Intra-cluster correlation coefficients in adults with diabetes
Author Information
Author(s): Benjamin Littenberg, Charles D MacLean
Primary Institution: University of Vermont
Hypothesis
What are the intra-cluster correlation coefficients (ICCs) for various patient characteristics in adults with diabetes in primary care practices?
Conclusion
The ICC values reported here may be useful in designing future studies that use clustered sampling from primary care practices.
Supporting Evidence
- The median ICC value was 0.0185, indicating low correlation for most variables.
- The highest ICC was 0.288 for the proportion of patients with a recent creatinine measurement.
- ICCs varied widely, with some characteristics showing high correlation within practices and others showing little to no correlation.
Takeaway
This study looked at how similar patients are within the same doctor’s office when it comes to their diabetes care. It found that some things, like getting a kidney test, are very similar among patients in the same office, while other things vary a lot.
Methodology
The study calculated the ICC for 112 variables measured in a cluster-randomized trial involving adults with diabetes from 73 primary care practices.
Limitations
The study's findings may not be generalizable to other populations and settings due to the specific demographic and geographic characteristics of the sample.
Participant Demographics
The study population was predominantly white and rural, consisting of adults with diabetes, mostly Type 2.
Statistical Information
P-Value
0.54
Confidence Interval
0.006, 0.037
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website