Original Research
Influence of Obesity Diagnosis With Organ Dysfunction, Mortality, and Resource Use Among Children Hospitalized With Infection in the United States
Journal of Intensive Care Medicine 1-7 ª The Author(s) 2016 Reprints and permission: sagepub.com/journalsPermissions.nav DOI: 10.1177/0885066616631325 jic.sagepub.com
Nidhi Maley, BA1, Achamyeleh Gebremariam, MS1, Folafoluwa Odetola, MD1, and Kanakadurga Singer, MD1
Abstract Background: Sepsis induces inflammation in response to infection and is a major cause of mortality and hospitalization in children. Obesity induces chronic inflammation leading to many clinical manifestations. Our understanding of the impact of obesity on diseases, such as infection and sepsis, is limited. The objective of this study was to evaluate the association of obesity with organ dysfunction, mortality, duration, and charges during among US children hospitalized with infection. Methods: Retrospective study of hospitalizations in children with infection aged 0 to 20 years, using the 2009 Kids’ Inpatient Database. Results: Of 3.4 million hospitalizations, 357 701 were for infection, 5685 of which were reported as obese children. Obese patients had higher rates of organ dysfunction (7.35% vs 5.5%, P < .01), longer hospital stays (4.1 vs 3.5 days, P < .001), and accrued higher charges (US$29 019 vs US$21 200, P < .001). In multivariable analysis, mortality did not differ by obesity status (odds ratio: 0.56, 95% confidence interval: 0.23-1.34), however severity of illness modified the association between obesity status and the other outcomes. Conclusions: While there was no difference in in-hospital mortality by obesity diagnosis, variation in organ dysfunction, hospital stay, and hospital charges according to obesity status was mediated by illness severity. Findings from this study have significant implications for targeted approaches to mitigate the burden of obesity on infection and sepsis. Keywords child, hospitalization, infection, length of stay, obesity, and illness severity
Background Obesity remains a significant problem in the pediatric population. Although the prevalence of childhood obesity has stabilized in the last 9 years, 17% of children in the United States are still considered obese.1 Obesity is associated with underlying inflammation that has also been linked to other metabolic and inflammatory diseases (cardiovascular disease, asthma, and diabetes).2 This phenomenon of increased inflammation is seen in the increased C-reactive protein levels and absolute neutrophil counts of children with increased body fat.3 Lifetime medical costs of obese children are estimated to be approximately US$19 000 higher than normal weight children.4 The prevalence of childhood obesity, and the medical costs associated with it, prompts the need to investigate its impact on resource use when it is present along with other common pediatric medical conditions that prompt hospitalization. In the United States, sepsis (severe infection) is a major cause of morbidity and mortality in children,5,6 especially in the presence of organ dysfunction.7 In sepsis, toxins from bacteria trigger an inflammatory response throughout the body,8 with patterns of activation similar to that observed in obesity.9
While obesity’s link with other inflammatory diseases such as asthma, cancer, and diabetes is well established,10-12 it is unclear whether the existing inflammatory milieu of an obese patient exacerbates the body’s response to infection. Existing literature suggests that when sepsis occurs in obese individuals and obese animal models, an exaggerated inflammatory response develops.13,14 Current research efforts are increasingly focused on the impact of obesity on sepsis in the adult population, but such studies are still nascent in the pediatric population. Among adults, the effects of obesity on the clinical course of infection are unclear as conflicting reports have
1 Department of Pediatrics and Communicable Diseases, University of Michigan Medical School, Ann Arbor, MI, USA
Received October 13, 2015, and in revised form January 15, 2016. Accepted for publication January 18, 2016. Corresponding Author: Kanakadurga Singer, Department of Pediatrics, University of Michigan, D1205 MPB 1500 East Medical Center Dr, Ann Arbor MI, 48109, USA. Email: [email protected]
Downloaded from jic.sagepub.com at FLORIDA ATLANTIC UNIV on March 6, 2016
2
Journal of Intensive Care Medicine
shown a detrimental effect,15 no significant effect,16 or a protective effect with those who are obese having improved outcomes, the so-called ‘‘obesity survival paradox.’’17-20 It is not known what, if any, impact obesity as a diagnosis has on clinical outcomes and resource use among children hospitalized with infection. Improved understanding of the influence of obesity on clinical outcomes and resource use among children hospitalized with infection might foster development of targeted treatment approaches that additionally focus on the patients’ obesity diagnosis. Current guidelines for treatment of children with infection and sepsis are generalized for all pediatric patients and do not account for potential alteration in therapeutics due to the patient’s obesity status. This study was conducted to investigate whether an obesity diagnosis in children hospitalized with infection influences the occurrence of organ dysfunction, hospital mortality, length of hospital stay, and hospital charges. We hypothesized that obese pediatric patients hospitalized with infection will have higher occurrence of organ dysfunction, higher in-hospital mortality, longer hospital stay, and higher charges when compared to nonobese patients.
