Patterns of missing data and the screening of African Americans for latent TB infection during state psychiatric hospital admissions
Abstract:Objective: The purpose of the present study was to demonstrate that evaluating patterns of missing purified protein derivative test data can help to identify gaps in the screening of African American psychiatric inpatients for latent tuberculosis infection. Design: The results of tuberculin skin tests were abstracted during chart reviews from patients' physical examination form after hospital admission. Two different approaches to missing test data of purified protein derivative were examined in regression models: (1) classifying patients with missing test data as having negative readings on the tuberculin skin test; and (2) a comparison of patients with missing versus nonmissing test results of purified protein derivative as the outcome. The data were from a case-control study of the psychiatric misdiagnosis of 118 Black inpatients in a psychiatric hospital in upstate New York. Results: Chi-square tests and logistic regression analyses were conducted on the missing TB test data in relation to sex, age, hospitalization for a physical health problem, homelessness, substance abuse history and chart diagnosis of schizophrenia. Logistic regression and supplemental analyses of the outcome of missing/nonmissing data revealed that Black female psychiatric patients without a diagnosis of schizophrenia, relative to their male counterparts, were significantly more likely to have evidence of missing PPD test results in their charts. Conclusion: The findings suggest that low risk groups in the African American psychiatric patient sample are more likely to have missing data. These findings have implications for improving screening of African American psychiatric inpatients for latent tuberculosis infection.
Document Type: Research Article
Affiliations: University of Texas, Austin, TX, USA
Publication date: March 1, 2007