Table 4.
Testing for Interactions Between Demographic Characteristics and Time
| Outcome | Interaction | p-Value |
|---|---|---|
| Opioids administered in ED | Time × Race | 0.256 |
| Time × Gender | 0.315 | |
| Time × Location | 0.880 | |
| Time × Insurance | 0.535 | |
| Opioids prescribed at discharge | Time × Race | 0.844 |
| Time × Gender | 0.107 | |
| Time × Location | 0.910 | |
| Time × Insurance | 0.493 | |
| Non-narcotics administered in ED | Time × Race | 0.344 |
| Time × Gender | 0.333 | |
| Time × Location | 0.237 | |
| Time × Insurance | 0.195 | |
| Non-narcotics prescribed at discharge | Time × Race | 0.948 |
| Time × Gender | 0.713 | |
| Time × Location | 0.636 | |
| Time × Insurance | 0.565 |
For each outcome, we tested for the presence of interactions between ED admission time period (September 1–August 31) and demographic factors. p-Values for the significance of the two-way interaction terms are shown below. There was no strong evidence that the changes in outcome by year depend on race, gender, location, or insurance status.