Introduction. Efforts to limit the commercial use of data on physician prescribing could have broad implications for regulatory programs that promote drug safety. At issue are a series of laws pursued by state legislators that would restrict access to information on the prescriptions written by individual doctors. This information has long been aggregated, and sold by a number of database companies, principal among them IMS Health. The data has commercial value, supporting the marketing efforts of drug makers. That has made the collection of this information a target of those who would seek to curtail pharmaceutical marketing practices. But the data has also become an integral part of regulatory efforts to promote drug safety.
The IMS data is a part of many drug safety analyses conducted by the Food and Drug Administration (FDA). The data also forms a core component of an increasing number of FDA's risk management plans that regulate the use of drugs after they are approved in order to mitigate the risks. If access to this information is restricted it will significantly undermine FDA's ability to monitor, and respond to, drug safety challenges.
IMS Health is a data miner. It buys prescription data from pharmacies. The information on the patients is removed to preserve patient privacy and comply with strict laws that safeguard patient medical data. But the identities of individual doctors remain attached to the prescribing data. Once aggregated, the information forms a valuable record of the drugs that are prescribed worldwide and the clinical conditions under which they are used.
A lot of this information is sold to drug companies. The drug makers use the details on specific doctors, their locations, and the drugs they prescribe to target education programs, conduct studies to evaluate the safety and effectiveness of their medicines, and monitor for drug side effects. The data is also integral to many risk management programs that are aimed at regulating the way drugs are prescribed to reduce safety risks.
But drug companies also use the data to target their marketing efforts. This is one of the principal commercial uses of such information. As a result, the IMS data has become subject to scrutiny by government officials who want to reduce healthcare costs and see pharmaceutical marketing as one driver of rising state and federal spending on prescription drugs.
The laws aimed at restricting the collection and sale of the prescribing data are born of these efforts to cut healthcare costs. These laws have also won support of some medical societies and doctors who believe that their prescription data should be kept private. Both of these motivations are, in many respects, lacking.
In oral arguments of April 2011, the United States Supreme Court Justices appeared to be highly skeptical of the constitutionality of a Vermont law challenging the ability of private firms to collect this kind of data. Supreme Court Chief Justice John Roberts signaled his views this way during oral arguments: The government, he said, wants to "lower your health care costs, not by direct regulation, but by restricting the flow of information to the doctors by, to use a pejorative word, censoring what they can hear to make sure they don't have full information."
As for doctors' concerns that the collection of prescribing data is an inappropriate intrusion into their practices, the bottom line is that little of the data describing clinical practice is private. Billing information on individual medical encounters has long been collected and sold for commercial purposes. Information on the care of Medicare recipients is made available to commercial and academic researchers who use the data to form detailed analyses on the care of individual patients. Using free Medicare billing data, The Wall Street Journal was recently able to write a series of articles on the clinical practices of individual physicians.1-4
While legislators' efforts to curtail collection of this data may fall short of their economic goals, such effort could have far-reaching implications for public health prerogatives. The prescribing data has many valuable uses beyond its commercial applications to pharmaceutical marketing efforts. Yet it is that business value to the drug makers that helps to underwrite the expensive collection of this information for non-commercial purposes.
The FDA's use of the IMS data to help in its drug safety evaluations illustrates many of the public health applications of this information, not only to worldwide drug regulators, but also to other public health authorities, manufacturers, and researchers.
FDA uses the prescriber-level data in three major ways:
- The prescribing data provides one of the most precise measures available to more accurately gauge the incidence of drug side effects (a measure of the risk of developing a given side effect within a specified period of time) and prevalence (the total number of cases of a side effect in the population at a given time).
- The prescribing information is also used to help target drug safety updates to doctors. It allows FDA and drug companies to direct warnings and safety updates to doctors who are most likely to prescribe a certain medicine.
- Finally, the data also helps companies target post-market safety programs. The FDA is mandating many of these programs. The aim of the risk management plans is to help make sure doctors prescribe drugs in ways that minimize known side effects.
This data is a vital tool for promoting public health goals. There is no viable substitute. If aggregation of this sort of information falls victim to some of the recent political efforts, many important programs for assessing and promoting drug safety could be jeopardized.
