Use of propensity score matching in administrative database research to study rare conditions /
This case study will discuss a study assessing an association between symptomatic congenital cytomegalovirus infection and bronchopulmonary dysplasia. It is aimed at illustrating the learning points on the use of propensity score matching in administrative database research for studying rare conditi...
| Main Author: | |
|---|---|
| Format: | eBook |
| Language: | English |
| Published: |
London :
SAGE Publications Ltd,
2020.
|
| Series: | SAGE Research Methods Cases: Medicine and Health.
|
| Subjects: | |
| Online Access: | Connect to the full text of this electronic book |
| Summary: | This case study will discuss a study assessing an association between symptomatic congenital cytomegalovirus infection and bronchopulmonary dysplasia. It is aimed at illustrating the learning points on the use of propensity score matching in administrative database research for studying rare conditions. Administrative databases are a powerful tool in answering important questions in biomedical research, particularly when the condition of interest is rare. Administrative databases provide advantages of containing a large amount of data and no need for enrolling patients. However, researchers must be familiar with the limitations, because the validity of the study can be undermined without thorough understanding of the properties of the databases. Such limitations may include the lack of detailed clinical information and the intrinsic retrospective nature of administrative databases, among others. The propensity score is the probability of the assignment of a certain variable, conditional on the covariates, and can help mitigate confounding in retrospective studies when used appropriately. Among the commonly used propensity score-based methods, we employed propensity score matching in this study because a comparison between infants with and without congenital cytomegalovirus was expected to be confounded, unless appropriate adjustment was performed. Correct specification of the variables included in the logistic regression model for estimating propensity scores is crucial in ensuring the maximum effectiveness of the method. Although large-scale randomized controlled trials are the most definitive method to assess associations, administrative databases with the use of rigorous statistical methodologies can be an effective research approach, when limitations are well accounted for. |
|---|---|
| Physical Description: | 1 online resource. |
| Bibliography: | Includes bibliographical references and index. |
| ISBN: | 9781529723571 1529723574 |