Star Wars: Darth Vader Vol. This is my story of survival. Unsere digitale Zukunft: In welcher Welt wollen wir leben? SAS Programming in the Pharmaceutical Industry, Second Edition: Edition 2 by Jack Shostak Synopsis: This comprehensive resource provides on-the-job training for statistical programmers who use SAS in the pharmaceutical industry This one-stop resource offers a complete review of what entry- to intermediate-level statistical programmers need to know in order to help with the analysis and reporting of clinical trial data in the pharmaceutical industry.
SAS Programming in the Pharmaceutical Industry, Second Edition begins with an introduction to the pharmaceutical industry and the work environment of a statistical programmer. Then it gives a chronological explanation of what you need to know to do the job. It includes information on importing and massaging data into analysis data sets, producing clinical trial output, and exporting data. This edition has been updated for SAS 9. Whether you're a novice seeking an introduction to SAS programming in the pharmaceutical industry or a junior-level programmer exploring new approaches to problem solving, this real-world reference guide offers a wealth of practical suggestions to help you sharpen your skills.
This book is part of the SAS Press program. Clinical Trial Study Designs There are many types of clinical trials, and there are some general trial design concepts that you need to understand. One key concept is the randomization of study therapy. When you randomly assign patients to study therapy, you reduce potential treatment bias. Another key concept is treatment blinding.
Blinding a patient to treatment means the patient does not know what treatment is being administered. In a single-blind trial, only the patient does not know what treatment is being administered. There are other trial design concepts for you to be aware of.
A clinical trial can be carried out at a single site or it can be a multi-center trial. In a single-site trial all of the patients are seen at the same clinical site, and in a multi-center trial several clinical sites are used.
Multi-center trials are needed sometimes to eliminate site-specific bias or because there are more patients required than a single site can enroll.
Trials may be designed to determine equivalence or superiority between therapies. An equivalence trial is designed to show that there is no clinically significant difference between therapies, and a superiority trial is intended to show that one therapy is significantly better than another. Finally, trials can follow parallel or crossover study designs. In a parallel trial, patients are assigned to a therapy that they remain on, and they are compared with patients in alternate therapy groups.
In a crossover trial, patients switch or change therapy assignments during the course of the trial. Industry Regulations and Standards Regulatory authorities govern and direct much of the work of the statistical programmer in the pharmaceutical industry. It is important for you to know about the following regulations, guidance, and standards organizations. The goal of the ICH is to define a common set of regulations so that a pharmaceutical regulatory application in one country can also be used in another.
CDISC has developed numerous data models that you should familiarize yourself with. The SDTM was originally designed to simplify the production of case report tabulations CRTs , and therefore the SDTM is listing friendly, but not necessarily friendly for creating statistical summaries and analysis. These data sets are designed for creating statistical summaries and analysis. Because define. This allows the FDA to work more easily with the data submitted to it.
You will be exporting, importing, and creating data for these models, so it is important that you learn about them. The FDA has begun to formally endorse the use of these data models in their guidance. Any work that you perform that contributes to a submission to the FDA is covered by these federal regulations.
There are a number of specific regulations and guidance you must know. This guidance is of major importance, as you are often required to generate tables, figures, case report tabulations, and perhaps clinical narrative support for the clinical study report. It details trial design, trial conduct, and data analysis and reporting.
Although most useful Chapter 1: Environment and Guiding Principles 7 to the statistician, this guidance gives an excellent overview of how a clinical trial should be conducted. Anyone who works on a clinical trial needs to understand this document. Of particular interest to the statistical programmer are the following parts of E6. The italics have been added for emphasis. Part This reporting requires you to create adverse event, death, and subject dropout summaries annually for any drug under an IND application.
The PDF page should be a standard 8. The guidance defines how the files in the electronic submission should be structured for FDA review. This specification was developed by the International Conference on Harmonization ICH as an open-standards solution for electronic submissions to worldwide regulatory authorities. The FDA has adopted the eCTD as the future replacement for its other e-submission guidance, although for now the older guidance is still in effect.
