Sundaram, Shikha S.*; Alonso, Estella M.†; Zeitler, Phil‡; Yin, Wanron§; Anand, Ravinder§; on Behalf of the SPLIT Research Group
Pediatric obesity, defined as a body mass index (BMI) greater than the 95% for age and sex, is an escalating public health crisis (1). National Health and Nutrition Examination Surveys (NHANES) data highlight alarming trends in pediatric obesity. NHANES data from 1971 to 1974 reported obesity rates of 5% in 2- to 5-year-olds, 4% in 6- to 11-year-olds and 6.1% in 12- to 19-year-olds. The latest NHANES data from 2007 to 2008 report that 10.4% of 2- to 5-year-olds, 19.6 % of 6- to 11-year-olds, and 18.1% of 12- to 19-year-olds are obese, with a 16.9% overall prevalence of pediatric obesity (2). Racial and sex disparities exist in the prevalence of obesity, with non-Hispanic black boys, Hispanic American boys, and non-Hispanic black girls most affected (2).
Obese patients are known to experience significant medical and psychosocial comorbidities. These comorbidities include diabetes, impaired glucose tolerance, hypertension, hyperlipidemia, nonalcoholic fatty liver disease, the metabolic syndrome, cardiovascular disease, and depression. Significant weight gain and obesity after LT in adults has been widely reported, ranging from 17% to 43% (3–7). There is a paucity of literature regarding obesity in the pediatric LT population.
Therefore, the objectives of the present study were to determine the prevalence of obesity in the pediatric LT population and compare it with that from the general pediatric population of the United States and characterize and determine risk factors for obesity in pediatric LT recipients.
Data were queried from the Studies of Pediatric Liver Transplantation (SPLIT) database, which prospectively collects data on children receiving LTs at 39 centers in the United States and Canada. The study population included patients who received an LT between 1995 and 2007, were ages 2 to 18 years at follow-up, and had BMI data available on at least 1 occasion between years 1 and 5 after LT.
Overweight was defined as a BMI between 85% and 95% for age and sex (1). Obesity was defined as a BMI of ≥95% for age and sex (1). Throughout the present study, the term obesity encompasses both obesity and severe obesity unless otherwise specified. Severe obesity was defined as a BMI of at least 99% for age and sex, in keeping with recommendations by an expert committee convened by the American Medical Association, the Centers for Disease Control and Prevention and the Department of Health and Human Services (1,8).
The χ2 tests were performed to compare baseline categorical variables between obese and nonobese patients. The generalized estimating equation (GEE) approach was used to perform logistic regression on longitudinal data to identify risk factors for obesity and to calculate odds ratios (ORs) and 95% confidence intervals (CIs) associated with risk factors. The initial multivariate model included risk factors with a P <0.10 on univariate analysis. Before including time invariant factors identified on univariate GEE analysis into the multivariate model, individual GEE analyses with repeated measures were performed to examine interactions between the factor and time to determine whether an interaction effect should be included in the multivariate model. The final multivariate model was derived using a stepwise backward elimination procedure. Factors remaining significant at the P ≤ 0.05 level were considered statistically significant and entered the final model. For both the univariate and multivariate GEE analyses, robust standard error estimates were used to provide ORs, 95% CI, and P values.
To compare SPLIT and NHANES obesity data, SPLIT follow-up visits were divided into three groups based on calendar year of follow-up, in keeping with NHANES data reports (1999–2000, 2001–2002, and 2003–2004). In cases of multiple visits by 1 individual, the time point reflecting the last collected BMI was used for data analysis. The proportion of obese individuals was then calculated for each group (1999–2000, 2001–2002 and 2003–2004) in parallel with published NHANES data (2).
