral opioids, opioids for moderately severe to severe pain represented 13.1% (19.7 DDD/1,000) and opioid maintenance treatments represented 36.7% (55.3 DDD/1,000) (Table 2). 91 Pain Physician: January/February 2013; 16:89-100 Table 1. Total dispensed quantity of oral opioids dispensed Provence-Alpes Côte-d’Azur Corsica, Rhône-Alpes and Midi-Pyrénées in 2008. PACA RA MP Dispensed quantity (DDD) Number of users Dispensed quantity (DDD) Number of users Dispensed quantity (DDD) Number of users Weak opioid analgesics Codeine combinations 3 976 731 88 529 5 785 583 123 540 2 532 913 75 964 Dextropropoxyphene combinations 14 931 528 652 785 15 771 840 656 949 4 789 028 238 685 41 920 375 40 448 319 70 336 252 Dihydrocodeine Tramadol 8 999 318 32 0021 10 137 602 340 129 3 811 368 145 749 Tramadol alone 5 382 128 107 700 6 196 532 147 872 2 329 906 47 304 Tramadol combinations 3 617 190 212 321 3 941 070 229 868 1 481 462 98 445 33 290 861 28 403 723 10 850 291 Strong opioid analgesics Buprenorphine painkiller Hydromorphone 38 878 330 37 719 267 20 622 138 1 447 789 18 216 1 452 723 22 871 474 215 9339 Morphine SR 1 157 120 10 860 1 111 412 13 720 356 132 5616 Morphine IR 279 723 12 224 325 151 15 656 111 369 6124 Morphine Syrup 10 946 644 16 160 915 6 714 396 254 651 2856 211 943 2289 109 725 1492 Oral morphine Oxycodone Oxycodone SR 191 358 2219 163 460 1745 85 557 1095 Oxycodone IR 63 293 1910 48 483 1631 24 168 1032 Opioid maintenance treatments Buprenorphine maintenance 2 885 892 8137 2 660 504 10 148 1 152 769 4117 Methadone 1 169 124 2421 1 064 538 2306 603 393 1260 Methadone syrup 1 058 762 2358 995 821 2280 550 754 1248 Methadone tablet 110 363 491 68 718 293 52 639 200 Doctor Shopping Indicator Doctor Shopping Quantity Opioids with the highest DSI were buprenorphine maintenance (8.0%), oral morphine (5.5%), dihydrocodeine (3.7%), buprenorphine painkiller (2.9%), and oxycodone (2.7%) (Table 2). The total DSQ for opioids was 213.3 DDD/1,000 in PACA, 115.1 DDD/1,000 in RA, and 106.2 DDD/1,000 in MP. According to Fig. 2, the 5 medications with the highest DSQ were buprenorphine maintenance (first in all regions), dextropropoxyphene (second in PACA and RA, and fourth in MP), codeine (second in MP, third in RA, and fifth in PACA), tramadol (third in PACA and MP, and fourth in RA) and oral morphine (fourth in PACA and fifth in RA and MP). PACA was the region with the highest DSQ for all medications except for oxycodone and dihydrocodeine, for which MP had the highest DSQ. Region Specific Analysis Dispensed Quantity PACA was the region with the highest total dispensed quantity of opioids per 1,000 insured inhabitants (8331 DDD/1,000), followed by RA (8030 DDD /1,000) and MP (6853 DDD/1,000). As shown in Fig. 1, PACA was the region with the highest dispensed quantity for each medication except for codeine (for which RA had the highest dispensed quantity), methadone, dihydrocodeine, and hydromorphone (for which MP had the highest quantities). 92 Doctor Shopping Indicator As shown in Fig. 3, PACA had the highest DSI for all opioids except oxycodone (for which MP had the high- www.painphysicianjournal.com Geographical Variations of Opioid Abuse Table 2. Dispensed quantity, doctor shopping quantity and doctor shopping Indicator of oral opioids in Provence-Alpes Côte-d’Azur Corsica, Rhône-Alpes and Midi-Pyrénées in 2008. Dispensed quantity (DDD/1000) Doctor Shopping Quantity (DDD/1000) Doctor Shopping Indicator (%) Weak opioid analgesics 6640 75.5 1.1% Dextropropoxyphene 1199 27.6 0.8% Codeine 3296 24.1 2.0% Tramadol 2131 23.3 1.1% 14 0.5 3.7% Strong opioid analgesics 395 19.7 5.0% Dihydrocodeine Oral morphine 324 17.8 5.5% Oxycodone 56 1.5 2.7% Buprenorphine painkiller 7 0.2 2.9% Hydromorphone 9 0.2 1.8% 891 55.3 6.2% Buprenorphine maintenance 626 50.3 8.0% Methadone 265 4.9 1.9% Opioid maintenance treatments DDD/1000 insured inhabitants 0 500 1000 1500 2000 2500 3000 3500 4000 3683 Dextropropoxyphene 3332 2418 2219 2142 Tramadol 1924 981 Codeine 1353 1279 712 570 582 Buprenorphine