"Monitoring of randomized clinical trials is important for protection of patients. This applies both to those who are included in the trial and to those future patients who may potentially be treated in that or similar trials or in clinical practice. There are necessary regulatory systems to ensure the validity of the data from trials used to inform their decisions relating to the monitoring. In some instances, this has been unnecessarily onerous and expensive. In the end, it is patients who pay the financial and other costs of such systems.
"Recently there has been recognition that monitoring can be done better, utilizing the accumulating data in electronic databases that will eventually be used for the analysis of the results of the trial. There are many books on how to analyze trial data, and quite a number on how to carry out data management and preparation for that analysis. Risk-Based Monitoring and Fraud Detection in Clinical Trials Using JMP® and SAS® is almost unique in focusing on how to monitor accumulating and accumulated data in order to help ensure that the trial is conducted properly and that the data have been collected properly.
"The data for these trials will vary in the risk that errors in the different variables have for the influence on the final trial interpretation. Monitoring may then be risk-based. Overall strategies have been set out for this approach by regulatory authorities, and the Transcelerate not-for-profit group have given a detailed set of suggestions both for the overall strategy and how to implement a risk-based approach. This book, utilizing the excellent facilities of SAS® and JMP® Clinical when analyzing a trial database, sets out to develop methods that ensure the key variables are monitored for validity. This includes not just possible fraud, which may be very rare, but also other aspects where quality of the data or accidental errors affect the results.
"In this book, there is a good introduction to the overall methods and how to use JMP Clinical to implement the monitoring of a risk score that highlights either particular centers or particular patients within a center, where there is a suggestion there could be an issue. This allows resources to be focused on the most important areas that could affect overall validity of the trial. Catching problems early and remedying them, for example, by providing extra training to staff collecting the data, will lead to more reliable trial results and avoid wasting data and the accompanying extra costs. The prevalence of fraud tends to be unknown, and the knowledge that such monitoring is being done may itself prevent or discourage fraud at all stages.
"The introduction by Richard Zink also gives a useful survey of fraud and misconduct detection using statistical methods. Looking for unusual patterns in data, even where fraud is neither suspected nor exists, has the gain that accidental problems in data collection will also be detected. These can include under-or over-reporting of adverse events, failure to calibrate instruments properly, and inefficiency in returning clinical record forms.
"Richard's writing is very clear and the example data set is necessarily limited, but he introduces errors into the data to show the performance of the system as a whole. The graphical outputs are very clear. The target of all of these is quality of the data and will usually be done when treatment allocation is unknown. The system assumes that regular snapshots of the data collected so far are taken. The frequency will depend on circumstances for a particular trial.
"There are two whole chapters on the detection of fraud. First at the level of a site where between-site comparisons in the summary data allow for concentration on where a particular site may have a problem. This may not necessarily be fraud, but understanding reasons for divergent patterns is important. I am prejudiced, but it is good to find my ideas about inliers being implemented!
"The next level is where individual patients have issues with the data, and the ability of the system to drill down to particular patients and particular variables where a problem may exist is very helpful. "Most of the literature around the techniques has been concept-based, but this book is very practical. It is unusual to find a book with a good statistical basis being written to show exactly how to implement the techniques in accessible software.
"As noted at the start, this book is unique in showing how to use good statistical methods to analyze a clinical trial data set for quality purposes. The fact that it is contingent on SAS and JMP Clinical may be helpful to promote the use of these products. It also uses the CDISC standards which is increasingly encouraged by the FDA.
"High standards in clinical trials are in everyone's interests, and this book is an important step, or even several steps, in the right direction."
Stephen J.W. Evans MSc FRCP (Edin) Hon. FRCP
Professor of Pharmacoepidemiology
Dept of Medical Statistics
London School of Hygiene & Tropical Medicine
"Careful quality control of clinical trials and their data has always been important in the pharmaceutical industry. This is necessary to ensure that valid results are produced and that fraud is minimized. But, with increasing trial costs, as well as increasing complexity and indeed competition, effective data quality control has become even more important—and challenging. New kinds of data-capture technology have been adopted, such as electronic case report forms. New kinds of data have arisen, such as genomic and proteomic indicators. And new trial methodologies have been developed, such as adaptive trials.
"Risk-Based Monitoring and Fraud Detection in Clinical Trials Using JMP and SAS describes the JMP Clinical software system, which allows centralized monitoring and interactive examination of clinical trials data, permitting easier identification of warning signals.
"Early chapters outline risk-based monitoring ideas, describing metrics that can be used to guide the use of quality examination tools. The next two chapters look at fraud detection, including both patient and investigator fraud.
"One of the difficulties with fraud detection in clinical trial work, in common with that in other areas, is that fraud is, thankfully, rare. That means that searching for fraud can be potentially unrewarding; most of the effort will fail to find any. The JMP Clinical package helps in this regard by presenting a wide variety of indicators of potential fraud, along with graphical displays that enable anomalies to be easily recognized. Indeed, researchers in other areas of fraud detection would benefit from reading the two chapters on fraud in this volume.
"The final chapter, on snapshot comparisons, notes that "clinical data review is extremely time-consuming." One cannot get away from that fundamental truth! Nonetheless, the JMP Clinical software brings powerful tools to bear on the problem.
"I noticed that the appendix to Chapter 2 begins by commenting that "with the numerous reports and analyses available... navigating the JMP Clinical Starter can at first seem a bit overwhelming." That is certainly true! But on the other hand, it's also equally apparent that effort made to get to grips with the software will be amply rewarded in terms of improved data quality—and hence in sound clinical trial results."
David J. Hand
"Risk-based monitoring and fraud detection are extremely important topics in clinical research. At heart, both are fundamentally concerned with providing patients timely access to medical products based on legitimate and compelling scientific evidence that they are safe and provide therapeutic benefit. Recent regulatory landscapes emphasize that biopharmaceutical sponsors do not need to adopt a one-size-fits-all formulaic view regarding the execution of monitoring activities during a clinical trial: flexibility to choose the optimal combination of monitoring strategies for a given trial is now granted. With regard to fraud detection, the negative impacts of fraudulent conduct are clear and reprehensible. How best, then, to design and implement strategies to optimize the benefits of risk-based monitoring and to control as much as possible the potentially dire consequences of fraud?
"Richard Zink is a professional statistician with over 20 years of programming experience. He has also taught university courses in statistical methodology to non-statisticians. This combination of complementary experiences is evident in his engaging writing style. In this new book, Richard has provided informative and very practical discussions of how best to address risk-based monitoring and fraud detection that will be of considerable interest to readers from many branches of clinical research. The book is both sufficiently sophisticated to inform those actively engaged in these pursuits and sufficiently accessible to reveal the importance of this work to experts in other aspects of clinical trial endeavors. Discussions are grouped into six chapters: Introduction; Risk-Based Monitoring— Basic Concepts; Risk-Based Monitoring— Customizing the Review Experience; Detecting Fraud at the Clinical Site; Detecting Patient Fraud; and Snapshot Comparisons. Chapters conclude with lists of references that are instructive while not being overwhelming. I found it very informative and well-written, and I would recommend it for anyone in clinical research."
J. Rick Turner, PhD
Senior Scientific Director
Durham, North Carolina