Original ArticleManitoba and Saskatchewan administrative health care utilization databases are used differently to answer epidemiologic research questions
Introduction
Administrative health care utilization databases are a useful resource for the evaluation of health service delivery and quality, and health policy development. If prescription drug dispensing data are an integral or linkable component, the data system also has the potential for pharmacoepidemiologic research. However, the value of some databases is limited by design and process characteristics [1], [2].
In Canada, the provision and delivery of almost all health care services is the responsibility of its provinces and territories. These services provide universal coverage of physician and hospital services for most needs and, in some of provinces, prescription drug use is recorded for almost all the population. This has led to the accumulation of population-based information about the services provided. Although the data are usually collected and maintained in separate files, the unique life-time identification numbers that residents of most provinces have allow file linkage and access to individual histories of health care utilization. In particular, the provinces of Manitoba and Saskatchewan have accrued large, longitudinal, administrative health care utilization databases that have been used extensively for research purposes. Some of this research has led to considerable impact on service delivery and drug safety, and significant media interest.
Significant similarities exist between Manitoba and Saskatchewan. They each have a population of about one million and comparable demographics [3]. In addition, both have similar health care needs, delivery and administration, as well as similar data recording processes for the registration of their residents and the health care services they receive. Nevertheless, the type of organization playing the role of data custodian is different. In Manitoba, a university research centre, the Manitoba Centre for Health Policy (MCHP) [4], acts as steward of the information controlling access to the data in return for an annual set of deliverables to the government [5], whereas, in Saskatchewan, the government operates a small liaison unit to manage data access and work with researchers on project development [6], [7]. Until the mid-1990s, when they became accessible in Manitoba [8], a difference also existed between the two provinces in the availability of prescription drug use data; drug information has been an integral part of the Saskatchewan system since the mid-1970s [9], [10].
A recent review of the uses of administrative health care utilization databases for epidemiologic research suggests that important differences between databases may lead to bias [11]. Researchers using these databases need to be aware of differences, as do policy makers who frequently base decisions on evidence arising from such data. Although differences between databases are thought to exist, the relative merits of using them for research purposes and issues surrounding their use have not been formally assessed. The objective of this research was to evaluate the utilization of the Manitoba and Saskatchewan databases as a case study of potential differences between databases through a systematic review.
Section snippets
Methods
A comprehensive search to identify studies that used one or both of the provincial systems was conducted in the PubMed, EMBASE, BIOSIS, and CINAHL literature databases. Medical subheadings used were purposely kept broad and included “Manitoba” or “Saskatchewan.” To ensure study reports were included from the establishment of each database, searches covered 1970 to July 2004 for Manitoba and 1969 to July 2004 for Saskatchewan. The MCHP and Saskatchewan government web sites [4], [6] were also
Results
Searches of the literature databases identified 3,637 records, which were screened for study eligibility (see Fig. 1 in Appendix on the journal's web site at www.elsevier.com/). For the 506 potentially relevant citations, the full-text articles were reviewed and 20 additional articles were identified from conference abstracts. From these 526 records, 303 reports relevant to the systematic review were included.
The MCHP web site [4] detailed 530 articles published between 1977 and mid-2004, of
Discussion
This systematic review has some limitations. Unpublished study reports, such as those on the MCHP web site [4], were not included. Scanning reference lists of the included studies was not done due to resource constraints in time and personnel. However, thorough searches of the respective provincial data custodian web sites [4], [6] resulted in 22 additional relevant study reports being identified.
The review included 325 studies, providing a comprehensive evaluation of the use of these
Acknowledgments
We thank Warren Chin, Theresa Chua, Leanne De Souza, Yang Liu, Alkesh Patel, Reid Robson, and Eric Rodrigues for their assistance with this project.
References (20)
- et al.
A review of uses of health care utilization databases for epidemiologic research on therapeutics
J Clin Epidemiol
(2005) - et al.
Using administrative data for longitudinal research: comparisons with primary data collection
J Chronic Dis
(1987) - et al.
Assessing the Saskatchewan database for outcomes research studies of depression and its treatment
J Clin Epidemiol
(2000) - et al.
Examples of automated databases
Health care utilization databases
Can J Clin Pharmacol
(1998)Canada e-Book
(2003)Manitoba Centre for Health Policy
(2006)Establishing a population data-based policy unit
Med Care
(1999)Epidemiology, Research and Evaluation Unit
(2003)- et al.
Health services databases in Saskatchewan
Cited by (39)
Identifying optic neuritis and transverse myelitis using administrative data
2018, Multiple Sclerosis and Related DisordersCitation Excerpt :In jurisdictions where health care is publicly funded for all residents, these data offer a cost-effective source of population-based information, often over extended periods. Therefore, researchers are increasingly using such data; (Tricco et al., 2008) however because they are not collected for research, approaches to identifying specific health conditions should be validated before they are used (Tricco et al., 2008). A clinically isolated syndrome (CIS) is an incident demyelinating event of the central nervous system that may represent the onset of multiple sclerosis (MS) if there is ultimately evidence of other demyelinating events disseminated in space and time (Miller et al., 2012).
Secular trends and population differences in the incidence of epilepsy. A population-based study from Saskatchewan, Canada
2018, SeizureCitation Excerpt :However, because secular analysis was used, we firmly believe that cases with SUS were eventually included in our analysis if they have finally been diagnosed with epilepsy. On the other hand, because of its high quality and comprehensiveness, Saskatchewan’s health database system has frequently been used by other researchers, including those outside the province, and has been demonstrated in several studies to be complete and accurate [44–47]. Therefore we believe that the quality and reliability of the information is high.
Validation of algorithms to identify stroke risk factors in patients with acute ischemic stroke, transient ischemic attack, or intracerebral hemorrhage in an administrative claims database
2016, International Journal of CardiologyCitation Excerpt :National Health Insurance Research Database (NHIRD), a database derived from the NHI claims, has become a rich source for healthcare research [7]. Nevertheless, the validity of administrative claims databases in research and surveillance must be evaluated because administrative claims data have not been collected for these purposes [8]. Understanding the extent of validity of an administrative claims database helps assess the potential biases of variables and outcomes derived from the database [9].
Status and perspectives of claims data analyses in Germany-A systematic review
2016, Health PolicyCitation Excerpt :The United States (US) were one of the first countries to initiate claims data analyses, for instance, by utilising drug data, for research purposes and health-related political decisions [1]. In addition, in Canada, administrative healthcare utilisation databases were used to support health policy development and the evaluation of health service delivery and quality [2]. As opposed to North America, claims data in Europe have been involved in health-related political decisions more recently.
Development and Internal Validation of a Disability Algorithm for Multiple Sclerosis in Administrative Data
2021, Frontiers in NeurologyEvidence-Based Decision Making 6: Administrative Databases as Secondary Data Source for Epidemiologic and Health Service Research
2021, Methods in Molecular Biology
This work was presented as poster presentations at the annual conferences of the Canadian Association for Population Therapeutics in April 2005 and the International Society for Pharmacoepidemiology in August 2005. The views expressed are those of the authors. GlaxoSmithKline provided partial support to the project in lieu of research time for the authors.
- 1
Andrea Tricco began this research while at GlaxoSmithKline and completed it during her current position at the Chalmers Research Group. She is also a PhD student at the Institute of Population Health, University of Ottawa.