Table 2.
Multivariable Regression Model of Premature Mortality, 3,139 US Counties, 2002-2006a
| County Characteristic | SD | β (SE)b | P Valuec | % Changed |
|---|---|---|---|---|
| Income | ||||
| Median household income | 8.9 | −0.02 (0.01) | <.001 | −2.7 |
| Income inequality (Gini indexe) | 3.5 | 0.03 (0.004) | <.001 | 2.1 |
| Region | ||||
| Northeast | 1 | −0.08 (0.01) | <.001 | −12.5 |
| Midwest | 1 | −0.09 (0.01) | <.001 | −16.4 |
| South | 1 | 0.04 (0.01) | <.001 | −6.1 |
| West | 1 | 1 [Reference] | ||
| Population density per square mile | ||||
| <17 | 1 | −0.03 (0.02) | .13 | −2.3 |
| 17-42 | 1 | −0.02 (0.01) | .01 | −1.3 |
| 43-104 | 1 | −0.01 (0.01) | .06 | −0.5 |
| >104 | 1 | 1 [Reference] | ||
| Race/ethnicity | ||||
| % Non-Hispanic white | 16.1 | 1 [Reference] | ||
| % Non-Hispanic black | 14.5 | 0.05 (0.003) | <.001 | 4.2 |
| % Nonwhite, nonblack | 8.0 | 0.01 (0.004) | <.001 | 0.8 |
| % Hispanic | 12.5 | −0.02 (0.004) | <.001 | −3.7 |
| Health care access and quality | ||||
| % Uninsured residents | 6.1 | 0.005 (0.004) | .14 | −1.7 |
| Preventable hospital stays per 1,000f | 36.0 | 0.04 (0.004) | <.001 | 3.2 |
| Primary care providers per 100,000 | 59.6 | 0.006 (0.005) | .24 | 2.8 |
| Socioenvironmental factors | ||||
| High school freshman graduation ratef | 12.3 | −0.02 (0.005) | .001 | −2.4 |
| % Adults with a 4-year college degree | 8.4 | −0.06 (0.004) | <.001 | −5.5 |
| % Children living below federal poverty guidelines | 9.0 | 0.02 (0.004) | <.001 | 1.7 |
| % Single-parent households | 2.8 | 0.01 (0.004) | <.001 | 1.4 |
| Behavioral factors | ||||
| % Residents who smokef | 5.9 | 0.07 (0.01) | <.001 | 6.7 |
| % Obese residents | 3.9 | 0.03 (0.003) | <.001 | 3.0 |
Abbreviations: SD, standard deviation; SE, standard error.
Model adjusted for variables listed in table. The model includes an intercept term. Premature mortality defined as all-cause, age-adjusted mortality rate per 100,000 population aged birth to 75 years, averaged from 2002 through 2006.
β coefficients for continuous covariates reflect the percentage change in the premature mortality rate associated with each SD change in the predictor.
Percentage change in premature mortality associated with a 1 SD increase in the covariate (or Northeast, Midwest, South vs West; <17, 17-42, 43-104 vs >104 people per square mile).
Calculated by using [EXP(Beta)-1]*100.
The Gini index scores income inequality in resources from 0 (complete equality) to 100 (complete inequality).
Some counties were missing values for this variable (see Methods for list).