Automated Detection of Drain-Related Meningitis Using Clinical Data
Author Information
Author(s): van Mourik Maaike S. M., Groenwold Rolf H. H., Berkelbach van der Sprenkel Jan Willem, van Solinge Wouter W., Troelstra Annet, Bonten Marc J. M.
Primary Institution: University Medical Centre Utrecht
Hypothesis
Can a prediction model using routinely collected clinical data accurately detect drain-related meningitis?
Conclusion
The study developed a prediction model that can accurately quantify rates of drain-related meningitis using multi-source data.
Supporting Evidence
- The model achieved 98.8% sensitivity and 87.9% specificity.
- Discriminatory power of the model was excellent with an area under the ROC curve of 0.97.
- The model reduced the number of manual chart reviews by 74.7% while identifying 98.8% of infections.
Takeaway
Doctors can use a computer program to find out if patients have a type of meningitis caused by drains, which saves time and helps catch infections better.
Methodology
Logistic regression was used to develop a model predicting the occurrence of drain-related meningitis based on clinical data.
Potential Biases
Potential bias due to reliance on retrospective data and imputation of missing values.
Limitations
The model does not account for infections occurring after discharge unless the patient is readmitted.
Participant Demographics
Median age of participants was 58.5 years, with a mix of genders and a significant portion admitted to the ICU.
Statistical Information
P-Value
p<0.05
Confidence Interval
95% CI 88.0% to 99.9%
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website