maintenance 357 330 239 Morphine 288 229 305 Methadone Oxycodone 63 49 55 Dihydrocodeine 10 9 35 Hydromorphone 10 9 10 Buprenorphine painkiller Provence-Alpes-Côte d'AzurCorse Rhône-Alpes Midi-Pyrénées 8 6 5 Fig. 1. Dispensed quantity in DDD/1000 of oral opioids Provence-Alpes Côte-d’Azur Corsica, Rhône-Alpes and Midi-Pyrénées in 2008. www.painphysicianjournal.com 93 Pain Physician: January/February 2013; 16:89-100 DDD/1000 insured inhabitants 0 10 20 30 40 50 60 70 80 90 83.7 Buprenorphine maintenance 25.6 32.1 36.6 Dextropropoxyphene 24.0 17.7 29.0 Tramadol 20.0 19.6 28.3 Morphine 11.3 12.1 26.2 22.8 22.6 Codeine 7.0 3.6 4.0 Methadone Oxycodone Buprenorphine painkiller 1.6 1.0 2.7 Provence-Alpes-Côte d'AzurCorse 0.4 0.1 0.1 Hydromorphone 0,3 0.03 0.1 Dihydrocodeine 0.2 0.4 1.7 Rhône-Alpes Midi-Pyrénées Fig. 2. Doctor shopping quantity of oral opioids in DDD/1000 Provence-Alpes Côte-d’Azur Corsica, Rhône-Alpes and Midi-Pyrénées in 2008. Doctor Shopping Indicator (%) 0% 2% 4% 6% 8% 10% 12% 14% 11.8 Buprenorphine maintenance 5.6 4.4 Morphine 7.9 3.4 5.1 4.8 Buprenorphine painkiller Hydromorphone 1.4 1.3 0.4 3.6 1.0 2.7 Codeine 1.7 1.8 2.0 Oxycodone Methadone 1.6 1.3 1.6 Dextropropoxyphene 4.8 2.4 Dihydrocodeine Tramadol 2.5 4.2 4.8 1.3 0.9 1.0 1.0 0.7 0.7 Provence-Alpes-Côte d'AzurCorse Rhône-Alpes Midi-Pyrénées Fig. 3. Doctor shopping indicator of oral opioids Provence-Alpes Côte-d’Azur Corsica, Rhône-Alpes and Midi-Pyrénées in 2008. 94 www.painphysicianjournal.com Geographical Variations of Opioid Abuse est DSI) and dihydrocodeine (for which RA and MP had higher DSI). In each region, the opioids with the highest DSI were buprenorphine maintenance in PACA (11.8%) and RA (5.6%) and oral morphine in MP (5.1%). Oral morphine had the second highest DSI in PACA (7.9%) and the third in RA (3.4%). Oxycodone was second in MP (4.8%), fourth in RA (2.0%) and sixth in PACA (2.5%). Dihydrocodeine was eighth in PACA (1.6%), second in RA (4.2%), and second in MP (4.8%). DISCUSSION The purpose of this study was to assess the geographical variations of opioid use and abuse in 3 French regions using doctor shopping to estimate abuse. Opioid abuse is a major public health issue, as one fifth of the population received at least one opioid in our study. The key findings of this study were that the total opioid DSQ per inhabitant of PACA (213.3 DDD/1,000) was twofold superior to that in RA (115.1 DDD/1,000) and in MP (106.2 DDD/1,000). The DSQ of opioids for mild to moderate pain was 75.5 DDD/ 1000 (DSI = 1.1%), 19.7 DDD/1,000 (DSI = 5.0%) for opioids for moderately severe to severe pain, and 55.3 DDD/1000 (DSI = 6.2%) for opioid maintenance treatments. Regional specificities were observed, such as the emergence of oxycodone abuse in MP and dihydrocodeine abuse in RA and MP. Geographically-specific Analysis Despite a comparable global level of opioid use across the 3 regions (approximately 8000 DDD/1,000 in PACA and RA, 7,000 DDD/1,000 in MP), the total opioid DSQ per inhabitant of PACA (213.3 DDD/1,000) was twofold superior to that in RA (115.1DDD/1000) and in MP (106.2DDD/1000). Moreover, PACA was the region with the highest DSI for all opioids except oxycodone (higher in MP) and dihydrocodeine (higher in RA and MP). A parallel could be drawn with socio-demographic and economic data presented in Table 3 (35). Indeed, several indicators, such as the number of crimes and offences/1,000 inhabitants, the proportion of the population living in difficult urban areas, the poverty rate, the unemployment rate, and the proportion of individuals covered by the universal complementary health insur- Table 3. Socio-demographic and economic characteristics of the general population living in 2008 in Provence-Alpes Côte-d’Azur Corsica, Rhône-Alpes and Midi-Pyrénées PACA RA MP 5 185 879 6 117 200 2 838 228 Gender (% of women) 52 51 51 Age>20 (%) 23 26 23 Age<60 (%) 25 21 25 Density (inhabitants/km²) 157 141 63 Urbanization indicator (%)* 59 35 36 Proportion of the population living in difficult urban