Interdisciplinary Safety Evaluation and Quantitative Safety Monitoring: Introduction to a Series of Papers

Safety monitoring and safety surveillance, also referred to as pharmacovigilance, is a key component of pharmaceutical product development life cycle. Pharmacovigilance is defined as “the science and activities relating to the detection, assessment, understanding and prevention of adverse effects or any drug-related problem.” ¹ In the past, most pharmacovigilance departments at biopharmaceutical companies focused on the handling of individual adverse event reports (called individual case safety reports [ICSRs]), with less attention paid to systematic analysis and review of aggregate adverse event data. However, this has shifted in recent years. Although individual adverse event case handling remains highly regulated and timely reporting of ICSRs continues to be an important compliance matter, the expectations for marketing authorization holders and clinical trial sponsors have increased in the areas of aggregate safety evaluation in drug development, life cycle management and postmarketing surveillance.

On the other hand, no one can deny the increasing complexity of clinical trials. Perhaps because of the flowering of molecular approaches to drug development and the increase in computing power, complexity is present in almost all areas of clinical trials. Scholarly articles have detailed arising issues in clinical trial knowledge and information system management, ² increased complexity in operational management, ³ trial design, ³ regulatory requirements, ^4,5 and even participant issues. ⁶

This increase in complexity has spawned new, innovative models of looking at efficacy: bayesian and frequentist methods for analyzing binary, ordinal, continuous data, meta-analysis, and network meta-analysis to explore efficacy across various pairs of treatments of comparisons. Moreover, longitudinal efficacy data are analyzed using mixed models, repeated measures analysis, generalized estimating equations, and other advanced methodologies. Since 2015, CDER/CBER has issued 37 guidance documents regarding the efficacy area. Importantly, 6 of them have specifically focused on generalized principles for efficacy detection/analysis.

Despite these advances in the efficacy setting, safety analysis remains stagnant—descriptive statistics and rudimentary statistical analysis. A simple search on Google Scholar for “new models for demonstrating efficacy” yields 1.2 M results whereas “new models for analysis of safety in clinical trials” yields a relatively paltry 218,000. Since 2015, the US Food and Drug Administration (FDA) has issued 13 guidance documents about drug safety, with only one specifically looking at analyses (clinical drug interaction) and no guidance documents focused on the general principles surrounding methodologies for signal detection in the arena of safety.

The current environment demands faster drug development, smaller studies, and reduced regulatory load. ⁷ Ideas like “right to try” and personalized medicine will naturally reduce the size of clinical trials, and therefore there is an increasing need for better, faster safety information. As an industry colleague says, “Safety is the New Efficacy.” Statistical methodology for safety evaluation will need to be further developed to match that for efficacy. ^8,9 In addition to evolving improved methodologies to support safety evaluation, our new, more complex world will require a multidisciplinary approach. No longer will the “safety doc” be able to get by with the simple aforementioned analysis used in years past. That said, statisticians do not have the clinical background to either develop or interpret the safety models they design. What’s needed is a true medical-statistical joint venture to develop the models the 21st century will require for safety evaluation.

These are indeed what happened in the last few years. In 2014, the American Statistical Association (ASA) Biopharmaceutical Section started a safety working group including members from both regulatory agency and industry. The working group initially focused on the design and analysis of cardiovascular safety outcome trials for Type 2 diabetes drugs, and later on expanded into a systematic review of multi-source safety data and corresponding analysis strategies. Several recent publications, ^{10 –14} including a mini-series in this journal in 2018, summarize the work of those initiatives.

In parallel, another dedicated working group was formed in 2015 to further help empower the biostatistics community in the field of quantitative safety monitoring. One initiative of this safety monitoring working group was to focus on systematic review of statistical methodologies on safety monitoring, ⁹ which included the following: Bayesian and frequentist methods; blinded versus unblinded safety monitoring; individual case analysis versus aggregate meta-analyses; pre-marketing versus postmarketing evaluation; static versus dynamic safety reviews; and methods of safety data visualization. Another initiative was to perform a thought-leader interview and industry survey on the current practices and future direction of statistical safety statistics practice, tools, and methods. In addition, a systematic review of safety regulation both at global level and at regional level (eg, US, EU, Japan, and China) was also conducted. Numerous presentations and short courses have been offered by the group at various scientific conferences and manuscripts have been developed to summarize the survey and reviews.

To cultivate interdisciplinary collaboration, the aforementioned two efforts have been integrated and expanded into one joint interdisciplinary working group between the ASA biopharmaceutical section and the DIA scientific communities. The collaboration between the ASA and DIA offers a great opportunity for cross-functional global innovations. In service of this cause, we’ve assembled academic, industry, and regulatory experts in the area of drug safety and statistics to develop a series of contributions. These include, but are not limited to, the following:

Develop interdisciplinary frameworks for aggregate safety assessment planning, and visual tools to enable/enhance cross-disciplinary collaboration.

This journal provides a multidisciplinary and global readership opportunity for the interdisciplinary deliverables from the ASA Biopharmaceutical section safety scientific working group. The following manuscript by Colopy, Statistical Practices of Safety Monitoring: An Industry Survey, provides a summary of our industry survey on safety monitoring. Future, near-term manuscripts will include (1) an overview of safety regulations by the ICH and regional regulatory authorities and (2) a framework for aggregate safety assessment planning. Over the next few years, this group and others will contribute a series of manuscripts to this journal, which will detail the historical pillars upon which drug safety is based as well as contribute ideas for process, tools, methods, and applications for the evaluation of drug safety.