Real-world data (RWD) offers a multitude of use cases that can benefit pharmaceutical companies throughout the product life cycle and, as such, is becoming increasingly prevalent in the industry. This has led to a robust market of companies offering patient de-identification (de-ID) or tokenization solutions. De-identification allows patient records to be matched across multiple data sources in a privacy-protected and HIPAA-compliant manner to follow the patient’s longitudinal healthcare journey. With the inherent noise in RWD from nicknames, name changes, typos and other errors, this can be a daunting challenge. As is ensuring that a patient cannot be re-identified. HealthVerity recently hosted a webinar discussing how you can overcome these challenges with a precise and economical solution, as well as how this precision can enhance go-to-market strategies. If you missed the webinar, here is a recap.
De-ID that masters the sync, not misses the link
With legacy de-identification solutions, the token, which is an alphanumeric string that replaces personally identifiable information (PII), is created locally and patients are compared against other individuals in the dataset to establish whether they are a unique person or the same individual. If deterministic techniques that require an exact match are used and an individual is encountered multiple times with different spellings or other inconsistencies, such as Elizabeth Wilson, Liz Wilson, E.S. Jones and Liz Wilson without a date of birth, four different tokens could potentially be created. This causes a false negative problem, fragmenting the patient journey. Instead of seeing one patient with four transactions, you would see four different patients with one transaction each, losing the longitudinality of the journey and overstating the patient count. Additionally, there is a risk of re-identification in the case of a hack, data breach or some other nefarious activity because the PII and the token are possessed by the same entity.
HealthVerity does things differently, managing identity centrally for a precise sync. All of the different data sources in the nation’s largest healthcare and consumer data ecosystem, plus the dataset from the client submitted for de-identification, are fed into a matching environment where billions of transactions are compared using probabilistic matching that takes into account the inherent noise in RWD to resolve the patient’s identity to a single HealthVerity ID (HVID). HealthVerity retains the parent ID and gives each client a child ID so the data will not be interoperable across clients, protecting the data from being shared improperly or hacked, but because we have the parent ID, the data can be crosswalked between the children if need be. With this synchronized process an individual is compared to over 300 million HVIDs, where we may see the same individual 100 times across 27 datasets, rather than locally within the dataset, and a single ID is created, rather than continually creating new tokens for the same individual. By comparing and contrasting against all of the data in our ecosystem and your data we are able to master the sync, not miss the link.
De-ID that locks it down, not locks you up
This synchronized process also ensures privacy. As discussed, legacy solutions create tokens locally, which opens up potential for re-identification because the same entity possesses both the PII and the token. With the HealthVerity de-identification solution, Identity Manager, customized software is deployed locally, behind the data owner’s firewall where the PII is housed and the data is run through the de-id engine, either in batch form or using an API. This creates a one-way cryptographic hash that is sent to a matching environment that HealthVerity built, but doesn’t have access to, and is run by a third-party who locks down that environment. The cryptographic hash is compared to every cryptographic hash in the HealthVerity data ecosystem to resolve and match the records to existing HVIDs (it is very rare these days that we create a new HVID for someone we have not seen in our data ecosystem before). Only the HVIDs are shared with HealthVerity proper. The data owner maintains the PII and hash, the locked environment possesses the hash and the HVID, and HealthVerity has the HVID alone, which is returned back to the client as a child HVID in a HIPAA-compliant manner, often in a separate environment. With this secure process, if any of the environments are subject to a data breach, the hacker would only be able to access the PII by going after the data owner where that risk already exists. At every step, there’s added levels of privacy and security to lock down the data and not lock you into a solution that doesn’t meet your needs and is increasingly expensive.
De-ID that’s consistent in an instant
Another concern with de-identification solutions is that they take a substantial amount of time and that can discourage pharmaceutical companies from looking into new solutions. HealthVerity can run 1 million rows of data per second. To put that in perspective, we could de-identify the entire population of New York City, which is 8.3 million people, in 8.3 seconds.
