Persistent food refusal and poor feeding pose a significant diagnostic and management challenge (1–7). Poor feeding can be a result of concurrent diseases, behavioral disorders, or a combination of both. Infantile feeding disorders (IFD) typically present as either refusal to take food or as poor food intake. Additional symptoms such as vomiting, gagging, retching, irritability, failure to thrive (FTT), or an apparent difficulty in swallowing food are common (1). Because these symptoms are common to many diseases, IFDs are often misdiagnosed or initially mistaken for other medical conditions, leading to diagnostic delay (1). Failure to swiftly identify and treat IFD may lead to unnecessary invasive tests and even tube feeding, progression of FTT, and sensory aversive behaviors.
At present, the diagnostic criteria for IFD require identification of food refusal together with FTT after the exclusion of organic cause for the symptoms (8–11). The Chatoor (or Chatoor-based DC 0–3R) criteria (Appendix 1) are cumbersome, with different diagnostic criteria in 6 different categories. The Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) criteria are much simpler, but do not facilitate differentiation between IFD and organic disorders (11), because they simply mandate that poor feeding and FTT in the absence of organic disease and adequate food supply can be considered an IFD.
We have previously identified multiple feeding behaviors that are characteristic of IFDs, but not of organic causes of food refusal or poor feeding (1). We subsequently developed the Wolfson criteria for diagnosing IFD in children with or without organic disease (1). These criteria were derived from statistical modeling of 2 retrospective cohorts (83 infants with IFD diagnosed by Chatoor criteria responsive to behavioral therapy and 68 controls with poor feeding resulting from clearly defined medical conditions) (1). A discriminant analysis of 21 clinical variables evaluating parental and infant feeding behaviors identified the following 3 as significantly different between the groups: a history of food refusal, pathological feeding (defined in Appendix 2), and anticipatory gagging. Using these data, we generated the Wolfson criteria, which correctly classified 88% of the entire cohort (89% sensitivity, 84% specificity, 79% positive predictive value, and 92% negative predictive value) (Table 1).
Despite the encouraging initial results, the original report often overestimates the validity of diagnostic criteria (12), necessitating duplication of the results in an independent, preferably prospectively enrolled cohort. We, therefore, tried to evaluate the validity of the Wolfson criteria on an independent, prospectively enrolled cohort, and to compare its performance with the DSM-IV and Chatoor classifications.
PATIENTS AND METHODS
This double-cohort prospective study evaluated infants who were referred to the feeding disorders clinic between February 2008 and July 2009 for complaints of feeding disorders, food refusal, and failure to thrive.
The parents of all of the included infants were interviewed by 1 of the 3 physicians experienced in managing IFD. Inclusion criteria included meeting any 1 of the 3 contending classification themes (Wolfson, Chatoor, or DSM-IV), age younger than 6 years, onset of symptoms before 2 years of age, and a history of persistent food aversion or poor oral intake (defined by a dietitian) lasting >1 month. Patients were excluded if they were tube fed or diagnosed as having organic disease either by medical history or by screening tests (see below). Infants with overt clinical milk protein allergy (ie, food refusal, weight loss, or vomiting shortly after being exposed to milk, with complete resolution after exclusion of dairy products) were also excluded. Written informed consent was obtained from all of the participants as per instructions of the local ethics board that approved this study.
Infants and their caregivers underwent a detailed structured medical examination and completed a 34-point qualitative questionnaire. The questionnaire included feeding history from birth, details of food transitions, history of the onset of avoidance behavior, feeding patterns (Appendix 2), and clinical symptoms including vomiting, gagging, retching, and anticipatory gagging (gagging or retching at the sight of food).
