Marketing, patient data, and privacy concerns
*Anonymous patient-level data (APLD) is an attractive information source for pharma marketers, but its acquisition and use are complex.*
By on Feb 14, 2011Anonymous patient-level data (APLD) is an attractive information source for pharma marketers, but its acquisition and use are complex. Angelo DePalma reports on what Europe can learn from Americas experience with APLD
Anonymous patient-level data (APLD) has been a staple of US healthcare utilization analysis for at least a decade.
Numerous firms provide this data, which is collected from cooperating physicians, hospitals, and pharmacies, and through patient opt-in.
Some firms provide raw numbers on prescriptions, procedures, and other utilizations; others offer data that has been massaged or collected from several sources to provide a more comprehensive picture of patient behavior.
APLD has become an attractive short cut for pharma marketers, but its acquisition and usefulness are by no means straightforward.
Data firms face hurdles connected with licensing, data processing, aggregation, analysis, and privacy management; customers considerations include the datas cost, market size for the target medicine, indication, and level of competition.
Antidepressants, for example, prescribed on- and off-label for several illnesses, are more difficult to track than cancer drugs or blood pressure medications.
David Bernard, managing director at DB Marketing Technologies, calls APLD essential for pharmaceutical marketers and the only measure of patient behavior over time and against the competition.
How APLD works
While APLD may be generated by any recorded or reimbursed health-related event, prescription drug data is by far the most advanced technologically.
Data is collected when patients present their benefit cards to a pharmacy.
Apothecaries or health plans sell the data to large data firms like IMS Health, which in turn offer them to pharmaceutical clients or aggregators who massage the data for resale.
Many APLD data sources exist in the US, each with differing coverage by product and region and variable levels of accuracy and data cleanliness.
Data from plans is considered top-tier since insurers have standardized how data is collected and presented and can offer cross-sections that aggregators cannot; for example: non-smoking type 2 diabetics over 50 years of age who are also taking blood pressure medication.
The downside is, such data is limited to that particular plans patients.
IMS provides pure, fill-level transactional data.
Marketers prefer data that is formatted, massaged, or annotated to meet their specific needs.
At the high end of granularity are firms like Medicx, an aggregator of aggregators that collects data from compendial sources like SDI, the US Census Office, and consumer marketing agencies.
Medicx claims to map medicine-related behavioral patterns down to the zip+4 level, roughly the geographic footprint of a city block.
Somewhere in the middle are healthcare information companies like DB Marketing Technologies and SDI.
SDI uses a linking code to identify patients uniquely, persistently, and anonymously at all points in their journeys through the US healthcare system.
SDI aggregates prescriptions, doctor visits, and hospital stays.
SDIs data is known as an open source dataset, available from anyone or any organization willing to provide it anonymously.
But completeness is not something that is achievable, or even desirable.
Aside from behaviors occurring outside the watchful eyes of aggregators, there are built-in limitations to how much information is capturable and issues of timeliness as well.
Were happy not to know how complete [the data] is to ensure patient privacy, says Jody Fisher, VP of marketing at SDI.
If you know its complete, youre that much closer to knowing everything there is to know about an individual.
The closest one can get to completeness are records at insurance companies that pay for every significant health-related expense for a particular patient.
No data aggregation scheme is iron clad, however.
Uncertainties arise when patients move, die, switch or drop out of plans, take a medication holiday, or purchase inexpensive drugs outside their plans.
The time between pharmacy fill and data acquisition can be several months, so months can pass before anyone realizes that patient 12345 has been lost.
There are also data aggregation errors and the issue of data arriving late, Bernard of DB Marketing Technologies notes.
APLD under fire
APLD is generally non-existent in Europe and under fire in the US.
In the US, the 1996 Health Insurance Portability and Accountability Act (HIPPA) governs how drug firms and data aggregators handle patient data.
HIPAA mandates that patients not be identifiable, either directly or by inference.
So, if a customer asks for analysis of patient data by granular segments, data providers must mask certain data fields in addition to those stripped out at the collection point.
The most recent issues with privacy and government regulations began with the privacy concerns of physician data.
Typically, pharmaceutical companies use prescribing data to help manage their sales force, including the distribution of free samples to physicians.
