Predictors of Length of Stay on the Pediatric Psychiatric Floors in the US Hospitals: A Nationwide Study Using NIS Database Open Access

Manocha, Pankaj (Spring 2020)

Permanent URL:


Objective: To determine the effect of clinical and demographical factors on the length of stay (LOS) in patients admitted to pediatric psychiatric floors across hospitals in the United States.  

Design: Analysis of Nationwide Inpatient Sample (NIS) dataset, which is a systematic sample of the Healthcare Cost and Utilization Project (HCUP) hospital care data.

Methods: We analyzed the 2014 NIS dataset with records of all the children and adolescents age six to eighteen years who were admitted to community hospitals. Candidate variables included in the model were age, gender, race, median household income, urbanicity of patient residence, geographic region, insurance provider information, admission type, primary psychiatric diagnosis, and hospital characteristics (HC) such as ownership, bed size, location, teaching status. Analyses were conducted using SPSS version 26.0 (IBM Corp, Armonk, NY). Means (SD) were calculated for continuous variables and bivariate analyses were conducted with ANOVA. Frequencies (%) were calculated for categorical variables and bivariate analyses were conducted with Pearson’s chi-square test. Linear regression with an outcome of log-transformed length of stay (LOS) was used to identify statistically significant predictors of prolonged LOS. A backward elimination method was used for multivariate regression. P values ≤ 0.05 were considered statistically significant. 

Results: A total of 184,340 patients age 6-18 years old were included. Of these, 24,840 were children (age 6-11 years) and 159,500 were adolescents (age 12-18 years). The median LOS was 5.6 days. There were significant demographic differences between the children and adolescent populations. The stepwise backward elimination method determined that median household income, sex, and urbanicity were not significant predictors of LOS. Compared to mood disorder, the LOS was as follows: schizophrenia and other psychotic disorders (41% longer), developmental disorder (18% longer), impulse-control disorder (16% longer), attention deficit hyperactivity disorder (15% longer), anxiety disorder (14% longer), suicide and intentional self-injury (6% longer), alcohol related disorder (14% shorter) and substance related disorder (3% shorter). Patients admitted voluntarily had 20% longer LOS. Compared to hospitals in the Northeastern (NE) region, all the other regions had shorter LOS. Compared to public hospitals, private hospitals have longer LOS. Patients in teaching hospitals stay 8% longer. Compared to public insurance, LOS was shortened in private insurance group (6%) and self-pay (7%).

Conclusion. Substantial regional differences were observed with LOS longer in hospitals in the NE. Additionally, HC, including owner (public vs private) and teaching status also affect LOS. This suggests potential disparities in care for patients based on where they access care. 

Table of Contents

Chapter 1 - Introduction 1

Chapter 2- Manuscript 8

Chapter 3 - Discussion 17

About this Master's Thesis

Rights statement
  • Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.
Subfield / Discipline
  • English
Research Field
Committee Chair / Thesis Advisor
Committee Members
Last modified

Primary PDF

Supplemental Files