Use of administrative medical databases in population-based research

Abstract

Administrative medical databases are massive repositories of data collected in healthcare for various purposes. Such databases are maintained in hospitals, health maintenance organisations and health insurance organisations. Administrative databases may contain medical claims for reimbursement, records of health services, medical procedures, prescriptions, and diagnoses information. It is clear that such systems may provide a valuable variety of clinical and demographic information as well as an on-going process of data collection. In general, information gathering in these databases does not initially presume and is not planned for research purposes. Nonetheless, administrative databases may be used as a robust research tool. In this article, we address the subject of public health research that employs administrative data. We discuss the biases and the limitations of such research, as well as other important epidemiological and biostatistical key points specific to administrative database studies.

Sociodemographic characteristics and chronic medical conditions as risk factors for herpes zoster

“…The smaller the chosen area of residence, the more homogeneous it is in terms of its sociodemographic composition. 56 Each health center of the MRPHS provides assistance to people living in its area of influence, ranging in size from 5000 to 25 000 patients. A deprivation index based on census data of 2001 was used to assign SES to each patient.…”

Response to Journal Club

“…4 As a result, diagnoses and pertinent therapies with high financial burden for health care systems, such as SAH, [5][6][7][8] are documented better than less costly diagnoses or medical interventions. 4 Conversely, less costly diseases may require pilot-scaled field investigations to support hypothesis testing. 4 Significant disadvantages exist when performing research based on administrative databases.…”

“…First, exposure and outcome misclassification. 4 This term is related to measurement error and is probably the most common form of bias in epidemiological research. Misclassification occurs frequently when broad definitions of exposure or outcome occur.…”

“…4 Conversely, less costly diseases may require pilot-scaled field investigations to support hypothesis testing. 4 Significant disadvantages exist when performing research based on administrative databases. In general, accrued data from administrative databases suffer from the same biases as observational data.…”

“…13 Finally, generalizability is a common claim made against the use of administrative databases for research purposes. 4 Epidemiologists believe that results from observational studies are meaningful only if the are applicable to large geographic definitions. The premise of generalizability may be misleading particularly because other variables beyond subject or case geographic area of residence such as age, race/ethnicity, and gender, are important for the definition of the cohort.…”