Methods Study Design This is a retrospective study of children 20 years hospitalized with infection. The source of data was the Agency for Healthcare Research and Quality’s 2009 Kids’ Inpatient Database (KID). This is the only national, all-payer database of children’s hospitalizations comprising more than 3 million discharge records obtained from 4121 hospitals in 44 states.21 The database contains 80% of the normal non-newborn discharges and is nationally representative with the inclusion of sampling weights of the data in analyses. Patient demographics, hospital characteristics, and diagnosis codes are included for each hospitalization.21 This database has previously been utilized to describe clinical outcomes and resource utilization among US children hospitalized with sepsis.22
Study Sample and Variable Identification Children with a primary diagnosis of infection were identified using the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes, applying the same methodology used in prior studies.22,23 The ICD-9-CM codes defined hospitalizations with infection and not severe sepsis which would have involved combining codes for organ dysfunction with those for sepsis which would have been problematic, given that organ dysfunction is an outcome variable for the current study. To identify which of these children with infection also had a diagnosis of obesity, we utilized ICD-9CM diagnosis codes of 278.00, 278.01, or 278.02. The all-patient refined diagnosis-related group (APRDRG) classification within KID was used to categorize patient-level illness severity into minor, moderate, major, and extreme loss of function, hereafter referred to as illness severity. This classification
is a proprietary, validated and extensively used measure of illness severity.24,25 Organ dysfunction was identified using ICD-9-CM codes as applied previously.23 Outcome variables in the study included in-hospital mortality, organ dysfunction, hospital length of stay (LOS), and hospital charges. Organ dysfunction was measured both according to actual organ systems involved and the number of organ systems involved, since prior literature suggest worse outcomes with increase in the number of organ systems involved in the disease state.5,7,26 The ICD-9-CM codes were used to determine the existence of organ dysfunction. Patient characteristics studied, other than organ dysfunction, were age, gender, presence of comorbid illness, and payer status. Comorbid illness was identified using ICD-9-CM codes applying methodology described previously.27 Hospital characteristics of study interest included rural–urban status and the census regions.
Statistical Analysis We obtained weighted percentages for categorical variables and weighted means and medians for continuous variables. Bivariate analysis was performed to identify factors, in addition to obesity, associated with the outcome variables (in-hospital mortality, organ dysfunction, hospital LOS, and total hospital charges). We used the log-transformed total charges due to skewness in the distribution of data for overall hospital charges. We fit a simple linear regression model for log-total charges, negative binomial regression for LOS, and simple logistic regression for binary outcomes (mortality and organ dysfunction). Additional variables associated with the outcome variables, with a P value .20 in bivariate analyses, were included in multivariable regression models. Obesity diagnosis was the primary exposure variable in all models. To assess for any effect modification by patient illness severity, an interaction term between illness severity and obesity status (Obesity Illness Severity) was created and included in each of the 4 regression models. In multivariable analyses, to investigate the relationship between obesity status and the outcome, variables were adjusted for by fitting multivariable logistic regression models for mortality and organ dysfunction, negative binomial regression for LOS, and multiple linear regression models for logtotal charges. Predicted values of outcomes and associated 95% confidence intervals (CIs), for obesity by illness severity combinations, were obtained with adjustment for other predictor variables (age, gender, presence of comorbid illness, APRDRG severity, payer type, hospital region, and hospital location). For overall hospital charges, we obtained the predicted values in the original scale by back-transforming and applying the smearing estimate.28 The number of hospitalizations in our results was unweighted, while all effect estimates and accompanying 95% CIs were calculated using sample weights to account for the complex survey design. All estimates used the survey command in Stata for Windows version 11 (Stata Corp, College Station, Texas), which accounted for the complex
Downloaded from jic.sagepub.com at FLORIDA ATLANTIC UNIV on March 6, 2016
Maley et al
3
Table 1. Characteristics of the Study Population. Characteristic
No.
