Long-term Outcome of Children Receiving Home Parenteral Nutrition: A 20-year Single-center Experience in 302 Patients

Abstract

Background: 

More information is needed regarding the prognosis of children receiving home parenteral nutrition (HPN). This article describes 20-year outcome data in children receiving HPN and provides separate profiles for the major pediatric diagnostic subgroups.

Patients and Methods: 

This retrospective study included children who started receiving HPN between January 1, 1980, and December 31, 1999, in a single pediatric HPN center.

Results: 

A total of 302 children were recruited, 230 (76%) with primary digestive disorders and 72 (24%) with nonprimary digestive disorders. Median age at HPN onset was 1.5 years. Median duration of HPN was 1.3 years. By January 1, 2000, 54% had weaned from HPN, 26% were still receiving HPN, 16% had died, and 4% had undergone intestinal transplantation. The survival probabilities at 2, 5, 10, and 15 years were 97%, 89%, 81%, and 72%, respectively. The likelihood and cause of death depended on the underlying diagnosis. Nine percent of children with primary digestive disorders died, 24% from their primary disease and 48% from liver disease or sepsis. Children with intractable diarrhea of infancy had the highest mortality rate (25%) and the highest incidence of liver disease (48%; P = 0.0002). Thirty-eight percent of children with primary nondigestive diseases died, 94% from their primary disease and 6% from liver disease or sepsis.

Conclusions: 

Outcome and survival of children receiving HPN are mainly determined by their underlying diagnosis. Nearly all children with primary digestive disease survive if referred early to an expert center.

INTRODUCTION

During the past 3 decades, use of parenteral nutrition (PN) has been increasing in pediatric patients with primary digestive and nondigestive disorders to support survival during a period of severe intestinal failure and to ensure normal or catch-up growth ⁽¹⁾. Home PN (HPN) is an alternative to prolonged hospitalization and is recognized as the best option for improving the quality of life of these children and their families. Advances in vascular access and nutrient formulas have decreased the morbidity and mortality rates associated with long-term PN and extended the indications for which HPN is considered beneficial ^(2–6).

During the past 15 years, the use of intestinal transplantation has also developed rapidly and has become a viable option for children who are dependent on lifelong PN, who currently represent approximately two thirds of all candidates for intestinal transplantation worldwide. Although combined intestinal and liver transplantation is required for patients with end-stage liver failure, children with mild to moderate liver disease may recover normal liver function after an isolated intestinal graft ^(7–9). The timing of transplantation and criteria for isolated intestinal or combined intestinal and liver transplantation remain controversial ^(7–10). To help specialized pediatricians with proper patient referral, the morbidity and mortality risks of long-term PN according to the underlying disease need to be known.

In the past 15 years, there have been several reports of children receiving HPN ^{(2–4,11–16)}, some of which focused on outcomes in subgroups of children with different diagnoses ^(12–16). This report, which describes a large cohort of children receiving HPN managed in a single specialized center, was conducted to describe outcomes for different underlying diagnoses, with the aim to analyze for each subgroup the potential for bowel adaptation and weaning off HPN, the likelihood of serious complications, and survival.

PATIENTS AND METHODS

This retrospective study included children (N = 302; 172 boys and 130 girls) younger than 16 years of age referred to our institution and discharged consecutively on HPN between January 1, 1980, and December 31, 1999. Children were judged eligible for HPN when they were unable to maintain growth and nutritional status without PN, and when pediatricians at the expert HPN center estimated that they would depend on PN for ≥3 months when they were in stable condition, on cyclic PN enabling discharge regardless of the duration of PN at the hospital before referral to the HPN center.

The parents' home infusion teaching program took place during the child's hospitalization and lasted approximately 2 weeks as previously described ^(3–5). At discharge, a 24-hour telephone hotline number was provided and regular outpatient follow-up visits were scheduled. The composition and volume of the nutrient solution were adapted at intervals to meet the child's requirements. Amino acid sources, lipid emulsions, trace elements, and vitamin solutions varied during the 20-year study period. As the patients' intestinal function improved, the numbers of nightly infusions were slowly decreased when possible.

Evaluation Methodology

Data collected from the medical charts included demographic characteristics (age and sex), the primary diagnosis, and other details such as the cause of short bowel syndrome (SBS), the length of remaining small bowel, the starting date of HPN, the date and the reason for HPN discontinuation, and, when applicable, the cause of death. The number, duration, and reason for all the hospitalizations were recorded. Abnormal liver function test results for more than 2 months were considered as evidence of liver disease. Hepatocellular injury was defined as a plasma aminotransferase concentration 1.5 times higher than the upper limit of the normal range. Cholestasis was defined as a bilirubin or alkaline phosphatase and γ-glutamyl transpeptidase plasma concentration 1.5 times higher than the upper limit of the normal range.

Data were updated on January 1, 2000. Probability of survival was calculated from HPN onset to January 1, 2000, using the Kaplan-Meier method. Survival distributions and occurrence of PN weaning were compared with the log-rank test according to age at PN onset, date of inclusion in the HPN program, primary disease, length of small bowel, presence of ileocecal valve, and presence of liver disease. To identify independent prognostic factors, these variables were entered in a Cox proportional-hazards regression model ⁽¹⁷⁾. The statistical analysis was performed using the last release of the standard BMDP statistical software (University of California, Berkeley), SAS software (SAS Institute, Cary, NC), and R statistical software (http://www.r-project.org). Quantitative variables were expressed as mean ± SD. All tests were 2-tailed. A P value <0.05 was considered significant.

RESULTS

HPN Incidence

The annual incidence of HPN (number of new patients starting HPN per year) was 17 ± 8 during the 20-year study period. The number of children referred to our institution for HPN increased regularly from 1980 to 1985 to 20 patients per year. Later on, the incidence of HPN programs was stable at a level of 20 to a maximum of 25 new patients per year from 1985 to 1999. Home PN was started between 1980 and 1984 for 28 children, between 1985 and 1989 for 106 children, between 1990 and 1994 for 105 children, and between 1995 and 1999 for 100 children. Some children were weaned off HPN and resumed HPN a few months or years later.

Indications for HPN

Table 1 lists the underlying diagnoses and the age at HPN onset. Two hundred thirty children (76%) had a primary digestive disease (PDD). The largest PDD subgroup included 141 children with SBS (47% of the whole cohort and 61% of the PDD cohort). The SBS was mostly from a neonatal origin. The primary causes were atresia (24% of SBS), late-onset volvulus (20%), gastroschisis (16%), necrotizing enterocolitis (16%), and long-segment Hirschsprung disease (13%). In the SBS cohort, the mean remaining small bowel length was 53 ± 3 cm (range 3–86 cm). The ileocecal valve was present in 51% of SBS cases. Other digestive disorders were inflammatory bowel disease (IBD; n = 33; 11% of the whole cohort and 14% of the PDD cohort); chronic intestinal pseudo-obstruction syndrome (CIPOS; n = 31; 10% of the whole cohort and 13.5% of the PDD cohort), and intractable diarrhea of infancy (IDI; n = 25; 8% of the whole cohort and 11% of the PDD cohort). Major causes of intractable diarrhea were microvillus atrophy and epithelial dysplasia. Less common causes included diarrhea with phenotypic abnormalities and autoimmune enteropathy ^(18,19).

TABLE 1: Indications for HPN in the 302 children studied and their age at HPN onset

Seventy-two children (24% of the whole cohort) had primary nondigestive diseases (PNDDs), which included immune deficiency syndrome (IDS) in 41 children. The other 31 cases of PNDD were cancer in 7 children, cystic fibrosis in 5 children, metabolic disorders in 5 children, radiation enteritis in 5 children, congenital liver diseases that required nutritional supportive therapy before liver transplantation in 4 children, lymphatic disease in 4 children, and extensive digestive polyposis in 1 child. The number of children receiving HPN for PNDD increased from 21% to 28% of patients receiving HPN from the first to the second decade of this study.

