Prescription Drug Advertising and Drug Utilization Part 4

FOR ONLINE PUBLICATION Appendix A A.1 Decomposition of Extensive and Intensive Margin Effects The increase in total drug utilization we observe due to advertising is driven both by increased take-up of treatment (extensive margin) as well as increased use among existing drug users (intensive margin). In this section we decompose the overall effect of advertising into the extensive and intensive margins. We re-estimate the main 2SLS regression for total prescriptions shutting down the extensive margin effect (i.e. we hold the take-up rate constant at pre-Part D levels). This allows us to estimate the proportion of the total effect that is due to intensive margin changes. The remaining proportion of the total effect is then due to extensive margin changes. We estimate: (A1) 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑚𝑚𝑚𝑚𝑚𝑚,<65 = 𝜎𝜎𝑚𝑚 + 𝜏𝜏𝑐𝑐 + 𝜖𝜖 𝑚𝑚𝑚𝑚𝑚𝑚 Where 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑚𝑚𝑚𝑚𝑚𝑚,<65 is the proportion of non-elderly individuals with any use of prescription drugs for condition 𝑐𝑐 in market 𝑚𝑚 and quarter 𝑡𝑡. We estimate this equation for the pre-Part D � 𝑚𝑚𝑚𝑚𝑚𝑚,<65 for the period from 2004-2005. We then use the estimated parameters to predict 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 � 𝑚𝑚𝑚𝑚𝑚𝑚,<65 varies only by condition and market, the entire sample from 2004-2010. Since 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 proportion that takes-up prescription drugs (i.e. extensive margin) is held constant over time. Next we construct a counterfactual measure of total prescriptions purchased which holds � 𝑚𝑚𝑚𝑚𝑚𝑚,<65 ∗ 𝑌𝑌𝑚𝑚𝑚𝑚𝑚𝑚,<65 , where 𝑌𝑌𝑚𝑚𝑚𝑚𝑚𝑚,<65 is the the extensive margin constant: 𝑌𝑌�𝑚𝑚𝑚𝑚𝑚𝑚,<65 = 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 average number of prescriptions purchased conditional on use. We use this constructed outcome variable to estimate the 2SLS model as before. The 2SLS results are presented in Columns 3 and 4 of Appendix Table B.4. The estimates using the counterfactual outcome, representing the intensive margin, are less than one-third the size of the baseline total effect (Columns 1 and 2). Using the preferred continuous instrument specification, this implies that extensive margin changes explain 71% to 73% of the total utilization effect. A straightforward back-of-theenvelope calculation using the pre-Part D take-up rate and estimated intensive margin effect from Table 8 produces similar estimates of the extensive margin effect. 46 46 Using the continuous instrument, the change in prescriptions purchased is 0.017 for one additional ad viewed. The change in prescriptions purchased along the intensive margin is predicted to be the fraction of the sample that 57 Appendix B Appendix Figure B.1 – Annual Views per Person of Ads for Top 200 Brand Name Drugs, for Elderly 65+ Notes: Sample means from Nielsen Ad*Views in above median elderly share markets relative to below median elderly share markets. The vertical lines represent the dates when Part D was signed into law (December 2003) and was implemented (January 2006). In Panel A, means are plotted for a balanced panel of the top 200 advertised brand-name drugs. There is a secular downward trend in overall views per person due to patent expirations of several of these drugs over this time period (in particular, four of the top 200 drugs went off patent around 2006: Pravachol, Wellbutrin XL, Zocor, and Zoloft). The downward trend in views matches the pattern in national advertising expenditures shown in Figure 1. used chronic drugs prior to Part D (0.063) x the estimated change in prescriptions purchased among users (0.057). Subtracting this from the total effect, we get the predicted extensive margin effect: 0.013 (or 76% of the total effect). 