Table 2. Illness Severity According to Obesity Diagnosis. %
Age
Influence of Obesity Diagnosis With Organ Dysfunction, Mortality, and Resource Use Among Children Hospitalized With Infection in the United States
Journal of Intensive Care Medicine 1-7 ª The Author(s) 2016 Reprints and permission: sagepub.com/journalsPermissions.nav DOI: 10.1177/0885066616631325 jic.sagepub.com
Nidhi Maley, BA1, Achamyeleh Gebremariam, MS1, Folafoluwa Odetola, MD1, and Kanakadurga Singer, MD1
Abstract Background: Sepsis induces inflammation in response to infection and is a major cause of mortality and hospitalization in children. Obesity induces chronic inflammation leading to many clinical manifestations. Our understanding of the impact of obesity on diseases, such as infection and sepsis, is limited. The objective of this study was to evaluate the association of obesity with organ dysfunction, mortality, duration, and charges during among US children hospitalized with infection. Methods: Retrospective study of hospitalizations in children with infection aged 0 to 20 years, using the 2009 Kids’ Inpatient Database. Results: Of 3.4 million hospitalizations, 357 701 were for infection, 5685 of which were reported as obese children. Obese patients had higher rates of organ dysfunction (7.35% vs 5.5%, P < .01), longer hospital stays (4.1 vs 3.5 days, P < .001), and accrued higher charges (US$29 019 vs US$21 200, P < .001). In multivariable analysis, mortality did not differ by obesity status (odds ratio: 0.56, 95% confidence interval: 0.23-1.34), however severity of illness modified the association between obesity status and the other outcomes. Conclusions: While there was no difference in in-hospital mortality by obesity diagnosis, variation in organ dysfunction, hospital stay, and hospital charges according to obesity status was mediated by illness severity. Findings from this study have significant implications for targeted approaches to mitigate the burden of obesity on infection and sepsis. Keywords child, hospitalization, infection, length of stay, obesity, and illness severity
Background Obesity remains a significant problem in the pediatric population. Although the prevalence of childhood obesity has stabilized in the last 9 years, 17% of children in the United States are still considered obese.1 Obesity is associated with underlying inflammation that has also been linked to other metabolic and inflammatory diseases (cardiovascular disease, asthma, and diabetes).2 This phenomenon of increased inflammation is seen in the increased C-reactive protein levels and absolute neutrophil counts of children with increased body fat.3 Lifetime medical costs of obese children are estimated to be approximately US$19 000 higher than normal weight children.4 The prevalence of childhood obesity, and the medical costs associated with it, prompts the need to investigate its impact on resource use when it is present along with other common pediatric medical conditions that prompt hospitalization. In the United States, sepsis (severe infection) is a major cause of morbidity and mortality in children,5,6 especially in the presence of organ dysfunction.7 In sepsis, toxins from bacteria trigger an inflammatory response throughout the body,8 with patterns of activation similar to that observed in obesity.9
While obesity’s link with other inflammatory diseases such as asthma, cancer, and diabetes is well established,10-12 it is unclear whether the existing inflammatory milieu of an obese patient exacerbates the body’s response to infection. Existing literature suggests that when sepsis occurs in obese individuals and obese animal models, an exaggerated inflammatory response develops.13,14 Current research efforts are increasingly focused on the impact of obesity on sepsis in the adult population, but such studies are still nascent in the pediatric population. Among adults, the effects of obesity on the clinical course of infection are unclear as conflicting reports have
1 Department of Pediatrics and Communicable Diseases, University of Michigan Medical School, Ann Arbor, MI, USA
Received October 13, 2015, and in revised form January 15, 2016. Accepted for publication January 18, 2016. Corresponding Author: Kanakadurga Singer, Department of Pediatrics, University of Michigan, D1205 MPB 1500 East Medical Center Dr, Ann Arbor MI, 48109, USA. Email: [email protected]
Downloaded from jic.sagepub.com at FLORIDA ATLANTIC UNIV on March 6, 2016
2
Journal of Intensive Care Medicine
shown a detrimental effect,15 no significant effect,16 or a protective effect with those who are obese having improved outcomes, the so-called ‘‘obesity survival paradox.’’17-20 It is not known what, if any, impact obesity as a diagnosis has on clinical outcomes and resource use among children hospitalized with infection. Improved understanding of the influence of obesity on clinical outcomes and resource use among children hospitalized with infection might foster development of targeted treatment approaches that additionally focus on the patients’ obesity diagnosis. Current guidelines for treatment of children with infection and sepsis are generalized for all pediatric patients and do not account for potential alteration in therapeutics due to the patient’s obesity status. This study was conducted to investigate whether an obesity diagnosis in children hospitalized with infection influences the occurrence of organ dysfunction, hospital mortality, length of hospital stay, and hospital charges. We hypothesized that obese pediatric patients hospitalized with infection will have higher occurrence of organ dysfunction, higher in-hospital mortality, longer hospital stay, and higher charges when compared to nonobese patients.