Prescribing Data as a Public Health Tool. One way that the IMS data serves as an important public health tool is through its use in the evaluation of adverse drug reactions. The FDA's Center for Drug Evaluation and Research (CDER) generally lacks quality data from government sources about the utilization of drugs. Without accurate data on utilization, even if FDA has numbers on reported adverse events, the agency is still unable to measure risk - the probability of a specific adverse event arising from the use of a drug. In drug safety evaluations by CDER's Office of Surveillance and Epidemiology, IMS data on drug prescriptions typically comprise the "denominator" in safety evaluations. In other words, FDA will have data on a certain number of observed events believed to represent adverse reactions to drugs - the "numerator." But without specific information on the magnitude of not only drugs prescribed, but also prescriptions actually filled and, in turn, taken by patients, FDA has no way to accurately gauge the proportion of adverse events to pills taken.
From the data that FDA collects through its Adverse Event Reporting System the agency has a measure of the count of adverse events, on a monthly or yearly basis, associated with the use of different drug products. But it is impossible for FDA to interpret changes in such counts over time, or differences in such counts between drugs, without also using a measure of utilization (or exposure) to the medicine.
For this purpose, IMS serves as the only reliable source of high quality data because it aggregates information on quantities sold and enables FDA to approximate utilization by looking at prescriptions filled and re-filled. The IMS information further allows FDA to look at adverse events by region to correlate the reports with specific sources of distribution. This can help in the rare instance where an adverse event might be related to the composition of the drug or a glitch in its manufacturing (i.e., contamination of the drug, or introduction of counterfeits into the supply chain that are adulterated and being confused with the real drug).
Among the unique data that IMS is able to provide to FDA is information on quantities of a drug sold. This is a good indicator of utilization that IMS can measure as either prescriptions written and filled (30 day, 60 day, and 90 day), units sold (both in terms of bottles or packages), or "eaches" (the number of individual pills, tablets or capsules or injections delivered). IMS is also able to measure and report on sales for different prescription strengths. Finally, IMS can calculate the total weight (or "kg") of a medicine sold by summing the dose across all of the different dispensed units and strengths of a drug.
This kind of data provides a very specific measure of the total exposure to an active pharmaceutical ingredient and has already been demonstrated in the medical literature,5,6 and by FDA, to provide an irreplaceable resource in establishing causality (the likelihood that exposure to a drug is responsible for a given side effect, and the circumstances and rate under which that side effect is likely to occur). Drug exposure is reported by IMS as number of outpatient prescriptions. By monitoring prescriptions, the IMS data can be trended over calendar time, and also stratified by age, gender, and indication for use.
When used in conjunction with supplemental data obtained from population-based claims or databases linked to medical records, it is possible for FDA to estimate the actual number of patients exposed to a drug. The FDA reports that it uses this data regularly in association with spontaneous case reports data to understand the context within which adverse drug reactions occur - how the side effect was associated with a drug's use.7 Such information can be important to teasing out the exact relationship between a drug and a side effect, for example, determining whether a side effect is dependent on total exposure to a drug (i.e., the absolute amount of drug) or on the length of time of the exposure.
Because of the quality and breadth of the IMS data, the information forms an ongoing component of many FDA analyses. Most presentations on drug safety to the agency's advisory committees include data on utilization that is drawn from IMS. Data on off-label use is typically also derived from IMS. This allows the FDA to evaluate drug safety signals, not just by the indication for which the drug is prescribed, but also by patient characteristics (for example, comorbid illnesses) that might contribute to unmasking a certain side effect, or by other drugs that a patient might be on that led to untoward drug-drug interactions.
Data on trends in use of unapproved (illegally imported) drugs also comes from IMS. Finally, among other noteworthy uses, the IMS data is also utilized for some very specific kinds of drug analyses. For example, the evaluations conducted by the FDA team that reviews drugs for use during pregnancy rely, in part, on information provided by IMS, such as data on the percentage of drugs in a certain class prescribed for women who are pregnant or lactating. This data enables FDA to correlate individual drug use with information on whether a patient was pregnant or lactating when prescribed a medicine.