Your Clinical Trial Colleagues Within any pharmaceutical company or contract research organization, there are groups and individuals outside the statistics department that you work with. Site Management The site management group is responsible for clinical site relations.
They recruit doctors at clinics to participate in clinical trials, train their staff in trial conduct, and monitor the sites for protocol compliance while serving as an all-around advocate for the clinical site. Site management can be your ally in helping to get the data entered in a clean and readily usable form.
Clean data at the start of the data collection process precludes the need for extra data queries from data management and helps prevent subsequent data analysis problems. With the arrival of electronic data capture EDC technology, the importance of site management has grown, because data entry has moved from the data management group to the clinical site itself. Data Management Next to the clinical trial statistician, the statistical programmer works most closely with the data management group.
The data management group is usually responsible for case report form CRF design, database design and setup, data entry, data cleaning, data coding, data quality control, and providing the clinical trial data for analysis by the statistics group. Cleaning the data involves scouring the data for problems by using programmatic and manual checks of the data.
Coding the data entails applying generic codes to freely entered text fields such as adverse events, medications, and medical histories. Quality control of the data involves auditing the data to make sure that it was entered properly. Finally, the data management group typically provides the data to the statistical programmer via some kind of relational database management system RDBMS , which can then be imported into SAS.
You save time when data management provides a well-cleaned and well-coded clinical database, because this means you do not have to program around dirty data. Chapter 1: Environment and Guiding Principles 9 Information Technology The information technology IT group has varying responsibilities, depending on the size of your organization. IT is usually responsible for computer systems infrastructure, maintenance, and general computer help desk support.
The IT group may also perform some level of software development. In small to midsize organizations IT may simply make application program interfaces APIs between off-the-shelf systems, while at large organizations IT may be responsible for full software applications architecture and development.
You need to work with the IT department within your organization as well as with external sponsors and vendors. Internally, you may work with IT for SAS configuration management and installation qualification, encryption technologies, and desktop publishing or report distribution concerns.
The most common reason for you to work with external IT staff is usually in regard to information exchange technologies such as FTP and encryption tools. Project Management Most contract research organizations and pharmaceutical companies are organized in a matrix management structure. This structure is called a matrix because there are project teams that span various functional departments.
The project manager is responsible for meeting the trial needs by enlisting the support of the various functional departments. He or she also works with the primary investigator, as well as with external vendors such as laboratories, pharmaceutical companies, and contract research organizations. The project manager needs to work with the statistical programmer over the course of a clinical trial.
As a statistical programmer, you may find that you answer to at least two managers during a trial. The statistical programming 10 SAS Programming in the Pharmaceutical Industry functional management serves as your skill-specific manager, while the project manager serves as your project-specific manager.
Quality Assurance The quality assurance QA group is your internal regulatory reference, and they are there to help you. The primary goal of QA is to see that operations in your organization meet regulatory standards. They can assist you in interpreting the various regulations and help you to prepare for customer and regulatory audits.
Quality assurance may also perform internal audits to make sure that your business processes meet regulatory standards. Medical Writing The medical writing group may assist in creating various documents for your organization. Medical writers may help with the writing of clinical study reports for the FDA. Medical writers may also get involved in writing an NDA submission. Clinical narrative safety reporting is another task that medical writers help with.
On occasion you will have to respond to requests for additional data from the medical writing group as they compile their reporting. Finally, a good medical writer can be a staunch ally in statistical reporting, as he or she may find any last-minute inconsistencies in your analysis before sending it along to the authorities. Guiding Principles for the Statistical Programmer The following are specific guiding principles for SAS programming in the pharmaceutical industry. These are high-level concepts that you should keep in mind while performing any of a broad range of tasks.
Chapter 1: Environment and Guiding Principles 11 Understand the Clinical Study A good statistical programmer takes time to understand the subject matter. If you were going to perform open-heart surgery and you were handy with a knife, you would not just roll up your sleeves and get to work. You would get formal training and obtain a medical degree first so that you understood what you were doing. The same can be said of SAS programming in the pharmaceutical industry.