areas (%) 8 6 2 Poverty rate† (%) 16 12 14 17 147 18 143 17 157 11 9 9 Demographic characteristics Population Demographic and economical characteristics Median income per year (€) Unemployment rate (%) People covered by the universal complementary health insurance‡ (%) 7 5 6 Number of crimes and offences per 10 000 inhabitants 81 58 49 Obesity (%) 12 12 14 Tobacco consumption over 1 cigarette/day (%) 29 27 31 Health characteristics Alcohol consumption over 10 times per month (%) 8 9 9 Drunkenness over 3 times/year (%) 24 28 27 Cannabis consumption over 10 times per month (%) 10 7 7 Sources: CNAMTS, RSI, CCMSA, INSEE *Proportion of the population living in the 3 principal cities †Proportion of individuals under the poverty threshold (60% of the median standing of living) ‡The universal complementary health insurance is a free complementary health insurance for poor people www.painphysicianjournal.com 95 Pain Physician: January/February 2013; 16:89-100 ance (a GHI program dedicated to people with little or no income) showed that the economic and social situation was more unfavorable in PACA than in RA and in MP in 2008 (Table 3). Many factors could influence drug abuse and traffic, one of them is the proximity of trade areas such as ports (like Marseille and Nice in PACA) and the borders with Italy for PACA and RA and Spain for MP. Results found in this study cannot be extrapolated to the whole of France even though areas under study are 3 nearby regions representing 14 million inhabitants and 22% of the French population. However, a future study could apply the doctor shopping method to the entire French territory in order to confirm that this method is efficient in detecting emergent abuse signal in regions. In such a study, geographical variations observed in this study are likely to be amplified and specific cases such as those observed with dihydrocodeine and oxycodone would be multiplied. Product-Specific Analysis Opioids for Mild to Moderate Pain The most used oral prescription opioids were dextropropoxyphene and tramadol. They were respectively second and fourth of all oral opioids regarding their DSQ. However, the DSI for dextropropoxyphene and tramadol was relatively low (respectively 0.8% and 1.1%). In fact, the threshold value of DSI is estimated at 1% with the doctor shopping method, therefore below this value, there is no signal of abuse (27,34). Dextropropoxyphene and tramadol DSI are close to this threshold. Further studies using other abuse indicators are needed in order to confirm or exclude a signal of abuse. In our study, dihydrocodeine has the second highest DSI in RA and MP. In a study by Pauly et al (36), using several drug abuse-related indicators, dihydrocodeine was first regarding the number of forged prescriptions per million reimbursed DDD in 2008. However, it was seventh regarding the rate of illegal acquisition by OPPIDUM users and fifth regarding the abuse/dependence suspicion rate by OPPIDUM users (36). It was only eighth regarding the DSI. However, first DSI was calculated based on data from PACA only, second the doctor shopping methods used in the 2 studies were not exactly the same. In fact, in the study by Pauly et al (31), a fixed interruption period threshold was used (35 days), while we used a threshold which varied according to the observed period between 2 dispensings. 96 Opioids for Moderately Severe to Severe Pain Our study showed that opioids for moderately severe to severe pain represented 5.0% of all opioids dispensed; contrary to the US, where 84.9% of the prescriptions of opioid analgesics are for hydrocodone and oxycodone-containing products (37). Oral morphine was the opioid for moderately severe to severe pain with the highest dispensed quantity, DSQ, and DSI (Table 2). This is consistent with results of a survey among patients seen in care centers, where 56% of the oral morphine was illegally obtained (38). This is also consistent with the multi-indicator study where morphine was the only opioid to obtain the highest