Pediatrics and Neonatology (2013) 54, 179e187
Available online at www.sciencedirect.com
journal homepage: http://www.pediatr-neonatol.com
ORIGINAL ARTICLE
Factors Affecting Length of Stay in the Pediatric Emergency Department Sung-Tse Li a,b,c,*, Nan-Chang Chiu c,d, Wen-Chuan Kung e, Juei-Chao Chen f a
Department of Pediatrics, Hsinchu Mackay Memorial Hospital, Hsinchu, Taiwan Department of Early Childhood Care and Education, Taoyuan Innovation Institute of Technology, Chungli, Taiwan c Mackay Medicine, Nursing and Management College, Taipei Taiwan d Department of Pediatrics, Mackay Memorial Hospital, Taipei, Taiwan e Nursing Department, Hsinchu Mackay Memorial Hospital, Hsinchu, Taiwan f Department of Statistics and Information Science, College of Management, Fu Jen Catholic University, Taipei, Taiwan b
Received Dec 5, 2011; received in revised form Jun 5, 2012; accepted Nov 21, 2012
Key Words classification and regression tree; length of stay; pediatric emergency department
Background: A large volume of visits can cause an emergency department (ED) to become overcrowded, resulting in a longer length of stay (LOS). The objective of this study was to analyze factors affecting the LOS in the pediatric ED. Methods: Records of all visits to the pediatric ED of the study hospital, from July 1, 2006 to June 31, 2007, were retrospectively retrieved. Data were collected from the hospital’s computerized records system. Eta-squared correlation ratio and Cramer’s V test evaluated the associations between variables. Two-thirds of the database was randomized for the classification and regression tree (CART) model-building dataset, and one-third was used for the validation dataset. Results: A total of 29,035 patients visited the pediatric ED during the evaluation period. Of the total visits, 61.1% were due to complaints of fever. The mean LOS was 2.6 4.67 hours, and 74.3% of visits had an LOS of shorter than 2 hours. The CART analysis selected five factors (waiting time for hospitalization, laboratory tests, door-to-physician time, gastrointestinal symptoms, and patient outcome) to produce a total of nine subgroups of patients. The mean LOS of the modelbuilding dataset closely correlated with that of the validation dataset (r2 Z 0.999). Conclusion: Patients who were waiting for hospitalization for less than 8 hours or were not admitted, those without any laboratory tests, those having door-to-physician time less than 60 minutes, and those without any gastrointestinal symptoms had the shortest LOS. Patients who waited for hospitalization for more than 16 hours had the longest LOS. Copyright ª 2012, Taiwan Pediatric Association. Published by Elsevier Taiwan LLC. All rights reserved.
* Corresponding author. Department of Pediatrics, Hsinchu Mackay Memorial Hospital, Number 690, Section 2, Guangfu Road, Hsinchu City 30070, Taiwan. E-mail address:
[email protected] (S.-T. Li). 1875-9572/$36 Copyright ª 2012, Taiwan Pediatric Association. Published by Elsevier Taiwan LLC. All rights reserved. http://dx.doi.org/10.1016/j.pedneo.2012.11.014
180
1. Introduction An increasing number of patients visit hospital emergency departments (EDs) for treatment.1,2 In the USA, children younger than 15 years have an ED utilization rate of approximately 35.1 per 100 persons, accounting for 23% of total ED visits.1 According to data from the Bureau of National Health Insurance in Taiwan, the annual number of ED visits increased from 4.99 million in 1996 to 7.23 million in 2010. A large volume of visits may lead to overcrowding of the ED, as reflected by longer patient waiting times. This may place patients at a greater risk of poor outcomes.3e5 Length of stay (LOS) is one of the important quality factors in the ED, where prolonged waiting time can increase the likelihood of patient dissatisfaction.6e8 Several factors influence the LOS of adult patients, including patient characteristics; the requirement for further evaluations, such as radiology studies and laboratory tests; the need for sedation or special procedures; and the waiting time for hospitalization.9e15 In one US study, investigators reported that factors independently associated with an LOS of more than 10 hours in a pediatric ED were longer waiting time, night shift arrival, high triage acuity, radiology studies, and subspecialty consultations.16 Risk factors may differ in various areas, and, as yet, no study in Taiwan has published data concerning factors associated with LOS in the pediatric ED. The present retrospective study thus aims at identifying the varying factors associated with the LOS in the pediatric ED. Results may prove useful for hospital strategy development and for improving quality in the pediatric ED. A classification and regression tree (CART) analysis is a core component of the decision tree tool for data mining and predictive modeling.17 CART is a nonparametric discriminant method based on statistical theory. It is structured in two parts: a classification tree and a regression tree. The classification tree uses a nominal variable as the dependent variable, and the regression tree uses a continuous variable as the dependent variable.18 Results of the CART analysis are presented as a decision tree, which separates patients into subgroups following the flowchart form. The flowchart structure and rules can easily be explained and are easy to understand. Recent studies have shown CART to provide as meaningful an analysis as other traditional statistical techniques, such as linear regression and logistic regression analyses.18e20 Some earlier studies had used CART to analyze factors associated with the LOS in a hospital,21,22 but this is the first study that used CART to construct a model to predict LOS and analyze the factors affecting LOS in the pediatric ED.