Methods Study Design This is a retrospective study of children 20 years hospitalized with infection. The source of data was the Agency for Healthcare Research and Quality’s 2009 Kids’ Inpatient Database (KID). This is the only national, all-payer database of children’s hospitalizations comprising more than 3 million discharge records obtained from 4121 hospitals in 44 states.21 The database contains 80% of the normal non-newborn discharges and is nationally representative with the inclusion of sampling weights of the data in analyses. Patient demographics, hospital characteristics, and diagnosis codes are included for each hospitalization.21 This database has previously been utilized to describe clinical outcomes and resource utilization among US children hospitalized with sepsis.22
Study Sample and Variable Identification Children with a primary diagnosis of infection were identified using the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes, applying the same methodology used in prior studies.22,23 The ICD-9-CM codes defined hospitalizations with infection and not severe sepsis which would have involved combining codes for organ dysfunction with those for sepsis which would have been problematic, given that organ dysfunction is an outcome variable for the current study. To identify which of these children with infection also had a diagnosis of obesity, we utilized ICD-9CM diagnosis codes of 278.00, 278.01, or 278.02. The all-patient refined diagnosis-related group (APRDRG) classification within KID was used to categorize patient-level illness severity into minor, moderate, major, and extreme loss of function, hereafter referred to as illness severity. This classification
is a proprietary, validated and extensively used measure of illness severity.24,25 Organ dysfunction was identified using ICD-9-CM codes as applied previously.23 Outcome variables in the study included in-hospital mortality, organ dysfunction, hospital length of stay (LOS), and hospital charges. Organ dysfunction was measured both according to actual organ systems involved and the number of organ systems involved, since prior literature suggest worse outcomes with increase in the number of organ systems involved in the disease state.5,7,26 The ICD-9-CM codes were used to determine the existence of organ dysfunction. Patient characteristics studied, other than organ dysfunction, were age, gender, presence of comorbid illness, and payer status. Comorbid illness was identified using ICD-9-CM codes applying methodology described previously.27 Hospital characteristics of study interest included rural–urban status and the census regions.
Statistical Analysis We obtained weighted percentages for categorical variables and weighted means and medians for continuous variables. Bivariate analysis was performed to identify factors, in addition to obesity, associated with the outcome variables (in-hospital mortality, organ dysfunction, hospital LOS, and total hospital charges). We used the log-transformed total charges due to skewness in the distribution of data for overall hospital charges. We fit a simple linear regression model for log-total charges, negative binomial regression for LOS, and simple logistic regression for binary outcomes (mortality and organ dysfunction). Additional variables associated with the outcome variables, with a P value .20 in bivariate analyses, were included in multivariable regression models. Obesity diagnosis was the primary exposure variable in all models. To assess for any effect modification by patient illness severity, an interaction term between illness severity and obesity status (Obesity Illness Severity) was created and included in each of the 4 regression models. In multivariable analyses, to investigate the relationship between obesity status and the outcome, variables were adjusted for by fitting multivariable logistic regression models for mortality and organ dysfunction, negative binomial regression for LOS, and multiple linear regression models for logtotal charges. Predicted values of outcomes and associated 95% confidence intervals (CIs), for obesity by illness severity combinations, were obtained with adjustment for other predictor variables (age, gender, presence of comorbid illness, APRDRG severity, payer type, hospital region, and hospital location). For overall hospital charges, we obtained the predicted values in the original scale by back-transforming and applying the smearing estimate.28 The number of hospitalizations in our results was unweighted, while all effect estimates and accompanying 95% CIs were calculated using sample weights to account for the complex survey design. All estimates used the survey command in Stata for Windows version 11 (Stata Corp, College Station, Texas), which accounted for the complex
Downloaded from jic.sagepub.com at FLORIDA ATLANTIC UNIV on March 6, 2016
Maley et al
3
Table 1. Characteristics of the Study Population. Characteristic
No.
Table 2. Illness Severity According to Obesity Diagnosis. %
Age