This solution, that provides consistent de-identification in an instant, can be applied to any data sources, consolidating them into an interoperable dataset that can be leveraged in a variety of use cases across the enterprise:
- Internal data that you own and control that’s housed on internal servers
- External data from acquisitions or legacy systems that is stored in disparate silos
- Partner data that has been shared with you through a collaboration
- Data collected from sources, such as patients or providers
- Other data
Synced beats linked
As shown, the synchronized HealthVerity process beats legacy linking solutions in a number of ways:
- Accuracy
- Speed of processing
- Enhanced privacy
- Patient longitudinality
- Price to value
The value of this synchronicity is that, once your data has been accurately, efficiently, securely and economically de-identified, you can retrieve RWD for those individuals in a HIPAA-compliant manner, gaining insights about the patient before the trial starts, monitoring the patient during trial and conducting long-term follow up after the trial. There are even commercial and go-to-market benefits.
New pathways to real-time physician outreach
One way the HealthVerity synchronized solution enhances go-to-market strategies is with comprehensive, accurate and timely physician alerts or triggers. Similar to the shortfalls other tokenization solutions face, legacy trigger programs can lead to lost opportunities, wasted promotional dollars and are difficult to implement. HealthVerity Precision Event Alerts takes advantage of the breadth of the nation’s largest synchronized healthcare data ecosystem to quickly and accurately find physicians treating patients of interest to provide outreach and education efforts before a treatment decision is made. This leads to up to 30% more alerts, one to two weeks faster than legacy solutions and with 20% greater accuracy.
Opening the door to more opportunities
Most trigger solutions use open medical claims from clearinghouses, such as Change Healthcare or Waystar, that physicians use to route claims to the appropriate payer because of the short latency of weekly updates or daily for HealthVerity. This means that open claims are physician-centric and you can only see claims for the physicians using a clearinghouse that shares its data. If a patient changes doctors or is seeing multiple providers that use different clearinghouses, there could be gaps in the view of the patient journey. Legacy trigger programs do not tend to use closed payer claims, such as Aetna or Humana, because of a lack of access and the long latency of monthly to quarterly updates. Closed claims are patient-centric and provide a longitudinal view of the patient journey, no matter what doctor the patient is seeing.
Because HealthVerity has the nation’s largest fully interoperable healthcare data system, we have access to closed claims from over 150 payers, as well as open claims from the largest U.S. clearinghouses, and lab data from Quest, Labcorp and an array of other sources, representing nearly 70% of the outpatient lab market that we can leverage for physician alerts. This means we can triangulate the data to generate up to 30% more alerts. As an example, when searching for patients with diabetes using just open claims, we found 2.9 million. By leveraging additional sources, we found over 381,000 more patients for a nearly 13% increase in alerts. With the rare disease transthyretin amyloidosis (ATTR), we found nearly 38,000 patients in open claims and 9,992 more with additional sources for a 26.5% increase in alerts.
Sync fast
When it comes to physician outreach, time to influence is crucial. Because of our synchronized solution, with HealthVerity, the time from diagnosis to reporting an alert to the client is eight days. For legacy solutions, the average time is 18 to 22 days because of processing times and weekly reporting. It is also important to have the right alert at the right time and that means ensuring the patient is actually newly diagnosed, otherwise they could already be on a treatment. To determine if a patient is “new,” most legacy solutions look back through open claims, however, as discussed, open claims can have gaps in the view of the patient’s healthcare journey. By triangulating open and closed claims and lab data, HealthVerity can provide 20% greater accuracy. As an example, using only open claims to verify that a diabetes patient is newly diagnosed, we found 5.9 million alerts. But using open and closed claims to validate that a patient is truly new, we found the number dropped to 3.2 million, making Precision Event Alerts 46% more accurate than legacy solutions, getting you more of the right patients at the right time.
Precision is the right decision
The HealthVerity synchronized solution to identity management has enabled us to build the nation’s largest, interoperable healthcare data ecosystem that can be leveraged to provide the precise solutions you need to master real-world data insights.
Learn how precision can help you master real-world data insights:
