Home » Additional Features

Kelly H. Zou on Real-World Evidence

1 October 2017 273 views No Comment
Richard Zink, JMP Life Sciences


In the Biopharmaceutical Section’s podcasts, key opinion leaders from the pharmaceutical industry and regulatory agencies talk about current issues and upcoming statistical conferences. In the most recent podcast, Richard Zink spoke with Kelly Zou real-world data.

Kelly H. Zou is senior director and analytic science lead at Pfizer Inc. She is a Fellow of the American Statistical Association and an Accredited Professional Statistician. She has published extensively on clinical and methodological topics.


How did you become interested in statistics?

I grew up in Asia in my hometown of Shanghai, China, where students tend to receive a solid quantitative education during their teenage years. I majored in mathematics and minored in physics during my undergraduate school years, followed by a combined master’s and PhD degree in statistics.

I stumbled onto statistics as a discipline by “chance,” although perhaps not as a completely random event. I took a course in probability as a math major and was very much intrigued by the concept of “uncertainty.” Navigating in the face of uncertainty toward decisiveness is the recurring theme when dealing with real-world data (RWD).

I recall that laboratory reports in my physics classes often contained linear and nonlinear regression analyses. This training to seek signals and patterns out of a set of data points was quite beneficial for becoming a statistical lead (as my last job function) and later an analytic science lead (as my current job function).

I had several jobs in two sectors from academia to industry. My research topics and applications range from receiver operating characteristic analysis, validation of predictive modeling, Bayesian hierarchical methods, image analysis, time series, pragmatic trials, and observational studies, just to name a few.

Can you give us a bit more detail about your current responsibilities at Pfizer?

Currently, I am senior director and analytic science lead in a center of excellence, named Real-World Data and Analytics, in the Patient and Health Impact organization within Pfizer Inc.

My usual days are filled with being part of various cross-functional teams and interacting with talented subject-matter experts, data scientists, statisticians, and programmers. I work closely with multiple stakeholders to leverage RWD by collaborating with health economics outcomes researchers, market access colleagues, medical and clinical colleagues, epidemiologists, and collaborative liaisons to other organizations.

Besides multiple therapeutic and product areas, I interact with colleagues in the Asia-Pacific region and in China as a large country. Understanding various policies on privacy protection, data access, storage, linkage, and regulatory landscapes is of great importance.

Collaborating and presenting on RWD-based topics at national and international conferences would take place from time to time.

You’re also very active with the ASA. Can you describe the various activities in which you’re involved?

I became an elected Fellow of the American Statistical Association in 2012. Currently, I chair the ASA Statistical Partnerships among Academe, Industry, and Government (SPAIG) Committee. It has a three-year term. The SPAIG committee aims to identify, lead, and promote initiatives that foster statistical partnerships or collaborations between two or more entities across academic, industry, and/or government sectors.

I also serve as the chair-elect and incoming chair of the ASA Health Care Policy Statistics Section (HPSS). The HPSS section focuses on strategies for improving the quality and reducing the cost of health care in the United States and abroad through the systematic use of quantitative statistical methods.

Since the ASA is an organizational affiliate of AcademyHealth, my current three-year role as a member of its Methods Council may bring extra interactions and knowledge in RWD and other related areas.

Big Data and RWE are two terms that are often used interchangeably, but what do these terms mean in practice?

Based on Section 3022 of the 21st Century Cures Act, “the term ‘real world evidence’ [RWE] means data regarding the usage, or the potential benefits or risks, of a drug derived from sources other than randomized clinical trials.”

According to the International Society for Pharmacoeconomics and Outcomes Research (ISPOR), RWD “reflects data used for decision-making that are not collected in conventional RCTs [randomized controlled trials].”

The term “big data” is a popular term to characterize the four Vs: volume, velocity, variety, and veracity.

The first documented use of the term “big data” appeared in an article by NASA scientists in October 1997.

A summary of the history of big data can be found in media.

Big data is defined by its volume, which tends to be too high to be readily processed by standard database management or processing tools … its variety, velocity, and veracity of its accrual. In practice, RWD can be big data when vast in quantity and multiple sources are combined.

There is increased interest in using RWE in medical product development. Can you discuss what has led to this interest and the improved acceptability of using these types of data in regulatory decision-making?

Medical product development and clinical research rely on sound and solid evidence. The gold standard is randomized controlled trials. The RCTs are designed and conducted to assess the efficacy and safety to support approvals.

Beyond traditional RCTs, comparative effectiveness research is an essential tool for generating, gathering, and comparing evidence on the effectiveness of therapies and products. Outside the RCT world, other designs are frequently seen in the real world. Such designs include pragmatic trials where the randomization agent (e.g., counseling) may not be active medications versus the placebo, besides observational studies with non-RCT data.

