## Overview The code in this replication package constructs the analysis file using multiple data sources using SAS, R and Stata. Two main files run all of the code to generate the data for the 2 figures and 3 tables in the paper, and the supplementary tables and figures. The replicator should expect the code to run for about 3 hours. ## Data Availability and Provenance Statements - [ ] This paper does not involve analysis of external data (i.e., no data are used or the only data are generated by the authors via simulation in their code). ### Statement about Rights - [x] I certify that the author(s) of the manuscript have legitimate access to and permission to use the data used in this manuscript. - [ ] I certify that the author(s) of the manuscript have documented permission to redistribute/publish the data contained within this replication package. Appropriate permission are documented in the [LICENSE.txt](LICENSE.txt) file. ### (Optional, but recommended) License for Data The data are licensed under a Creative Commons/CC-BY-NC license. ### Summary of Availability - [ ] All data **are** publicly available. - [x] Some data **cannot be made** publicly available. - [ ] **No data can be made** publicly available. ### Details on each Data Source ### Public use data sourced from elsewhere #### Market report on newly available prescription medicines, 1996-2018 We extracted data on 573 prescription medicine therapies that were newly approved for use in Germany between 1996 and 2014 using a regular annual report of the German pharmaceutical market (Arzneiverordnungs-Report), editions 1996-2015 [@schwabeArzneiverordnungen2015Im2016]. This report relies on administrative data provided by [ABDATA-Pharma-Daten-Service](https://abdata.de/datenangebot/), a data provider that evaluates all available sources for pack-related and medical-pharmaceutical information. Personal communication with authors of Arzneiverordnungs-Report reassured no changes in data extraction methods for the data used in this analysis since start of the report. According to this report, approval is the first time a medicine was available for prescription in the German market, which may deviate from the date of market approval by the EMA or the German national marketing authorization agency because drug presentations need to be registered in a particular country for marketing. We clustered active ingredients by approval date in four groups, that means medicines authorized 1996-2000, 2001-2004, 2005-2010 and, 2011-2014. Active ingredient names reported in the annual reports were mapped to the WHO ATC classification level 5 to be able to link information about availability to other data sources used. #### Market report on multisource markets To identify multisource markets for the physician level analysis, we relied on a market report [albrechtAnalyseGenerikawettbewerbs200520152015a]. This report includes active ingredients with generic entry after January 2006 up to 2015 that were of high volume in the German prescription medicine market. We restricted this data to generic entries before January 2009 to allow that generic uptake has occurred at substantial scale by the time our observation period starts. #### MARKETING AUTHORIZATION OF ACTIVE INGREDIENTS To validate the date of first approval of brand name medicines, complementary information was collected from the European Medicines Agency (EMA) public assessment reports. We verified information such as timing of approval of brand name medicines, orphan and biologic/biosimilar drug status. We use the downloadable version (.csv-file) of the public assessment reports as available on the [EMA's website](https://www.ema.europa.eu/en/medicines/download-medicine-data) as of November 23, 2017. ### Confidential data #### PRESCRIBER PRESCRIPTION PANEL To observe physician prescribing, we rely on data from the CEGEDIM MEDIMED prescriber panel, which covers complete prescription data of 3,026 outpatient physicians between January 2011 and June 2014, available via <https://www.medimed.info/>, last accessed April 29, 2021. The data provider may be contacted via service@medimed.info or phone +49 6251 8484-522. The panel is representative of physicians balanced across regions, specialties, and prescription volumes present in the German health care system. It includes data on physician characteristics and selected patient attributes. A difference of our data compared to administrative data from sickness funds is that we observe the prescription decisions of the physicians instead of the medicine presentations that have been finally dispensed by the pharmacy. We used the information on physician and patient characteristics available from the panel to construct variables that capture physician age, physician sex, the share of publicly compared to the privately insured patients in a practice, the share of patients older than 65 years of age and the full set of prescriptions by patients to calculate disease prevalence based on prescription data. #### PROMOTIONAL ACTIVITY OF PHARMCEUTICAL INDUSTRY Data on promotional activties to measure promotional effort by manufacturers was obtained from