values for several drug abuse-related indicators (36). The second opioid for moderately severe to severe pain according to its DSQ and the third according to its DSI was oxycodone. In MP it had the highest DSI of all oral opioids. It has been on the market in France since 2001. Its use increased fourfold from 2004 to 2008 (29). To our knowledge, no abuse signal has ever been detected regarding oxycodone in France. In 2008, oxycodone was fifth of all opioid analgesics regarding the number of forged prescriptions per million reimbursed DDD and its use was not declared by any patients seen in centers dedicated to drug users in the OPPIDUM survey (36). If our results are validated by further analyses on oxycodone and dihydrocodeine abuse, they could suggest that the doctor shopping method allowed the detection of an emerging signal of abuse at a geographically specific level. Moreover, further research could assess whether the signals of abuse are transient or not using data from 2009 and 2010. Opioid Maintenance Treatments Concerning opioid maintenance treatments, buprenorphine maintenance had the highest magnitude of abuse (DSQ=50.3DDD/1,000) and abuse corrected for use (DSI=8.0%) of all opioids. In France, abuse of buprenorphine is acknowledged and has been extensively studied (39,40). Several reasons could explain the higher DSQ and DSI of buprenorphine compared to methadone. Firstly, methadone is registered as a narcotic whereas buprenorphine is not. Secondly, buprenorphine maintenance can be prescribed by every physician without any training. On the contrary, the initiation of methadone treatment is only authorized in specialized care centers for substance abuse or in hospitals. Third, the buprenorphine maintenance formula- www.painphysicianjournal.com Geographical Variations of Opioid Abuse tion is a tablet (which can be crushed, snorted, or injected) whereas methadone was only available as syrup until April 2008, when a tablet form was introduced. As a consequence, methadone is less used in France than buprenorphine. Strengths and Limitations The GHI reimbursement database is a large database that includes 77% of the French population (32). We cannot exclude the risk of underestimation of doctor shopping if doctor shoppers do not ask for opioid reimbursement to avoid checks by the GHI fraud department. Moreover, it is probable that poor people could not afford to pay the entire cost of their medication. So, people living in a lower socio-economic area may request reimbursement more frequently than those living in a higher socio-economic area, leading to a risk of selection bias. However, the general health insurance and other public health insurances cover every French inhabitant, whatever the socio-economic status. Consequently, to pay for medication in cash and not ask for reimbursement would be highly suspect for a pharmacist, particularly in the case of opioid dispensing. Therefore, we assumed that selection bias has a negligible impact on our results. Additional validity regarding dispensed quantities is provided by a study that assessed the trend in opioid use from 2004 to 2008 using data from the national GHI database. In this study, the total reimbursed opioid quantity was 8712 DDD/1,000 in 2008 (29), whereas in our study it was 7851DDD/1000. The difference corresponds to the non-oral reimbursed opioids quantity. The doctor shopping method has been slightly modified in this study. In the previous studies using the doctor shopping method, the main assumption was the threshold defining prescription interruption, fixed at 35 days (27). In a study where the doctor shopping method was applied to benzodiazepines, sensitive analyses using different threshold values showed no major variations (27). However, we consider that this threshold value should not be applied to all opioids. Indeed their indications are very different, which suggests that the modalities of use could vary between opioids. Moreover, the maximal dispensing duration for opioids is limited to 28 