The Multivariate Analysis of Factors Affecting Hospitalization Among Veterans at VHA Primary Care Clinics Across the United States and in King County, WA in 2015
| Variables | Odds Ratio; P-Value (95% CI) | |||||
|---|---|---|---|---|---|---|
| United States | King County, WA | |||||
| Model 1 | Model 2 | Model 3 | Model 1 | Model 2 | Model 3 | |
| Age, years | 1.0032; <.001 (1.003 to 1.0034) | 1.0032; <.001 (1.003 to 1.0035) | 1.02; <.001 (1.01 to 1.026) | 1.02; <.001 (1.01 to 1.029) | ||
| Sex, female as reference | 1.17; <.001 (1.15 to 1.18) | 1.16; <.001 (1.15 to 1.18) | 1.18; .13 (0.95 to 1.47) | 1.17; .15 (0.94 to 1.46) | ||
| Reported Race, non-white as reference | 0.85; <.001 (0.84 to 0.853) | 0.84; <.001 (0.83 to 0.85) | 0.91; .13 (0.81 to 1.03) | 0.92; .17 (0.81 to 1.04) | ||
| Gagne Comorbidity Score*, 0/ below 0 as Reference | 3.33; <.001 (3.31 to 3.35) | 3.32; <.001 (3.30 to 3.35) | 3.72; <.001 (3.34 to 4.15) | 3.68; <.001 (3.30 to 4.11) | ||
| NSES Index† | ||||||
| 1st Quartile as reference | ||||||
| 2nd Quartile | 0.80; <.001 (0.79 to 0.81) | 0.83; <.001 (0.82 to 0.84) | 0.96; 0.66 (0.81 to 1.14) | 1.06; 0.49 (0.89 to 1.27) | ||
| 3rd Quartile | 0.71; <.001 (0.70 to 0.72) | 0.75; <.001 (0.74 to 0.76) | 0.84; .05 (0.70 to 0.99) | 0.90; .27 (0.75 to 1.08) | ||
| 4th Quartile | 0.64; <.001 (0.63 to 0.643) | 0.69; <.001 (0.68 to 0.70) | 0.65; <.001 (0.55 to 0.77) | 0.71; <.001 (0.59 to 0.86) | ||
| Housing issues | ||||||
| Housing instability‡ | 1.03; <.001 (1.027 to 1.029) | 1.01; <.001 (1.007 to 1.014) | 1.02; <.001 (1.00 to 1.03) | 1.0003; .96 (0.99 to 1.01) | ||
| Home vacancy rate§ | 1.02; <.001 (1.018 to 1.020) | 1.01; <.001 (1.005 to 1.017) | 1.003; .085 (0.97 to 1.03) | 0.99; .52 (0.96 to 1.02) | ||
| Characteristics of the house | ||||||
| Percentage of houses with no plumbing | 1.01; <.001 (1.010 to 1.015) | 1.004; <.01 (1.001 to 1.007) | 1.06; <.05 (1.01 to 1.11) | 1.05; .05 (0.99 to 1.10) | ||
| Percentage of houses with no heating | 0.991; <.001 (0.99 to 0.992) | 0.99; <.001 (0.98 to 0.996) | 1.08; <.05 (1.03 to 1.13) | 1.09; <.001 (1.04 to 1.14) | ||
| QIC‖ | 2610527 | 2464538 | 2463445 | 10813.3 | 9958.248 | 9945.153 |
↵* Gagne comorbidity score was presented as a binary variable (above 0 vs. 0 and below 0) in the GEE model.
↵† The NSES index was a summary measure of six geographic-level census-based variables linked to the census tract of a participant's residence. The higher values corresponded to higher socioeconomic status.
↵‡ Percentage of households that moved across census tracts in the past year and had an income below 100% poverty line.
↵§ Percentage of vacant houses in each census tract.
↵‖ The QIC or Quasi likelihood under the Independence model Criterion statistic is analogous to the AIC (Akaike's Information Criterion) statistic used for comparing models fit with likelihood-based methods. Since the GEE method is not a likelihood-based method, the AIC statistic is not available. Smaller QIC value presents a better model.
Model 1. GEE Model—housing characteristics only.
Model 2. GEE Model—patient-level characteristics and NSES index.
Model 3. GEE Model—patient-level characteristics, NSES index, housing issues.
GEE, Generalized Estimating Equation; NSES, neighborhood socio-economic status; QIC, quasi likelihood under the Independence Model Criterion; VHA, Veterans Health Administration; CI, confidence interval.