the CEGEDIM PROMO database covering the years 2004-2014. This proprietary database which was provided by CEGEDIM in 2014, includes monthly spending of pharmaceutical companies in Germany at product level for detailing, sampling and multimedia channels. Direct-to-consumer advertising is not permitted in Germany and excluded from the study. We captured promotional spending by molecule corresponding to the active ingredient names according to the ATC classification, level 5, recorded in the CEGEDIM PROMO database and by market segment status. Promotional activity was cumulated to allow for that marketing activity in the past carries over to current periods. For the active ingredients identified as multisource markets, we calculated the cumulated quarterly promotional activity as a spending value between 2011 and 2014. #### PHARMACEUTICAL REFERENCE DATA, provided by Wido We used the 2021 version of the [GKV Arzneimittelindex](https://www.wido.de/forschung-projekte/arzneimittel/gkv-arzneimittelindex/) which provides pharmaceutical reference data of products available in Germany. We linked data on medicines at presentation level (identified by a product number, Pharmazentralnummer) to additional information including the classification of the active ingredient by the WHO ATC Classification, Level 5 and, whether the medicine by active ingredient was subject to reference pricing. For reference pricing, we used the 2012 version to capture the reference pricing status at the beginning of our observation period. The original classification is available from [BFARM](https://www.bfarm.de/EN/Medicinal-products/Information-on-medicinal-products/Reference-Pricing/_node.html). We used the 2020 version for additional information. #### PHARMACEUTICAL REFERENCE DATA, ABDATA Artikelstamm provided by ABDATA We used the complete version of pharmaceutical reference data ([Artikelstamm](https://abdata.de/produkte/abda-artikelstamm/)) provided by ABDATA Pharma-Daten- -Service as of April 15, 2020. The dataset contains pharmaceutical reference data of products available in Germany. We linked data on medicines at presentation level (identified by a product number, Pharmazentralnummer) to additional information including the classification of the active ingredient by the WHO ATC Classification, Level 5 and, whether the medicine by active ingredient was subject to reference pricing. The data can be accessed via Avoxa-Mediengruppe, Deutscher Apotheker GmbH, Apothekerhaus Eschborn, Carl-Mannich-Straße 26, 65760 Eschborn, Germany. ## Dataset list The following data set list contains all derived data sets listed from the sources above. | Data file | Source | Notes | Provided | |--------------------|-----------------|------------------|-----------------| | `01_Data\generic_entries.xlsx` | IGES Market Report on Generic Competition, 2005-2015: Albrecht M, de Millas C, Berkemeier F. Analyse des Generikawettbewerbs (2005-2015). Berlin: IGES Institut 2015 Dec | | Yes | | `01_Data\EMAentries.xls` | [EMA European Public Assessment reports](https://www.ema.europa.eu/en/medicines/download-medicine-data#european-public-assessment-reports-(epar)-section) | downloaded on 2017-11-23 | Yes | | `01_Data\180214 Leitsubstanzen national.xlsx` | National preferred drug status data | hand searched | Yes | | `roh.medimedall.sas7bdat` | CEGEDIM Medimed | Confidential | No | | `CSDdetail.sas7bdat` | CEGEDIM Promotional data - detailing | Confidential | No | | `CSDmulti.sas7bdat` | CEGEDIM Promotional data - multi channel | Confidential | No | | `Stammdatei Nov 2020.xlsx` | Wido Stammdatei 2020 | Confidential | No | | `wido_2012.sas7bdat` | Wido Stammdatei 2012 | Confidential | No | | `abdata.art_med.sas7bdat`, `abdata.atc_med.sas7bdat` | ABDATA Datenstamm | Confidential | No | | `AVRverord.csv` | Derived data used for analysis | Confidential | No | : List of analysis data sets used ## Computational requirements ### Software Requirements Data preparation was performed using SAS Enterprise Guide (code was run with version 9.04.01M3P062415, WX64_WKS). Econometric analyses were performed using Stata 17. We first estimated the physician level expenditure and prescription medicine use models that account for adoption bias using the regress, probit and heckman command. Using the identical estimation models, we performed the Heckman-adjusted Oaxaca-Blinder decomposition using the `oaxaca` package (as of 2022-08-31), manually estimating the Mills-ratio and the threefold decomposition (44). Results tables were processesd using the `estout` and the `outreg2` packages (as of 2022-08-31). Stata packages used (code was last run with version 17) - `estout` (as of 2023-11-20) - `outreg2` (as of 2023-11-20) - `oaxaca` (as of 2023-11-20) - the program "`00_setup.do`" will install all dependencies locally, and should be run once. Figures of the descriptive analyses and estimates of the reressions were compiled using R version 4.2.2 (2021-10-31) and the 'tidyverse', 'cowplot', 'ggplot2', 'lubridate', 'readxl', 'stringr', and 'scales' packages (as of 2023-04-25). The file "`00_setup.R`" will install all dependencies (latest version), and should be run once