Trend control charts for multiple sclerosis case definitions

Area of Science:

• Health Informatics
• Epidemiological Surveillance
• Statistical Process Control

Background:

• Administrative health data validity for chronic diseases can change over time.
• Trend control charts can identify unexpected changes in time-series data, signaling potential data quality issues.
• Monitoring disease estimates is crucial for accurate public health surveillance.

Purpose of the Study:

• To apply and compare trend control chart methods for multiple sclerosis (MS) incidence and prevalence.
• To assess the impact of different statistical control limits on identifying out-of-control (OOC) observations.
• To evaluate the utility of trend control charts for MS surveillance using administrative health data.

Main Methods:

• Eight validated MS case definitions were applied to Manitoba administrative health data (1972-2018).
• Incidence and prevalence trends were modeled, and trend control charts plotted predicted versus observed case counts.
• Out-of-control (OOC) observations were identified using two control limit methods: predicted count ±0.8*standard deviation (SD) and ±2*SD.

Main Results:

• The proportion of OOC observations varied significantly based on the control limit method used (0.8*SD vs. 2*SD).
• The 2*SD method yielded a lower proportion of OOC observations compared to the 0.8*SD method for both incidence and prevalence.
• Neither control limit method showed statistically significant differences in OOC observations across the evaluated MS case definitions.

Conclusions:

• Trend control charts are a potentially valuable tool for developing disease surveillance methods.
• The choice of control limit significantly impacts the proportion of identified OOC observations.
• Disease-specific calibrated control limits may enhance the effectiveness of trend control charts for chronic disease surveillance.