Direct-to-Patient Registries: A New Approach to Pharmacovigilance

Understanding drug safety is a tricky thing. It starts when new drugs are being developed and tested through the clinical trial paradigm. In the highly structured, homogeneous setting of a randomized controlled clinical trial, it is reasonably likely that safety events occurring in a treated group, but not in comparable patients who receive a placebo, are caused by the product under study. Once new products come to market, however, they are generally used quite broadly. Unlike the optimal patients, practices and settings that are used in drug development, products that have gained market authorization may be used in types of people that were never directly studied during drug development – like children, the elderly and people with multiple comorbidities requiring regular medication use.

Consider pregnant women, for example. We know that most pregnant women take at least one prescription drug at some time during pregnancy, and yet little systematically evaluated information is available about the safety of such use for the mothers and their unborn babies. Unless a product is being developed specifically for use in pregnancy, it is generally considered unethical to conduct experiments in pregnancy because the benefits rarely outweigh the potential risks of birth defects. So while researchers need to be prudent in terms of deliberately exposing women to potential teratogens, this doesn’t help women faced with decisions about medical treatments during pregnancy. This is particularly important for women with chronic diseases in need of regular treatment, who face tough choices about trade-offs between their health and that of their unborn children, without much evidence to guide their decision-making. Right now, the “best” guides they have to understanding drug safety are from animal experiments, but most would consider this to be a shaky ground for predicting outcomes in humans.

Currently, most postmarketing drug safety information regarding drug use during pregnancy comes either from astute clinical observations of case series or more often, from analysis of large databases like those with health insurance claims where mothers and babies can be linked, and using data from electronic medical records. These data sources generally contain information about medications used during pregnancy and doctor visits for the mother and newborn. Even using epidemiologic approaches based only on these scant bits of information, we have been able to identify strong safety signals about dangers of medication use that were not detected in clinical development. However, these tools are blunt instruments, in that they can detect strong effects, like those observed with maternal stilbestrol use (a relative risk of infinity for exposed cases) but otherwise, they only provide hints that require confirmation through replication and confirmation.

Read More: The Race to the Cloud - Changes Afoot for Life Science Research

Further, studying records of prescribed medications and events that come to medical attention only addresses part of the story about whether a medication is safe to use. For example, using these existing data like health insurance claims information, you cannot tell whether a person actually took the prescribed medication and what other factors about their lives and personal habits, if any, that may have contributed to any adverse events that may be observed. The risks that come from non-prescription medications, lifestyle and other factors, which patients may or may not share with their doctors and which may affect the outcome of pregnancy, are largely unknown.

At Quintiles, we are seeing promise in ‘direct-to-patient registries’ as a means to offset this challenge. Direct-to-patient registries allow consumers to record health information outside of the clinical study experience, enabling researchers to glean previously untapped insights that account for information derived from the patient’s own experience, such as details about what treatments they use, including those that were prescribed as well as any other treatments they may be using (for example, non-prescriptions drugs, homeopathic treatments or herbal products or meditation); how well they adhere to treatments prescribed by their doctors, and why or why not; as well as to capture information about their personal habits, home environment and workplace. Using electronic data capture approaches, we find that patients will report information that they may not share with their health care providers.

Direct-to-patient registries in action: The PROTECT case study

The use of direct-to-patient (or consumer) registry data is a new approach, and one that is quite different from the highly controlled world of clinical trials. Some worry about whether patients will be accurate and honest reporters. Will these patients be reliable, and will they provide clinically generalizable information? With these questions in mind, Quintiles embarked on the PROTECT (Pharmacoepidemiological Research on Outcomes of Therapeutics by a European ConsorTium) study in collaboration with the European Medicines Agency, and a number of academic centers and pharmaceutical companies. Funded by the European Commission, the PROTECT program addressed many aspects of understanding drug safety. One part of the PROTECT program was a study to assess the extent to which data collected directly from pregnant women could provide information on medication use and other potential risk factors throughout pregnancy, and would be suitable for research purposes.

Quintiles’ part of this PROTECT program was as co-leader of this direct-to-patient non-interventional study in which we invited pregnant women in Denmark, the UK, the Netherlands and Poland to share information in a registry about their medical history, medications taken during their pregnancy, lifestyle factors that may have an impact on birth outcome (like use of alcohol, recreational drugs, etc.), and family history. These patients had the option to report via the IVRS voice system or the Internet, and they could choose to report every two weeks or every four. With a small budget for advertisement, we utilized leaflets in pharmacies near pregnancy testing kits that directed women to a website where they could learn about the study in their predominant language, as well as Internet ads and use of email lists from pregnancy clubs.

The first thing we found with this approach was that recruitment was not as simple – or inexpensive – as we thought. Of the 2,000-plus patients recruited, fewer than 200 came from unpaid recruitment efforts. We also found that although respondents were split fairly evenly on the preference to report every two weeks versus every four, those who reported more frequently were slightly less likely to drop out of the study. Internet reporting was heavily preferred, with only a single respondent using the IVRS system.

However, the “when” and the “what” reported by these women was particularly promising. Five hundred and twenty-nine of our respondents were recruited during the first trimester of pregnancy, an important time in terms of fetal development and a time when they may not have even presented to a physician yet. We found that these women were happy to report the medicines they were and were not taking. For example, nearly five percent reported that they filled a prescription and then decided not to take it, while more than 20 percent used non-prescription drugs and herbal medicines. There also appeared to be little hesitancy to report exposure to potentially harmful substances; around five percent continued smoking throughout the pregnancy, and when we addressed marijuana, cocaine and other recreational drugs, we saw a surprising relative consistency of women across all four countries who were willing to report such use. For those who kept reporting throughout the duration of their pregnancy, we also collected information on birth outcome – which was consistent with traditional physician reports.

And yet, even with this wealth of information, we must ask ourselves: Is this data reasonably accurate? Are the results generalizable and reliable enough for pharmacovigilance purposes? We compared self-reported prescription medication use and birth outcome with national registries in Denmark and with electronic medical record data in the UK, and found reasonably good consistency with self-reported data. Based on our findings, the answer to all of these questions appears to be “yes, these data are reasonably accurate” – although that does not eliminate the need for medical validation of these outcomes to obtain clinical information from a medically trained reporter.

With the rapid advance of technology, we have seen a number of non-traditional approaches to AE reporting cropping up, including text messaging. In Nigeria for example, the National Pharmacovigilance Centre has implemented a mobile phone pharmacovigilance alert system for consumers aimed at improving knowledge about the 6.3 to 49.5 percent of hospitalized patients there that reported AE reactions to drugs. In addition, the web-based solution offered by MyMeds&Me; offers an intuitive, standardized method of capturing AE data via SaaS. Patients can simply point and click on “where it hurts” to generate direct-from-patient data nearly instantly.

What we can say for certain is that used in conjunction with clinical trial research, direct-to-patient registries can be a tremendously valuable tool to inform drug safety by obtaining information that is hard to come by otherwise – particularly co-medications, medical compliance, illicit drug use and other behaviors that patients may not be willing to tell health care professionals. With the era of Big Data and new technology has come opportunities to approach pharmacovigilance with a fresh perspective and glean insights into AE that have eluded us in the past. We would be wise to embrace and explore these opportunities for the sake of improved patient outcomes.