Table 3. Effect of Covariates in Accounting for Spatial Clustering of Diabetes Prevalence Among 59,767 Group Health Members in King County, Washington, 2005–2006.
| Model | Moran’s I Statistic of Residuals (95% CI)a | P Value Compared With Crude | P Value Compared With Model 3 |
|---|---|---|---|
| Crude (intercept-only model) | 0.46 (0.40–0.52) | Reference | <.001 |
| Model 1: Age-adjusted | 0.48 (0.43–0.54) | .54 | <.001 |
| Model 2: Model 1 + race/ethnicity | 0.40 (0.34–0.46) | .19 | .001 |
| Model 3: Model 1 + population density | 0.27 (0.21–0.32) | <.001 | Reference |
| Model 4: Model 3 + income | 0.27 (0.21–0.32) | <.001 | .99 |
| Model 5: Model 3 + home value | 0.17 (0.11–0.22) | <.001 | .02 |
| Model 6: Model 3 + college | 0.18 (0.13–0.23) | <.001 | .03 |
| Model 7 – Model 6 + income + home value | 0.12 (0.07–0.18) | <.001 | <.001 |
Abbreviation: CI, confidence interval.
a
Higher values for Moran’s I statistic indicate greater extent of spatial clustering of diabetes prevalence. A value of 0 indicates a random spatial pattern, a value of −1 indicates perfect dispersion, and a value of 1 indicates perfect clustering.