Food refusal and aversion to feeding are frequently encountered in paediatric practice. It is estimated that up to 25% of otherwise normally developing infants and 80% of those with developmental handicaps have feeding problems (1,2). When this pattern of behaviour persists, it warrants an investigation because it may stem from conditions that lead to a painful response to food, anorexia, or difficulty in feeding. Frequently diagnosed conditions that can cause painful or difficult feeding include gastroesophageal reflux disease (GERD), food allergy, airway obstruction, or swallowing disorders (3–6) to name but a few.
Infantile feeding disorders (IFD) usually present as refusal to feed or low food intake with refusal after a few bites or swallows. A variety of symptoms or outcomes have been associated with IFD, such as vomiting, gagging or retching, irritability, failure to thrive (FTT) (7), or apparent difficulty in swallowing, thus mimicking many of the previously mentioned organic conditions. In addition, the vulnerable child syndrome may also present with severe FTT and food refusal in the absence of an IFD or organic disease (8). The outcome of this situation is confusion, misdiagnosis, and underdiagnosis because many patients with IFD may initially obtain a diagnosis of another medical condition, and IFD is often diagnosed after significant delays.
The diagnosis of IFD relies on identifying intrusive feeding behaviour and a high index of suspicion, and diagnosis by mental health caregivers (most commonly diagnosed by psychiatrists or psychologists using psychiatric classifications as ICD-10 or DC 0-3R) (9,10). Certain classifications (DSM-IV) (11) are simplistic and do not help differentiate between IFD and organic disorders, whereas others (Chatoor, DC 0-3R) do not have a single set of criteria for diagnosis; rather each subset of IFD has different diagnostic criteria classifications (12,13). Furthermore, these criteria often require FTT or weight loss to make the diagnosis and do not account for children who maintain weight by pathological feeding methods or regressive age-inappropriate feeding methods. Lastly, criteria often require exclusion of other disorders, which means that every child must have an exhaustive workup before the diagnosis of a nonorganic disorder can be entertained.
The possibility of initiating tube feeding in the child who does not grow well and refuses to eat may be tempting if no cause is found and other therapies fail. This in turn can reinforce refusal to feed.
The paediatric FTT and IFD clinic at the Wolfson Medical Centre evaluates all children referred for difficulty in feeding, swallowing, or gaining weight by a multidisciplinary team. Infants and young children with IFD are subsequently enrolled in a treatment program (Appendix 3). This multidisciplinary effort led to identification of several common clinical characteristics for patients with feeding disorders as a cause for food refusal, gastrointestinal symptoms, or FTT. We, therefore, attempted to characterize maternal or infant child behaviour patterns, signs, and symptoms that were characteristic for IFD, which would help distinguish these patients from patients with symptomatic medical disorders and similar presentations.
The study was approved by a hospital ethical committee and was performed on 2 outpatient cohorts, referred to our clinics between mid-2005 and 2007. The nonorganic food refusal or poor feeding cohort consisted of children younger than 6 years of age, referred and evaluated for food refusal or IFD, without an underlying disorder, or of children with additional medical pathologies, unresponsive to medical, nutritional, or surgical interventions, who subsequently responded to behavioural intervention. Inclusion criteria for this group included onset of symptoms before 2 years of age, a history of persistent food aversion or poor oral intake (defined by a dietician) lasting more than 1 month, abnormal feeding patterns at present or during the period of refusal, and response to behavioural therapy with initiating of eating by the infant or child, increased oral intake, and maintenance of weight curves. Patients who were diagnosed as being picky eaters or patients who did not respond to behavioural therapy were excluded. We also excluded patients who were tube fed at the time of referral. We used these criteria to ensure that food refusal or poor feeding in the study cohort was due to behavioural and not other medical reasons. Patients with coexisting medical disorders were not excluded if they responded as above to this therapy.
The second group consisted of patients presenting with similar characteristics, in which a diagnosis of GERD, milk allergy, or FTT of other origins (idiopathic or genetic or vulnerable child syndrome–related FTT) syndrome was the final diagnosis, and who subsequently responded to medical or nutritional therapy without requiring behavioural intervention. We excluded patients who had other symptoms that clearly indicated the organic origin of symptoms, such as allergic rashes or atopy, diarrhoea, rectal bleeding, aspiration, or a previously diagnosed swallowing disorder. We further excluded patients with easily recognized organic causes of disease such as celiac disease, eosinophilic esophagitis, or evidence of malabsorption. We also excluded infants who had developed signs of refusal, weight loss, or vomiting shortly after being exposed to milk products, and resolved completely with a change in formula, because these patients do not present a diagnostic dilemma.