prior to running other programs. ### Memory and Runtime Requirements #### Summary Approximate time needed to reproduce the analyses on a standard (CURRENT YEAR) desktop machine: - [ ] \<10 minutes - [ ] 10-60 minutes - [ ] 1-2 hours - [x] 2-8 hours - [ ] 8-24 hours - [ ] 1-3 days - [ ] 3-14 days - [ ] \> 14 days - [ ] Not feasible to run on a desktop machine, as described below. #### Details The code was last run on a Intel(R) Core(TM) i5-6200U CPU \@ 2.30GHz 2.40 GHz4-core, 8 GB RAM, x86_64-w64-mingw32/x64 (64-bit), with Windows 10 Enterprise OS. Portions of the code were last run on a Intel(R) Xeon(R) CPU E5-2690 v4 \@2.6GHz, 32,0 RAM, 64 bit, x64 Processor, with Windows 10 OS. Computation took several (\<10) hours. ## Description of programs/code ### License for Code The code is licensed under a BSD license. ## Instructions to Replicators ### Details If running the code, the following order is important to keep. - The analysis data set was generated using SAS 9.4 / Enterprise Guide. - `02_Code/01_Directories.sas` will assign directories to source files. - `02_Code/02_Formats.sas` defines global formats used throughout the programming steps. - `02_Code/02_Macros.sas` assigns macro variables. - `02_Code/03_Data_Prep_Identifying_drugs_year-2000.sas` constructs a cross-sectional dataset of basket of 423 active ingredients by market entry date matched with prescription data - `02_Code/02_Data_Prep_Identifying-drugs.sas` identifies medicines by market entry date and generic / brand name status and matches the IGES market report data to the prescription data. - `02_Code/02_Data_Prep_High-volume-physicians.sas` identifies which physician categories are high volume prescribers to be included for analysis. - `02_Code/02_Data_Prep_Presc-based-disease-prevalence.sas` captures disease prevalence based on prescription data according to Huber, C. A., Szucs, T. D., Rapold, R., & Reich, O. (2013). Identifying patients with chronic conditions using pharmacy data in Switzerland: An updated mapping approach to the classification of medications. BMC Public Health, 13(1), 1030. <https://doi.org/10.1186/1471-2458-13-1030> - `02_Code/02_Data_Prep_Promo.sas` calculates promotional spending per physician using marketing and detailing data. - `02_Code/03_Data_Prep_Prescriptions-setup.sas` generates the baseline analysis dataset for first differences between 2011 and 2014. - `02_Code/03_Data_Prep_Prescriptions-time-series.sas` generates multiple datasets for each quarter from 2011 and 2014. - `02_Code/02_Code_P Q decomp.sas` performs the price and quantity decomposition according to Bundorf et al. - Data analysis was performed using Stata 17/MP: - `02_Code/04_Oaxaca Blinder Decomposition.do` calculates descriptive statistics and performs the regression analysis for the baseline analysis dataset. It generates all tables. - `02_Code/04_OaxacaBlinder-time-series.do` performs regression analyses for the time series analyses - Figures were produced using R.4.2.2: - `02_Code/04_OutcomesbyMarketApprovalYear.R` generates Figure 1 based on the data generated in `03_Data_Prep_Identifying_drugs_year-2000.sas` - `02_Code/04_UtilizationDifferential.R` generates Figure 2 based on the results generated in `03_Data_Prep_Prescriptions-time-series.sas` ## List of tables and programs The provided code reproduces: - [x] All numbers provided in text in the paper - [x] All tables and figures in the paper - [ ] Selected tables and figures in the paper, as explained and justified below. | Figure/Table \# | Program | Line Number | Output file | Note | |--------------|--------------|--------------|----------------|---------------| | Table 1 | 04_Oaxaca Blinder Decomposition.do | 142 | Table1-descriptives.csv | | | Table 2 | 04_Oaxaca Blinder Decomposition.do | 296,320 | Heckman-full-ref-ref.csv.csv | | | Table 3 | 04_Oaxaca Blinder Decomposition.do | 378 | Oaxaca-mills.csv.csv | | | Figure 1 | 04_OutcomesbyMarketApprovalYear.R | | Figure 1-Market.png | | | Figure 2 | 04_UtilizationDifferential.R | | Figure2.png | | ## References 1. Bundorf MK, Royalty A, Baker LC. Health Care Cost Growth Among The Privately Insured. Health Affairs. 2009 Sep 1;28(5):1294--304. 2. Huber CA, Szucs TD, Rapold R, Reich O. Identifying patients with chronic conditions using pharmacy data in Switzerland: an updated mapping approach to the classification of medications. BMC Public Health. 2013 Oct 30;13(1):1030. 3. Schwabe U, Paffrath D. Arzneiverordnungen 2015 im Überblick. In: Schwabe U, Paffrath D, editors. Arzneiverordnungs-Report 2016 [Internet]. Berlin, Heidelberg: Springer Berlin Heidelberg; 2016 [cited 2020 Feb 24]. p. 3--27. Available from: <http://link.springer.com/10.1007/978-3-662-50351-5_1> 4. Albrecht M, de Millas C, Berkemeier F. Analyse des Generikawettbewerbs (2005-2015). Berlin: IGES Institut; 2015 Dec. ------------------------------------------------------------------------ ## Acknowledgements Some content on this page was copied from [Hindawi](https://www.hindawi.com/research.data/#statement.templates). Other content was adapted from [Fort (2016)](https://doi.org/10.1093/restud/rdw057), Supplementary data, with the author's permission.