Testing for differences in the proportion of obese individuals in the SPLIT and NHANES samples requires calculation of standard error (SE) under the assumption that the true proportion is the same in both groups. This would require calculation of a pooled SE combining information from both the SPLIT and NHANES samples. Data required to calculate the pooled SEs are not included in data from the NHANES cohort reported by Ogden et al (2). Therefore, a 95% CI approach was used, which does not require assumptions about equality of proportions. The SE for the NHANES sample proportion is available, which can be combined with the SPLIT SE to determine the SE for the difference and calculate the 95% CI. This CI can be used to understand whether the 2 sample proportions are significantly different; if the value 0 is absent from the CI, then the data suggest that the true proportions are different. All of the statistical analyses were performed using SAS for Windows, version 9.2 (SAS Institute Inc, Cary, NC).
Demographic data for the 1706 individuals included in the study are presented in Table 1. Figure 1 demonstrates the weight status of this cohort from 1 to 5 years after LT. The majority of children are of healthy weight during the follow-up period. However, at 1, 2, and 3 years after transplant, 19.2% (of whom 10.5% were severely obese), 17.6% (of whom 9.1% were severely obese), and 17.5% (of whom 8.9% were severely obese) of individuals were obese, respectively. At 4 years after transplant, 12.7% of patients were obese (of whom 5.8% were severely obese) and at 5 years after transplant, 10.9% of patients were obese (of whom 4.6% were severely obese). Children were more likely to be obese 1 year after transplant (OR 1.34, 95% CI 1.09–1.65; P = 0.005) and less likely to be obese 5 years after transplant (OR 0.63, 95% CI 1.48–0.84; P = 0.002) than at the time of transplant, with no differences noted 2 to 4 years after transplant. Rejection in the first 3 months after transplant, biliary and vascular complications, diabetes, and full-time school attendance were similar among overweight/obese children and those of a healthy weight.
Figure 2 shows the prevalence of overweight and obesity in the posttransplant population over time compared with the general pediatric population of the United States from 2003 to 2004 (2). At 1 and 3 years after LT, patients were statistically more likely to be overweight or obese, as compared with the general pediatric population. At 5 years after transplant, however, the degree of overweight and obesity was similar to the general pediatric population.
The prevalence of obesity in the LT population was compared with NHANES data for 3 time periods (1999–2000, 2001–2002, and 2003–2004). According to NHANES data, 13.9% of children ages 2 to 19 years were obese in 1999 to 2000, 15.4% in 2001 to 2002, and 17.1% in 2003 to 2004 (2,9). As seen in Table 2, a similar proportion of the overall LT patients and the general pediatric population were obese across all of the periods studied. However, there was a significantly increased prevalence of obesity in 2- to 5-year-olds in transplant follow-up compared with the general population across all of the periods studied. The percentage of obese blacks, Hispanics, and whites were similar in the posttransplant and NHANES population.
A univariate analysis was conducted to identify risk factors for obesity after liver transplantation, at a significance level of ≤0.1. Identified factors included ALT, age younger than 2 years or 2 to 5 years at transplant, Hispanic ethnicity, Medicaid insurance, overweight, obese or severely obese weight status at the time of transplant, height deficit (height z score <−2) at transplant, steroid use at follow-up, and elapsed time interval between transplant and follow-up. Children who were 5 years old or younger when receiving transplant did not, however, have shorter lengths of follow-up, P = NS. Nonsignificant factors included sex, primary diagnosis, year of transplant, primary immunosuppression (cyclosporine versus tacrolimus) at transplant, early rejection after transplant, donor organ type, and AST, total bilirubin, and INR at time of BMI measurement. The present study had insufficient power to determine the effect steroid use at transplant had on obesity because 93% of all of the individuals received steroids at transplant.
Significant risk factors identified on univariate analysis were entered into a stepwise logistic regression model to further understand predictors of obesity after LT. Three time invariant factors were identified as potential risk factors for obesity on univariate analysis: age at transplant, ethnicity, and insurance status. No significant interaction was noted between these factors and time on GEE analyses with repeated measures, and therefore, an interaction term was not included in the multivariate analysis. The results of this multivariate analysis (Table 3) demonstrate that patients of Hispanic ethnicity, those using steroids at follow-up, those 3 or fewer years after transplant, and those who were overweight or obese at the time of transplant are at increased risk of posttransplant obesity.
