Summary of Relevant Prior Systematic Reviews

The good quality systematic review,³⁴ conducted by Snider and colleagues, focused on whether regular application of feeding interventions improves feeding efficiency and safety and results in enhanced weight gain in children with CP. We assessed the review as good quality as it reported search procedures, assessed and reported quality of studies, and appropriately synthesized results. The review investigators searched 12 electronic databases including MEDLINE, PsycINFO, CINAHL, ERIC, and HealthStar. After analysis, 21 studies (two case reports, four case series, one single-subject ABA design, six experimental, six quasi-experimental, one prospective matched pairs, and one RCT) were included in the review. Most (13 of 21) studies would have met inclusion criteria for our review as well; those that did not were either case reports, published prior to 1980, or did not address interventions of interest. Included studies were assessed as oral sensorimotor facilitation (“techniques specific to the enhancement of oral-motor control aim[ing] to decrease or increase tone and inhibit abnormal reflexes that interfere with safe feeding”³⁴), food consistency, positioning, oral appliances, or adaptive equipment. Some studies assessed more than one intervention (e.g., electric feeder + positioning,⁶⁷ oral device + motor-sensory therapy⁶⁸ or positioning,⁶⁹ feeding skills + positioning,⁷⁰ oral-motor exercises + attention and parent training,⁷¹ and oral-motor skills + thermal stimulation + dietary modification⁷²), with results presented in the aggregate. Methodological quality was assessed using the Physiotherapy Evidence Database rating scale (PEDro). A PEDro score of 9–10 points is considered to be of excellent quality, a score of 6–8 points is good, 4–5 points is fair and a score below 4 points is of poor quality. A total of 414 participants aged 1.6 to 17 years were included in the reviewed studies. Severity, type of CP, nutrition status and feeding skills varied among participants. We summarize key findings of the studies included in the review below:

Feeding Devices

Two studies analyzed effectiveness of an electric feeder in 20 participants (17 with CP) using a pre-post design.^67,86 Neither demonstrated positive effects in outcomes of interest for our review and improvements in components of oral-motor behaviors were not observed at followup. Electric feeding devices were not included in our review. Table 7 outlines key findings as summarized in the review's evidence table.³⁴

Table 7

Key findings summarized in Snider et al. systematic review.

Summary of Primary Research

In updating the Snider review described above, we identified one case series addressing caregiver training.⁶⁶ The study was conducted in child-caregiver pairs who received in-home training and support in Dhaka, Bangladesh (Table 8). All of the children had moderate to severe CP, assessed as levels III to IV on the Gross Motor Function Classification System (GMFCS). Thirty-seven pairs were enrolled in the study, but only 22 completed it. Pairs received advice and completed a baseline assessment during an initial home session and then participated in four to six sessions focused on improving dietary intake and ease and efficiency of feeding, including introduction of a high calorie diet, adaptation of food consistency, use of appropriate utensils and provision of appropriate postural and physical support. The intervention was delivered over 2.5 months and evaluated 4 to 6 months after completion.

Table 8

Key outcomes of studies assessing behavioral interventions.

Children had significantly fewer episodes of chest-related illness after 3 months (15 vs. 6, p=0.005), had improved mean weight-for-age z-scores (−4.83 vs. −4.07, p=0.02), mean mid-upper-arm circumference (14.75 cm vs. 15.46 cm, p=0.02 and 0.001), and fluid intake during meals (173.7 ml vs. 300.2 ml, p<0.01) 4 to 6 months after the end of intervention. Observed child feeding skills and affect also improved, with a significant decrease in child fussiness and food refusal and improvement in general mood and child feeding skills (p<0.05). At baseline, nine children always refused food, no children were involved in self-feeding, and six were observed munching or chewing. At 4 to 6 months after the caregiver training, only one child was observed refusing food and six and eight children were involved in self-feeding and demonstrated munching or chewing, respectively. The authors also reported a significant improvement in caregiver self-reported stress and reduction in mealtime length (p=0.005).

