Table 2

Multivariate regression results

Individual DM status – multivariate logistic regression results
CategoryFactorMean (SD) (range)Coefficient (SE)AUROC
DemographicAge49.4 (15.9) (20.0–80.0)0.0467 (0.00185)*0.6950.711
Female50.9−0.221 (0.0544)*
African-American10.20.633 (0.0688)*
Hispanic14.20.445 (0.0672)*
SocioeconomicHousehold incomeNA−0.474 (0.127)*0.653
Some college61.1−0.200 (0.0595)†
County-level DM prevalence – multivariate linear regression results
CategoryFactorMean (SD) (range), %Coefficient (SE)R2
DemographicPer cent <18 years old22.3 (3.5) (0.0–40.9)0.0567 (0.0112)*0.3770.619
Per cent >65 years old18.4 (4.6) (4.6–56.3)0.0894 (0.00879)*
Per cent female49.9 (2.3) (27.8–56.6)0.220 (0.0143)*
Per cent African-American9.0 (14.3) (0.0–85.2)0.0465 (0.00230)*
Per cent Hispanic9.3 (13.7) (0.5–96.3)−0.0502 (0.00223)*
SocioeconomicMedian household income91.3 (2.1) (84.6–100.0)−0.284 (0.0191)*0.4
Percent some college57.2 (11.6) (15.1–94.1)−0.0869 (0.00339)*
  • Multiple regression results for models using factors shared between NHANES (individual level) and CHR (county level) datasets (ie, sociodemographic factors). For female gender, Hispanic and African-American race/ethnicity, and education level factors from the individual-level (NHANES) data, summary characteristics are expressed in terms of percent of total sample; not all summary statistics could be calculated. Similarly, for household income from the individual-level data, values were collected and stored as ranges of income; summary statistics could not be calculated. Finally, for county-level median household income, variables were normalized and scaled to have a maximum value of 100.

  • *P<0.05.

  • †P<0.001.

  • AUROC, area under the receiver operating characteristic; CHR, county health rankings; DM, diabetes mellitus; NHANES, national health and nutrition examination survey.