A subgroup analysis was conducted on individuals who were obese 3 and/or 5 years after transplant, because these individuals were unlikely to be affected by postoperative factors and were judged by the authors of the present study to be at increased risk for persistent obesity. Follow-up BMI data were available at 3 and/or 5 years after transplant for 564 patients, 117 (20.7%) of whom were obese 3 and/or 5 years after transplant. These patients were predominantly girls (52%), white (56%), and received transplants for biliary atresia (61%), primarily before 2001 (79%). Triglycerides were available for 475 patients (84%) at 3 years after transplant. The mean levels of triglycerides were as follows: 77 ± 12 mg/dL in 12 underweight patients, 82 ± 3 mg/dL in 287 healthy weight patients, 81 ± 4 mg/dL in 100 overweight patients, 105 ± 20 mg/dL in 40 obese patients, and 88 ± 7 mg/dL in 36 morbidly obese patients (P = NS). Cholesterol values were available for 496 patients (88%) at 3 years after transplant. The mean cholesterol values were as follows: 108 ± 12 mg/dL in 12 underweight patients, 132 ± 2 mg/dL in 301 healthy weight patients, 134 ± 4 mg/dL in 105 overweight patients, 127 ± 6 mg/dL in 40 obese patients, and 134 ± 4 mg/dL in 38 morbidly obese patients (P = NS).
A univariate analysis was conducted to identify risk factors for obesity at 3 and/or 5 years after LT, at a significance of P ≤ 0.1. Identified factors included Hispanic ethnicity, metabolic disease requiring transplant, and obesity or severe obesity at transplant. Nonsignificant risk factors for chronic obesity included age at transplant, sex, choice of primary immunosuppression, insurance coverage, year of transplant, steroid use at the time of transplant or at follow-up, early rejection after transplant, height deficit (height z score <−2) at transplant, and AST, ALT, INR, and total bilirubin at time of BMI measurement.
The risk factors identified on univariate analysis were entered into a stepwise logistic regression model to further understand predictors of obesity 3 and/or 5 years after LT. The results of this multivariate analysis (Table 3) show that Hispanic patients and those who were obese or morbidly obese at transplant are at increased risk of obesity 3 and/or 5 years after LT.
We have shown in a large, nationally representative cohort of pediatric LT recipients that the prevalence of obesity after transplant is extremely high. In the first year after LT, 19.2% of pediatric LT recipients were obese. Three years after transplant, 17.5% of individuals were obese and 5 years after transplant, 10.9% of individuals remained obese. Surprisingly, we report a higher prevalence of obesity among post-LT patients who are 2 to 5 years old as compared with a cohort of similar age among the general pediatric population of the United States. The prevalence of obesity after LT in children 6 to 19 years old, however, is similar to a comparable age group in the general pediatric population. More important, the prevalence of obesity in children 5 years after transplant, a sufficient duration after transplant to minimize the effect of factors such as posttransplant medications and activity, is similar to that in the general pediatric population. Therefore, these findings reveal a disturbing trend in pediatric obesity in a special pediatric population typically perceived as underweight.