While this had been going on for years, a vocal minority of doctors realized that pharmaceutical companies were using this data to targetand, in some cases, punishthem for their prescribing behavior.
So, in 2006, New Hampshire passed the New Hampshire Prescription Confidentiality Act, which bans the sale of physician-level data to pharmaceutical companies.
The other justification for restricting data, also physician-based but rationalized on the basis of cost, is to prevent branded products from achieving an unfair advantage over generics.
The downside to the majority of physicians, however, is a drop in the number of free samples they receive.
Doctors dont like to see detailers, but they do like their samples, DB Marketing Technologies Bernard says.
Ironically, one of the biggest suppliers of physician data is the American Medical Association, by far the largest US physician group.
According to data from the competing American Medical Student Association, AMA receives close to $50 million per year selling physician data to aggregators.
The top data providersIMS, Verispan (now SDI), and Wolters Kluwersued New Hampshire, claiming its law was unconstitutional.
A Federal Court overturned the law, but its decision was reversed on appeal, so the law now stands.
The upsides of APLD
While initial challenges predominantly focused on physician data, other states drafted laws that restricted the use of all data, not just from doctors.
So far, only Maine and Vermont have passed laws similar to the New Hampshire statute, but these are being disputed.
Meanwhile, US aggregators and physician groups are looking for alternatives, such as providing physician-level data without identifying specific doctors, allowing doctors to opt out, or giving them more control over what types of data may be acquired and distributed.
One could argue that generic drug houses could also use the data to their benefit.
But, in fact, generic drug houses have almost no incentive for acquiring APLD since they have numerous competitors selling precisely the same drug, whereas branded pharmaceuticals are unique.
Nothing, after all, prevents doctors from prescribing a generic that is different from that of the company purchasing the data.
The premise for banning physician data is questionable.
If they need such information, drug companies will obtain it one way or another, either through aggregators or by more-aggressive detailing, just as they did before aggregators began collecting it.
Another objection to the objectors is the touchy subject of improved health care delivery.
Tracking prescribing habits, some say, supports post-marketing safety studies that have increasingly become contingencies for new US drug approvals.
EU and US differences
The success of patient data aggregation in the US is due to a unique confluence of factors.
The fact that you have electronic data available at all is a byproduct of how our insurance system works, SDIs Fisher notes.
Even if European laws permitted APLD on the scale that US regulators do, it would take many years for APLD systems to take hold on the Continent.
The reason, Fisher says, is that the US system is highly computerized and uniform, a fact that every aggregator exploits fully.
The entire data flow within an insurance company is standardized, regardless of where the service or product was acquired.
European data streams are more diverse, both within and among payers.
There are also significant cultural differences.
Europeans appear to be less tolerant of large-scale pharmaceutical marketing.
Europe, moreover, has no reason to change due to its stronger sense of personal privacy and built-in differences among systems, according to Fisher.
One possible solution for cracking the European data wall is through patient-driven opt-in services, like the Internet-only OptimizeRx.
Patients browsing the companys website receive free drug samples, disease information, and substantial price reductions on common medicines in return for providing basic information on health and drug-taking habits.
Another opt-in service, CloserLook, collects information from physicians who spend time online.
Neither service currently operates in the EU.
Europeans are much more likely to obtain their prescriptions and medications from their physicians; US patients receive at most a teaser prescription during an office visit, but fill the script at drug stores.
Simply having to walk from the doctors office to the pharmacist creates adherence barriers that become magnified with each prescription refill.
Co-pay surprises are responsible for approximately one in five unfilled prescriptions in the US.
Patients who might otherwise leave pharmacy prescriptions unfilled are much less likely, Fisher says, to say no to their doctors.
What can drug companies do to overcome the absence of APLD in Europe?
In some instances, it is possible to extrapolate data from one jurisdiction to another, but the validity of such exercises depends on the product, market, and demographics, as all European payer systems are unique.
Fisher believes data acquired in US markets can enable drug firms to make educated guesses about behavior overseas.
But they wouldnt be able to narrow down the conclusions due to regional differences, Fisher says.
There certainly are things companies can learn from US data, but there isnt necessarily a direct correlation to other jurisdictions, concludes Bernard of DB Marketing Technologies.