Neighborhood Poverty and Pediatric Intensive Care Use

Abstract

Background: Disparities in health service use have been described across a range of sociodemographic factors. Patterns of PICU use have not been thoroughly assessed.

Methods: This was a population-level, retrospective analysis of admissions to the Cincinnati Children's Hospital Medical Center PICU between 2011 and 2016. Residential addresses of patients were geocoded and spatially joined to census tracts. Pediatric patients were eligible for inclusion if they resided within Hamilton County, Ohio. PICU admission and bed-day rates were calculated by using numerators of admissions and bed days, respectively, over a denominator of tract child population. Relationships between tract-level PICU use and child poverty were assessed by using Spearman's ρ and analysis of variance. Analyses were event based; children admitted multiple times were counted as discrete admissions.

Results: There were 4071 included admissions involving 3129 unique children contributing a total of 12 297 PICU bed days. Child poverty was positively associated with PICU admission rates (r = 0.59; P < .001) and bed-day rates (r = 0.47; P < .001). When tracts were grouped into quintiles based on child poverty rates, the PICU bed-day rate ranged from 23.4 days per 1000 children in the lowest poverty quintile to 81.9 days in the highest poverty quintile (P < .001).

Conclusions: The association between poverty and poor health outcomes includes pediatric intensive care use. This association exists for children who grow up in poverty and around poverty. Future efforts should characterize the interplay between patient- and neighborhood-level risk factors and explore neighborhood-level interventions to improve child health.

FIGURE 1

A, PICU admission rate. B, PICU IPBD rate. Scatter plots demonstrate the relationships between PICU use measures and neighborhood poverty. Each point is an individual census tract. PICU admission rate and neighborhood poverty are strongly positively correlated (upper; r = 0.59; P < .001), as are PICU bed-day rate and neighborhood poverty (lower; r = 0.47; P < .001). Both axes are on the log₁₀ scale.

FIGURE 2

A, PICU admission rate. B, PICU IPBD rate. Plots of PICU admission and bed-day rates by poverty quintile are shown. Census tracts were aggregated into quintiles based on child poverty rate from lowest (Quintile 1) to highest (Quintile 5). Bold horizontal lines represent the median admission (upper figure) and PICU bed-day (lower figure) rates in each quintile (P < .0001 for both variables), whereas gray boxes represent the interquartile range. Whiskers represent 1.5 times the interquartile range for each quintile, and dots represent census tracts that were outliers.

FIGURE 3

Maps of Greater Cincinnati demonstrating PICU bed-day (upper left), child poverty (upper right), and PICU admission rates (lower left) of census tracts. Shades correspond to the rank, rather than the absolute number, of each continuous variable. Poverty quintiles (lower right) range from Quintile 1 (lowest poverty; lightest shade) to Quintile 5 (highest poverty; darkest shade).