Similarly, a study with a parent training component⁷¹ reviewed by Snider et al.³⁴ reported an increase in parents' feeding effectiveness, but as noted, these studies are of high risk of bias (typically pre-post design) and report high attrition.

Effectiveness Data

Effectiveness data for gastrostomy come from one prospective cohort study,⁹⁵ four prospective,^87-89,92-94 and two retrospective case series^90,91 (Table 9). The prospective cohort study was intended to assess potential harms of treatment and is therefore discussed in detail in that section, below. The one multicenter, prospective case series provides the most detailed outcomes data available.^92-94 This study includes data collected systematically prior to the intervention, establishes primary outcomes a priori, controls for age and sex in reporting results, and uses standards for typically developing children as a comparison in z-score analyses. The study (reported in three publications^92-94) included 57 children (age range: 5 months–17 years, median: 4 years) with CP and severe motor difficulties (75% with spastic quadriplegia, 93% unable to feed independently, 77% with global developmental delay) undergoing gastrostomy (percutaneous endoscopic gastrostomy [PEG] in 53 children and open gastrostomy in 4). Eighteen children with significant reflux (determined by a reflux index > 10% as proposed by Sullivan 1999⁹⁷) had fundoplication performed at the same time as gastrostomy. The majority of children (n=53/57) had severe CP, assessed to be equivalent to level V GMFCS, with 95 percent unable to walk independently. All had substantial oral-motor impairment and feeding difficulty. None of the children had experienced adequate (not defined) weight gain for at least 12 months prior to surgery. Some had experienced no weight gain at all. Outcome measures included changes in body weight and skinfold measurement; dietary intake; general health (measured by infections requiring hospitalization or antibiotics); harms, including respiratory morbidity; and quality of life for caregivers (measured using the validated Short-Form 36, version II [SF-36]⁹⁸). Although there was no active comparison group in this study, standards for typically developing children were used as a reference.

Table 9

Summary of key outcomes of studies of surgical interventions reporting effectiveness data.

Participants improved on all measures of growth. Median weight z-scores (relative to standards for typically developing children of the same age and sex) increased from −3 at baseline to −1.6 at 12 months post-gastrostomy, and the proportion of children more than 2 standard deviations below the mean decreased from 2/3 to under half at 12 months. Linear growth (lower leg length, median baseline z-score −1.31 vs. −0.44 at followup) and head circumference (median baseline z-score among those with 12 month data available 0.15 vs. 0.52 at followup) measures improved significantly at followup as did arm circumference (median baseline z-score −0.14 vs. 0.58 at followup) and skinfold z-scores (median baseline z-score −1.71 vs. −0.59 at followup). Thus, improvements were observed and were statistically significant for every measure of growth (p values ranging from <0.0001 to 0.004).

Energy intake as measured by 3-day dietary records maintained by caregivers increased, and the proportion of children experiencing at least one infection requiring hospitalization decreased from 26 to 7 percent (p=0.021). A separate analysis of chest infections showed a significant decrease in the mean number of infections requiring antibiotics from baseline to 12 month followup (mean at baseline=1.8 [SD=2.7], mean at followup=0.9 [SD=1.7], p=0.07) and in hospitalizations for chest infection (mean at baseline=0.5 [SD=1.0], mean at followup=0.09 [SD=0.4], p=0.04).⁹⁴

An Australian prospective case series including 21 children with quadriplegic CP (median age=8 years 5 months, all GMFCS level V) similarly reported significant increases in weight (median baseline weight in kg=15.4, followup=23.3, p<0.05) and height (median baseline height in cm=105.4, followup=118.3, p<0.05) after a median 19.4 months of gastrostomy tube feeding.⁸⁷ Standard deviation scores for weight, but not height, also increased significantly (p<0.05). Body fat percentage increased from a median 10.7 percent at baseline to 16.3 percent at followup (p<0.05). Another prospective case series conducted in Brazil included 16 children (mean age 6.5 years) with CP and dysphagia or recurrent aspiration undergoing PEG tube placement; the study also included 4 children with swallowing dysfunction but not CP.⁸⁸ Children were followed for a mean of 5.9 months (range: 2–10 months). Weight, but not height or skinfold, z-scores significantly increased from baseline across all participants (data presented graphically only).

