Pharmaceutical Industry Payments and Physician Drug Selection in Oncology Aaron Mitchell, MD Aaron Winn, MPP Stacie Dusetzina, PhD 1
Background: conflicts of interest • Common in US • Potential unintended consequences DeJong et al, JAMA Internal Medicine, 2016 2
Overall Research Goals • Conflicts within oncology practice • Granularity: different types of conflicts 3
Study Goals 1) Test for an association between receipt of payments from a pharmaceutical company and prescription of that company’s cancer drug 2) Test for an association with different payment categories, General Payments ( GP ) and Research Payments ( RP ) • Sponsored meals • Direct research grants • Consulting • Funding going to PI • Speaker fees institution • Travel/lodging • Gifts 4
Data Sources 1. Data on oncologists’ prescribing patterns: 2. Data on industry payments to oncologists: 5
Methods 1. Define cancer types of interest Those for which oncologists have several FDA-approved, guideline-recommended, orally-administered drugs to choose from Renal Cell Carcinoma (RCC): Chronic Myeloid Leukemia (CML): Sunitinib Imatinib Sorafenib Dasatinib Pazopanib Nilotinib 2. Define cohort of oncologists prescribing these drugs At least 20 filled claims among the “ drugs of interest ” for each cancer type, during 2014 Specialty listed as “oncology” 6
Analytic Strategy Model structure: McFadden conditional logit Primary independent variable: Receipt of payment (binary yes/no) from the manufacturer of one of the cancer drugs in 2013 Primary dependent variable: Odds of prescribing, in 2014, the drug made by the manufacturer from which physician had received payment[s] during 2013 Adjusted covariates: Geographic region, physician gender, year of medical school graduation, practice size, prescribing volume, GP or RP 7
Analytic Strategy Control Index Drug A Drug A Drug B Drug B Drug C Drug C 8
Results Physicians in Open Payments Physicians in Medicare Part D - Received payments from - 20 or more claims among manufacturer of drug of interest drugs of interest - Specialty is oncology - Specialty is oncology - Duplicates excluded - Duplicates excluded N = 12,991 N = 2,440 Merge by name + practice location In Open Payments, In Part D, NOT in NOT in Part D In both datasets Open Payments N = 11,351 N = 1,634 N = 803 Physicians prescribing Physicians prescribing Physicians who received RCC and/or CML drugs RCC and/or CML drugs money from the relevant who DID receive who DID NOT receive drug companies, but did not industry payments industry payments prescribe RCC or CML drugs 9 9
Results Physician cohort RCC (n = 356) CML (n = 2,140) Group practice size by number of physicians, % 1 7.3 6.8 2-10 12.9 17.4 11-50 15.5 20.5 >50 64.3 55.3 Received general payments in 2013, % 36.0 45.1 Received general payments in 2014, % 53.7 52.4 Received research payments in 2013, % 9.0 6.3 Received research payments in 2014, % 13.8 8.9 Mean dollar value of all general payments in 2013 (SD)* $ 566 (1,143) $ 166 (775) Mean dollar value of all research payments in 2013 (SD)* $ 33,391 (60,950) $ 185,763 (718,595) * Among those physicians who received payments 10
Results: base case analysis Odds ratio (95% CI) Received payment in: 2013 2014 RCC CML +/- Control group 1.78 1.29 GP (1.23 – 2.57) (1.13 – 1.48) +/- Index group +/- Control group 2.13 1.10 RP (1.13 – 4.00) (0.83 – 1.45) +/- Index group 11