days except for methadone (14 days) and buprenorphine painkiller (30 days). Thus, in this study, www.painphysicianjournal.com the threshold value was a function of the observed periods between 2 consecutive dispensings (and therefore less arbitrary). A limitation of doctor shopping to estimate abuse is that part of the DSQ may have been received by individuals for legitimate reasons, such as loss of prescription or the patient or physician being on vacation for instance. In addition, doctor shopping is not the only source for prescription drug diversion, although most studies suggested that it is one of the principal means (12,13,41). Moreover, federal agencies in the US considered that diverted drugs enter the illegal market primarily through “doctor shoppers”, inappropriate prescribing practices by physicians, and improper dispensing by pharmacists (1). CONCLUSION Magnitude of abuse and abuse corrected for use (estimated respectively by DSQ and DSI) provide different and complementary information. First, these results confirm important variations among the 3 French regions although they are geographically close. Next, they highlight different rates of opioid abuse between opioids for mild to moderate pain, opioids for moderately severe to severe pain, and opioid maintenance treatments, as well as differences within these groups. This methodology should be extended to a wider geographical area including the northern half of France, and even overseas territories, to assess these variations between all French regions. Should oxycodone and dihydrocodeine abuse be confirmed by these analyses, it would confirm that the doctor shopping method is efficient in detecting regional emergent abuse signals. ACKNOWLEDGEMENTS The authors would like to thank Dr Vincent Sciortino (Head of the PACA CNAM-TS medical office), Dr Gérard Dubial (Head of the Midi-Pyrénées CNAMTS office), Dr Gilbert Weill (Head of the Rhône-Alpes CNAM-TS office), and their respective teams (Dr Véronique Allaria Lapierre and Dr François Natali from the PACA CNAMT-TS office; Dr Robert Bourrel and Carole Suarez from the Midi-Pyrénées CNAM-TS office; Valérie Tainturier and Philippe Dufour from the Rhône-Alpes CNAMT-TS office ) 97 Pain Physician: January/February 2013; 16:89-100 Appendix Example of calculation of the Doctor Shopping Quantity 8 mg of the prescription drug M dispensed (according to prescription by prescriber A) 14 0 Prescriber A 28 Prescriber A Prescriber B 8 mg of M dispensed (according to prescription by prescriber B) 42 56 70 Time (days) Prescriber A Prescriber B Prescriber B Step 1: determination of prescription periods (from the first to the last prescription by a prescriber) Period of prescription by prescriber A = 56 days 0 14 28 42 56 70 Time (days) Period of prescription by prescriber B = 56 days Step 2: calculation of ni (number of simultaneous prescription periods at the date of dispensing i) for each dispensing For example, ni =2 at days 14, 28, 42 and 56 because of overlap of prescription periods from prescribers A and B and ni = 1 at days 0 and day 70. 0 ni = 1 14 28 42 56 70 2 2 2 2 1 Time (days) Step 3: calculation of quantities Dispensed quantity (Qi) Prescriber A : 3 * 8mg = 24mg Prescriber B : 3 * 8mg = 24mg Qi = 24 + 24 = 48mg Doctor Shopping Quantity (DSQ) To take into account that a proportion of the quantity of M is medically legitimate, at each dispensation date i, the DSQ is computed using this formula DSQ = [(ni – 1) / ni] Qi It is null when ni =1 (at dates of dispensing 0 and 70) It is equal to ½ * Qi when ni =2 (at dates of dispening 14, 28, 42 and 56) For each patient DSQ = ∑ ( ni -1) Qi = (1/2*8) (day 14) + (1/2*8) (day 28) + (1/2*8) (day 42) + (1/2*8) (day 56) = 16mg ni 98 www.painphysicianjournal.com Geographical Variations of Opioid Abuse REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. Manchikanti L, Helm S II, Fellows B, Janata J, Pampati V, Grider JS, Boswell MV. Opioid epidemic in the United States. Pain Physician 2012; 15;ES9-ES38. Maxwell JC. The