Age at Onset of PN and HPN

The mean age at PN onset was 3.6 ± 0.6 years (median 0.5) and the mean age at HPN onset was 4.4 ± 0.6 years (median 1.5). Age distribution at HPN onset is shown in Fig. 1.

FIG. 1: Age distribution at HPN onset.

HPN Duration and Patient Outcomes

Among the 302 children, the overall HPN duration was 2.6 ± 0.2 years (range 4 days to 14.6 years; median 1.3 years). Table 2 shows the mean and median durations of HPN for each diagnostic group according to the outcome, weaning off HPN, death, transplantation, or ongoing HPN.

TABLE 2: Duration of HPN for each diagnostic group

In January 2000, at the end of the study period, 54% of the total HPN cohort (n = 163) had been weaned off HPN, 20% (n = 60) were continuing HPN in the same center, 6% (n = 18) were continuing HPN in another center, 16% (n = 48) had died, and 4.3% (n = 13) had undergone an isolated intestinal (n = 6) or combined liver/intestinal transplantation (n = 7). Table 3 summarizes outcomes for each of the main diagnostic groups and compares patients with PDD versus those with PNDD. The rate of weaning off HPN among survivors varies with the underlying diagnoses: 65% for SBS, 38% for CIPOS, 87% for IBD, and 15% for IDI. In patients with SBS, weaning was more likely to occur for children with longer small bowel length (P < 0.05) and an intact ileocecal valve (P < 0.05). Eighty percent of PNDD survivors were weaned off HPN.

TABLE 3: Outcome of patients according to the primary diagnosis

Complications

The mean number of hospitalizations was similar for patients with PDD (2.3 per HPN patient-year) and those with PNDD (2.5 per HPN patient-year). Nineteen percent of hospitalizations were a result of catheter-related bloodstream infections (CRBSIs) in both groups, whereas 16% were a result of fever without evidence of CRBSI in both groups. The incidence of CRBSI was 0.44 per patient-year or 1.20/1000 HPN days. Other causes of hospitalization were the necessary tests related to the primary disease (23% of hospitalizations in PDD, 21% in PNDD), the noninfectious complications of the primary disease or HPN (16% of hospitalizations in both groups), the surgical procedures related to the primary disease (8% in PDD, 4% in PNDD), and central venous catheter replacements (8% in both groups). The duration of hospitalization was 88 ± 16 days per patient HPN course and 39 ± 7 days per HPN patient-year.

As shown in Table 4, 23% of the children receiving HPN developed liver disease (20% in PDD, 35% in PNDD). In the PDD group those with IDI had the highest rate of liver disease compared with other children with PDD (48%; P = 0.0002).

TABLE 4: Prevalence of liver disease in the whole HPN cohort in children with primary digestive disease and primary non-digestive disease

Small Bowel and/or Liver Transplantation

Thirteen children (4% of the whole cohort and 6% of children with PDD) underwent intestinal transplantation. Seven children (2% of the whole cohort) who presented with irreversible intestinal failure and liver cirrhosis underwent combined intestinal and liver transplantation (3 of 141 patients with SBS and 4 of 25 patients with IDI). One child with SBS underwent an isolated liver transplantation before completing intestinal adaptation. Six children with irreversible intestinal failure and normal liver function underwent an isolated intestinal transplantation as a result of life-threatening HPN-related complications such as recurrent CRBSI, onset of liver disease, or lack of venous access (5 patients with SBS and 1 with CIPOS).

Of the 13 children who received transplants, 6 were weaned off PN and 2 died at the early stage of the postoperative period (these deaths are not taken into account in the HPN mortality rate). The other 5 restarted HPN after transplantation failure; 2 died from late complications of transplantation.

Mortality

Forty-eight patients died (16% of the whole cohort). Probabilities of survival given by the Kaplan-Meier method at 2, 5, and 10 years were 97%, 89%, and 81%, respectively as shown in Fig. 2A.