Results: sensitivity analysis Odds ratio (95% CI) Received payment in: 2013 2014 RCC CML +/- Control group 2.64 1.25 GP (1.51 – 4.61) (1.08 – 1.47) +/- Index group Control group 2.14 0.95 RP (0.93 – 4.91) (0.70 – 1.30) Index group 12
Results: sensitivity analysis Odds ratio (95% CI) Received payment in: 2013 2014 RCC CML Control group 2.87 1.21 GP (1.70 – 4.86) (1.05 – 1.42) Index group Control group 2.48 1.08 RP (1.15 – 5.34) (0.82 – 1.43) Index group 13
Results: individual drug effects 0.8 RCC CML General 0.7 Adjusted fraction of prescriptions for 0.6 Payments 0.5 each drug * 0.4 0.3 * 0.2 * No 0.1 Payments 0.0 Received PAZOPANIB SORAFENIB SUNITINIB DASATINIB IMATINIB NILOTINIB Payments 0.8 * Adjusted fraction of 0.7 prescriptions for Research 0.6 each drug 0.5 Payments 0.4 0.3 0.2 0.1 0 PAZOPANIB SORAFENIB SUNITINIB DASATINIB IMATINIB NILOTINIB 14
Limitations Accuracy of Open Payments Generalizability Drug indications Correlation and causation 15
Conclusions Consistent association between general payments and increased prescribing Inconsistent association between research payments and increased prescribing 16
Questions? @TheWonkologist This research was partially supported by a National Service Research Award Post- Doctoral Traineeship from the Agency for Healthcare Research and Quality sponsored by the Cecil G. Sheps Center for Health Services Research, University of North Carolina at Chapel Hill, Grant No. 5T32 HS000032-28. 17
18
19
20
Slides in reserve 21
Results: continuous variable Industry payments modeled as a continuous variable on logged-odds scale Odds ratio* (95% CI) Received payment in: RCC CML 2013 2014 1.13 1.06 Index group +/- GP (1.05 – 1.22) (1.03 – 1.10) 1.08 1.01 Index group RP +/- (1.01 – 1.16) (0.98 – 1.03) * Odds ratio associated with a 10-fold increase in dollar value of payments 22
Financial relationships with industry among National Comprehensive Cancer Network (NCCN) guideline authors 35% Percentage of Authors 29% 30% 24% 25% 18% 20% 16% 15% 10% 10% 3% 5% 0% 0 0 - 99 100 - 999 1,000 - 9,999 10,000 - >=50,000 49,999 General payments received (inclusive of consulting fees, meals, travel, lodging), USD Mitchell et al, JAMA Oncology, 2016 23
Perlis RH and Perlis CS, PLOS One, 2016 24
25
Iria Puyosa, 2009 26
Analytic Strategy Identify scenarios in which oncologists have several drugs to choose from, and hence any preferences are more likely to be detectable Surgical resection + Stage III colon cancer FOLFOX chemo Sunitinib or Sorafenib Stage IV kidney cancer or Pazopanib 27
mRCC (n = 356) CML (n = 2,140) Male sex, % 82.3 81.5 Group practice size by number of physicians, % Physician 1 7.3 6.8 2-10 12.9 17.4 11-50 15.5 20.5 characteristics >50 64.3 55.3 Year of Medical School Graduation, mean (SD of number of years) 1990 (9.5) 1987 (9.8) US Geographical Region, % Northeast 14.0 15.3 Midwest 20.5 26.5 South 39.3 39.8 West 26.1 18.4 Mean number of claims for all drugs of interest per MD (SD) 33.4 (18.3) 36.3 (17.5) Received GP in 2013, % 36.0 45.1 Received GP in 2014, % 53.7 52.4 Received GP in 2013 & 2014, % 31.7 39.0 Mean dollar value of all GP in 2013 from the manufacturer of the drug of interest, among MDs who received at least one GP (SD) $ 566 (1,143) $ 166 (775) 25th percentile $ 16 $ 16 50th percentile $ 30 $ 20 75th percentile $ 235 $ 24 Received RP in 2013, % 9.0 6.3 Received RP in 2014, % 13.8 8.9 Received RP in 2013 & 2014, % 8.7 5.2 Mean dollar value of all RP in 2013 among MDs who received at least one RP (SD) $ 33,391 (60,950) $ 185,763 (718,595) 28
Recommend
More recommend