Growth curves were plotted on Centers for Disease Control and Prevention growth charts, and height and weight were recorded as standard deviation scores (ie, z scores). FTT was defined as a drop of at least 2 channels in Centers for Disease Control and Prevention weight curves or curves below the third percentile. Standard laboratory examinations were performed in all of the cases and included complete blood count, liver enzymes, glucose, electrolytes, creatinine, urea, albumin, and thyrotropin-stimulating hormone. Celiac serology (tissue transglutaminase and deamidated gliadin peptide antibody) was performed in all of the cases in which a history of gluten exposure was apparent. The decision to perform additional testing was dictated by board-certified, experienced, pediatric gastroenterologists after a thorough medical review. A gastroscopy was performed on patients with a high clinical suspicion of esophagitis or dysphagia.
Following the initial evaluation, a previously published treatment scheme was used to modify parental behavior (1). This scheme, termed role reversal or avoidance transfer, was performed on an outpatient basis. Patients were seen initially every 2 to 4 weeks according to the severity of refusal and then followed up monthly. Response to therapy required all of the following: initiation of feeding, disappearance of refusal, increased variety of foods, increased weight gain or stabilization of growth curves, and, when applicable, cessation of vomiting.
Patients younger than 6 years referred to the feeding disorders clinic due to food refusal or poor intake with an organic origin of their symptoms and who responded to medical or nutritional therapy without behavioral intervention served as a reference group. This group was used to compare behavior patterns and determine the specificity of the criteria of IFD. Categories were assigned by the Chatoor categorization for patients defined as IFDs by Chatoor criteria (Appendix 1) to 1 of the 5 categories based on the cause that triggered refusal according to the Wolfson criteria, previously described (1). These include size (small infants, premature infants, and small-for-gestational age infants are perceived to be too small by parents or the medical team, 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 causing 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 (a choking episode, after painful hospital feeding experience, tubes, etc). Pathological feeding was present when any 1 of the criteria outlined in Appendix 2 was present, including clockwork (mechanistic) feeding, forced feeding, nocturnal feeding, conditional distraction, and feeding persecution.
Analytic Approach and Statistics
Validity is the degree to which the instrument measures the concept that it purports to measure and can be evaluated in many ways, including criterion and predictive assessments (13). In psychosocial science, a criterion is rarely found. We therefore estimated the classification agreement between the Wolfson criteria and Chatoor and, separately, DSM-IV criteria. FTT is not required by the Wolfson criteria to diagnose IFD; therefore, we expected the Wolfson criteria to be more sensitive than the existing measures.
We also used predictive validity, in which children who responded to generic treatment for IFD without other interventions can be assumed to have had IFD. All of the children were classified as responders or nonresponders to treatment. Predictive validity was assessed by calculating diagnostic statistics using response to therapy as the reference standard. Overall diagnostic accuracy was used to compare the predictive validity of all 3 classification criteria, calculated as the sum of true positive and true negative cases divided by the total number of patients in the analysis. Intention-to-treat (ITT) analysis was used in which all of the infants entering the treatment program were analyzed, considering those who were lost to follow-up in the IFD group as failures. In a conservative sensitivity analysis, dropouts were considered in the Wolfson criteria as failures and in the Chatoor and DSM-IV criteria as successes. Finally, the ability of each of the classification schemes to differentiate infants with organic disease from those with IFD was assessed by calculating diagnostic utility statistics.
Analysis of the data was carried out using SPSS 11.0 statistical analysis software (SPSS Inc, Chicago, IL). Continuous variables are expressed as mean ± standard deviation, and categorical variables are presented as proportions. Point estimates are shown with 95% confidence intervals (95% CI). Comparison of continuous variables between the IFD and the comparison group was performed using the unpaired t test or the Mann-Whitney test according to the distribution normality. Categorical variables were compared using the chi square test or Fisher exact test as appropriate. The validity of diagnosis was assessed by calculating sensitivity, specificity, positive and negative predictive values, and prevalence and diagnostic accuracy using Chatoor's method and, separately, DSM-IV, as the reference. In addition, response to treatment was used as the reference in an analysis of prediction validity. A value of 1 was added to cells that contained no counts (ie, the value 0) to permit calculations. All of the tests were 2-sided and considered significant at P < 0.05.