A retrospective case series conducted in the United States assessed growth in gastrostomy-fed children with CP between the ages of 9 months and 23 years (mean=10.7 years). At the time of data collection, most participants had severe motor dysfunction, and over 70 percent had signs of GER.⁹⁰ No data are provided on the proportion that had GER prior to surgery and it is not possible to assess whether surgery ameliorated or created cases of GER. Fifty-six percent (n=32) of participants had Nissen fundoplication as well as gastrostomy, suggesting that at least half had substantial reflux prior to surgery; two of these later required jejunostomy after failed fundoplication. Baseline and followup data were available for 35 participants; among these, roughly 70 percent remained below the 5^th percentile for height and weight post-gastrostomy. Roughly half (46%) attained appropriate weight for height standards, 21 percent were overweight, and 33 percent were underweight. In subgroup analyses, at least 50 percent of children with gastrostomy before age 2; those with gastrostomy for more than 2 years; and those with fundoplications reached appropriate weight for height. Caution should be used, however, in interpreting these data as reference standards in the literature are often for typically developing children and not an appropriate comparison for individuals with CP. Caregivers generally reported satisfaction with gastrostomy and noted increased ease of feeding and improvements in child affect.

An additional retrospective case series evaluated outcomes for 19 children with CP (mean age: 60.4 months) and reported increases in growth overall (baseline mean weight/length z-scores: −2.71, followup: −1.18).⁹¹ Two studies assessed quality of life. One, the Sullivan case series, assessed quality of life (QOL) of caregivers using the SF-36 and interviews with caregivers of these 57 children.⁹³ Participants completed the questionnaire prior to their child's gastrostomy and at 6 and 12 months post-placement; roughly 35 individuals completed the 12-month followup. Baseline scores on all measures were lower than those for the population norm at baseline. By 12 months, scores had improved on all measures and improved significantly (p values <0.01) on the domains of Role limitations—emotional (baseline: 68.06 ± 31.05, followup: 73.54 ± 27.27), Energy/vitality (baseline: 41.59 ± 23.58, followup: 51.56 ± 22.46), and on the Mental Component summary (baseline: 40.10 ± 15.60, followup: 46.37 ± 11.84). Caregivers reported declines in feeding time from baseline (median: 2.5 hours/day) to 12 month followup (1 hour/day) and greater ease in medication administration. Caregiver concern/worry over their child's nutritional state also lessened post-gastrostomy, with 15 percent indicating concern at 12-month followup compared with 78 percent at baseline. While SF-36 scores improved overall, not all participants reported positive changes; scores on social functioning declined post-gastrostomy for one in four caregivers, and scores for 13 of 39 respondents were lower for the Energy/vitality domain.

The other study assessing QOL was also prospective, but focused on parental perception of the child's quality of life. In this series, 50 children underwent gastrostomy, 42 of whom had static neurologic disorders, presumably CP (overall median age: 591 days).⁸⁹ Investigators used a 10 cm visual analog scale (VAS) to assess parental perception of the child's life and health, an unvalidated questionnaire adapted from the CHQ-PF50 to assess QOL, and collected anthropometric data at baseline and 12 months post-gastrostomy. The study also documented harms. Most parents (98%) expected QOL to improve for their children prior to gastrostomy; however, parent-rated VAS QOL scores did not differ between baseline and followup. Ratings on questions related to global health, impact on activities, and impact on feeding improved at 6 months, but improvements were not sustained at the final, 12-month followup. Ratings on questions related to impact on parental time and medications improved significantly at 12 months (p≤0.05), and ratings on bodily pain, mental health, health perceptions, and parental impact on emotions and respite care did not change. Weight for age z-scores increased significantly at 12 months (p<0.01) though height did not change significantly. Triceps skinfold measurements were increased or maintained.