All patients underwent a detailed structured medical history and examination. Weight, height, and feeding history were recorded. A structured symptom and feeding history as well as a treatment scheme developed by the senior author was used in all cases. This scheme and the treatment method we termed avoidance transfer appear below and in Appendices 1 to 3. The medical history will not be elaborated upon. The focused feeding history concentrated on commonsense abnormalities that we had previously observed to appear frequently in these children, and are described below.
In brief, feeding history from birth was recorded, including age of transition from foods or types of feeding, with special emphasis on transition of type of feeding (breast → bottle → spoon → self-feed) and consistency of foods (liquid → mash → soft → solids). Special attention was given to first onset of avoidance behaviour by the infant or child, and sequence of normal or abnormal feeding patterns by the feeding caregiver. The external events or etiology identified by the team leading to food refusal or abnormal feeding patterns was defined as the trigger. On the basis of experience in our clinic, there are 5 categories of triggers, explained in more detail in Appendix 2. These include size (small infants, premature infants, and small-for-gestational age infants are perceived by parents or the medical team to be too small, leading to persistent attempts to feed the infant beyond the infant's hunger cues), transition (the feeding disorder is precipitated by a traumatic transition from 1 type of feeding method to another, such as breast-feeding to bottle or bottle to spoon feeding), organic causes (a disease causes decreased hunger or painful feeding and food refusal, which in turn leads to intrusive or persecutory feeding and subsequent food refusal, even if the initial disease has been resolved), mechanistic feeding (usually feeding at fixed intervals without a hunger cue, or emotionally detached feeding), and posttraumatic causes. The acronym STOMP (size, transition, organic, mechanistic, and posttraumatic) was used to recall triggers.
Dysfunctional Caretaker Feeding Patterns
These appear in Appendix 1 and included forced feeding, feeding during sleeping (the child is fed while somnolent and without a hunger cue, because the child did not feed well), mechanistic feeding, excessively prolonged feeding at mealtimes, conditional distraction, feeding persecution, and parental response to refusal (do the parents respect refusal or override refusal?).
Abnormal Infant Feeding Patterns
Abnormal feeding patterns included type of refusal (global or selective to mode of feeding, global or selective to a specific parent, global or selective for consistency or type of food), as well as presence of fixation (the willingness to ingest only 1 type of food or texture of food), initiation of feeding (parent initiates feeding or child indicates desire for food), and autonomy of feeding in age-appropriate patients.
In addition to the above, we solicited a history of clinical symptoms such as vomiting, gagging, or retching, anticipatory gagging (gagging or retching at the sight of food, a spoon, or a bottle), perceived abdominal pain or irritability, and head turning when trying to feed the child. We also inquired whether vomiting was postprandial, occurred hours after meals, or during sleep. Growth curves were plotted on Centers for Disease Control and Prevention growth charts. Standard laboratory examinations included a complete blood count, biochemical tests (alanine aminotransferase, aspartate aminotransferase, glucose, electrolytes, creatinine urea, albumin, calcium, and phosphorus), and thyrotropin stimulating hormone. Celiac serologies (tissue transglutaminase, anti-endomysial antibody, or deamidated anti-gliadin antibodies) with IgA levels were performed when appropriate (gluten exposure). Additional tests such as a sweat test, fecal elastase, barium meal, modified barium swallow, gastroscopy, pH study or pH impedance, gastric emptying scintigraphy, allergy testing, or switching to a different formula (hydrolysed or elemental formula) were performed on an individual basis depending on the presentation, age of onset, symptoms, growth, and response to therapy. Nutritional enrichment programs were instituted by a dietician as needed.