A total of 85 infants who fulfilled the eligibility criteria were enrolled in the IFD group and compared with 55 children diagnosed as having organic disease (comparison group). The basic characteristic data of both groups are presented in Table 2. The majority of the children in the IFD group (70%) were either first (42%) or second born (28%). Subjects identified as IFD were similar in terms of age at presentation and sex regardless of the classification method used.
Feeding Patterns in the Study Group
As expected from the inclusion criteria, pathological feeding was identified in all 85 cases. Food refusal had presented in 10 (12%) infants during breast-feeding, in 33 infants (39%) during bottle feeding, in 34 infants (40%) during spoon feeding, and in 10 infants (12%) during semisolids or solids feeding. Consistent food refusal was present in 78 (91%); intermittent refusal lasting at least 1 month occurred in 8 (%) subjects. Eighty-seven percent of parents reported that the initiative for feeding was always or almost always from the parent and not the child (Table 2).
Evaluation of Validity
All 85 infants were classed as IFD according to the Wolfson criteria (100%), 65 (76%) by the Chatoor criteria, and 48 (56%) by the DSM-IV criteria. The most frequent IFD type identified by Chatoor was “infantile anorexia” (55%), and 45% of IFD cases identified by the Wolfson criteria were considered “transitional.” Both infantile anorexia and transitional types suggest that IFD arises during the transition from 1 feeding method to another (Table 3).
The sensitivity and specificity of the Wolfson criteria for detection of IFD were 98% (95% CI 96%–100%) and 5% (0%–14%) using Chatoor as the reference and 65% (95% CI 53%–76%) and 70% (50%–90%) using DSM-IV as the standard reference.
Treatment directed at IFD was initiated in all 85 infants by a multidisciplinary team (1). Sixty-five (77%) responded to treatment and 12 (14%) were lost to follow-up (defined as patients attending only 1 or 2 treatment visits and not returning).
The 3 classification criteria were compared with respect to treatment outcome as a reference standard (Table 2). The ITT response for patients identified by the 3 classification themes was similar (DSM-IV 73% vs Chatoor 75% vs Wolfson 76%).
In a worst-case scenario analysis in which all of the lost-to-follow-up cases were assumed to be treatment failures, Wolfson criteria had 75% (66%–84%) sensitivity and 98% (95%–100%) specificity, and Chatoor criteria had 75% (66%–84%) sensitivity and 75% (69%–86%) specificity. In a best-case scenario analysis in which all of the lost-to-follow-up cases were assumed to be treatment successes using response to treatment as the reference standard, Wolfson criteria had 99% (96%–100%) sensitivity and 33% specificity (20%–83%), and Chatoor criteria had 76% (67%–85%) sensitivity and 33% (20%–85%) specificity.
There were 20 patients classified as having IFD by the Wolfson criteria but not by Chatoor's criteria, of whom 16 (80%) had complete clinical response to therapy, similar to the overall response rate (Fig. 1). Twenty infants (24%) from the entire cohort did not respond to therapy, of whom 16 (80%) were classified as having IFD by Chatoor, similar to the positive fraction of Chatoor in the entire cohort (76%). These findings indicate that although response to therapy is the same, the Wolfson criteria will identify a substantial proportion of treatable patients that Chatoor would not identify (16/85, 18% of the whole cohort).
Infants who failed to respond tended to be older at presentation (mean age 30.69 ± 7.0 months vs 22 ± 15.98 months for responders, P = 0.069) and had a younger age of onset (6.6 ± 4.5 vs 9.6 ± 4.9 months, P = 0.016).
Discriminative Validity: Wolfson Criteria in the Comparison Group
The control group comprised 55 infants younger than 6 years referred to the IFD clinic for food refusal or poor feeding, but in whom an organic symptom etiology was discovered (Table 1). Patients in the control group were significantly younger at onset than IFD cases. Symptoms such as vomiting and FTT did not differ significantly between cases and controls. Consistent food refusal (91% vs 18%, P < 0.001), forced or aggressive feeding (65% vs 7%, P < 0.001), and nocturnal feeding (59% vs 11%, P < 0.001) were significantly more prevalent among patients with IFD than among controls. As in our previous study, anticipatory gagging was much more common among patients with IFD compared to the organic food refusal group (47% versus 2%, P < 0.001).
