Introduction Prompt:
This text introduces the article titled “Towards using administrative databases to measure population-based indicators of quality of end-of-life care: testing the methodology” published in Palliative Medicine in 2006, authored by Eva Grunfeld and colleagues.
The primary objective of this study was to assess the feasibility, validity, and reliability of utilizing routinely collected administrative data to measure population-based indicators for the quality of end-of-life care. Recognizing the growing importance of defining, measuring, and monitoring quality in health services research, the authors aimed to address the slower development in measuring quality end-of-life care compared to other phases of the cancer care continuum.
The research employed a retrospective cohort study design, focusing on all females who died of breast cancer in Nova Scotia or Ontario, Canada, between January 1, 1998, and December 31, 2002. The “end-of-life” period was specifically defined as the last six months of life, measured backward from the time of death. This approach allowed for an efficient monitoring of end-of-life care across different jurisdictions, demographic groups, and time periods. The selection of subjects relied on Vital Statistics files and provincial cancer registries, ensuring a homogeneous group whose death was attributable to advanced breast cancer.
From an initial list of 19 quality indicators selected from existing literature and deemed potentially measurable by an expert panel, the study determined that seven indicators were fully measurable in both provinces. Additionally, seven indicators in Nova Scotia and three in Ontario were found to be partially measurable. Examples of fully measurable indicators included the interval between last chemotherapy and death, site of death, frequency of emergency room visits, and hospital days near the end of life. Partially measurable indicators included enrollment in palliative care, access to palliative care, and radiotherapy for uncontrolled bone pain.
The study concluded that using administrative databases for this purpose is feasible, valid, and reliable. Key findings supporting this conclusion include:
- High levels of agreement between administrative and chart data, with most measures showing a match of 80% or higher.
- Kappa statistics indicated good to excellent agreement for many categorical variables, with seven out of thirteen exceeding the 0.75 threshold in Nova Scotia data and two out of eight in Ontario data.
- Intraclass correlation coefficients (ICC) showed strong agreement for continuous variables like days in hospital (ICC of 0.98 for NS and 0.72 for Ontario).
- High inter-rater reliability in chart abstraction further confirmed consistency in the data collection process.
The authors highlight that administrative data offers several advantages: they are efficient, population-based, and continuously collected, providing consistent data over time and reflecting actual occurrences without sampling variability. They also allow for flexibility in statistical calculations and can be extracted and reformulated relatively quickly.
However, the study also acknowledges limitations of administrative databases. They may not capture crucial information valued by patients, families, and providers, such as psychosocial care, pain and symptom management, spiritual well-being, advanced directives, empathy, dignity, and communication. Furthermore, issues like the shift from fee-for-service to other physician payment methods (e.g., salaried physicians submitting ‘shadow billings’) could impact the completeness and reliability of billing databases for research purposes. The authors emphasize that administrative databases should be used in conjunction with other methods, such as qualitative and purposefully collected clinical data, to provide a comprehensive profile of care.
APA Reference:
Grunfeld, E., Lethbridge, L., Dewar, R., Lawson, B., Paszat, L. F., Johnston, G., Burge, F., McIntyre, P., & Earle, C. C. (2006). Towards using administrative databases to measure population-based indicators of quality of end-of-life care: testing the methodology. Palliative Medicine, 20, 769–777. 10.1177/0269216306072553
