In this issue of CMRO, the article entitled ‘Expanding the use of administrative claims databases in conducting clinical real-world evidence studies in multiple sclerosis’Citation1 provides a concrete example of an exciting vision of the future – a future in which we can effectively harness the power of large automated multipurpose population-based systems (LAMPS) to enlighten our decision making with evidence derived from the real-world, day-to-day experiences of our medical and health activities. Many of these LAMPS have evolved from the old days of automated accounting systems, created solely for the purposes of management and billing, often for a single function such as pharmacy. Now, many of these elegant systems describe the full spectrum of medical transactions linked across entire enrolled populations and across the whole spectrum of service, from prevention and early diagnosis through presumption and work-up to care and disposition. Key to the usefulness of these data for purposes other than billing and accounting is the ability to link these bits of information to the individual patients receiving the services, and to their experiences over time. And key to their actual use for real-time research is the assurance that LAMPS are now equipped with the protection that will assure privacy for individual patients, while the world learns from their ‘real world experience’.
Capkun et al. make it clear that we are not quite ‘there’ yet. They analyze two administrative databases and depict the differences in rates of ascertainment of something as ostensibly clear as an incident diagnosis of a chronic disease and monitoring of its recurrence, using the ‘tracer’ of multiple sclerosis. For some such diagnoses, a single coded event or event pair, e.g. cataract and surgery for cataract extraction, might be enough to permit confidence that the experience being portrayed actually represents the illness being treated.
Likewise, a diagnosis for a treatable disease and the prescription for a targeted medication might be convincing, e.g. herpes and acyclovir.
The authors illustrate the complexity of ascertainment and propose an algorithm (or two) which should be used going forward to help the world of real-world researchers to apply consistent definitions and hence develop data worthy of comparison. And they point out that, as we become better able to do these things, the world will also be better able to describe with much greater confidence the natural history of treated and untreated diseases. No longer will we need to rely on personal anecdote and recall, ‘conventional wisdom’, unsubstantiated claims even in definitive textbooks, or reports from small studies of short duration.
The duration of the ‘natural history study’ is limited by the duration of the individual within a single population database. The ‘power’ of the study is limited by the prevalence of the disease, engagement of the treaters and their medical practices, and the vagaries of each specific database. Not all LAMPS are created equal!
Regarding the limitations of many of the administrative databases, the authors rightly point to the US Department of Defense (DoD) database as a ‘cradle-to-grave’ system (not quite, but a good turn of phrase), meaning whole populations and their offspring/extended families receive medical coverage and payments for long periods of time. The DoD system has moved aggressively toward full automation of their medical records, which permits much richer and more nuanced evidence to be brought to bear on longitudinal analysis, and, as they attest, ‘confirmation’ of associations such as those drawn from administrative data alone. The automated medical record is also more likely to capture important personal information about patients, such as tobacco or alcohol use, or consumption of non-prescription drugs, vitamins, or herbal remedies – attributes which might otherwise confound analyses.
As these LAMPS evolve, so too will the users and uses of the data emerging from them. Indeed, the next decade will be one of the ‘learning health system’. In the learning health system, actual experience (real world evidence) can be examined and used to test hypotheses, follow up signals of potential problems (e.g. adverse drug events), clarify expectations and outcomes of diseases and even evaluate interventions in actual use populations. By extension, it won’t be long before the data from LAMPS can be used for a paperless clinical trial system, in which all parameters under study – even randomization status – will be captured (along with any other medical information) in the same automated patient database. LAMPS will permit the care-giving system to learn from ‘real’ practice patterns about the effective uses of differing approaches to delivery of care among diverse practitioners, and allow for large scale comparative studies at costs lower than ever before imagined. This is already the case with the revolutionary PCORI (Patient Centered Outcome Research Institute) program in the US.
In short, instead of throwing away administrative and other transaction data after the books are balanced or the year-end report is written, we will harness our experiences, link them with other powerful sources of information on the same persons, and plow what we find back into the field as dedicated continuing students of a vastly improved ‘real world’!
Transparency
Declaration of funding
This editorial was not funded.
Declaration of financial/other relationships
H.T. has disclosed that he is a member of the Novartis-sponsored Gilenya multiple sclerosis pregnancy advisory board, a registry that does not use claims data.
Reference
- Capkun G, Lahoz R, Verdun E, et al. Expanding the use of administrative claims databases in conducting clinical real-world evidence studies in multiple sclerosis. Curr Med Res Opin 2015: published online 7 February 2015, doi: 10.1185/03007995.2015.1014029