Pathologic feeding was present in 100% in the IFD group versus 38% of the comparison group (P < 0.001). Ten infants (18%) from the control group could have been diagnosed as having IFD based on Wolfson criteria if the underlying medical disorder had not been diagnosed. Of these, 5 (50%) had food allergies, 1 had gastroesophageal reflux disease, and 4 had nutrition-related issues (vulnerable child syndrome). Eight of 10 would have been diagnosed as having IFD by Chatoor or DSM-IV under similar circumstances, 2 would have been excluded because of the absence of significant weight loss.
This study shows that the recently developed Wolfson criteria are a more accurate screening tool for diagnosing IFD than the 2 existing ones. Wolfson criteria are simple to score, making them attractive as an accurate screening tool for primary care physicians and have high predictive validity in identifying patients likely to respond to therapy with better sensitivity than either DSM-IV or Chatoor.
In this study, we have corroborated our earlier finding that IFD almost always involves a history of intrusive feeding. The types of pathological feeding are relatively easy to recall and elicit during a feeding history (Appendix 2). In fact, pathological feeding was detected in all of our patients in this study. This was not because of bias related to inclusion in our criteria because it was true for all of the patients diagnosed by the other 2 methods as well. Anticipatory gagging, although present in slightly less than 50% of patients, was rare in the comparison group and should be regarded as a specific diagnostic clue for differentiating IFD from organic disease.
We believe that our findings, and especially the pathological feeding, also bear significance regarding the possible pathogenesis of IFDs. Although there are many reasons a child may develop IFD (14), we hypothesize that the final common pathway is always an interaction between food refusal and intrusive feeding and that 2 different scenarios may lead to the same outcome. The first begins with intrusive feeding that provokes food refusal. The subsequent decline in intake starts a vigorous vicious cycle. The scenario is typically seen not only with abusive feeding practices or aberrant maternal bonding but also with mechanistic feeding patterns in infancy or even medical supervision. Parents of infants growing along the bottom percentiles are often instructed to increase the volume of intake and this can lead to mechanistic feeding (ie, feeding every 3 hours or getting a child to finish a given quantity of formula at every meal irrespective of hunger cues or refusal). The second scenario starts with feeding refusal and an inappropriate parental response to the refusal (7). This is seen typically in transitional feeding disorders primarily from breast-feeding to bottle feeding or bottle feeding to spoon feeding (45% of our cohort). The infant's refusal leads to increasing persecution and even nocturnal and forced feeding. In the Millennium Infant Study that studied 923 infants from birth for feeding behavior, avoidant feeding behavior was common, but it was not associated with decreased appetite, weight gain, or FTT (7). The manner in which caregivers responded to avoidant behavior was 1 of the strongest predictors of poor weight gain by 12 months, supporting the premise that an abnormal parental response to avoidant feeding behavior is required for a child to develop IFD.
The common wisdom that an underlying organic disorder is usually responsible for poor feeding is untrue in many cases (15). However, comorbidity with IFD is not rare, occurring in 35% of patients in our prior study (1) and approximately 18% of patients in this study. Therefore, the presence of IFD cannot be excluded on the basis of the presence of a medical condition associated with poor feeding. On the contrary, the presence of comorbidity can trigger avoidant feeding and subsequently IFD. Our screening criteria thus permit comorbidities with the caveat that they have not responded to treatment.
