Table 1. Coefficients for Estimated Trajectories From Group-Based Trajectory Models Using 1999–2016 US County Annual All-Cause Mortality Dataa .
Trajectoryb | Interceptc (P Value) | Sloped (P Value) | Quadratice (P Value) | Cubicf (P Value) | % of US Counties | No. of Counties |
---|---|---|---|---|---|---|
1 | 628.51 (<.001) | −19.92 (<.001) | 0.51 (.02) | NA | 0.5 | 14 |
2 | 775.94 (<.001) | −17.78 (<.001) | 0.47 (<.001) | NA | 9.4 | 290 |
3 | 826.74 (<.001) | −8.19 (<.001) | −0.48 (.03) | 0.03 (<.001) | 19.7 | 608 |
4 | 901.42 (<.001) | −5.31 (.002) | −0.96 (<.001) | 0.05 (<.001) | 25.2 | 780 |
5 | 968.73 (<.001) | −1.53 (.42) | −1.25 (<.001) | .05 (<.001) | 20.3 | 626 |
6 | 1,020.59 (<.001) | 5.95 (.007) | −1.88 (<.001) | 0.08 (<.001) | 15.1 | 467 |
7 | 1,087.72 (<.001) | 13.58 (<.001) | −2.55 (<.001) | 0.10 (<.001) | 8.2 | 252 |
8 | 1,273.51 (<.001) | −7.73 (.002) | 0.28 (.03) | NA | 1.7 | 54 |
Abbreviation: NA, not applicable.
Coefficients are from an 8-group model; coefficients were added or removed from models if P < .05 for the coefficient. Note that if a term became nonsignificant when a higher-order term was added to the model and significant, the nonsignificant lower-order term remained in the model. Data are from the Centers for Disease Control and Prevention’s Compressed Mortality File (14).
The numeric ordering of trajectories reflects mortality rates from most favorable (lowest baseline rate/largest decline in rate) to least favorable (highest baseline rate/largest decline in rate).
Baseline mortality rate estimated by the model.
First order term estimated by model; represents the linear component of change in rate per year.
Second order term estimated by model; represents the quadratic component of change in rate per year.
Third order term estimated by model (if significant); represents the cubic component of change in rate per year.