Morse AR, Yatzkan E, Berberich B, Arons RR. Acute Care Hospital Utilization by Patients With Visual Impairment. Arch Ophthalmol. 1999;117(7):943-949. doi:10.1001/archopht.117.7.943
Copyright 1999 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.1999
To assess whether visual impairment contributes to average length of stay (ALOS) within inpatient care facilities.
We used the New York State Department of Health's Statewide Planning and Research Cooperative System (SPARCS) data for 1993, containing 1 principal diagnosis code and up to 8 secondary diagnosis codes for approximately 2.6 million hospital discharges. We evaluated ALOS differences in patients with and without visual impairment and in patients with eye pathologic conditions, including eye surgery. Visual impairment is not a primary admitting diagnosis, but may be coded as a secondary diagnosis. Eye pathology comprises a large variety of conditions, including corneal ulcers, abscesses, corneal deposits, edema, cataracts, vitreous hemorrhages, and many other eye disorders (ICD-9-CMcodes 360-368.9 and 370-379).
The ALOS was 13.4 days for patients with visual impairment (N=5764), 11.9 days for patients with either eye pathology or visual impairment (N=60,085), and 8.2 days for patients with no visual impairment (N=2,546,586). Using a series of multivariate models that controlled for the variables of age, sex, and payer source, as well as disease, disorders, and ophthalmology procedures, we found that the existence of visual impairment added 2.4 days to the ALOS (P<.001), while eye pathology combined with a secondary diagnosis of visual impairment added 1.8 days to the ALOS (P<.001).
Visual impairment contributes significantly to hospital length of stay. A better understanding of the functional care needs of patients with visual impairment in an acute care setting and at the time of discharge from the hospital may contribute to reducing excess ALOS and its related costs while improving the quality of patient care.
VISUAL IMPAIRMENTS affect approximately 8.6 million people, or 33.1 per 1000, in the United States.1 There is a paucity of hospital length of stay (LOS) studies on this population. Studies of visual impairment are generally clinically related, while LOS studies are typically focused on chronic disabilities, psychiatric disorders, orthopedic problems, patients undergoing rehabilitation, or elderly patients.2- 4 Retrospective analysis of inpatient care hospital discharge abstracts of patients with visual impairment suggests important questions that are not addressed by clinical studies. For example, does visual impairment, combined with other conditions, result in increased LOS? To address this question, this study examined the approximately 2.6 million acute care discharges from the hospital that occurred in New York State in 1993.5 Analysis of large-scale hospital data sets is essential to provide answers to such complex questions and to measure the effects on treatment strategies of nonacute disorders, such as visual impairment, coupled with acute diseases.6,7 Geographical disease and procedure treatment patterns are increasingly available, allowing for better understanding of how low-incidence diseases and comorbid conditions, combined with major diagnoses, influence LOS within a changing and unpredictable health care environment.8- 12 This study evaluates the differences in hospital average LOS (ALOS) in patients with and without visual impairment and patients with eye pathologic conditions. Eye pathology as a primary diagnosis includes acute conditions, such as corneal ulcers, abscesses, corneal deposits, edema, cataracts, and vitreous hemorrhages, as well as many other eye disorders and eyelid disorders. A patient with eye pathology as a diagnosis has been admitted for primary treatment of a particular eye condition. These patients are included in the analysis because our underlying assumption is that an acute eye condition requiring hospitalization will affect vision, even if only on a short-term basis. Applying descriptive statistics and multivariate analysis, we measured ALOS differences between patients with both visual impairment and combined visual impairment and eye pathology.
The goal of this study was to determine the contribution of visual impairment to hospital ALOS.
All hospital discharge abstracts for 1993 were obtained from the New York State Department of Health's Statewide Planning and Research Cooperative Systems (SPARCS) data. Multivariate models were tested with LOS as the dependent variable and age, sex, race, hospital discharge disposition, third-party payer, and major diagnostic category as independent variables. Descriptive statistics are included to measure ALOS differences between individual International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM), codes, diagnosis related groups (DRGs), and clinical subspecialty categories.
We defined visually impairedto include all discharges containing ICD-9-CMcodes 369.0 (Profound Impairment in Both Eyes) through 369.9 (Unspecified Visual Loss). These codes are defined as functional limitations of the eye(s). We defined eye pathologyto include all discharges containing ICD-9-CMdisease codes 360.0 (Purulent Endophthalmitis) through 379.99 (Other Ill-defined Disorders of the Eye). We defined eye surgeryto include ICD-9-CMprocedure codes 08.0 (Operation of the Eyelid) through 16.99 (Other Operations on Eyeball).13 All other discharges from the hospital were designated as patients with no visual impairment. We assumed that eye pathologic conditions, such as Operation of the Eyelid or cataract extraction, would be day surgery procedures done in ambulatory care settings.