Harms Data

Three studies were focused specifically on potential harms of gastrostomy; two were related to the potential to overfeed,^41,95 and one on the potential to induce GER post-gastrostomy⁹⁶ (Table 10).

Table 10

Summary of surgical papers focused on harms.

A fair quality prospective cohort study included 40 consecutive children (median age: 8 years 6 months, range: 1–18 years) at a tertiary feeding clinic for children with neurological impairment in the United Kingdom.⁹⁵ Given the reduced energy requirements of quadriplegic children with CP, it is conceivable that they might be overfed, which would be indicated by an imbalance in body composition (i.e., excessive fat mass to fat-free mass) and this was the focus of this study. Investigators also used a reference population of normally developing children for comparison. Twenty-two children were gastrostomy fed (approximately 2/3 also underwent preoperative nasogastric tube feeding prior to gastrostomy), and 18 were fed orally. Investigators took anthropometric (weight, skinfold, height), body composition, and energy expenditure measurements at baseline and after 12 months (median duration of tube feeding: 12 months, range: 4–60 months). As expected, children in the orally fed group had lower (less severe) GMFCS scores, lower weight, higher fat-free mass index, and lower skinfold thickness (p=0.01) z-scores than the gastrostomy group at baseline. Children in the gastrostomy group had lower energy expenditure (p=0.04) and resting metabolic rates and higher fat mass index (p=0.02) than the orally fed group at baseline.

At followup, the gastrostomy group had greater fat mass than the orally fed group, and had lower fat-free mass, but the differences were not statistically significant. Compared with data from the reference population, both CP groups had higher body fat content and lower fat-free content, with the gastrostomy group having greater fat mass relative to the reference population than the orally fed participants. The investigators note that results may be confounded by preoperative nasogastric tube feeding in the gastrostomy group and ongoing nutritional therapy (caloric supplementation and/or enteral feeds) in the orally fed participants, and the degree to which gastrostomy may be associated with poor outcomes from overfeeding remains unclear.⁹⁵

One case series similarly addressing the potential for overfeeding assessed the use of a low-energy feed on weight, body fat, and other anthropometric measures (assessed using a doubly labeled water method); micronutrient status; and general health in 14 gastrostomy-fed children (median age=2 years).⁴¹ Participants had severe disability: all had spastic quadriplegic CP, a GMFCS level of V (n=13) or IV (n=1), and weight between 8 and 30kg at baseline (median=12.1 kg). All participants were solely tube-fed, and seven of 14 had undergone fundoplication in addition to gastrostomy. Six month followup data were available for eight of the 14 participants; three participants died of respiratory infections in the study period.

Participants' weight, mid-upper arm circumference, and lower leg length increased significantly over the 6-month feeding period. Fat mass and body fat percentage increased, but not significantly (fat mass at baseline=3.1 kg, at followup=5.0; body fat percentage at baseline=28.3, at followup=32.5). Fat mass was not significantly different from that of a reference group of children without disability, though these data should be interpreted with caution given the limited comparability of the two groups. Micronutrient levels were largely within normal limits both prior to and after gastrostomy feeding, with levels of zinc, coper, vitamin B1 and chromium elevated beyond recommended ranges at followup. Fiber intake remained below recommended levels, and bowel movement frequency was generally unchanged over the study period. Chest infections requiring antibiotics decreased in half of participants at followup; 25 percent had no chest infections pre- or post-gastrostomy.