Just because you are a SAS expert does not mean you know all there is to know about a particular drug or device or the disease state it intends to cure. There are several areas of study that will help you understand the research topic. As a first step you will want to read the clinical protocol. The protocol describes the device or medication to be used, the patient populations under study, the statistical plan of the clinical trial, and the details of the disease state.
If you want to understand the disease state or indication further, you may want to seek out a clinical investigator of the clinical trial or do some further reading about the disease.
Understanding the patient population is a good way to understand the data that you will see and whether there is reason for concern when viewing the data.
For example, if you were studying a medication to reduce hypertension, you would not be as worried if patient blood pressure data were elevated at baseline. You would expect to see this because you understand that hypertensive patients have high blood pressure.
The next step in understanding the topic of study is to read the statistical analysis plan SAP. The SAP is a very detailed document, separate from the protocol, describing how the clinical trial data will be analyzed. Although the protocol usually has only a few paragraphs on the statistical analysis, the SAP presents the entire statistical analysis in considerable detail.
The SAP describes what inferential analyses will be done, defines the study population, presents data windowing or other special data handling rules, and often includes draft output shells that show precisely what tables, listings, and graphs will be provided in the reporting. The SAP is where the majority of your work is defined. Thus, you need to understand the SAP in exquisite detail, so it is beneficial to study it well in advance of programming.
There are additional documents describing the operations of the clinical trial that you may want to review. The site monitoring plan describes how the site management staff ensures that the clinical sites are conducting the protocol and completing the CRF properly. The clinical data management plan used by the data management staff may be useful to review. The clinical data management plan contains data entry instructions, data coding instructions, data review instructions, and a data quality control plan.
Finally, there is the very important annotated CRF, which shows you where the variables in the clinical database come from on the CRF. Also note that there may be external data from the laboratory, ECG Holter monitor, etc. Program a Task Once and Reuse Your Code Everywhere One of the main reasons that you use computers is to perform repetitive tasks for you.
We can look at a demonstration of modular programming by examining the SAS libref. Program 1. You had to copy three common lines of SAS code into different places. You now have a code maintenance problem. Chapter 1: Environment and Guiding Principles 13 The code maintenance problem surfaces when you realize that you need to change one of those SAS librefs. Then you have to edit many SAS programs to make this simple change.
An alternative to having those three SAS librefs everywhere is to have them in a single location. The SAS macro facility provides two simple ways to do this. First, set up a SAS macro. If you ever have to make a change to one of those SAS librefs across all programs, you can easily change it in a single place.
This practice is fundamental to good programming, and although it is possible to be overly modular, it is better to err on the side of making your SAS code more modular than to create SAS code maintenance problems over the long term.
The clinical trial protocol and clinical trial staff make the best effort to guide the patient through a common treatment protocol, but this is often not enough to control the data coming from the patient.
It is also often the case that the case report form used to collect the data turns out to be a less than perfect instrument for collecting what is needed for analyses. Finally, despite the best efforts of data management to provide a clean database, not all data fields are scoured. Therefore, you may be faced with a sometimes deviant and heterogeneous clinical trial database, so you need to be on guard for dirty or discrepant data. In other words, you should write SAS code that accounts for all possible data permutations.
Imagine you have a SAS data set that contains adverse event data for patients in a trial. To extract data for the patients who had an adverse event, you might set up a SAS data set as in the following program. The code in Program 1. Anywhere you have conditional logic is another place for defensive programming techniques. When there is conditional logic, there should be a catch-all follow-up statement. Assume you have SAS code such as the following.
However, to be a good statistical programmer in the clinical trial arena, you must always keep a lookout for errant data and program defensively. Defensive programming lets you account for all possible clinical data permutations. Unfortunately, with such great power comes the potential for great abuse. A SAS macro can become unreadable when it is too dense with macro invocations, is poorly documented, or involves too many nested macro calls.
For instance, examine the following SAS code. The SAS macro language can also be abused when it is used in place of a built-in facility of SAS designed to solve the given task. Examine the following SAS code, which prints out demographic data patient by patient. Chapter 1: Environment and Guiding Principles 17 Program 1.
0コメント