Syndicated, cloud-based KOL databases are great – but only for specific use cases

Until a few years ago KOL Mapping was a lengthy and costly undertaking for Medical Affairs Managers in Pharma.

Depending on the geographic scope, identifying and profiling let’s say the top 200 KOLs in Europe for a certain indication could easily take 6-8 weeks and cost a mid to high five-figure dollar amount for a one-off project with a KOL Data vendor.

Everyone in the KOL data industry and of course on the customer side always dreamed of a ready to use global database with HCP/KOL profiles updated in real time. And with the rapid developments in big data analytics and AI/ML this dream came true. Several companies have revolutionized the KOL data market and offer syndicated, cloud-based global KOL databases claiming to cover millions of HCPs/KOLs with real time profile updates.

More and more customers are adopting this new approach to KOL Mapping and introduce enterprise solutions (with or without CRM integration of the KOL databases), making regularly updated HCP/KOL profiles available to a large variety of different types of users in Medical, Commercial, Market Access, etc.

I have talked to several stakeholders in the global KOL Data business over the past few months, collected their opinions/potential issues on these syndicated KOL Data offerings and aggregated them to a few bullet points:

- Data privacy, particularly for Europe with EU-GDPR, could be an issue

There are different legal opinions whether or not Art. 14 EU-GDPR applies to this type of data processing, if so it would require the KOL database provider to notify all EU HCPs/KOLs (and other HCPs/KOLs from countries that have a similar legislation) in a syndicated database e.g. about their rights including the right ?to be erased“, and opt-outs would need to be deleted from the database so that a user would not know who opted out and is not included. As a user you’re in trouble if the #1 Top KOL opted out and you are not aware that this particular KOL is missing in your selection. If your provider has a different legal opinion and does not notify the data subjects I would recommend to check with your data privacy officer.?

- Geographic and therapeutic area data coverage can vary?

There are obviously regions in the world which are not as ?medical research data rich“ as the US and Europe, so if you are interested in KOL data for example in the Sub-Saharan region, a syndicated KOL database might not have sufficient data coverage. Also for rare/ultra-rare diseases you would only find a very small group of Top KOLs based on narrow, indication specific search terms. Additional experts can be very hard to find in a syndicated database because you would need to take other search parameters into consideration like ?related diseases“, ?centers of excellence“, etc. Although this information might be available in a syndicated databases, it requires a lot of experience and therapeutic area insights to be able to find the right group of experts from a syndicated universe.

- Important data fields like medical congresses, guideline involvement or simply KOL address and affiliation are sometimes not covered well, complete or validated

You can imagine that it is close to impossible to have up to date address/affiliation information in a global database with millions of HCPs, particularly if you are primarily using automated methods to capture these data fields. And even if a KOL cloud providers work with a specialized HCP address company you would

1. Still not have full global address/affiliation coverage?

2. Still have 15-30% outdated addresses because that’s the range of workplace changes per year in healthcare systems

With regards to activity data sources and coverage it is a matter of effort and investment of the data provider because most of the data sources beyond publications and clinical trials are highly unstructered and better coverage is directly related to higher effort/investment, why e.g Medical Congress coverage is typically limited.

-Name matching accuracy is key for user acceptance

With HCP/KOL data coming from a variety of structured and unstructered data sources, the name matching algorithm becomes the key quality parameter. With real time data updates for millions of individuals name matching can?only be mastered automatically. And keep in mind even a very good name matching algorithm is far away from 100% accuracy, which means that millions of activities are matched with the wrong experts. If it happens for the TOP KOLs you might find yourself in trouble. Practical example Martine Piccart, a Top Global Breast Cancer KOL who also publishes/speaks under Martine Piccart-Gebhart, MJ Piccart, M Piccart, M Piccart-Gebhart, etc. The name matching algorithm must understand that tghese are all the same person, if not maybe 50% of her activities are matched with Martine Piccart and 50% are matched with Martine Piccart-Gebhart which would then probably make her no longer rank in the TOP 5 Breast Cancer experts globally (based on a quantitative ranking) and if you are only considering Top 5, e.g for an advisory board, you would miss her. And I am not even talking about name matching in South Korea or China…

- Scoring, ranking and searching functionalities require expertise?

Although cloud platforms offer a lot of analytics and different types of search functionalities it can still be difficult to find the right HCP/KOL universe, see rare/ultra-rare disease example above. It also requires time and experience to perform searches that actually lead to the right results and the risk of missing important experts is quite high if you are not aware of the potential pitfalls of such a powerful database with millions of individuals. Rule of thumb: the more targeted and specific your desired expert universe is, the higher is the risk of missing important experts with your filter settings.

What does that all mean for your daily business and good use cases for syndicated cloud-based KOL databases?

Any use cases where you need a quick, broad overview of experts in established markets and large/larger indications are perfect for syndicated KOL databases. You have information on a large universe of potential experts across many countries and indications at your fingertips and can get a first idea of top KOLs.

Also if you want to track (e.g. as an MSL) what KOLs in your territory and other KOLs you know are doing, this approach is great (even better with a CRM integration) because you know the therapeutic area and any potential data coverage and name matching shortfalls of the KOL database are not so serious for you because you are tracking a well known expert universe and can easily spot any mismatches and treat them accordingly.

However the more specific the research questions become and the smaller the targeted expert universes are, the risk of missing important individuals increases significantly due to the effects described above. And this is also what we see in reality, many customers who have access to an enterprise KOL cloud solution continue to commission specialized agencies to help with specific KOL data use cases.?

So the new technology in KOL research really made the dream come true, but it is still not perfect, and personally I doubt it will ever be with reasonable effort and cost...