A detailed explanation of the treatment methods is beyond the scope of this article. In brief, therapy for all patients-caregivers was performed on an ambulatory basis, and was based on modification of parental behaviour and not by direct infant behavioural modification, similar to the Graz model (14,15). Once the diagnosis was entertained, we explained to the parents that child feeding behaviour, was often a response to parental feeding behaviour which led to a vicious cycle of refusal → circumventing refusal by abnormal parental feeding → reinforcement of refusal → increased intrusive feeding behaviour and so on. We attempted to transfer infant–child avoidance behaviours from the child to the parents or caregivers (avoidance transfer). We asked the parents to avoid all identified pathological feeding habits, avoid placing emphasis on weight or weighing child, avoid taking an interest in feeding, avoid reinforcing feeding behaviour by awards punishment or explanations, avoid expressing emotions related to the child's feeding behaviour, and avoid intervening in food choices or autonomy. We defined response to therapy as response of the infant or child to therapy, although parental improvement was noted as well. Response to therapy by the child included initiation of feeding, disappearance of refusal, increased variety of foods, increased weight gain or stabilization of growth curves, and for those children with vomiting at presentation, cessation of vomiting. Response for the above parameters included improved feeding behaviour for all infants and improvement in somatic symptoms or weight gain whether a problem had been present at onset. Weight gain was defined as weight gain that led to maintenance or improvement of a weight curve.
We did not perform any scoring for parental response, but recorded parental compliance with our treatment scheme.
Analysis of data was carried out using SPSS 9.0 statistical analysis software (SPSS Inc, Chicago, IL). Normality of distribution of continuous variables was assessed using the Kolmogorov-Smirnov test (cutoff at P = 0.01). All continuous variables had distributions significantly deviating from normal, so are described as median (min–max) in addition to mean ± standard deviation. Comparisons of continuous variables were made by group using the Mann-Whitney U test. Categorical variables such as sex, types of aversion behaviours, and disorder symptoms were described using frequency distributions and are presented as frequency (%). The χ2 test (exact as needed) was used to assess associations between group and other categorical variables. All tests are 2-sided and considered significant at P < 0.05.
We initially evaluated 226 children presenting with food refusal and low energy intake FTT of unknown origin. We excluded 75 children because of lack of follow-up (n = 29), lack of response (n = 12, 6 were complete failures, 6 partial responders), a clear organic diagnosis from onset, or other exclusion criteria mentioned above (n = 34). The study cohorts consisted of 151 children. The behavioural group (group 1) contained 83 children, and the control group of organic disorders with similar symptoms (group 2) contained 68 children. Group 2 included 21 with idiopathic low intake FTT, 27 with nonobvious milk allergy at presentation, 14 with GERD and 6 with combined pathologies. Demographic data for both groups are presented in Table 1.
No significant differences were found for birth weight, gestational age, or age of onset of symptoms (median onset of symptoms in group 1 was 5 months, in group 2 was 4.5 months; means presented in Table 1).
Patients with milk allergy and reflux symptoms presented earlier to medical care than did patients in group 1 (P = 0.001 and P = 0.035, respectively), although age of onset of symptoms was not significantly different as mentioned above (P = 0.17).
A history of tube feeding was seen in both groups, and although found more frequently more in group 1 than in group 2, there was no statistical difference (P = 0.09).
Thirty-five percent of children in group 1 had a history of an underlying organic disorder (milk allergy = 10%, GERD = 20%, other = 5%).
With respect to infant feeding patterns, significant differences were found between the 2 groups, as described in Table 2.
Food aversion was common in group 1 but not in group 2 (only 17.7%). This difference was significant also when comparing the different types of refusal (ie, bottle, spoon, or solids) between the 2 groups. Similarly, head turning away from food was seen in the vast majority of children in group 1 but in only one-quarter of children in group 2. Similarly, fixation (the willingness to ingest only 1 type of food or texture of food), anticipatory gagging (gagging or retching at the sight of food, a spoon, or a bottle), and meal-related gagging were also significantly more frequent in group 1.
Poor weight gain was significantly more frequent in patients with organic disorders, although more than 50% of patients in group 1 had poor weight gain defined as FTT. On the other hand, the number of children with low intake and those with vomiting (Table 2) did not differ between the 2 groups (P = 0.82 and P = 0.66, respectively).
Data regarding abnormal parental feeding practices are portrayed in Table 3. These practices were observed more frequently in group 1. Appendix 1 offers a brief description of each of these abnormal feeding behaviours. Triggers, as defined in Appendix 2, are presented in Table 4.
The mean length of follow-up in both groups was similar (5.7 ± 5.5 months for group 1 vs 5.1 ± 4.3 in group 2). Follow-up data for 11 patients in group 2 who responded to therapy were not available. Cessation of vomiting occurred in 76.8% of patients with vomiting in group 1, vs 69.2% of patients with vomiting in group 2; this was not statistically significant. Cessation of vomiting was checked only for those presenting with vomiting on admission. Children in group 1 initiated feeding on their own more than children in group 2 (88.9% vs 69.2% (P = 0.01), but both groups showed a good response.