As is evident in this study, there is a significant delay in referral for IFD (mean onset of refusal was 8 months, but the mean age of presentation was 24 months,). Several reasons can explain the delay in diagnosis by the primary care physicians, including lack of familiarity with diagnostic criteria and tolerance of poor infant-feeding patterns among parents or physicians. This may have ramifications for treatment success because patients in our study who did not respond to treatment were likely to develop symptoms earlier and present later than responders, indicating a longer duration of unsuccessful treatment. Our Wolfson criteria reflect a much simpler version than Chatoor. We hope that the implementation of these user-friendly criteria will facilitate more rapid diagnosis and treatment.
It is important to recognize the limitations of this study. We did not evaluate our criteria in patients who were tube fed at the time of enrollment (this was an exclusion criterion). Patients with prolonged tube feeding may develop tube dependency (16), requiring a different treatment strategy in our institution, which would have affected our ability to evaluate predictive validity with a treatment specifically geared to treat only IFDs. This subgroup needs to be investigated further in an independent study.
In conclusion, our screening criteria propose a simple set of feeding patterns for IFD, even if other medical conditions have not yet been excluded, with a detection rate that is as good as the present criteria. We believe that these criteria would simplify and facilitate earlier detection of the disorder and educate primary care physicians about the negative consequences of pathological feeding.
1. Levy Y, Levy A, Zangen T, et al. Diagnostic clues for identification of nonorganic versus organic causes of food refusal and poor feeding. J Pediatr Gastroenterol Nutr 2009; 48:355–362.
2. Chatoor I, Ganabian J. Food refusal by infants and young children: diagnosis and treatment. Cogn Behav Pract 2003; 10:138–146.
3. Mathisen B, Worrall L, Masel J, et al. Feeding problems in infants with gastro-oesophageal reflux disease: a controlled study. J Pediatr Child Health 1999; 35:163–169.
4. Heyman PE. Gastroesophageal reflux: one reason why baby won't eat. J Pediatr 1994; 125:S103–S109.
5. Hill DJ, Heine RG, Cameron DJ, et al. The natural history of intolerance to soy and extensively hydrolyzed formula in infants with multiple food protein intolerance. J Pediatr 1999; 135:118–121.
6. Staino A. Food refusal in toddlers with chronic diseases. J Pediatr Gastroenterol Nutr 2003; 37:225–227.
7. Wright CM, Parkinson KN, Drewett RF. How does maternal and child feeding behavior relate to weight gain and failure to thrive? Data from a prospective birth cohort. Pediatrics 2006; 117:1262–1269.
9. Zero to Three. Diagnostic Classification of Mental Health and Development Disorders of Infancy and Early Childhood (DC:0-3R Revised). Washington, DC: Zero to Three; 2004.
10. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 4th ed (text revision). Washington, DC: American Psychiatric Association; 2000.
11. Chatoor I. Feeding and eating disorders of infancy and early childhood. In: Weiner JM, Dulcan MK, editors. Textbook of Child and Adolescent Psychiatry. Arlington, VA: American Psychiatric Publishing Inc; 2004. pp. 639–652.
12. DeVellis RF. A consumer's guide to finding, evaluating, and reporting on measurement instruments. Arthritis Care Res 1996; 9:239–245.
13. Chatoor I. Feeding disorders in infants and toddlers: diagnosis and treatment. Child Adolesc Psychiatr Clin North Am 2002; 2:163–183.
14. Matson JL, Fodstad JC. Issues in identifying the etiology of food refusal in young children. J Pediatr Gastroenterol Nutr 2009; 48:274–275.
15. Zangen T, Ciarla C, Zangen S, et al. Gastrointestinal motility and sensory abnormalities may contribute to food refusal in medically fragile toddlers. J Pediatr Gastroenterol Nutr 2003; 37:287–293.
16. Dunitz-Scheer M, Levine A, Roth Y, et al. Prevention and treatment of tube dependency in infancy and early childhood. ICAN 2009; 1:73.
Diagnostic Criteria of DSM-IV
The feeding disorder of infancy or early childhood from the DSM-IV-R contains the following criteria (11):
* Criterion A. Persistent failure to eat adequately, as reflected in significant failure to gain weight or significant weight loss over at least 1 month.