FIG. 2: Probabilities of survival in children on HPN. A, Probabilities of survival in the whole cohort and in patients with primary digestive disease. B, Probabilities of survival in the main HPN diagnostic subgroups.

The date of HPN onset (in the first or second decade of the study period) did not influence the survival rate, even in the PDD group. The younger the child was at the start of HPN, the higher was the risk of death (likelihood ratio test, P = 0.0008; Wald test, P = 0.00364; log-rank test, P = 0.00246). Probability of survival was significantly influenced by the underlying disease (Fig. 2B). Patients with IDS and other forms of PNDD had a much higher risk of death compared with most of those with PDD (P < 0.05). Parenteral nutrition–associated liver disease appeared to be a significant risk factor for mortality (P < 0.05), whereas mortality from liver dysfunction was not common (1% of children in the PNDD group, 2% of children in the PDD group, and 2% of children with SBS). Table 5 shows the number of deaths in the different diagnostic groups and their direct cause. Among the 141 children with SBS, transplantation accounted for 3 of the 10 deaths (30%). The direct causes of mortality were sepsis, liver failure, and lymphoma (ie, other category). Deaths related to transplantation are not separately identified in Table 5.

TABLE 5: Direct causes of death in the whole HPN cohort in children with primary digestive disease and in children with primary nondigestive disease

Status at Study Update

In January 2000, 68 patients (59 children and 9 young adults) were still receiving HPN, which represented 22% of the cohort studied. Their primary diagnoses were SBS (50%), CIPOS (24%), IDI (18%), IBD (5%), and others (3%). Forty of these 68 patients (60%) had been receiving HPN for more than 4 years and 27 (40%) had been receiving HPN for more than 8 years. Fifteen children and 4 adults (for a total of 19 patients, which represents 6% of the whole cohort) seemed likely to require lifelong HPN because of severe and irreversible intestinal failure (SBS, n = 10; IBD, n = 6; CIPOS, n = 3).

DISCUSSION

Our institution has the largest pediatric HPN program in France and has taken care of approximately half of all children on HPN in France over the study period. It is also the center where pediatric small bowel transplantation started in France ⁽²⁰⁾. The present report focuses on long-term outcome in the first 302 children enrolled in our HPN program.

Our large cohort enabled the creation of subgroups according to the underlying diagnoses. Separation of outcome profiles according to the primary disease is primordial because HPN is not so much a treatment for a disease as it is a treatment for a complication, notably intestinal failure, common to a wide spectrum of diseases. Large adult studies have shown that HPN survival, duration of use, and rehabilitation on therapy are strongly determined by the underlying diagnosis ^(21–24). Like earlier pediatric studies from the United States ⁽²⁾ and European countries ^(3,16), this report shows that SBS resulting from neonatal causes accounts for almost half of all children receiving HPN. As experience has grown, the uses of HPN have been extended to younger children and to those with a PNDD, such as congenital and acquired immune deficiencies. This gradual shift toward the treatment of children in more fragile condition may explain in part why this pediatric center has not witnessed overall improvement in survival over time, whereas a learning curve leading to better survival and fewer complications over time has been reported in a specialized HPN adult center ⁽²¹⁾. Another explanation may be the mortality related to the complications of transplantation procedures in the second decade of the study period ⁽²⁰⁾. Our study shows that children have a better probability of survival than adults, 89% after 5 years versus approximately 60% in adults ^(21–24). Moreover, it confirms that the influence of primary disease is equally strong in pediatric HPN outcomes. Adults with IBD have the best chances of survival, at more than 80% after 5 years, whereas patients with chronic obstruction have a higher mortality rate ^(21–24). Similarly, in our series, the best survival rate was associated with children with IBD and SBS. The primary diagnosis influences the likelihood and cause of death. Only 9% of children with a primary digestive disorder died. As described in a recent Italian study ⁽¹⁶⁾, the poorest outcome was seen in children with IDI. Children with CIPOS also have a high mortality rate. However, in the present series, the mortality rate in children with CIPOS was 13%, compared with the 32% mortality rate described in another pediatric study ⁽²⁵⁾. In children with a PNDD, deaths were overwhelmingly related to the primary disorder (94%). A significant death rate (98%) caused by underlying diseases such as cancer and HIV has been also described in adults, whereas patients with more chronic disorders who survive for many years on HPN have a 15% to 23% likelihood of eventually dying from an HPN complication ^(21,22). Likewise, duration of HPN varied with the primary diagnosis and tended to be shorter in children with a PNDD. Weaning off HPN and resuming full enteral autonomy is most likely in children who survive PNDD and those with IBD. The present study also confirms that children have a better chance of weaning off HPN than adults with comparable diagnoses ⁽²²⁾. As many as 54% of our patients have been weaned off HPN. This recovery rate is 25% higher than that reported by the same team 16 years ago ⁽³⁾, and also 30% higher than the recovery rate shown in a contemporary European HPN adult survey ⁽²¹⁾. Children with SBS (141 cases) had a 93% survival rate, with a 65% rate of PN weaning among survivors. In a recent multicenter Italian study, survival rate was 92% and only 40% adapted, but this low rate may be attributed to the shorter study period of 5 years ⁽¹⁶⁾. In a study of 78 children with SBS, survival rate was 73%, but 77% of survivors adapted ⁽¹²⁾. Finally, 60% of children with SBS adapted in that study and in the current series. Both studies show that the length of residual small bowel and the presence of ileocecal valve are important factors for adaptation and HPN weaning.