To provide more accurate evaluations of drug adverse events, IMS data also comes into play. IMS provides FDA with data on panels of patients with known conditions. This enables FDA to follow how adverse event profiles might change as patients switch among various drug products approved for the same indication. It gives FDA a tool for "active" surveillance of drug adverse events, instead of being forced to rely on a "passive" collecting system where FDA is dependent on busy physicians filing reports to the agency.
Examples of FDA Analyses that Rely on Provider Prescribing Data. One tangible example of how IMS data is used by FDA relates to the databases of prescriptions collected from nearly all of the retail pharmacies in the U.S. These databases - the National Disease and Therapeutic Index (NDTI) and the National Prescription Audit (NPA) - are the principal components of FDA's active postmarketing surveillance activities. The NPA provides national estimates of prescription drug use in the U.S. based on data collected from over 37,000 computerized retail, chain, grocery, and mail order pharmacies nationwide. The FDA uses this data to track prescribing trends over calendar time and for calculation of adverse drug reaction reporting rates. The NDTI database provides information on the age and gender distribution of patients using outpatient drug products, as well as information on the duration of treatment course and indication for use.
The FDA uses these large population-based databases in a variety of specific areas, including hypothesis testing, signal refinement, and for the confirmation of signals emerging from spontaneous case reports filed by doctors and patients. In addition, they are also used by the agency to assemble case series and retrospective cohort and case-control studies to estimate relative risks and identify important risk factors. Finally, this kind of real-time monitoring has become an important tool, not only for drug safety management, but also for disease surveillance. For example, seasonal tracking of infectious disease such as flu or gastrointestinal illnesses can be done using prescribing trends that might be an early tip off to an emerging epidemic. This information is also vital to biodefense monitoring.
Provider Prescribing Data as a Tool of Risk Management. The prescriber-level data provided by IMS also forms an important tool for enabling targeted risk management programs. The FDA mandates many of these risk management programs as part of new drug approval requirements created by drug safety legislation that Congress passed in 2007. The Food and Drug Administration Amendments Act authorizes FDA to require manufacturers to implement programs aimed at managing how doctors prescribe drugs thought to have modifiable risk factors. These programs are referred to as Risk Evaluation and Mitigation Strategies (REMS).
Today, more than 160 drugs have REMS associated with their use. More than 40 of these risk management programs require that drug companies target specific drug safety information to those providers most likely to prescribe the particular medicine. Over 20 of these programs have more prescriptive features that require drug makers to actively manage how doctors prescribe the medicines with interventions aimed at, for example, patient education or specific clinical monitoring of patients while on the drug. These features are referred to in the 2007 legislation as Elements to Assure Safe Use (ETASU).
The REMS programs end up being highly individualized for each dug. They are based on the ability of the drug makers to target prescribing physicians with informational resources as well as other risk mitigation programs that are mandated by FDA. These REMS programs encompass a broad range of tools, including the provision of Medication Guides or Patient Package Inserts; structured communication plans (methods for how information about risks of a medication and steps that can be taken to mitigate them should be shared with physicians and/or patients); and the more prescriptive Elements to Assure Safe Use. These ETASU might include: programs to ensure that health providers who prescribe or dispense a medication have particular training or experience; the restriction of distribution of a drug only to doctors who are trained in using the medicine and agree to certain prescribing requirements such as scheduled blood tests to monitor patients for side effects; and the tracking of those patients prescribed a certain drug through registries or other kinds of monitoring programs.
The Key Role of Data in Recent Safety Programs. Among the more recent, notable REMS programs and safety assessments that were dependent to a significant degree on data collected from IMS were: a REMS to help manage the safe use of narcotic painkillers,8 a recent analysis by FDA of the safety of Flovent® (fluticasone propionate) which formed the basis of new post-market safety requirements imposed by the agency on that and similar medicines,9 and a risk management plan implemented upon approval of the novel bone drug Prolia® (denosumab)10 which is used for treatment of osteoporosis as well as cancer.