The New York State Department of Health's SPARCS data contain 1 principal diagnosis and 8 secondary diagnosis codes. For our study, we used 1 principal procedure code and 4 secondary procedure codes. When any of the designated ICD-9-CMcodes was identified as either the principal or secondary diagnosis or procedure, the hospital discharge was classified as either visual impairment or eye pathology. The 20 independent sociodemographic and clinical variables included age, race (African American vs all other), sex, third-party payer (Medicare vs non-Medicare), discharge disposition (alive or dead), visual impairment vs no visual impairment, eye pathology vs no eye pathology, and eye surgery vs no eye surgery. Disease and disorder variables (used as surrogates for disease severity) included infection; malignant neoplasm; metabolic and blood disorders; mental health; nervous, circulatory, digestive, genitourinary, musculoskeletal, and respiratory systems; and injury/poisoning, and were developed by grouping ICD-9-CMcodes according to major body systems.13
We used version 6.07 of the SAS application software (SAS Inc, Cary, NC) for the statistical analyses. Data were accessed using IBM mainframe languages.14- 16 The SAS procedures used were the general linear models PROC GLM and PROC MEANS. Multivariate linear regression models were built to test the relationships among ALOS, the dependent variable, and the sociodemographic and clinical variables. Models tested are represented by the following equation17: Y=b0+b1X1+b2X2 . . . biXi+E where Yis the dependent variable ALOS, X1. . . Xiare the independent variables, and Eis the error, while the coefficients b0, b1, b2, and bimeasure the strength of the variables when significance is achieved. The distribution of ALOS for patients discharged from the hospital is not normally distributed and is positively skewed. While typically a log-linear transformation is used, this has been shown to have an insignificant impact on the coefficients and their significance.18 Patient age is a continuous variable; all other variables are treated as either indicators or dummy variables (R.R.A., unpublished data from the Office of Case Mix Studies, New York Presbyterian Hospital, New York, NY, 1994). When the total number and the numbers in each group vary, this is attributable to missing values associated with the variables within the particular model.
This study included 2,552,350 inpatient hospital discharges in New York State for 1993. Table 1presents the values for 20 variables. The ALOS for all hospital discharges was 8.2 days, and the mean age was 43.6 years (mean age of patients with visual impairment, 66.3 years). The contribution of age to ALOS was significant in both the visual impairment and eye pathology groups. (Table 2). Patients with visual impairment accounted for 0.23% of the total number of hospital discharges. The 0.23% identified as patients with visual impairment primarily included patients who were legally blind. When the eye pathology group was included with the visual impairment group, 2.4% of total hospital discharges were represented. Eye surgery accounted for 0.9% of all discharges from the hospital.
A group of 5764 patients discharged from the hospital were identified as patients with visual impairment. For these patients, the ALOS was 5.2 days longer than for patients with no visual impairment (13.4 vs 8.2 days). The ALOS was 11.9 days for patients with visual impairment and eye pathology (N=60,085).
Table 3presents the top 25 ICD-9-CMprincipal diagnosis codes by number of cases with visual impairment as a secondary diagnosis, listed in descending order, and reports the ALOS differences between patients with and without visual impairment. The most common principal diagnosis for visual impairment was ICD-9-CMcode 428.0 (Congestive Heart Failure), followed by ICD-9-CMcode 486 (Pneumonia, Not Otherwise Specified). The total numbers of cases accounted for by these 25 diagnoses were 1956 for patients with visual impairment and 450,608 for those with no visual impairment.
Table 4presents the ALOS differences by clinical subspecialties. We used DRG-mapping methods that distributed DRGs into meaningful clinical areas related to the clinical service of the physician responsible for the overall care of a patient (R.R.A., unpublished data from the Office of Case Mix Studies, New York Presbyterian Hospital, 1994). The most common clinical subspecialty of physicians caring for patients with visual impairment was general medicine (1198 hospital discharges in which visual impairment was a secondary diagnosis). In the general medicine group, the LOS was 1.7 days longer for patients with visual impairment than for patients with no visual impairment.
Table 5presents the top 25 ICD-9-CMprincipal diagnosis codes by number of cases in which eye pathology or visual impairment was either the principal or secondary diagnosis, listed in descending order. The table shows differences in ALOS between patients with eye pathology or visual impairment and those with no visual impairment. The most common principal diagnosis for patients with eye pathology or visual impairment was ICD-9-CMcode 428.0 (Congestive Heart Failure), followed by ICD-9-CMcode 486 (Pneumonia, Not Otherwise Specified).
General linear multivariate models were used to measure the influence of visual impairment and visual impairment or eye pathology, with or without related surgery, on hospital ALOS, controlling for disease and sociodemographic factors. Although the total variability of ALOS explained by these models is small, the coefficients shown in Table 2support these findings: Visual impairment adds 2.4 days to ALOS (P<.001). When eye pathology, including visual impairment, exists, ALOS increases by 1.8 days (P<.001). When eye surgery accompanies eye pathology, the ALOS decreases by 0.4 days (P<.001). For every year above the mean age of 59.0 years, the ALOS increases by 0.08 days (P<.001). Men stay 0.4 days longer than women (P<.001). African American patients stay 1.8 days longer than patients of other races (P<.001). Medicare patients stay 2.4 days longer than non-Medicare patients (P<.001).
Table 6summarizes the major findings of this study.
Patients with visual impairment had an ALOS 2.4 days longer than that of patients with no visual impairment (P<.001). The diagnoses that contributed most to this difference, as well as the economic impact of the estimated costs associated with the additional inpatient days, are discussed below.
Rehabilitative care was prescribed for 2.7% of patients with visual impairment compared with only 0.75% of patients with no visual impairment on discharge from the hospital. This suggests a need for more effective hospital discharge planning and disease management for patients with visual impairment and all other patients with special health care requirements. Existing community-based programs specific to patients with disablities, blindness, and visual impairment can offer acute care hospitals resources to assure prompt hospital discharge and more appropriate care at lower cost to the facility and payers.19
Patients with visual impairment in New York State had a significantly longer ALOS (2.4 days, P<.001) than patients with no visual impairment after adjusting for sociodemographic and disease variables. For the 5764 patients with visual impairment discharged from the hospital, 13,834 additional patient days were used. When patients with eye pathology are included with patients with visual impairment, the ALOS was 1.8 days longer than for patients with no visual impairment (P<.001). Thus, 60,085 hospital discharges resulted in 108,153 additional patient days.
We conclude that visual impairment contributes significantly to excess inpatient LOS. A better understanding of the specific needs of patients with visual impairment in an acute care setting might help to reduce these additional days and their related costs. The reasons for increased ALOS for patients with visual impairment seem to be a function of visual impairment and not a result of a comorbid condition or complicating factor associated with the admitting diagnosis. However, the absence of information on the accuracy of coding of eye conditions is a limitation of this study.
While this study focuses on visual impairment, it can be viewed as a paradigm for other disabilities or special conditions. Much of the current health care system is driven by diagnosis, with minimal regard for concomitant conditions, such as visual impairment, that may adversely affect the LOS of the patient. An admitting assessment of conditions that could predictably affect the patient's LOS should be viewed with the same importance as the admitting diagnosis in controlling costs and assuring appropriate and adequate patient care.
While this study does not identify the reasons for the increased ALOS of patients with visual impairment, we hypothesize that patients with visual impairment may be catheterized longer than patients without visual impairment because of staff concerns about persons with visual impairment safely finding the bathroom and using it appropriately. Prolonged catheterization may lead to more infections and may prolong the ALOS. The ALOS for those undergoing cesarean sections was longer for patients with eye pathology and visual impairment than for patients with no visual impairment. This may be because the hospital staff are not familiar with the proper care for mothers with visual impairment. Other examples include a lack of mobility during hospitalization because of problems in room orientation, confusion in finding the bathroom or safely walking in the corridors, the inability to find call bells to summon help, or finding food trays that are covered and unfamiliar. These may all be factors that contribute to increased ALOS.
Physicians and hospital discharge planners need to be sensitive to the functional needs of patients with visual impairment. Patients with significant visual impairment may be unable to obtain and manage new prescriptions or to properly care for themselves without immediate vision rehabilitation and other community-based support. Patients with eye pathologic conditions may have short-term functional deficits that need to be planned for and addressed in the hospital discharge process even if their long-term vision prognosis is favorable. We speculate that these factors are at the core of the differences in LOS between patients with and without visual impairment demonstrated in this study. Further research may determine the exact reasons for the increased ALOS of patients with visual impairment.
Accepted for publication January 9, 1999.
Corresponding author: Alan R. Morse, PhD, The Jewish Guild for the Blind, 15 W 65th St, New York, NY 10023 (e-mail: email@example.com).