Applying control charts – why a one size fits all approach isn’t the way to go.
Trend control charts are a potentially useful tool for developing chronic disease surveillance methods. They can tell us when and where there may be changes in the quality of administrative health data for identifying disease cases. But, in order to make them even more efficient, disease-specific calibrated control limits are likely needed – and to do that, understanding the disease of interest is key.
Case definitions are used to estimate new and existing cases of chronic disease from administrative health data. However, data quality for determining how many cases there are in the population may change over time. Figuring out when and where these changes in data quality happen isn’t as easy as one might think.
Control charts, a tool often used for quality research, use control limits to identify where changes in data quality occur, and calculating control limits depends on the process of interest. Whilst previous research has applied trend control charts to study trends in juvenile diabetes, researchers from Universities of Manitoba and British Columbia have applied control charts to multiple sclerosis trends and compared control limit calculations to see if these can be applied across different diseases.
In a new study ‘Trend control charts for multiple sclerosis case definitions’ published in the International Journal of Population Data Science (IJPDS), the proportion of out-of-control observations (i.e. those that fell outside of the control limits) for multiple sclerosis were much higher for a given control limit calculation. Considering the different diagnosis and treatment patterns between diabetes and multiple sclerosis, the research team have determined that wider control limits are a better choice for multiple sclerosis.
Despite minimal application within the chronic disease surveillance setting so far, there is no doubt that control charts have great potential when disease-specific control limits can be calculated.
Lead author Naomi Hamm from the University of Manitoba added “Control charts are a great tool for assessing potential changes in data quality - they were originally developed for quality control purposes, so the application makes sense! But there are additional methodological considerations when applying them to administrative health data, like disease presentation and treatment protocols. This paper demonstrates the importance of those considerations.”
Click here to view the full article
Naomi C. Hamm, PhD ABD, Department of Community Health Sciences, Max Rady College of Medicine, University of Manitoba, Canada
Hamm, N., Marrie, R. A., Jiang, D., Irani, P. and Lix, L. (2024) “Trend control charts for multiple sclerosis case definitions”, International Journal of Population Data Science, 9(1). doi: 10.23889/ijpds.v9i1.2358.