Infants who present with poor feeding, difficulty in feeding, or persistent vomiting are among the most vexing patients physicians may encounter in both primary and tertiary care settings (6,14,16,17).
In our study, we indeed found that FTT, low intake with poor weight gain, or vomiting were present in the majority of patients with a behavioural source of these symptoms, and that these causes for referral do not help discriminate between the organic and nonorganic causes. Interestingly, a significant portion of the poor feeders did not meet criteria for FTT, probably because of the abnormal feeding patterns adapted by the parents or use of tube feeding. We believe that this study also highlights the fact that infants and children can have persistent vomiting because of underlying feeding disorders and abnormal feeding patterns, and this symptom as well as weight gain and willingness to eat can be restored by ceasing these abnormal feeding practices and avoidance transfer. Our findings also indicate that the present criteria for IFD would lead to underdiagnosis of children with behavioural causes for gastrointestinal symptoms and food refusal because they did not have nutritional deficiencies or FTT, which may be a requirement for diagnosis of IFD (9,11,13).
A common denominator for many of these patients with IFD was the presence of particular feeding behaviours by the infant, abnormal feeding practices by the parents, or a trigger that predated the onset of refusal.
An organic disease triggering food refusal or poor feeding (ie, GERD, milk allergy, swallowing problems, congenital or structural abnormalities), was found in almost 70% of the children in the control group, but also in one-third of the children in group 1. These findings illustrate the high prevalence of comorbidity (ie, that an organic disorder may lead to an IFD that may in turn be responsible for the symptoms even if the original medical condition is corrected).
The perception that the child was too small or that the child's growth rate was inadequate was more frequently found in group 2, but was the major trigger for abnormal feeding practices and subsequent refusal or poor feeding in 23.5% of patients in group 1. This trigger was commonly found in children born prematurely, small for gestational age, or who grew adequately but near third to fifth percentiles. The strong emphasis placed on catch-up growth and the importance of adequate nutrition can actually boomerang and lead to subsequent FTT if the emphasis on feeding to obtain growth becomes too intrusive. Parents and medical caretakers often place a great deal of emphasis on volume and frequent feeding in these children. Although children who do not feed well may not grow, the reverse is true as well. Children who are not growing for other reasons may feed less, because they require fewer calories for growth. Mei et al (18) demonstrated that a significant proportion of normal children will drop growth curves between 1 and 2 standard deviations. The first response by caregivers is often to try and increase feeding, irrespective of hunger cues or satiety, leading to aversion. This in turn may accentuate the drop in growth curves by causing the child to avoid feeding because of parental intrusive feeding behaviours.
Patients with FTT, low intake, and no clear underlying cause can be challenging. Wright et al (7) prospectively evaluated 923 infants for avoidant feeding behaviour. They found that avoidant behaviour on its own did not correlate with weight gain or loss. On the contrary, parental response to avoidant behaviour was inversely related to weight gain. In our study, patients in group 2 with idiopathic or nutritional FTT resolved their FTT with more nutritional intervention and no behavioural intervention, whereas patients in group 1 responded to less nutritional intervention and behavioural modification. We believe that the majority of patients in group 1 could be identified by identifying parental intrusive feeding behaviour as well as the other observations noted in this study, and the majority in group 2 with length of observation and response to nutritional support. Although there may be overlap, we often feel comfortable with observation and nutritional support if we do not find clues to raise the suspicion of an IFD.
Three categories of history taking provided strong clinical clues to the presence of a behavioural cause for refusal. The first was the presence of a nonorganic trigger for the subsequent feeding difficulty. The triggers most likely to be identified at an initial interview were onset of refusal or decrease in intake during a transition in feeding method, a traumatic event, or mechanistic feeding (usually feeding at fixed intervals without a hunger cue) by a parent. The second and possibly most important category was the presence of abnormal and intrusive feeding practices, which we found to be surprisingly uncommon in patients with an underlying organic cause for difficulty in feeding or food refusal. Among these, nocturnal feeding, mechanistic feeding, conditional distraction, and prolonged mealtimes are relatively easy to obtain during a short interview and do not usually arouse antagonism. The third category was infant behaviour. Multiple infantile or child behaviours were more common in the behavioural group. We have noted that fixation, anticipatory gagging (gagging at the sight of food or as food enters the mouth and before swallowing) and head turning away from food when offered, are particularly helpful clues to the presence of an IFD, and are unlikely to be found with refusal or poor feeding because of other causes.