Obesity in adult LT recipients has been reported in 17% to 43% of patients, depending on the definition of obesity used and the duration from transplant captured (3–7). The present study illustrates that obesity after pediatric LT, although less prevalent than in adults, is remarkably common. High recipient BMI at the time of transplant influences posttransplant obesity in adults (3,4,10). Similarly, in this large pediatric cohort, we have demonstrated that the most prominent predictor of obesity after LT was body habitus at the time of transplant. LT recipients who were obese were 10 times as likely and those who were severely obese were 14 times more likely, to be obese after transplant compared with children who were of healthy weight at the time of transplant. SPLIT does not, however, collect data on pretransplant medical therapies, such as steroid use, which have the potential to affect pretransplant BMI. It is also possible that some children were misclassified as obese at the time of transplant due to the presence of ascites or anasarca. Unfortunately, SPLIT does not collect data on ascites or anasarca at transplant, nor does it collect anthropometric data such as triceps skinfold or mid-arm circumference, which may allow insight into this potential problem. Clinically, the majority of children at risk of such misclassification are infants or toddlers, and age at transplant (including younger than 2 years) was not found to be predictive of obesity. Therefore, body habitus at the time of transplant remains a crucial risk factor for post-LT obesity. In addition, these findings highlight the opportunity for primary care and subspecialty providers to collaboratively improve the posttransplant health of pediatric patients by focusing on pretransplant weight optimization when possible.
Conflicting data exist regarding the affect of calcineurin inhibitor choice on obesity in adult transplant recipients (3,4,10–12). We did not find that choice of calcineurin inhibitor affected the presence of posttransplant obesity in pediatric LT recipients. Both steroid use at the time of transplant and cumulative prednisone dose have been shown to affect obesity in adult LT recipients (3,4,10). The nearly universal use of steroids at transplant in this population precluded our ability to assess their affect on obesity. The SPLIT database also does not capture data on cumulative steroid dose immediately after transplant. Children who were 3 years or younger at transplant, however, were at increased risk for obesity, suggesting a potential role for steroid-related weight gain. In addition, the use of steroids in the posttransplant period was noted to be a risk factor for obesity. Therefore, particularly in overweight or obese patients, transplant teams should consider prioritizing a rapid wean of steroids into therapeutic paradigms. Further investigation is necessary to determine optimal immunosuppressive regimens for overweight and obese pediatric patients in the posttransplant period, including steroid-free protocols.
In the present study, Hispanic ethnicity was also a significant risk factor for posttransplant obesity. This parallels racial trends in obesity seen in 2003 to 2004 NHANES data, in which 19.2% of Hispanics were obese compared with 16.3% of whites (2). In addition to ethnic disparities, geographic disparities in pediatric obesity exist, with the highest rates seen in the southeastern United States (13). Although SPLIT does not collect data on the geographic area from which patients originate, this too may be an important risk factor. Recognition of these demographic risk factors may allow primary care providers and transplant teams to collectively focus intensive nutritional and exercise counseling into the routine care of high-risk groups. In addition, high-risk demographic groups may particularly benefit from steroid minimization.
Adults with obesity after transplant are at increased risk for the long-term complications of cardiovascular disease, type 2 diabetes, and the metabolic syndrome. In addition, they may have lower overall survival after transplant (14). Children who are obese before transplant are reported to have decreased long-term survival compared with normal weight or underweight children (15). Our study shows that children who are overweight or obese at transplant are likely to remain so after transplant. As such, it is possible that decreased long-term survival in obese patients may be related to obesity-related comorbidities. Transplant teams and primary care providers together must remain cognizant of the long-term cardiovascular and metabolic risks in these overweight and obese patients (1,16). We suggest incorporating screening of overweight and obese children for type 2 diabetes mellitus, hypertension, and hyperlipidemia into routine transplant follow-up as recommended in the 2007 summary report of the Expert Committee on Prevention, Assessment, and Treatment of Child and Adolescent Overweight and Obesity (1,17,18). Such screening may be best facilitated by a partnership between primary care providers and transplant teams. Unfortunately, the SPLIT study lacks robust data on these obesity-related comorbidities. Systematic study of these parameters in the future may enhance our ability to understand the affect these potential risk factors have on posttransplant outcomes.