58 Appendix Figure B.2 – Trends in Composition of Claims Data Sample Notes: Sample means of Census 2000 characteristics linked to individuals in claims sample by 3-digit ZIP code. 59 Appendix Figure B.3 – Adherence of Chronic Drugs: Medication Possession Ratio Notes: Sample means from claims (mean MPR, ages 40-60) in above median elderly share markets relative to below median elderly share markets. Includes the 50 drugs that advertised during the study period and the drugs that did not advertise for 5 chronic conditions: depressions, diabetes, hyperlipidemia, hypertension, and osteoporosis. The vertical line represents the implementation date of Medicare Part D. 60 Appendix Table B.1—Top Advertised Chronic Drugs, 2001-2010 Condi ti on Bra nd-Na me Drug DEPRESSION CYMBALTA DEPRESSION EFFEXOR DEPRESSION PAXIL DEPRESSION PRISTIQ DEPRESSION PROZAC DEPRESSION SARAFEM DEPRESSION WELLBUTRIN DEPRESSION ZOLOFT DIABETES ACTOS DIABETES AVANDIA DIABETES BYETTA DIABETES EXUBERA DIABETES HUMALOG DIABETES JANUVIA DIABETES LANTUS DIABETES LEVEMIR DIABETES METAGLIP DIABETES ONGLYZA DIABETES NOVOLIN DIABETES NOVOLOG HYPERLIPIDEMIA ALTOCOR HYPERLIPIDEMIA BIDIL HYPERLIPIDEMIA CADUET HYPERLIPIDEMIA CRESTOR HYPERLIPIDEMIA LESCOL HYPERLIPIDEMIA LIPITOR HYPERLIPIDEMIA LOVAZA HYPERLIPIDEMIA NIASPAN HYPERLIPIDEMIA PRAVACHOL HYPERLIPIDEMIA TRILIPIX HYPERLIPIDEMIA VYTORIN HYPERLIPIDEMIA WELCHOL HYPERLIPIDEMIA ZETIA HYPERLIPIDEMIA ZOCOR HYPERTENSION ALTACE HYPERTENSION AVAPRO HYPERTENSION COREG HYPERTENSION DIOVAN HYPERTENSION INNOPRAN HYPERTENSION TEKTURNA HYPERTENSION TOPROL OSTEOPOROSIS ACTIVELLA OSTEOPOROSIS ACTONEL OSTEOPOROSIS BONIVA OSTEOPOROSIS EVISTA OSTEOPOROSIS FORTEO OSTEOPOROSIS FOSAMAX OSTEOPOROSIS PREMARIN OSTEOPOROSIS PREMPRO OSTEOPOROSIS RECLAST 61 Appendix Table B.2— Effect of Part D on Views Per Person for Top 200 Drugs Dependent Variable: Views Per Person A. Instrument=Share65+*Post Post*Share65+ B. Instrument=High Elderly Share*Post High Elderly Share*Post DMA x Quarter Obs Views per Person (Non-Elderly) (1) Views per Person (Elderly) (2) 64.379 (50.69) 263.830* (138.34) 6.233*** (1.63) 18.055*** (5.04) 3,991 3,991 Notes: *** p<0.01, ** p<0.05, * p<0.1. Clustered standard errors at the DMA level; all specifications include quarter fixed effects and DMA fixed effects. Data is from 2001-2010. 62 Appendix Table B.3— Top 10 Advertised Classes and Non-Advertised Classes Advertised Drug Classes -- Top 10 in Terms of Utilization by Ages 40-60 ANTIHYPERLIPIDEMICS ANTIHYPERTENSIVES ANTIDEPRESSANTS ANTIDIABETICS ULCER DRUGS BETA BLOCKERS ANALGESICS - ANTI-INFLAMMATORY ANTIASTHMATIC AND BRONCHODILATOR AGENTS DERMATOLOGICALS ANTICONVULSANTS Non-Advertised Drug Classes -- Top 10 in Terms of Utilization by Ages 40-60 ANALGESICS - OPIOID THYROID AGENTS DIURETICS ANTIANXIETY AGENTS CALCIUM CHANNEL BLOCKERS PENICILLINS MACROLIDES CORTICOSTEROIDS FLUOROQUINOLONES ANTI-INFECTIVE AGENTS - MISC. 63 Appendix Table B.4— Decomposition of Extensive and Intensive Margin Effects, 2SLS Dependent Variable: # of Prescriptions Full Sample 2004-2007 (1) (2) A. Instrument=Share65+*Post Views per Person (Non-Elderly) Counterfactual # of