* Criterion B. The disturbance is not due to gastrointestinal or other general medical condition (eg, esophageal reflux).
* Criterion C. The disturbance is not better accounted for by another mental disorder (eg, rumination disorder) or by lack of available food.
* Criterion D. The onset must be before age 6.
Chatoor Diagnostic Criteria for Feeding Disorder
1. Diagnostic criteria for feeding disorder under state regulation
* Infant has difficulty reaching and maintaining a state of calm alertness for feeding; is either too sleepy or too agitated or distressed to feed
* Infant's feeding difficulties start in the newborn period
* Infant fails to gain adequate weight or shows loss of weight
2. Diagnostic criteria for feeding disorder of reciprocity (neglect)
* Infant shows lack of developmentally appropriate signs of social responsivity (eg, visual engagement, smiling, and babbling) during feeding with primary caregiver
* Infant shows significant growth deficiency
* The growth deficiency and lack of relatedness are not solely caused by a physical disorder or a pervasive developmental disorder
3. Diagnostic criteria for infantile anorexia
* Child refuses to eat adequate amounts of food for at least 1 month
* Onset of food refusal often occurs during the transition to spoon and self-feeding, typically between 6 months and 3 years of age
* Child does not communicate hunger and lacks interest in food, but shows strong interest in exploration and interaction across caregiver contacts
* Child shows significant growth deficiency (see measurement of growth deficiency above)
* The food refusal did not follow a traumatic event
* The food refusal is not caused by an underlying medical illness
4. Diagnostic criteria for sensory food aversions
* Child refuses to eat specific foods with specific tastes, textures, smells, or appearances
* Onset of the food refusal occurs during the introduction of a different type of food (eg, the child may drink one type of milk, but may refuse another; may eat carrots, but may refuse green beans; may eat crunchy foods, but may refuse pureed food or baby food)
* Child eats better when offered preferred foods. Child must have specific nutritional deficiencies or oral motor delay or both
5. Diagnostic criteria for feeding disorder associated with concurrent medical condition
* Child readily initiates feeding, but during the course of feeding shows distress and refuses to continue feeding
* Child has concurrent medical condition that is believed to cause the distress
* Medical management improves but not fully alleviates the feeding problems
* Child fails to gain adequate weight or may even lose weight
6. Diagnostic criteria of posttraumatic feeding disorder
* Food refusal follows a traumatic event or repeated traumatic insults to the oropharynx or gastrointestinal tract (eg, choking, severe vomiting, insertion of nasogastric or endotracheal tubes, suctioning) that trigger intense distress in the infant
* Consistent refusal to eat manifests in one of the following ways:
- Child refuses to drink from the bottle, but may accept food offered by spoon (although consistently refuses to drink from the bottle when awake, may drink from the bottle when sleepy or asleep)
- Child refuses solid food, but may accept the bottle
- Child refuses all oral feeding
* Reminders of the traumatic events cause distress, as manifested by one or more of the following:
- Child may show anticipatory distress when positioned for feeding
- Child shows intense resistance when approached with bottle or food
- Child shows intense resistance to swallowing food placed in mouth
- The food refusal poses an acute or long-term threat to the child's nutrition
Abnormal Feeding Behaviors
1. Nocturnal feeding— feeding an infant while somnolent because the child refuses food or eats small quantities while awake.
2. Persecutory feeding— constant (often unsuccessful) attempts to feed an infant by overriding refusal, frequent attempts to offer food to get the child to take another suck on the bottle or ingest another spoonful of food even though refusal has taken place.
3. Forced feeding— forcefully feeding a child even against his or her will by prying open the child's jaws.
4. Mechanistic feeding— feeding precisely at regular scheduled times (clockwork feeds, ie, precisely every 3 hours, usually a given exact amount), getting a child to finish a given quantity of formula at every meal irrespective of hunger cues, or treating the child like an inanimate object during feeds (best identified while observing a feed).
5. Conditional distraction— all meals take place with distraction, the child will not eat without distraction and does not show interest in food.