Registries in chronic disease: coming your way soon?

Registries—problems, solutions and the future The introduction of biologic agents, such as TNF-α antagonists and drugs that block co-stimulation or affect CD20 cells, has transformed the outlook for patients with many inflammatory disorders as well as those primarily affecting the musculoskeletal system. Evidence of efficacy is largely from time-honoured randomized, double-blind, placebo-controlled trials, and yet inclusion criteria as well as length of follow-up are often inadequate to inform our use of these agents in the real world outside clinical trials. Rheumatologists have been at the forefront of specialists attempting to use registries of patients with chronic diseases to study some of the difficult questions associated with the clinical use of these drugs. The editors of Rheumatology are to be congratulated on calling for submissions to fill an issue themed around the subject of registries. They have in their call for submissions correctly identified several key facets, such as problems and solutions of registry design, funding of registries and how registries can best capture treatment and treatment failures, including long-term side effects. Real-world information on the use of these agents is needed because of the frequent mismatch between research and clinical criteria of severity: an example from the world of IBD is that many patients entered into trials of anti-TNF antibodies may have high clinical activity scores (that are highly influenced by subscores of abdominal pain and frequency of bowel movements) but a normal serum CRP. For a summary of the differences between randomized controlled trials (RCTs) and registries, the reader is referred to an excellent review by Kremer et al. [1]. But there are other key aspects of registry design and maintenance that remain problematic. A crucial feature is the method of data capture and how data are recorded prospectively. Employment of dedicated staff to maintain a registry is an expensive operation that requires validation and cross-checking: the only long-term solution is operational data capture—that is, the recording of data by the clinician or other health care professional at the point of clinical contact. This allows for recording of longitudinal data much more easily and renders a registry less of a snapshot and able to record changes over time. Clearly this relates to the purpose of a registry: a specialized multinational registry of patients with very rare diseases (for instance, see the contribution of Lainka et al. [2]) will have a different structure and aim from a registry like the British Society for Rheumatology Biologics Register (BSRBR) that collects 6-monthly data on efficacy, safety, quality of life and resource use on ∼4000 patients on each of the three anti-TNF agents and a similar number of controls. Convincing busy clinicians to enter data at the time of a clinical encounter is a challenge: some consortia such as the Consortium of Rheumatology Researchers of North America (CORRONA) have addressed this by making a direct payment to participating physicians as well as a Web-based interactive reporting system that provides participating physicians with the ability to receive detailed reports on data entered on patients at their site. It is clear that the willingness of clinicians to spend time on the collection of registry data depends in part on the utility of this information in clinical practice, whether to provide evidence of meeting standards of care or to populate electronic patient records (EPRs) and facilitate generation of clinic letters where the information gathered is retrievable rather than being locked away in the medical notes. What is the evidence that registries make a difference in our ability to evaluate treatments? One important area is the ability of registries to reveal events such as infections or malignancies sometimes associated with prolonged administration of biologic agents. Several papers in the current issue, such as that by Gladman and Chandran [3], illustrate this point. Registries can also yield useful information on the time course of serious events, such as the risk of infection within the first 90 days of commencing anti-TNF therapy. There are many other areas that can usefully be addressed by a well-designed registry: these may include issues such as regional variation in access to biological therapy (post-code prescribing), waits for treatment and other aspects of inequity of access. Registries could have a role in assessing adherence to guidelines and provide useful information to inform health economic analyses and health technology appraisals. Registries can be designed around a specific disease or a specific drug or class of drugs: the choice of model depends largely on the primary objectives of the registry. As Kremer points out; ‘a drug registry with a carefully selected drug comparator cohort may be well positioned to answer pharmacoepidemiologic questions of the drug of interest, but may be less well suited to study the disease epidemiology and natural history. A registry which collects data on only one drug may have significant channeling bias, and thus be difficult to derive comparative efficacy or toxicity comparisons with other agents … regardless of whether the study uses a drug versus disease registry, it is critical to carefully consider the comorbidities and risk factors of patients in the comparator group in evaluating a drug’s toxicities’ [1]. Consent by patients is a key consideration in registry design. While data collected locally for the purposes of patient management do not constitute a registry as otherwise referred to here and consent is not mandatory; export of patient data for inclusion on a network or on the Web is usually held to necessitate both consent and anonymization preventing patient identification. If registry data might be used for research, then informed consent must be sought. Even when explicit consent is not sought (e.g. because data are anonymized), efforts should still be made to ‘inform’ patients of data collection so that they have the option to opt out if they wish (and this remains their right). Consent for non-research data collection, if obtained, may be verbal, but must be recorded in the medical records. Useful advice and further resources are available from the Medical Research Council (MRC) data and tissues toolkit [4]. Funding of registries is a contentious issue: funding by pharmaceutical companies leads to allegations of potential bias and also the question of preferential access to the data. However, there are examples in several countries where pharmaceutical sponsors subscribe for access to aggregate data while maintaining complete independence of any longitudinal or other research studies. In conclusion, registries have an emerging role in the management of patients with chronic diseases. Whether a registry becomes of lasting value depends on it being embedded in the routine process of clinical care, and this in turn depends on a careful decision as to the purpose, together with a design that allows for operational data capture and easy utilization by clinicians with the aim of informing and improving patient care. Disclosure statement: The author has declared no conflicts of interest.