2. Methods This study was approved by the Mackay Memorial Hospital Institutional Review Board. Computerized records of all visits to the pediatric ED of the study hospital, from July 1, 2006 to June 31, 2007, were retrieved. All patients younger than 18 years who registered at the pediatric ED were included in the analyses. The study hospital is a secondary teaching hospital in northern Taiwan, which receives approximately 30,000
S.-T. Li et al emergency visits per year. Triage nurses were required to enter patient registration data, including triage level, age, body weight, sex, arrival time, chief complaint, vital signs, and medication allergy history, into the hospital computer network. After the waiting time, children were evaluated and treated by the attending or resident pediatric physicians. Pediatric physicians recorded patients’ clinical conditions, physical examination results, medications, and necessary laboratory or radiology tests. Data collected included information on triage level, age, sex, registration and discharge time, time of physician’s first order, time of transfer to observation unit, chief complaint, image study, laboratory tests, consultation, and final patient outcome. All computerized medical records of every revisit were reviewed by one of the authors. The time of registration at the ED was considered as the time of the visit. LOS was calculated as the time period between registration of patients and the time of their physically leaving the ED, whether admitted or discharged. Door-to-physician time was calculated as the time period between the arrival at triage and the first evaluation by a physician. If no beds were available when patients needed hospitalization for more treatment, they had to be sent to the observation unit at the ED, where they should wait for admission. The waiting time for hospitalization was calculated as the time between when they were sent to the observation unit and the time of their admission. Age was divided into five groups: (1) younger than 3 months, (2) 3 months to 1 year, (3) 1e6 years, (4) 7e12 years, and (5) 13e18 years. The time of patients’ registration was decided based on the three shifts of a day: day shift (8:00 AM to 4:00 PM), evening shift (4:00 PM to 12:00 midnight), and night shift (12:00 midnight to 8:00 AM). The days of a week were divided into weekdays (Monday to Friday; outpatient department available in the mornings and afternoons), Saturdays (outpatient department available in the mornings only), and holidays (outpatient department closed). According to the body system, chief complaints were categorized into six groups: (1) fever as the predominant symptom, (2) gastrointestinal symptoms, (3) respiratory symptoms, (4) dermatological symptoms, (5) neurological symptoms, and (6) other symptoms or complaints. One patient may have had one symptom or more than one symptom occurring simultaneously. Door-to-physician time was categorized into five groups: (1) less than 10 minutes, (2) 10e30 minutes, (3) 30e60 minutes, (4) 60e120 minutes, and (5) more than 120 minutes. Waiting time for hospitalization was categorized into five groups: (1) nonadmission or no bed available, (2) less than 8 hours, (3) 8e16 hours, (4) 16e24 hours, and (5) more than 24 hours. Patient outcome was categorized into six groups: (1) discharge, (2) ward admission, (3) intensive care unit (ICU) admission, (4) transfer to another hospital, (5) discharge against medical advice, and (6) death. This study also collected information on whether these patients had undergone consultation by other subspecialty physicians, and had image and laboratory studies.
Factors affecting LOS in pediatric patients Data were collected from the computerized records system of the hospital and entered into Microsoft Excel (Microsoft Corp., Redmond, WA, USA). Statistical analysis was performed using SPSS 17.0 Windows, version 7, software. Data for categorical variables were analyzed using the chi-square or Fisher’s exact test. Data for continuous variables were analyzed using one analysis of variance, as appropriate. A p value of