Dataset Information

Analysis of racial differences in hospital stays in the presence of geographic confounding.

ABSTRACT: Using recent methods for spatial propensity score modeling, we examine differences in hospital stays between non-Hispanic black and non-Hispanic white veterans with type 2 diabetes. We augment a traditional patient-level propensity score model with a spatial random effect to create a matched sample based on the estimated propensity score. We then use a spatial negative binomial hurdle model to estimate differences in both hospital admissions and inpatient days. We demonstrate that in the presence of unmeasured geographic confounding, spatial propensity score matching in addition to the spatial negative binomial hurdle outcome model yields improved performance compared to the outcome model alone. In the motivating application, we construct three estimates of racial differences in hospitalizations: the risk difference in admission, the mean difference in number of inpatient days among those hospitalized, and the mean difference in number of inpatient days across all patients (hospitalized and non-hospitalized). Results indicate that non-Hispanic black veterans with type 2 diabetes have a lower risk of hospital admission and a greater number of inpatient days on average. The latter result is especially important considering that we observed much smaller effect sizes in analyses that did not incorporate spatial matching. These results emphasize the need to address geographic confounding in health disparity studies.

SUBMITTER: Davis ML

PROVIDER: S-EPMC7359673 | biostudies-literature | 2019 Aug

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Analysis of racial differences in hospital stays in the presence of geographic confounding.

Davis Melanie L ML Neelon Brian B Nietert Paul J PJ Burgette Lane F LF Hunt Kelly J KJ Lawson Andrew B AB Egede Leonard E LE

Spatial and spatio-temporal epidemiology 20190705

Using recent methods for spatial propensity score modeling, we examine differences in hospital stays between non-Hispanic black and non-Hispanic white veterans with type 2 diabetes. We augment a traditional patient-level propensity score model with a spatial random effect to create a matched sample based on the estimated propensity score. We then use a spatial negative binomial hurdle model to estimate differences in both hospital admissions and inpatient days. We demonstrate that in the presenc ...[more]

PMID: 31421795

Similar Datasets

Project description:ImportanceWhile racial disparities in prostate cancer mortality are well documented, it is not well known how these disparities vary geographically within the US.ObjectiveTo characterize geographic variation in prostate cancer-specific mortality differences between black and white men.Design, setting, and participantsThis cohort study included data from 17 geographic registries within the Surveillance, Epidemiology, and End Results (SEER) database from January 1, 2007, to December 31, 2014. Inclusion criteria were men 18 years and older with biopsy-confirmed prostate cancer. Men missing data on key variables (ie, cancer stage, Gleason grade group, prostate-specific antigen level, and survival follow-up data) were excluded. Analysis was performed from September 5 to December 25, 2018.ExposurePatient SEER-designated race (ie, black, white, or other).Main outcomes and measuresFine and Gray competing-risks regression analyses were used to evaluate the difference in prostate-cancer specific mortality between black and white men. A stratified analysis by Gleason grade group was performed stratified as grade group 1 and grade groups 2 through 5.ResultsThe final cohort consisted of 229 771 men, including 178 204 white men (77.6%), 35 006 black men (15.2%), and 16 561 men of other or unknown race (7.2%). Mean (SD) age at diagnosis was 64.9 (8.8) years. There were 4773 prostate cancer deaths among white men and 1250 prostate cancer deaths among black men. Compared with white men, black men had a higher risk of mortality overall (adjusted hazard ratio [AHR], 1.39 [95% CI, 1.30-1.48]). In the stratified analysis, there were 4 registries in which black men had worse prostate cancer-specific survival in both Gleason grade group 1 (Atlanta, Georgia: AHR, 5.49 [95% CI, 2.03-14.87]; Greater Georgia: AHR, 1.88 [95% CI, 1.10-3.22]; Louisiana: AHR, 1.80 [95% CI, 1.06-3.07]; New Jersey: AHR, 2.60 [95% CI, 1.53-4.40]) and Gleason grade groups 2 through 5 (Atlanta: AHR, 1.88 [95% CI, 1.46-2.45]; Greater Georgia: AHR, 1.29 [95% CI, 1.07-1.56]; Louisiana: AHR, 1.28 [95% CI, 1.07-1.54]; New Jersey: AHR, 1.52 [95% CI, 1.24-1.87]), although the magnitude of survival difference was lower than for Gleason grade group 1 in each of these registries. The greatest race-based survival difference for men with Gleason grade group 1 disease was in the Atlanta registry.Conclusions and relevanceThese findings suggest that population-level differences in prostate cancer survival among black and white men were associated with a small set of geographic areas and with low-risk prostate cancer. Targeted interventions in these areas may help to mitigate prostate cancer care disparities at the national level.

Dataset Information

Analysis of racial differences in hospital stays in the presence of geographic confounding.

Publications

Analysis of racial differences in hospital stays in the presence of geographic confounding.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets