Lupus (2014) 23, 714–720 http://lup.sagepub.com

LUPUS AROUND THE WORLD

The effect of comorbidity on hospital mortality in patients with SLE from an Asian tertiary hospital Y Yang1,2, J Thumboo3,5, A Earnest2, SL Yong4 and KY Fong3,5 1

Department of Epidemiology, Medical Board, Singapore General Hospital, Singapore; 2Centre for Quantitative Medicine, Office of Clinical Sciences, Duke-NUS Graduate Medical School, Singapore; 3Department of Rheumatology & Immunology, Singapore General Hospital, Singapore; 4Medical Board, Singapore General Hospital, Singapore; and 5Duke-NUS Graduate Medical School, Singapore

Objectives: The objective of the study was to assess the disease burden of systemic lupus erythematosus (SLE) and the usefulness of the Charlson Comorbidity Index (CCI) as riskadjusted hospital mortality predictors in patients with SLE using a hospital administrative database. Methods: A historical cohort study of a hospital discharge database from 2004 to 2011 was used to identify cases with SLE and comorbidity using the International Statistical Classification of Diseases and Related Health Problems, ninth revision, Australian modification (ICD-9-AM) codes. Results: Over the eight years, 841 patients met the criteria of SLE with a hospital mortality rate of 9.2%. The hospital mortality rates (2.4%, 15.7%, 25.0%, and 30.4%, respectively, p < 0.001) and hospital length of stay (geometric mean, 3.5, 5.6, 8.8, and 7.5 days, respectively, p < 0.001) were consistently increased for patients with CCI ranging from none, low, moderate to high grade, respectively. Cox proportional hazards model analysis showed that CCI (hazard ratio (HR) 7.8 high vs. none, p < 0.001) and infectious disease (HR 2.0, p ¼ 0.016) were significant and independent predictors of hospital mortality. Similar results were also seen with hospital length of stay by zero-truncated negative binomial regression model analysis. Conclusion: The SLE burden is high in this population. Comorbidities and infectious disease were some of the most important contributors to hospital mortality and resource utilization. Lupus (2014) 23, 714–720. Key words: SLE; Charlson Comorbidity Index; hospital mortality; length of stay; ICD-9-AM

Introduction Systemic lupus erythematosus (SLE) is a chronic inflammatory autoimmune disease, with an estimated prevalence ranging from 103.0 to 149.5 per 100,000 people in the United States (US).1,2 In Asians, SLE has been reported to be more prevalent,3,4 with a rate approximately two to three times higher than the Caucasian populations.1,5 Also, Asian patients with SLE have been reported to have higher mortality rates.6–8 The economic burden is high for the management of SLE, too. In a US study from 2008, the annual direct health care cost for a single patient was estimated to be approximately US$13,000, with the indirect cost exceeding US$8600.9 Others also reported an Correspondence to: Yang Yong, Epidemiology Department, Medical Board, Singapore General Hospital, Outram Road, 169608 Singapore. Email: [email protected] Received 30 October 2013; accepted 8 January 2014

estimated two to three times higher indirect cost compared with the corresponding direct costs.10 Over the past 50 years, survival has improved dramatically in patients with SLE.11 However, tremendous improvements have been seen only in short-term and medium-term survival, and the long-term prognosis of SLE remains poor.10 However, even with improved care, patients with SLE have up to a five-fold higher mortality rate than the general population.12 Risk adjustment has been identified as a necessary strategy for evaluating health care outcomes, and the severity of patients’ comorbidities is part of this adjustment.13,14 The Charlson Comorbidity Index (CCI)15 is an algorithm developed to classify a patient’s severity of comorbidities that uses recorded data on secondary diagnoses to assign a weight to morbidity, thereby generating an index of the patient’s risk of death. The index has been widely utilized by health researchers to measure burden of disease. Since the first study by

! The Author(s), 2014. Reprints and permissions: http://www.sagepub.co.uk/journalsPermissions.nav

Downloaded from lup.sagepub.com at NANYANG TECH UNIV LIBRARY on May 26, 2015

10.1177/0961203314522340

Factors affecting mortality in Asian SLE patients Y Yang et al.

715 15

Charlson et al., the index has been validated for its ability to predict mortality in various disease subgroups.13,16–20 These studies consistently demonstrate that the CCI is a valid prognostic indicator for mortality.19 Causes and prognostic indicators of death in SLE have been well studied in Western countries but are lacking in Asian countries. There is also scarce clinical information on mortality studies among Asian lupus patients.7,8 Thus, this study aims to assess the disease burden of SLE in patients from a tertiary hospital in Singapore and assess the usefulness of CCI and infection as a hospital risk adjustment method in patients with SLE.

Methods Data source Singapore General Hospital (SGH) is the biggest acute tertiary-care hospital in Singapore, with 1775 beds and serving approximately one-third of its population of 5.31 million.21 Each year, SGH handles approximately 80,000 patient discharges. Data of all hospitalized patients aged 21 years who were admitted to SGH from January 1, 2004 to December 31, 2011, were collected from a hospital data warehouse at the Information Technology Department, SingHealth Group, Singapore. The collected data included demographic information such as race, gender, age and marital status, and clinical characteristics including hospital admission and discharge date, admission ward class, intensive care unit (ICU) admission and discharge date, admission ICU type, up to 10 International Statistical Classification of Diseases and Related Health Problems, ninth revision, Australian modification (ICD-9-AM) diagnosis codes, up to 10 ICD-9-AM procedure codes, discharge status, and disposition at discharge. In Singapore, admission ward classes are categorized based on different levels of government subsidies. The government subsidies are 0%, 20%, 65% and 80% for ward classes A, B1, B2 and C, respectively.22 In general, patients in the lower socioeconomic groups are more likely to be admitted to beds in wards with a higher level of government subsidy. Therefore, we chose admission ward class as a surrogate to measure the socioeconomic status of the patients. The protocol for this study was approved by the ethics committee of SGH. The informed consent process was exempted as this study involved only reviewing hospital databases.