For these risk management purposes, the provider-level data provided by IMS is an essential tool. The FDA details on its website a brief sampling of some of the specific circumstances under which provider-level prescribing data is used to support risk mitigation plans. The FDA cites the following examples of how the agency's drug program makes routine collaborative use of data from IMS Health to support regulatory actions:
- Used extensively in the review of terfenadine prior to its withdrawal from the market
- Used routinely to evaluate effect of prescription-to-OTC switches
- Used to evaluate Diprivan® (propofol), a generic drug marketed with a different preservative from the innovator product
- Used to provide data regarding schedule II and other drugs with abuse potential for use in determining manufacturing quotas
- Used to provide sales figures to support cost/benefit arguments for Prescription Drug User Fee Act (PDUFA) II. Demographic data has been used to provide support decisions concerning the Medicare formulary
- Used in providing demographic patterns of use in support of congressionally-mandated initiatives to encourage study in various subpopulations.7
The life science trade association BIOCOM recently published a white paper on the role that provider-level data played in enabling a complex REMS program that was implemented around the approval of the diabetes drug Symlin®.11 Symlin is a highly novel drug that works as an insulin sensitizer. This means that when it is used in conjunction with insulin, Symlin helps potentiate the body to the effects of insulin.12 But Symlin also has a narrow therapeutic margin; when insulin isn't prescribed carefully in conjunction with Symlin, the combination of the two can cause dangerously low blood sugars.
As a result, FDA was reluctant to approve Symlin unless it was used under strict conditions that enabled careful patient selection and education of both doctors and patients on how to use the new medicine. Using physician-level data, the drug's manufacturer, Amylin Pharmaceuticals, was able to identify physicians who were the highest prescribers of insulin. These physicians were then surveyed to determine whether or not their practices included diabetes educators who could help assist in educating patients on how to use Symlin. Approximately 22,000 physicians were determined to be the most appropriate recipients for information about the new drug, providing the right direction for representatives to reach the right physicians. The new drug was targeted to these doctors on the assumption that they would have the training and resources to prescribe Symlin under safe conditions.
Indeed, the design of REMS by FDA, and the entire concept for that matter, was predicated on the availability of this kind of prescribing information.13 Absent provider-level data on prescriptions written, many of these risk management programs simply could not exist.
Safety Issues Blur Line Between Non-Commercial and Commercial Use. While some may consider safety communication to be a non-commercial use of data, the legislation being developed in some states to restrict access to provider-level data does not easily accommodate these kinds of distinctions. Even if legislation were written to allow provider-level data to be collected and used for drug safety purposes, but not for marketing, it is unlikely such a distinction could be easily interpreted, or that the resources would be available for the information to be aggregated in the first place.
Moreover, drug sales representatives are increasingly called upon as part of REMS to deliver targeted safety information to physicians. Even if a distinction between commercial and non-commercial uses of provider prescribing data could be written and interpreted, the tools for distributing marketing versus safety information are increasingly being blurred in the marketplace as drug companies and regulatory authorities alike struggle to find opportunities to interact with doctors in order to deliver safety information.
The Bottom Line. The bottom line is that collection of this kind of provider-level data is a difficult and resource-intensive endeavor. That is why programs like Medicare and Medicaid struggle to obtain data on how their own members are medically managed. Medicare has tried for years just to get information on how a particular patient courses through a single episode of care. This challenge has stymied the program from implementing popular reimbursement initiatives that tie payment more closely to outcomes. The fact is that the commercial value of the IMS data underwrites the tools and systems that enable the aggregation of this information in the first place. Absent that commercial use, it is unlikely IMS - or any similar provider - will be able to invest the resources necessary to enable the continued collection of this information. The data will simply not be available, whether it is for marketing uses, or targeted drug safety programs.
It is widely acknowledged that efforts to restrict access to provider-level prescribing data result from a focus on drug spending and rising healthcare costs. Whether or not this information promotes more effective marketing, which in turn drives higher use of more expensive drugs, is beyond the scope of this paper and really is not the point. The point here is to make note of the fact that the provider-level prescribing data fulfills many ambitions, some of them with significant public health implications.
There is no alternative source for this information. The failure of government agencies to develop similar data tools is one testament to the cost and difficulty of implementing the systems needed to reliably track this sort of information. If pharmaceutical critics aim to challenge drug industry marketing practices by restricting access to this kind of information, they should consider the fallout, as well as the likelihood that once collection and access to this information is restricted through legislation, there is unlikely to be any substitute.
Scott Gottlieb is a resident fellow at AEI.