Overview of North American Databases

Overview of North American Databases

INTRODUCTION

Large electronic databases can often meet the need for a cost-effective and efficient means of conducting surveillance after a new drug is marketed, and to establish baseline data prior to marketing. Such databases can be used to assess signals from trials and from spontaneous reports, and, given adequate power and design, the results of such assessments generally are more credible than evidence from spon-taneous reports of the same problem (Edwards, Faich and Tilson, 2005). Large databases are often needed to address acute and serious regulatory, commercial and public health crises. Post-marketing studies of drug effects must generally include at least 10 000 exposed persons within a definable population base.

In North America, databases able to meet this need are primarily administrative in origin, generated by the request for payments, or claims, for clinical services and therapies. Large databases of electronic medical records, like the General Practice Research Database (GPRD) in the UK (Gelfand, Margolis and Dattani, 2005), do not exist yet in North America. The resulting databases of health insurance claims are inherently different from medical record databases (Strom, 2005). Health insurance in the United States is typically obtained through one’s place of employment, and does not always include coverage for prescrip-tion drugs. The instability of this system is caused by employees’ changing jobs and employers’ chang-ing health plans and coverage for specific employees and their families. The opportunity for longitudinal analyses is thereby hindered by the continual enroll-ment and disenrollment of plan members. However, strategies have been developed for selecting stable populations within a specific database and for address-ing compliance, such as examining patterns of refills for chronically used medications. Most of the US health care programmes described in this chapter are employee-based, but may offer coverage to Medi-caid and Medicare recipients as well, providing some representation of the elderly and the economically disadvantaged in its databases.

Beyond the employer-based health insurance programmes are the US Medicaid programme, which provides medical care, including prescription drug coverage, for economically disadvantaged and disabled persons, and the Medicare programme, which provides health care to persons aged 65 and older. The latter is now undergoing changes to include a prescription drug programme. In contrast to the vari-ety of health care systems for selected eligible subsets of populations in the United States, Saskatchewan, a province in Canada, provides a publicly funded health system for all of its residents.

To meet the needs of drug surveillance and phar-macoepidemiologic studies, claims data from multiple sources (drug purchases, visits to physicians, hospi-tal stays, etc.) must be linked on a per patient basis. Depending on the nature of the study, records from the following sources may need to be included: inpa-tient and outpatient care, emergency care, mental health care, laboratory and radiological tests, alter-native therapies and prescribed and over-the-counter medications. The size of the population covered by the database must be large enough to permit discov-ery of rare events for the drug(s) (in surveillance studies); questions such as the stability of the popu-lation and the completeness of therapies and clini-cal services obtained solely through the health plan may be considerations in study design. Although it is generally preferable for the population included in the database to be representative of the general population from which it is drawn, it may sometimes be advan-tageous to emphasize the more disadvantaged groups that may have been absent from pre-marketing test-ing. The drug(s) under investigation must, of course, be present in the formulary and be prescribed in suffi-cient quantity to provide adequate power for analyses.

Additional considerations are that the records are verifiable and are reliable. The ability to conduct chart review to confirm outcomes is a necessity for most studies, as diagnoses entered into an electronic database of paid claims may include interim diagnoses and recurrent or chronic, as opposed to acute, events. Information on potential confounders, such as smok-ing and alcohol consumption, and such information as time of menarche and menopause, may only be available through chart review or, more consistently, through patient interviews. With appropriate permis-sions and confidentiality safeguards in place, access to patients is sometimes possible and useful for assessing compliance with the medication regimen as well as for obtaining information on other factors that may relate to drug effects. Information on drugs taken intermit-tently for symptom relief, over-the-counter drugs and drugs not on the formulary must also be obtained directly from the patient.

The advantages of a claims database remain, that is, data do not have to be collected de novo, investigations can be completed more efficiently and more economically, and data on exposures are not subject to recall or interviewer bias. Although data on drugs dispensed are of extremely high quality (West, Strom and Poole, 2005), the quality of disease data may be less so. With the caveat of the need to confirm outcomes, the availability of such databases is an important asset for post-marketing surveillance.

In the following sections we will discuss the databases associated with four major US health plans, one Canadian health plan, a unique consortium of health plans created to meet the needs of the research communities, and a special-purpose database.