Related Experiment Video
Updated: Mar 15, 2026
Quantifying the Effects of Antimicrobials on In vitro Biofilm Architecture using COMSTAT Software
Published on: December 14, 2020
Capturing the Central Line Bundle Infection Prevention Interventions: Comparison of Reflective and Composite Modeling
Heather M Gilmartin1, Karen H Sousa, Catherine Battaglia
1Heather M. Gilmartin, PhD, NP, is Postdoctoral Nurse Fellow, Denver-Seattle Center of Innovation, Department of Veterans Affairs, Denver VA Medical Center, Colorado. Karen H. Sousa, PhD, RN, FAAN, is Professor and Associate Dean for Research and Extramural Affairs, University of Colorado College of Nursing, Anschutz Medical Campus, Aurora. Catherine Battaglia, PhD, RN, is Nurse Scientist, Denver-Seattle Center of Innovation, Department of Veterans Affairs, Denver VA Medical Center, Colorado.
This study compared two modeling approaches for central line (CL) bundle interventions to prevent bloodstream infections. Both models showed adherence to CL bundles impacts infections, but conceptual differences between methods were noted.
Area of Science:
- Health Systems Science
- Infection Prevention
- Statistical Modeling
Background:
- Central line (CL) bundle interventions are crucial for preventing central line-associated bloodstream infections (CLABSIs).
- A validated modeling method to test CL bundle interventions within a health systems framework is currently lacking.
Purpose of the Study:
- To test CL bundle interventions using reflective and composite latent variable measurement models.
- To assess the impact of these modeling approaches on the relationships between adherence to CL bundle interventions, organizational context, and CLABSIs.
- To evaluate the models within the Quality Health Outcomes Model (QHOM) framework.
Main Methods:
- Secondary data analysis of 614 U.S. hospitals from the Prevention of Nosocomial Infection and Cost-Effectiveness Refined study.
- Random splitting of the sample into exploration and validation subsets.
- Testing of reflective and composite latent variable measurement models.
Main Results:
- Both modeling approaches yielded well-fitting structural models (RMSEA = .04; CFI = .94).
- Adherence to CL bundles significantly impacted organizational context (p = .01) and CLABSIs (p = .01).
- The relationship between organizational context and CLABSIs was not statistically significant.
Conclusions:
- Little statistical but significant conceptual differences exist between reflective and composite modeling approaches.
- Both models provided partial support for the QHOM.
- Comparing modeling approaches is recommended for novel or ambiguous variables to enhance transparency and confidence in findings.
