Thoughts on Global Medical Affairs
Data has always been the backbone of medical affairs. Understanding the data that underlies the company’s products and the products of the competition and understanding the prevalence and treatment data available about the disease state has always been a requirement. And, being able to summarize and explain the meaning of the data is one of the greatest values that MA brings to the organization.
But as a generator of data analysis and manager of data, most MA organizations have been passengers not drivers. Most data sets are being generated by clinical development and analyzed by statistical staff within the clinical function. Even after the product is on the market and the data generation turns from a clinical responsibility to a medical responsibility, most MA organizations still rely on these resources or the resources of outsourcers to define, manage and analyze their data.
However, as MA groups grow more sophisticated in their use of data, and as real world data sets available for analysis continue to increase in size and importance, it may become in the best interest of MA to begin developing some data science capability of their own.
Some MA organization already have their own stats staff or stats staff assigned to them but working in clinical, but in my experience this is still the exception not the rule. Instead, most organization rely on stats people internally who are not primarily focused on MA or on outsourced stats resources. There are a number of challenges with this environment. First, for the internally loaned people, most of them are not that familiar with what MA does and they are also usually not familiar with using real world data sets. Leveraging them requires bringing them up the learning curve, sometimes at the expense of time and effectiveness of the analysis.
Relying on to a large degree on outside statistical help is also very problematic. In these outsourcing situations, the cost can be high and the learning curve that you are paying for becomes the property of the outsourcer to resell to your competition. Additionally, when a non-stats person hires and manages a stats outsourcer, it is very difficult for them to understand if they are getting the best thinking out of the outsourcer and suggest other alternate directions if they feel the outsourcer is taking a less than optimal path. It is this very difficulty that led many clinical organizations, even virtual clinical organizations, to realize that they always needed some stats capability in-house, even if it was simply to manage the outsourcers. Finally, working with outsourcers makes it very difficult to answer quick, smaller “what if” questions that always seem to come up after the main analysis is complete.
Given the importance of these analysis for MA and the need to be flexible, I believe that more and more MA organizations will realize that they need their own stats capability on the MA team – focused full time on the data sets and analyses that are most relevant to the post-marketing world.
MA Data Managers
While some MA organizations already have stats, I have yet to see one that has their own data management function. Nevertheless, I am going to suggest that this will be less rare in the future. Data managers are responsible for the “care and feeding” of the databases that the stats team analyzes. A common function on clinical.
As medical affairs becomes more data sophisticated and as cutting edge groups decide to build huge repositories of real world data to perform ongoing analysis, the need for professional MA focused data managers will grow. These data managers will be more focused on the collating and cleaning of external data then their clinical counterparts and that is why I think the need for an MA specialist group will take hold.
What do you think? Does your MA team have its own stats function today? Is data sciences in the plans for the future? If you would like to leave a comment, click here and scroll down.
As evidence, did you see this article in the NEJM? It’s a research analysis in which they pulled the entire ClinicalTrials.gov database and analyzed it to determine who has been publishing summarized clinical trials results as required by FDAAA from 2007. This analysis was done by cardiologists at Duke, and what got the headlines was that only 38% of completed studies had their data posted as required by the law.
However, dig a little deeper and with more of our focus and you can learn some interesting things about the biopharma industry. Of all the 13,300 (all numbers rough for discussion purposes) completed clinical trials analyzed, 66% were industry sponsored or roughly 8700 trials. Only 5100 of the 13,300 completed clinical trials reported any results, and of 3600 of those 5100 reporting trials, or 71%, were industry trials. So, in a world of terrible compliance, industry was punching above its weight. That still leaves some 5100 industry trials with not results posted despite the law.
BUT, you say, isn’t there some exception for holding back results until after FDA approval/rejection. Yes, a company can file a certificate which allows it to delay posting results in that circumstances. Of the 5100 completed industry trials missing data only 2000 certificates were filed – leaving 3100 or 36% of all industry trials unreported or uncertified per the law.
This is where big data and company reputation risk raises its ugly head. There are big data sources already out there that read all the data from ClinicalTrials.gov. With today’s big data tools it is a straight forward exercise to determine which industry company sponsored which trial and whether it reported/certified per the law. They will also know which products have failed to report/certify if the products are approved. It is not long from now that some reporter will put this together and produce a list of the “best and worst biopharma companies for publishing results” – some uncharitable media outlets (and which ones are charitable to biopharma) may even imply sinister intent at “denying their legal obligation to share this data – what are they hiding.”
Implications for Medical Affairs and Clinical Development
There is a unique opportunity to head this potential distraction off at the pass. We don’t want our MA field team’s spending time justifying why the company is not posting data.