The prevalence of obesity after pediatric liver transplant is high and similar to that in the general pediatric population. Obesity at the time of liver transplant, a potentially modifiable factor, confers a high risk of posttransplant obesity. These findings suggest a need to broaden standard care to include obesity assessment and intervention in routine pre- and posttransplant care. Pediatric transplant teams are accustomed to the malnourished state that is commonplace in many children listed for liver transplant. As such, well-accepted treatment algorithms exist to improve nutritional health and BMI in undernourished patients. Our study demonstrates the critical nature of focused attention on the nutritional status of children who are overweight or obese before transplant as well. A carefully devised weight loss program, with interventions by both nutritionists and exercise therapists may stimulate pretransplant weight loss in a rigorous and safe manner. In addition, families should be educated and empowered to make good nutritional choices with their children and emphasize active lifestyles, even as they move forward with transplant.
1. Barlow SE. Expert committee recommendations regarding the prevention, assessment, and treatment of child and adolescent overweight and obesity: summary report. Pediatrics 2007; 120 (Suppl 4):S164–S192.
2. Ogden CL, Carroll MD, Curtin LR, et al. Prevalence of overweight and obesity in the United States, 1999–2004. JAMA 2006; 295:1549–1555.
3. Canzanello VJ, Schwartz L, Taler SJ, et al. Evolution of cardiovascular risk after liver transplantation: A comparison of cyclosporine A and tacrolimus (FK506). Liver Transpl Surg 1997; 3:1–9.
4. Mor E, Facklam D, Hasse J, et al. Weight gain and lipid profile changes in liver transplant recipients: Long-term results of the American FK506 Multicenter Study. Transplant Proc 1995; 27:1126.
5. Munoz SJ, Deems RO, Moritz MJ, et al. Hyperlipidemia and obesity after orthotopic liver transplantation. Transplant Proc 1991; 23 (1 Pt 2):1480–1483.
6. Reuben A. Long-term management of the liver transplant patient: Diabetes, hyperlipidemia, and obesity. Liver Transpl 2001; 7 (11 Suppl 1):S13–S21.
7. Stegall MD, Everson G, Schroter G, et al. Metabolic complications after liver transplantation. Diabetes, hypercholesterolemia, hypertension, and obesity. Transplantation 1995; 60:1057–1060.
8. Skelton JA, Cook SR, Auinger P, et al. Prevalence and trends of severe obesity among US children and adolescents. Acad Pediatr 2009; 9:322–329.
9. Hedley AA, Ogden CL, Johnson CL, et al. Prevalence of overweight and obesity among US children, adolescents, and adults, 1999–2002. JAMA 2004; 291:2847–2850.
10. Everhart JE, Lombardero M, Lake JR, et al. Weight change and obesity after liver transplantation: Incidence and risk factors. Liver Transpl Surg 1998; 4:285–296.
11. Neal DA, Gimson AE, Gibbs P, et al. Beneficial effects of converting liver transplant recipients from cyclosporine to tacrolimus on blood pressure, serum lipids, and weight. Liver Transpl 2001; 7:533–539.
12. Richards J, Gunson B, Johnson J, et al. Weight gain and obesity after liver transplantation. Transpl Int 2005; 18:461–466.
13. Singh GK, Kogan MD, Van Dyck PC, et al. Racial/ethnic, socioeconomic, and behavioral determinants of childhood and adolescent obesity in the United States: Analyzing independent and joint associations. Ann Epidemiol 2008; 18:682–695.
14. Dick AA, Spitzer AL, Seifert CF, et al. Liver transplantation at the extremes of the body mass index. Liver Transpl 2009; 15:968–977.
15. Dick AA, Perkins JD, Spitzer AL, et al. Impact of obesity on children undergoing liver transplantation. Liver Transpl 2010;16:1296–302.
16. Munoz SJ, Elgenaidi H. Cardiovascular risk factors after liver transplantation. Liver Transpl 2005;(11 Suppl 2):S52–6.
17. Pinhas-Hamiel O, Dolan LM, Daniels SR, et al. Increased incidence of non-insulin-dependent diabetes mellitus among adolescents. J Pediatr 1996; 128 (5 Pt 1):608–615.
18. Type 2 diabetes in children and adolescents. American Diabetes Association. Pediatrics 2000;105(3 Pt 1):671–80.