Prescriptions Holding Extensive Margin Constant Full Sample 2004-2007 (3) (4) % of Total Effect is Extensive Margin 1-(3)/(1) 1-(4)/(2) Full Sample 2004-2007 (5) (6) 0.017*** (0.004) 0.011*** (0.003) 0.005*** (0.001) 0.003*** (0.001) 71% 73% B. Instrument=High Elderly Share*Post Views per Person (Non-Elderly) 0.014*** (0.005) 0.008*** (0.003) 0.006*** (0.002) 0.003* (0.001) 57% 63% Notes: *** p<0.01, ** p<0.05, * p<0.1. Clustered standard errors at the 3-digit ZIP code level; all specifications include quarter fixed effects, 3-digit ZIP code fixed effects, condition fixed effects. Columns 1 and 2 repeat the main baseline results for total number of prescriptions purchased; Columns 3 and 4 show the effect on total prescriptions purchased coming from intensive margin changes (i.e. assuming that there are no changes in the extensive margin) as described in Appendix A; Columns 5 and 6 compute the percentage of the effect on total prescriptions purchased due to extensive margin effects. Includes the 50 drugs that advertised during the study period for 5 chronic conditions: depression, diabetes, hyperlipidemia, hypertension, and osteoporosis. 64 Appendix Table B.5— Adherence of Chronic Drugs – Mean Medication Possession Ratio Reduced Form Dependent Variable: Medication Possession Ratio A. Instrument=Share65+*Post Post*Share65+ 2SLS Full Sample (1) 2004-2007 (2) 2005-2007 (3) 0.195*** (0.067) 0.219*** (0.073) 0.414*** (0.140) Views per Person (Non-Elderly) B. Instrument=High Elderly Share*Post High Elderly Share*Post 0.005* (0.003) 0.007** (0.003) 0.75 102,477 0.75 59,252 2004-2007 (5) 2005-2007 (6) 0.018** (0.008) 0.016** (0.007) 0.034** (0.015) 0.009* (0.005) 0.009** (0.004) 0.020** (0.009) 102,477 59,252 44,519 0.012*** (0.004) Views per Person (Non-Elderly) Mean of Dep. Var (pre- Part D) Zipcode x Condition x Quarter Obs Full Sample (4) 0.71 44,519 Notes: *** p<0.01, ** p<0.05, * p<0.1. Clustered standard errors at the 3-digit ZIP code level; all specifications include quarter fixed effects, 3-digit ZIP code fixed effects, condition fixed effects. The outcome variable is the medication possession ratio (MPR). Includes the 50 drugs that advertised during the study period and the drugs that did not advertise for 5 chronic conditions: depression, diabetes, hyperlipidemia, hypertension, and osteoporosis. 65 Appendix Table B.6 – Effects on Total Chronic Drug Spending Reduced Form NonAdvertised Advertised Dependent Variable: Total Expenditures Drugs Drugs (2) (1) A. Instrument=Share65+*Post Post*Share65+ 12.488*** 5.878*** (3.399) (2.033) Views per Person (Non-Elderly) B. Instrument=High Elderly Share*Post High Elderly Share*Post 0.532*** (0.205) 0.092 (0.121) Total (3) 14.61 107,345 4.75 107,345 Total (6) 1.964*** (0.537) 0.925** (0.424) 2.889*** (0.779) 1.526*** (0.576) 0.265 (0.350) 1.790** (0.716) 107,345 107,345 107,345 18.367*** (3.801) 0.624** (0.248) Views per Person (Non-Elderly) Mean of Dep. Var (pre- Part D) Zipcode x Condition x Quarter Obs Advertised Drugs (4) 2SLS NonAdvertised Drugs (5) 19.36 107,345 Notes: *** p<0.01, ** p<0.05, * p<0.1. Clustered standard errors at the 3-digit ZIP code level; all specifications include quarter fixed effects, 3-digit ZIP code fixed effects, condition fixed effects. Includes the 50 drugs that advertised during the study period and the drugs that did not advertise for 5 chronic conditions: depression, diabetes, hyperlipidemia, hypertension, and osteoporosis. Data is from 2004-2010. 66