prescription drug epidemic in the United States: A perfect storm. Drug Alcohol Rev 2011; 30:264270. Brownstein JS, Green TC, Cassidy TA, Butler SF. Geographic information systems and pharmacoepidemiology: Using spatial cluster detection to monitor local patterns of prescription opioid abuse. Pharmacoepidemiol Drug Saf 2010; 19:627-637. Cicero TJ, Surratt H, Inciardi JA, Munoz A. Relationship between therapeutic use and abuse of opioid analgesics in rural, suburban, and urban locations in the United States. Pharmacoepidemiol Drug Saf 2007; 16:827-840. Butler S, Budman S, Licari A, Cassidy T, Lioy K, Dickinson J, Brownstein J, Benneyan JC, Craig Green T, Katz N. National addictions vigilance intervention and prevention program (NAVIPPRO): a real-time, product-specific, public health surveillance system for monitoring prescription drug abuse. Pharmacoepidemiol Drug Saf 2008; 17:1142-1154. Curtis LH, Stoddard J, Radeva JI, Hutchison S, Dans PE, Wright A, Woosley RL, Schulman KA. Geographic variation in the prescription of schedule II opioid analgesics among outpatients in the United States. Health Serv Res 2006; 41:837-855. Paulozzi LJ, Ryan GW. Opioid analgesics and rates of fatal drug poisoning in the United States. Am J Prev Med 2006; 31:506-1051. Smith MY, Irish W, Wang J, Haddox JD, Dart RC. Detecting signals of opioid analgesic abuse: Application of a spatial mixed effect poisson regression model using data from a network of poison control centers. Pharmacoepidemiol Drug Saf 2008; 17:1050-1059. Gerada C, Ashworth M. ABC of mental health. Addiction and dependence--I: Illicit drugs. BMJ 1997; 315:297-300 Cicero TJ, Kurtz SP, Surratt HL, Ibanez GE, Ellis MS, Levi-Minzi MA, Inciardi JA. Multiple determinants of specific modes of prescription opioid diversion. J Drug Issues 2011; 41:283-304. Fischer B, Bibby M, Bouchard M. The global diversion of pharmaceutical www.painphysicianjournal.com drugs non-medical use and diversion of psychotropic prescription drugs in North America: A review of sourcing routes and control measures. Addiction 2010; 105:2062-2070. 12. Inciardi JA, Surratt HL, Cicero TJ, Beard RA. Prescription opioid abuse and diversion in an urban community: The results of an ultrarapid assessment. Pain Med 2009; 10:537-548. 13. Inciardi JA, Surratt HL, Kurtz SP, Cicero TJ. Mechanisms of prescription drug diversion among drug-involved club- and street-based populations. Pain Med 2007; 8:171-183. 14. Pradel V, Frauger E, Thirion X, Ronfle E, Lapierre V, Masut A, Coudert C, BlinO, Micallef J. Impact of a prescription monitoring program on doctor-shopping for high dosage buprenorphine. Pharmacoepidemiol Drug Saf 2009; 18:36-43. 15. Cepeda MS, Fife D, Chow W, Mastrogiovanni G, Henderson SC. Opioid shopping behavior: How often, how soon, which drugs, and what payment method. J Clin Pharmacol 2012; In Press. 16. Cepeda MS, Fife D, Chow W, Mastrogiovanni G, Henderson SC. Assessing opioid shopping behaviour: A large cohort study from a medication dispensing database in the US. Drug Saf 2012; 35:325334. 17. Martyres RF, Clode D, Burns JM. Seeking drugs or seeking help? Escalating “doctor shopping” by young heroin users before fatal overdose. Med J Aust 2004; 180:211-214. 18. Dhalla IA, Mamdani MM, Sivilotti ML, Kopp A, Qureshi O, Juurlink DN. Prescribing of opioid analgesics and related mortality before and after the introduction of long-acting oxycodone. CMAJ 2009; 181:891-896. 19. Simeone R. Holland L. An evaluation of prescription drug monitoring programs. 2006. Retrieved from http://pmpexcellence.org/pdfs/simeone pdmp eval2 2006.pdf. 20. Reisman, RM, Shenoy PJ, Atherly AJ, Flowers CR. Prescription opioid usage and abuse relationships: An evaluation of state prescription drug monitoring program efficiency. Substance Abuse: Research and Treatment 2009; 3:41-51. 21. Paulozzi LJ, Kilbourne EM, Desai HA. Prescription drug monitoring programs and death rates from drug overdose. Pain Medicine 2011; 12:747-754. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. Baehren DF, Marco CA, Droz DE, Sinha SS, Callan EM, Akpunonu BS. A statewide prescription monitoring program affects emergency department prescribing behaviors. Ann Emerg Med 2010; 56:19–23. Perrone J, Nelson LS. Medication reconciliation for controlled substances - an “ideal” prescription-drug monitoring program. N Engl J Med 2012; 366:23412343. Worley J. Prescription drug monitoring programs, a response to doctor shopping: Purpose, effectiveness, and directions for future research. Issues Ment Health Nurs 2012; 33:319-328. Pradel V, Thirion X, Ronfle E, Masut A, Micallef J, Bégaud B. Assessment of doctor-shopping for high dosage buprenorphine maintenance treatment in a French region: Development of a new method for prescription database. Pharmacoepidemiol Drug Saf 2004; 13:473481. Pauly V, Frauger E, Pradel V, Nordmann S, Pourcel L, Natali F, Sciortino V, Lapeyre-Mestre M, Micallef J, Thirion X. Monitoring of benzodiazepine diversion using a multi-indicator approach. Int Clin Psychopharmacol 2011; 26:268-277. Pradel V, Delga C, Rouby F, Micallef J, Lapeyre-Mestre M. Assessment of abuse potential of benzodiazepines from a prescription database using “doctor shopping” as an indicator. CNS Drugs 2010; 24:611-620. Wilsey BL, Fishman SM, Gilson AM, Casamalhuapa C, Baxi H, Lin TC, Li CS. An analysis of the number of multiple prescribers for opioids utilizing data from the California Prescription Monitoring Program. Pharmacoepidemiol Drug Saf 2011; 20:1262-1268. Pauly V, Pradel V, Frauger E, Micallef J, Thirion X. Evolution of opioids reimbursement since 2004 from the National Database of the General Health Insurance System. Therapie 2011; 66:369-372. De Conno F, Ripamonti C, Brunelli C. Opioid purchases and expenditure in nine western European countries: “Are we killing off morphine?”.Palliat Med 2005; 19:179-184. Fredheim OM, Skurtveit S, Breivik H, Borchgrevink PC. Increasing use of opioids from 2004 to 2007 - pharmacoepidemiological data from a complete national prescription database in Norway. 99 Pain Physician: January/February 2013; 16:89-100 32. 33. 34. 35. 100 View publication stats Eur J Pain 2010; 14:289-294. Tuppin P, de Roquefeuil L, Weill A, Ricordeau P, Merlière Y. French national health insurance information system and the permanent beneficiaries sample. Rev Epidemiol Sante Publique 2010; 58:286-290. WHO Collaborating Centre for Drug Statistics Methodology. Guidelines for ATC classification and DDD assignment, 2010. Oslo, 2009. Rouby F, Pradel V, Frauger E, Pauly V, Natali F, Reggio P, Thirion X, Micallef J. Assessment of abuse of tianeptine from a reimbursement database using “doctor-shopping” as an indicator. Fundam Clin Pharmacol 2012; 26:286-294. Ministère des Affaires sociales et de la 36. 37. 38. Santé. Regional indicators collection. 2010. www.drees.sante.gouv.fr/recueild-indicateurs-regionaux-2010,1739. html. (Last accessed on May, 21, 2012). Pauly V, Pradel V, Pourcel L, Nordmann S, Frauger E, Lapeyre-Mestre M, Micallef J, Thirion X. Estimated magnitude of diversion and abuse of opioids relative to benzodiazepines in France. Drug Alcohol Depend 2012; 126:13-20. Volkow ND, McLellan TA. Curtailing diversion and abuse of opioid analgesics without jeopardizing pain treatment. JAMA 2011; 305:1346-1347. Frauger E, Nordmann S, Orleans V, Pradel V, Pauly V, Thirion X, Micallef J. Which psychoactive prescription drugs are illegally obtained and through which ways of acquisition? About OPPIDUM survey. Fundam Clin Pharmacol. 2012; 26:549-556. 39. Nordmann S, Frauger E, Pauly V, Orléans V, Pradel V, Mallaret M, Thirion X, Micallef J. Misuse of buprenorphine maintenance treatment since introduction of its generic forms: OPPIDUM survey. Pharmacoepidemiol Drug Saf 2012; 21:184-190. 40. Carrieri MP, Amass L, Lucas GM, Vlahov D, Wodak A, Woody GE. Buprenorphine use : the international experience. Clin Infect Dis 2006; 43:S197-S215. 41. McLellan AT, Turner B. Prescription opioids, overdose deaths, and physician responsibility. JAMA 2008; 300:2672-2673. www.painphysicianjournal.com