Table 2.
Summary of analyses of spatial patterns of dependence, heterogeneity, and spillover
| Method | Test statistic | P-value |
|---|---|---|
| Spatial correlation of intervention allocation | ||
| Join count—control | 10.105 | 0.979 |
| Join count—intervention | 11.282 | 0.744 |
| Spatial correlation of SMR | ||
| Moran’s I—entire study area | 0.237 | < 0.001 |
| Moran’s I—control | 0.396 | < 0.001 |
| Moran’s I—intervention | 0.095 | 0.176 |
| Moran’s I—residuals | 0.134 | 0.0198 |
| Spatial heterogeneity of intervention | SMR | |
| Geographically weighted regression, median (interquartile range) | – | 0.94 (0.77–1.09) |
| Impact of spillover on intervention effect* | – | SMR |
| Cluster reallocation method | ||
| Original cluster definition | – | 0.827 |
| Controls cluster larger, mean (min, max) | – | 0.885 (0.855–0.916) |
| Intervention clusters larger, mean (min, max) | – | 0.781 (0.742–0.827) |
SMR = standardized mortality ratio.
* This summary represents the mean of the intervention estimates that derive from increasing either the control or intervention cluster boundaries based on the cluster reallocation method.