Repeat hospitalization for complications has a serious impact on quality of life for the children and their families and on health care costs. The mean annual number of hospitalizations per patient, <3/year, was similar to that reported in the United States for children receiving HPN with congenital bowel disorders ⁽²²⁾. Approximately half of the readmissions were a result of suspected or confirmed CRBSI. The incidence of CRBSI depicted in this cohort, 1.2/1000 HPN days, is low, whereas children usually have a higher incidence of CRBSI ^(12,26–28) than adults ⁽²⁹⁾. Parenteral nutrition–associated liver failure is another potentially life-threatening complication frequently described in children ^(30–33). In this study, using the hepatocellular injury and cholestatic biological criteria described, 23% of children developed PN liver disease, similar to the findings of recent adult surveys ^(23,34). Prevalence depends in part on the criteria used because there is reluctance to perform serial liver biopsies for a more definitive histological diagnosis, especially in children. In the present series, liver dysfunction was especially prevalent in children with IDI, in whom risk factors are prolonged bowel rest and long total PN dependence. Excessive parenteral doses of lipids are also associated with liver dysfunction in adults and children ^(34,35). Liver dysfunction may also reflect bacterial overgrowth. However, in this large cohort, children with motility disorders had the same prevalence of liver disease as children with SBS. Although liver dysfunction was common in our series, the mortality rate associated with liver failure and the rate of liver transplantation were low.

Survival of children receiving prolonged HPN depends chiefly on the underlying diagnosis, but complications such as recurrent CRBSI, liver disease, and lack of venous access should not be underestimated. The low rate of HPN complications described in this report suggests that expert HPN centers can keep life-threatening complications to a minimum, as stated in previous reports ⁽³⁶⁾. This has an impact on the duration and the quality of life while reducing the need for transplantation referral. The rate of liver transplantations prompted by life-threatening liver failure should decrease in the future as a result of the better knowledge concerning risk factors and prevention of PN-associated liver disease. The results of small bowel transplantation are improving rapidly ^(7–10,20). However, survival and quality of life after transplantation should be compared with survival and quality of life of well-managed patients receiving long-term HPN ^(11,37).

This study shows that outcome and survival of children receiving HPN are mainly determined by their underlying diagnoses. The role of expert HPN centers is to improve the knowledge and the quality of care of children with intestinal failure and thereby decrease the number of intestinal and liver transplantations resulting from the complications of long-term PN.

Acknowledgment

The authors thank Dr Lyn Howard for her considerable help in writing the manuscript.