Another interesting outcome is the delay in diagnosis for the diagnosis and treatment of IFD. Patients in group 1 presented to our clinic at a mean age of 21 months, whereas food refusal or poor feeding occurred at a mean age of 5.6 months. This emphasizes the difficulty primary care physicians have in recognizing IFD, and the prolonged workups performed by subspecialists before patients were referred.
Many of our observations regarding patients in group 1 correlate with alternative definitions used in Chatoor's classification (12,13,19). Thus, in the subcategory of infantile anorexia, Chatoor's method requires that “food refusal be present for more than 1 month, that the refusal typically occurs from spoon to self feeding, the child does not communicate hunger, lacks interest in food, shows a significant growth deficiency, food refusal did not follow a traumatic event, and food refusal is not due to an underlying illness.”
However, Chatoor's classification depends primarily on a physician's impression of feeding behaviour and not on specific questions or clinical observations that can be translated into a diagnosis. The onset of anorexia can also appear after transition from breast-feeding to bottle feeding, or bottle feeding to spoon. These descriptive criteria are helpful for more advanced patients probably seen by psychiatrists specializing in IFD, who do not have to exclude organic diseases, and reflect an advanced state of affairs. We believe that the categories that we have described can be used at an earlier stage, and may be more helpful in distinguishing IFD from medical illnesses, which is what a paediatrician requires when evaluating infants and children who present with food refusal.
However, the pioneering work of Chatoor and others has led to the ability to identify these children in specialised settings, and instigate appropriate therapy. Various previously described clinical signs not dealt with in our study may suggest the presence of an IFD. For instance, patients with growth failure because of other nonorganic reasons such as emotional deprivation often exhibit apathy and developmental delay, whereas infants with FTT because of IFD are likely to be willful, communicative, and appropriately developed (20).
The present study was not geared to evaluate outcome measures of our treatment method because this study involves inherent bias in regard to treatment. We excluded patients who did not respond to therapy to guarantee that we were truly dealing with a behavioural or psychological cause for food refusal. A comparison of different diagnostic models as well as the efficacy of this treatment method is being studied. It has been claimed that each different type of category of feeding disorder (eg, infantile anorexia, posttraumatic, sensory aversion) requires a different therapeutic approach (2,21). We have not found this to be true, and use of avoidance transfer with frequent follow-up was often effective for the majority of our patients irrespective of the trigger or manifestation. Infantile anorexia, sensory aversion, and posttraumatic IFD may improve with regression and avoidance of offending foods, although severe cases of sensory aversion may not respond even with prolonged follow-up and multidisciplinary interventions.
Optimal identification of IFD and behavioural causes of poor feeding require a multidisciplinary approach and observation of infant feeding (22). However, for most paediatricians and subspecialists who encounter these patients, this is not a realistic office scenario. We believe that by identifying a trigger, abnormal parental feeding practices, and infantile behaviours, physicians are more likely to be capable of recognizing these patients early, without specialized teams. Earlier recognition could lead to earlier referral to specialized caregivers and early intervention for the more difficult cases. Studies are under way to try to identify and validate a simpler set of clinical criteria for identifying IFD, but until that time, a detailed history and a high index of suspicion may help facilitate earlier diagnosis of nonorganic causes of food refusal.
APPENDIX 1: ABNORMAL FEEDING BEHAVIOURS
Nocturnal feeding: feeding an infant while somnolent because the child refuses food or eats small quantities while awake.
Persecutory feeding: constant (often unsuccessful) attempts to feed an infant or child despite refusal, frequent attempts to offer food to get the child to take another suck on the bottle, or ingest another spoonful of food.
Forced feeding: forcefully feeding a child even against his or her will, prying open the child's jaws, and so on.
Mechanistic feeding: feeding precisely at regular scheduled times (clockwork feeds, ie, precisely every 3 hours), ignoring absent hunger cues, or treating the child like an inanimate object during feeds (best identified while observing a feed).
Conditional distraction: all meals take place with distraction, the child will not eat without distraction, and does not show interest in food.
