State level variations in nephrology workforce and timing and incidence of dialysis in the United States among children and adults: a retrospective cohort study

Table 6 Adjusted linear regression models of the association between nephrologist workforce density and incident number of ESRD cases per 100,000 population by state

Linear regression model for number of cases of ESRD/100,000 population	Adult	95% CI	Children	95% CI
	Change in incident ESRD cases/100,000 population per 1 unit increase in workforce density^*		Change in incident ESRD cases/100,000 population per 1 unit increase in workforce density^*
Unadjusted	5.88	[2.30, 9.47]	−0.22	[−0.86, 0.42]
Adjusted for provider characteristics¹	5.32	[1.63, 9.01]	−0.80	[−1.41, −0.18]
Adjusted for state population characteristics²	1.21	[−1.45, 3.86]	−0.35	[−0.86, 0.16]
Adjusted for patient characteristics at incident ESRD³	−0.70	[−3.43, 2.03]	−0.34	[−0.93, 0.25]

¹Adjusted for age of providers in 2008, sex, US versus foreign medical graduate, DO versus MD, > or ≤ 10 years since training, primary role (direct patient care versus other), and present employment (private practice or HMO versus other); N = 43 states for pediatric analyses with exclusion of 7 states without providers where provider characteristics could not be determined.
²Adjusted for percentage of state population with obesity, African American, diabetes, ≥ 65 years of age, rural population in adults and percentage obesity, African American, and rural population in children.
³Adjusted for race (African American versus other), diabetes as cause of ESRD (yes/no), CHF (yes/no), Medicaid status (yes/no), mean hemoglobin, mean albumin, and age ≥ 65 years (in adult model only); N = 49 states for pediatric analysis due to missing comorbidity and demographic data on the 1 ESRD case in Wyoming.
*Per 1 additional nephrologist/100,000.

ISSN: 1471-2369