The increasing digitization of health records, health insurance plans, and over-the-counter transactions and patients expressing their preferences through diaries or surveys have resulted in an accumulation of data. The recent explosion of digitized health care information and the wealth of databases call for efforts to build collaborative consortia, integrated delivery networks, and distributed networks. These non-RCT data can be considered as RWD for the purpose of generating insights and evidence.

What are some of the challenges of RWD (quality, interoperability, IC/privacy/security, bias/randomization, missingness, others)?

There are several challenges from the “evidence” standpoint. They not only require subject-matter knowledge, but also quantitative analytic expertise to deal with unstructured and structured data. Missing data mechanisms, data qualities, and other potential biases would occur.

Beyond statistical considerations, RWD require advances in technology, infrastructure, access, storage, linkage, algorithms, tools, connectivity, linkage, and—above all—a vision for the future in terms of information flow and management.

Despite the challenges of RWD, what are some of the perceived benefits of incorporating this information into the development process, both from a sponsor and regulatory perspective (if you can provide it)? In other words, how will the inclusion of RWE improve upon the current medical product development process?

Below are a few observations in the era of RWD and big data:

  • Recent explosion of digitized health care information would call for efforts to build collaborative consortia, integrated delivery networks, and distributed networks.
  • RWD is essential for supporting value-based agreements such as outcomes-based contracting and indication-specific pricing.
  • Payers and health technology assessment (HTA) agencies have been using RWD to inform formularies, pricing, and market access.
  • Regulatory agencies and HTA bodies still have mixed views on how to leverage RWE in their decision-making processes.

What are some of the sources of RWE (EHR, claims, hospital, registries, and surveillance systems)?

An ISPOR taskforce has defined real-world data, which “reflects data used for decision-making that are not collected in conventional RCTs.”

The sources of RWD are: (1) supplements to traditional registration RCTs; (2) large simple trials (also called practical clinical trials); (3) registries; (4) administrative data; (5) health surveys; (6) electronic health records (EHRs); and medical chart reviews.

There are several ways in which one might characterize RWD. One characterization is by type of outcome—clinical, economic, and patient-reported outcomes. The other characterization relies on evidence hierarchies.

There are numerous wearable technologies that are available to collect data. How has the proliferation of these new technologies impacted data collection and analysis (patient adherence)?

Digital innovations may enable direct feeds from devices and wearables for capturing RWD. The Internet of Things has provided great opportunities to collect data. For example, micro-randomized trials in mobile health (mHealth) may be designed prospectively for real-time experiments in which treatment assignments may be randomized at fractions of occasions over time.

There are legal challenges such as the definition of the age of majority across different states in the US. Technologies due to IoT, social media, sensors, devices, and gadgets will require novel software tools, analytic approaches, and statistical algorithms.

Can you summarize some of the challenges regarding analysis of RWD?

In short, as RWD can and tend to be big data, analytic challenges alone arise from the four Vs: (1) volume; (2) velocity; (3) variety; and (4) veracity.

Several other challenges are needed for policies, regulations, infrastructure, business environment, generalizability, and insights to actions.

Deloitte Center for Health Solutions has recommended the following next steps following its RWE benchmark survey results:

  • Developing an end-to-end evidence strategy that cuts across the entire product lifecycle
  • Designing and implementing a platform and operating model that are grounded in an enterprise strategy to support working with RWE across functions and franchises
  • Developing a data strategy and organizational capability to engage in external partnerships with health care system stakeholders to gain access to and integration of the right data
  • Employing data scientists with diverse backgrounds to challenge conventional ways of doing things

Final question, and you get to look into the future and tell us what you see. What do you envision are the major changes to medical product development and the regulatory environment in the next 5–10 years?

RWE is a key component of the 21st Century Cures Act. In the next decade, it may increasingly be used to support research and development such as RCT optimization and patient recruitment.

Regulatory guidelines on RWE may be expanded.

RWE may also be used in combination with artificial intelligence to define and target the right patient population and subpopulations via precision medicine.

The access of and linkage among RWD across various sources may pose challenges to governments and health care providers. Data lakes and cloud storage may bring more collaborative opportunities across different sectors (e.g., public-private or government-academia), as well as patient privacy and data security concerns and debates.

New technologies and automation may be necessary to aid processes and workflows for gaining insights.

Editor’s Note: Kelly H. Zou is an employee of Pfizer Inc. Views and opinions expressed in this interview are Zou’s own and do not necessarily reflect those of Pfizer Inc.

1 Star2 Stars3 Stars4 Stars5 Stars (No Ratings Yet)

Leave your response!

Add your comment below, or trackback from your own site. You can also subscribe to these comments via RSS.

Be nice. Keep it clean. Stay on topic. No spam.

You can use these tags:
<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>

This is a Gravatar-enabled weblog. To get your own globally-recognized-avatar, please register at Gravatar.