Prolonged meals: meals are long (usually more than 30 minutes), child eats minute quantities, parent continues with the meal despite lack of success, with or without other abnormal feeding behaviours listed above.
APPENDIX 2: TRIGGERS (STOMP ACRONYM)
Size: meaning that the caregiver of the child believes that the child growth rate is inappropriate to the child's age, and is often accompanied by size anxiety. This starts a viscous cycle in which food and weight gain become the main issue in the house, leading eventually to a feeding disorder. This trigger is often initiated by medical caregivers in small infants, by recommending increased volume or feeding irrespective of infantile cues.
Transitional feeding: this type of trigger refers to the case in which the refusal began in close association with the transition from one mode of feeding to another, and involves decreased intake or refusal of the new feeding method. For instance, refusal to drink formula from a bottle after switching from breast-feeding to bottle, refusing spoon feeds, or eating minute quantities after transition from bottle to spoon, from mash to particulate food, from particulate to solids and so on.
Organic: meaning that the child's feeding disorder is due to an organic problem, leading to an objective feeding difficulty. For example, GERD, swallowing problems, respiratory disease (bronchopulmonary dysplasia), congenital abnormalities, structural abnormalities.
Mechanistic feeding: the main caregiver of the child fails to interpret the basic hunger cues of the child, either because of parent–child attachment problems or to the caregiver's psychological difficulties (depression and anxiety). This type of behaviour is often characterized by precise timing of feeds without waiting for or respecting hunger cues, lack of emotional rapport or parental–infant communication during feeding, and attempts to get a specific quantity ingested during the meal.
Posttraumatic: the child suffers a traumatic event concerning his or her oral cavity (forced feeding, nasogastric tube, intubation, choking episodes, and so on). This event, influences the way the child perceives food and feeding, leading to avoidance of food.
APPENDIX 3: WOLFSON AVOIDANCE TRANSFER METHOD
Once the diagnosis was entertained, we explained to the parents that child avoidant feeding behaviour was often a response to parental feeding behaviour, which led to a vicious cycle of refusal → circumventing refusal by abnormal parental feeding → reinforcement of refusal → increased intrusive feeding behaviour, and so on. We attempted to transfer infant–child avoidance behaviours from the child to the parents or caregivers (avoidance transfer). Therapists identified as many abnormal parental feeding habits as possible. We asked the parents to avoid all identified pathological feeding habits, avoid placing emphasis on weight or weighing child, avoid taking an interest in feeding, avoid reinforcing feeding behaviour by awards punishment or explanations, avoid expressing emotions related to the child's feeding behaviour, and avoid intervening in food choices or autonomy.
Meals were to be kept short, and if not eaten removed from the table, and not offered without an infant–child hunger cue until the next meal. In cases of selective refusal such as to a spoon or bottle, we often performed regression to the previous mode of feeding (solid → mash, or spoon to bottle) until the next meeting or for up to 1 month, to decrease infant and parental anxiety and encourage more rewarding attitudes of the child to feeding. The problem feeding method was then re-introduced, and parents were instructed to accept infant–child cues. At the first sign of refusal, parents were asked to stop the feeding, take a short break, and then try again. If the child refused, then the meal was over. Reintroduction of a problem feeding was always preferred when the child was hungry. Reintroduction was often performed in day care initially (using peers as an adjuvant tool), and we took care to instruct the parents and caregivers on respecting the child's hunger cues. Parents were instructed on how to calmly cue a child that the meal was over, thus allowing the child to understand that if he or she did not eat the child would not be fed until the next meal. We encouraged autonomy in feeding and access to finger foods in the case of infants beyond 10 months. Access to juices and non-nutritious snacking were reduced.
Parents were encouraged to support the children for any other good behaviour unrelated to food or feeding by words of encouragement and increasing attention to other aspects of parent–child relationships, and to eat meals with the children. We encouraged parents to let children eat as much as they wanted and whatever they wanted without intervening. Children with global refusal or with parents who had difficulty in applying parental modification entered a food play group run by a swallowing therapist, whereas parents (and grandparents when possible) entered a parent support group run jointly by a dietician and a psychiatrist. Children were encouraged to play with food and interact with each other without adult intervention. The therapist observed infant–child feeding behaviour, whereas parents were absent, including whether food was eaten, and the ability to eat or swallow various types of foods. Follow-up was scheduled either weekly or every other week initially.