While maintaining ClinicalTrials.gov is generally the responsibility of the CD in most organizations, MA has a strong vested interest to ensuring that the company is bullet proof in this area. MA and CD need to collaborate to make sure that data is posted or the certificates are filed. A process audit to confirm that the processes are in place and are working, as well as verifying that the company does not have any missing posting or filings, is a small bit of work that can save a huge amount of distraction for the entire organization in the future.
What has been your experience in this area? Leave a comment
A couple of years ago I wrote a post (check it out here) on the emergence of big data for Medical Affairs. Given the rapid evolution of big data, two years is a long time ago so it’s worth revisiting this topic.
Let’s recap what we mean by “big data.” It is a broad concept, but for our discussion today we will be using big data to refer to the new capability to pull together huge quantities of data that were not directly generated for the purpose they are now being applied. Biopharma has excelled at generating proprietary data sets for a specific purpose, but big data take advantage of non-proprietary data that was generated for a different purpose by applying it in a new way.
These external data sources range in structure, format and value. The real trick to big data is pulling the data from disparate sources, efficiently cleaning it and standardizing it to allow it to be cross-referenced, then finding novel ways to use it.
Example of Big Data in MA
In the last couple of years we have seen examples of companies set up to provide big data services to MA. I will single out one here as an example, but this is not intended as an endorsement. I have no relationship with this company or practical experience with their products.
The company, Med’meme, is a case study of big data in MA. Based on their website, Med’meme takes large, public data sets – in this case lists of scientific presentations from medical meetings and peer-reviewed journals and clinical trial information at least – and in their backroom they apparently standardize it to make all those data cross referenceable. How well they do this, how complete and how accurate the data is, I can’t say. But, when you think about that data source as an MA professional I am sure you are jumping to a bunch of potential uses – like the ability to rank KOLs, to identify new KOLs, to track TA trends in publishing, to identify potential investigators, to be alerted to new publication identification, etc.
And that is the beauty of big data – there does not appear to be anything in their data set that has not been available (with some costs) to biopharma for years. Their service is finding a way to scrape it all together, standardize it and allow it to be searched effectively.
Buy v Build in Big Data
When I first published the article about big data I had a number of “buy vs. build” questions. The reality of big data in its current form is about re-using publically available data in novel ways, so building it internally is unlikely to produce proprietary value. However, combining these data sets with proprietary data, or asking interesting and unique questions of the data is something that can remain proprietary – so some hybrid solutions may be valuable.
If big data is not a part of the MA information technology planning it should be. This capability represents an opportunity for strategic advantage in the short-term until it is widely adopted.
Big data is a new reality. A huge new data set, the Sunshine Act database, has just come on-line, and other data sources are increasingly making their data available for these types of analysis. Expect to see major development in this area in the coming couple of years.
What has been your experience with big data in MA? Leave a comment.
“The era of Big Data is here!” That may be true but what does that mean for Medical Affairs? As in all of biopharma, MA is comfortable working with data. So much of our work revolves around discussing data and the implications of data that many people may think that we were already living in the era of Big Data.
But for most MA organizations, the data sets we have focused on are purpose generated – either our own data or data from similarly-scaled studies conducted by others. Big Data refers to something different. I like the differentiation that SAS uses when comparing Big Data to the past data sets. They break it down to four “V”s and a C:
This is not your grandfather’s data sets. What are some examples of Big Data as relevant to biopharma and MA:
Contained within these and many Big Data sources are key tools for MA:
But, none of these benefits can be achieved unless the question is asked and the data is analyzed. I would suggest that effective MA organizations of the future will need to have the capacity to ask and answer these types of questions.
In order to do so, MA organizations will either need to build or have access to increased levels of biostatistical and epidemiological resources. And these resources need to have skills directly related to Big Data. The characteristics that differentiate Big Data from existing data sets also means that many existing biostats and epi staff do not have the expertise or confidence working with these large, external data sets. MA organizations need to ensure that people with exactly these skills sets are available within their organizations or from outside vendors and that these resources have the capacity to support MA.
Then, MA needs to improve its overall level of confidence defining Big Data questions, conducting Big Data analysis, and discussing the results with others. Given the difference in the source of this type of data, the way this data is presented and discussed must be different too. Everyone in MA, but especially the MSLs, must become more comfortable understanding the nuance of this type of data analysis and discussing both the strengths and weaknesses of working with Big Data.
The era of Big Data is here. MA has a long history of effectively using data and explaining data in support of its organization. MA leaders must investigate and embrace Big Data to take advantage of all the tools available today. The questions unasked are always the questions unanswered.
What is your experience with Big Data? Please leave a comment.