Boston Collaborative Drug Surveillance Program

History of the BCDSP

    The Boston Collaborative Drug Surveillance Program (BCDSP) was established in 1966. In these early days, post-marketing drug safety studies conducted by BCDSP researchers were still primarily paper-based. Nurse monitors were trained to collect information on standardized forms, conducted interviews of hospitalized patients and collected information on drug exposure, medical problems and descriptive characteristics. Information collected was then reviewed, validated and put on computer. Although this approach was labor-intensive, by utilizing this in-hospital monitoring method the BCDSP was for the first time able to use this information to conduct formal epidemiology research on a large scale and to quantify the potential adverse effects of prescription drugs.


    Some of the clinically most relevant discoveries of the BCDSP date back to these early days including the finding of the protective effect of aspirin on cardiovascular events (Jick and Miettinen 1976Jick and Elwood 1976), and ever since the BCDSP has had a pioneering role in the development and application of methods in drug epidemiology. By the mid 1970s it became apparent that other more efficient methods would be available in the future as medical information was started to be directly recorded on computer by health care providers. It was soon realized that this electronic data could be used to create large databases and increase the accuracy and efficiency of drug safety research.


    One of the first organizations taking advantage of this development was Group Health Cooperative of Puget Sound (GHC), a health maintenance organization located in Seattle, Washington, that has been recording hospital discharge diagnoses on computer since 1972. By 1976 GHC also began recording prescriptions dispensed on computer. A unique identifier allows information to be linked between files. The BCDSP began using information from GHC to conduct research in 1978. Anonymised information from GHC on demographics, hospitalizations and dispensed prescriptions was available to the researchers at the BCDSP, and the BCDSP worked with GHC data for 20 years and published 90 studies using the GHC resource.


    Another important progress was the collection of the most comprehensive population-based patient information from a true “gate-keeper” health system. The characteristics of the United Kingdom health system are indeed very close to such an ideal system, where all patient information is collected by the general practitioners, making the UK an ideal setting to conduct epidemiologic research. In the late 1980s computers in doctors’ offices were becoming available and an organization named VAMP Health designed software for the offices of general practitioners. These circumstances began to change the medical record from a paper-based system to a computer-based system, and thus opened the door to the creation of the most valuable database for pharmacoepidemiologic research that exists today. General practitioners were trained to use this software and agreed to allow their anonymized patient records to be used for research purposes. The resulting database, the Clinical Practice Research Datalink (CPRD), formerly known as the General Practice Database (GPRD), is currently administered the UK Medicines and Healthcare products Regulatory Agency (MHRA).


    The BCDSP uses a special subset of this database for research purposes that has undergone extensive and ongoing validation processes. Anonymized information from the CPRD on diagnoses, drug prescriptions, demographics, outpatient visits and hospitalizations is available to the researchers at the BCDSP. Validation of diagnoses, reports of lab tests and anonymized notes from hospitalizations and referrals are available upon request. The CPRD has over twelve million subjects and over 25 million person-years of data. To date the BCDSP has worked with the CPRD for over 20 years and has published over 200 studies using the CPRD resource.