Case definition We defined individuals with SLE as having 2 ICD-9-AM diagnostic codes for SLE (710.0) from the hospital discharge database, as described previously.23 The method has been demonstrated to have a sensitivity of 98.2% and a specificity of 72.5% for identifying true cases of SLE.23 As patients with SLE had an average of two or more admissions recorded over the study period, patients’ data pertaining to the latest hospitalization were selected for analysis to avoid double counting of any patient. Infectious disease was defined by any ICD-9-AM codes 001-139.8, 480486, and 996.62. Surgical and medical conditions were determined based on the respective treatment departments. Chronic comorbid conditions were identified by ICD-9-AM codes and were then calculated by an established Charlson index.14,15,24 This index assigns a score between one and six points to a range of diseases (one point for myocardial infarction, congestive heart failure, peripheral arterial disease, cerebrovascular disease, dementia, chronic pulmonary disease, connective tissue disease, ulcer disease, mild liver disease, and diabetes without organ damage; two points for diabetes with organ damage, hemiplegia, severe renal disease, and nonmetastatic cancer; three points for severe liver disease; six points for metastatic cancer and human immunodeficiency virus infection), and the sum of these points serves as a measure of the burden of comorbidity.14 The Charlson score was then classified into four previously defined grades known as the CCI: zero points (none), one to two points (low), three to four points (moderate), and 5 points (high).14,15,24 The CCI has been widely used to predict mortality in hospitalized patients, including hospitalized sepsis patients from the same hospital.15,25 Statistical analysis Categorical variables were reported as percentages, and continuous variables were reported as mean and SD except for hospital length of stay (LOS), which was reported by geometric mean (GM) and 95% confidence interval (CI) because of its skewed distribution as a summary statistic. Hospital mortality rate was defined as the ratio of SLE cases ending in death to the total number of SLE cases. Age was categorized into three groups (21–44, 45– 74, and >74 years) to capture the nonlinear effect of age on mortality. Parameters were compared among different CCI groups by chi square test for hospital mortality and by Kruskal-Wallis rank test Lupus

Downloaded from lup.sagepub.com at NANYANG TECH UNIV LIBRARY on May 26, 2015

Factors affecting mortality in Asian SLE patients Y Yang et al.

716

for LOS. Multivariate analysis were used to access the association of hospital mortality with CCI groups using the Cox proportional hazards model with adjustment for gender, ethnicity, medical management, admission class and infectious disease. Hazard ratios (HRs) were reported with 95% CIs. The test of proportional-hazards assumption was used for model calibration. Because of over-dispersion and the absence of zero values in hospital LOS, a zero-truncated negative binomial regression model was used to estimate the association between hospital LOS and patient CCI after adjusting for covariates and interactions as listed above. Relative LOS was described in terms of incidence-rate ratios (IRR).26 All p values were two sided. The level of statistical significance is considered to be the conventional alpha ¼0.05. The data analysis was performed using STATA Version 12.0 (StataCorp, College Station, TX, USA).

Results Incidence and mortality Of 301,568 hospitalized patients aged 21 years and older from 2004 to 2011, 841 (0.28%) patients met the criteria of SLE, of whom 77 (9.2%) patients died. Patients who died (51.8 vs. 44.0 years, p < 0.001) were older and less likely to be females (80.5% vs. 86.8%, p ¼ 0.129) than survivor patients. Nonsurvivor patients (10.4% vs. 17.7%, p < 0.001) were less likely to be surgically managed and more likely to be admitted to C class beds (54.5% vs. 31.4%) and less likely to be admitted to A or B class beds (2.6% and 42.9% vs. 12.3% and 56.3%, p < 0.001) than survivor patients. Nonsurvivor patients had significantly higher rates of infectious disease (70.1% vs. 28.1%, p < 0.001) and were more likely to be admitted to the ICU (63.6% vs. 3.5%, p < 0.001) than survivor patients. The demographic and clinical characteristics and outcomes of all patients with SLE are shown in Table 1. Clinical outcomes As for the clinical outcomes, patients with SLE had increased mortality rates with increased CCI grades (2.4% in no CCI, 15.7% in low CCI, 25.0% in moderate CCI, and 30.4% in high CCI, p < 0.001). A similar trend was observed in hospital LOS (3.5, 5.6, 8.8, and 7.5 days, respectively,

Table 1 Demographic and clinical characteristics of hospitalized SLE patients

Age, mean year (SD)

The effect of comorbidity on hospital mortality in patients with SLE from an Asian tertiary hospital.

The objective of the study was to assess the disease burden of systemic lupus erythematosus (SLE) and the usefulness of the Charlson Comorbidity Index...
103KB Sizes 0 Downloads 3 Views