In one case series, investigators used impedance monitoring to assess episodes of GER pre- and post-PEG in children with CP (n=9) and Down syndrome (n=1) and normal reflux indices.⁹⁶ The intent of the study was to assess the degree to which tube feeding increased rates of GER. Participants' median age was 4.9 years (range: 0.5–16.8 years), and six were taking gastric pH/motility medications. The number of reflux events increased significantly post-PEG (183 at baseline to 355 post-PEG, p=0.047), with the percentage of events reaching the pharyngeal space increasing from 56 percent pre-procedure to 82 percent. The number of nonacidic events at followup was 182, while 173 were acidic. The proportion of reflux events that were acidic dropped from 85 percent prior to PEG to 51 percent after PEG. Weight increased over time (median weight gain: 2.53 kg, range: 0.8–7.24 kg), but the study was not designed to clearly ascertain the degree to which weight gain was associated with the PEG, and no comparisons were made to any reference group.

Table 11 outlines harms included in other surgical studies. In the Sullivan case series,^92-94 four of 57 children died during the study period, which the investigators note is comparable with death rates reported in other studies of tube-fed children. Importantly, given the observed decreases in hospitalizations and no change in antibiotic use, gastrostomy did not appear to increase respiratory morbidity.⁹⁴ No child in this study developed reflux that could not be managed medically. Three children died during the course of the quality of life study⁹³; the authors note that deaths were due to the children's underlying condition rather than the gastrostomy. In the Brazilian case series,⁸⁸ three participants died of causes thought to be unrelated to gastrostomy. In one retrospective case series, 23 percent of all participants had major complications including gastrointestinal bleeding (n=5), peritonitis (n=3), bowel obstruction (n=3), tube migration (n=2), wound dehiscence (n=1). Rates of harms did not differ among subgroups. Eight children died after gastrostomy, with 5 deaths occurring within 1 year of placement.⁹⁰

Table 11

Harms reported in case series assessing effectiveness of feeding interventions in CP.

Overall, rates of peritonitis were low, ranging from 2 to 5 percent, one study reported minor site infections at 59 percent and leakage at 30 percent. Deaths ranged from 6 to 15 percent, but were considered not to be related to gastrostomy in all studies. Mortality is high among individuals with CP, and it is impossible to know whether the observed deaths were causally related to treatment or to the course of the condition.

Effectiveness Data

No studies directly compared the use of g-tube with fundoplication to oral feeding. One fair quality RCT compared Nissen fundoplication (in which the upper part of the stomach is wrapped around the lower esophagus to create a one-way valve) with vertical gastric plication (in which the esophagus is lengthened) at two hospitals in Brazil.⁹⁹ GERD was established in the participants based on esophageal pH. Both groups had clinically significant improvement in reflux symptoms. The Nissen fundoplication group had a statistically significant decrease in the total number of reflux episodes (p=0.012), percentage of pH (p=0.002), and composite of pH parameters (p<0.001), whereas the vertical gastric plication group had a significant decrease in pH parameters (p=0.041) and percentage (p=0.042). There was no difference between the groups in the length of the procedure or hospitalization.

In one prospective case series, reflux and pH monitoring were reported to be improved after gastrostomy plus Nissen fundoplication, regardless of laparoscopic versus open approach (baseline median reflux index=5.7% vs. 0.15%, p=0.009).¹⁰⁰ Within one year after surgery, 30 percent of the patients had a relapse in reflux symptoms but were responsive to medical intervention. Table 12 summarizes key outcomes of these studies.

Table 12

Summary of key outcomes of studies of surgical interventions in individuals with CP and significant reflux.

Harms Data

The RCT comparing Nissen fundoplication with vertical gastric plication reported a rate of major complications (valve migration, hernia) of 14.3 percent in both the Nissen and vertical plication groups. Other harms reported are listed in Table 13; two participants died during the study (one in each group, considered unrelated to surgery).⁹⁹ In the case series early complications (within one week of fundoplication) included pneumothorax, stoma ulcers, and cellulitis, all in patients undergoing laparoscopic fundoplication. Late complications (i.e., more than one week post-fundoplication) included bloating, diarrhea (dumping syndrome), and intestinal obstruction; these events were all reported in individuals undergoing open fundoplication.¹⁰⁰

Table 13

Harms reported in primary research studies of surgical interventions in individuals with CP and significant reflux.