A systematic review of ordinal scales used to classify the eating and drinking abilities of individuals with cerebral palsy

Authors


Abstract

Aim

The aim of this review was to examine systematically the scope, validity, and reliability of ordinal scales used to classify the eating and drinking ability of people with cerebral palsy (CP).

Method

Six electronic databases were searched to identify measures used to classify eating and drinking ability; in addition, two databases were used to track citations of key texts. The constructs assessed by each measure were examined in relation to the World Health Organization International Classification of Functioning, Disability and Health. Evidence of validity and reliability of the identified scales was appraised from peer-reviewed studies using standard criteria.

Results

Fifteen scales were identified in 23 papers. Clinician or researcher assessment was required for 13 scales; nine scales made use of information from parents and carers through interviews or questionnaires. Eight scales used the terms mild, moderate, and severe (with varying definitions) to describe different aspects of eating and drinking impairment. There was an assessment of either content validity and/or reliability for five scales; however, none met the recommended psychometric quality standards.

Interpretation

Currently, there is a lack of evidence of the validity and reliability of ordinal scales of functional eating and drinking abilities of people with CP.

Abbreviations
ICF

International Classification of Functioning, Disability and Health

PEDI

Pediatric Evaluation of Disability Inventory

Individuals with cerebral palsy (CP) experience activity limitations including limitations in sitting, standing, walking, handling objects, and speaking. Impairments can also interfere with the oral functions required for eating, drinking, and swallowing,[1, 2] and the ability to bring food and drink to the mouth.[3] Limitations in the ability to bite, chew and swallow, and self-feed are often associated with prolonged mealtimes and loss of both food and fluid from the mouth; this can lead to insufficient food and fluid intake to ensure growth and good health,[3-7] as well as adverse respiratory consequences such as episodes of choking and aspiration.[1, 8-10]

The prevalence of eating and drinking difficulties in individuals with CP is unclear.[11] Estimates range from 27%[12] to 90%,[13] depending on the definitions and measurement tools used. It has been proposed that prevalence is related to the severity of motor impairment,[14] although eating and drinking difficulties also occur in individuals with mildly affected gross motor function.[5, 15]

Although valid and reliable systems are available to classify movement,[16, 17] manual,[18] and communication ability[19] in CP, there is inconsistent use of the measures of eating, drinking, and feeding difficulties.[3, 20, 21] A survey of international CP surveillance registers revealed that, in 2009, 13 out of 21 active CP registries collected eating and drinking data using 11 different measures.[22] The use of a consistent indicator of eating and drinking ability would enable more rigorous investigation of the prevalence of feeding disorders, and of associations between the severity of eating and drinking limitations and other health indicators such as growth, respiratory health, and gastrostomy use.[23]

A recent systematic review examined the psychometric performance and clinical utility of quantitative measures of oropharyngeal dysphagia in children with neurodevelopmental disabilities.[21] The aim of the current review was to examine the evidence for the validity and reliability of descriptive ordinal scales used to classify the eating and drinking ability of individuals with CP. These scales could then potentially be adopted in clinical and population-based research.

Method

Search strategy

A systematic search was conducted using bibliographic databases MEDLINE, EMBASE, CINAHL, PsycINFO, BNI, and AMED. An example of the search strategy used in MEDLINE and modified for other databases is given in Table SI (online supporting information). The searches were conducted up to 14 June 2013. Additionally, forward chasing of citations of key texts (listed in Table SII, online supporting information) was tracked through Web of Knowledge and Scirus; backward chasing (two generations) of references cited in these key texts was also performed.[24]

Inclusion and exclusion criteria

Studies were included in the review if they described an ordinal scale used to classify the eating and drinking ability of people with CP. An ‘ordinal scale’ was identified when descriptions of eating and drinking ability were placed in three or more ordered categories. Ordinal scales were eligible whether they were derived from clinical or technical assessments, surveys, or generic classification systems. Papers were excluded if they were not related to individuals with CP. We excluded scales if an English-language version of the instrument was not reported.

Study selection

Titles and abstracts were screened by one author (DS); full texts of papers that appeared to meet the inclusion criteria were retrieved and reviewed by one reviewer (DS), and 15% were checked independently by a second reviewer (LP). Agreement between reviewers was checked for quality assurance; absolute agreement and chance-corrected agreement (kappa)[25] were calculated. Authors were contacted for further information about the instruments if details were not explicit in the source papers.

Data extraction

Candidate measures were checked to ensure that they met the requirements of ordinal scales.

Each identified measure was classified according to its type:[26] (1) clinical measures – clinician-administered assessments and checklists; (2) technical measures such as the dynamic fluoroscopic imaging of swallowing (videofluoroscopy), electromyography, flexible endoscopic evaluation of swallowing, or measures of respiration; or (3) patient-reported measures such as patient and/or proxy questionnaires.

Appraisal

The validity and reliability of the selected measures were examined using defined quality criteria.[27-29] Evidence of the psychometric properties of scales was eligible for appraisal if results were published in peer-reviewed publications. The scope and content of the included instruments were coded with reference to a core set of identified categories defined within the World Health Organization International Classification of Functioning, Disability and Health (ICF;[30] Table 1).

Table 1. World Health Organization International Classification of Functioning, Disability and Health (ICF) domains and categories relating to eating and drinking
ICF domainCategories
Body structure (s)s320: Structure of mouth including teeth
Body function (b)b510: Ingestion functions. Functions related to taking in and manipulating solids or liquids through the mouth into the body including b5100, sucking; b5101, biting; b5102, chewing; b5103, manipulation of food in the mouth; b5104, salivation; and b5106, regurgitation or vomiting
b5105: Swallowing. Functions of clearing the food and drink through the oral cavity, pharynx, and oesophagus into the stomach at an appropriate rate and speed
b535: Sensations associated with the digestive system
b539: Functions related to the digestive system, including feeding tube
b440: Respiration functions. Functions of inhaling air into the lungs, the exchange of gases between air and blood, and exhaling air, including aspiration
b450: Additional respiratory functions. Additional functions related to breathing, such as coughing, sneezing, and yawning
b126: Temperament and personality functions including attention and awareness. General mental functions of constitutional disposition of the individual to react in a particular way to situations, including the set of mental characteristics that makes the individual distinct from others
b130: Energy and drive functions including appetite. General mental functions of physiological and psychological mechanisms that cause the individual to move towards satisfying specific needs and general goals in a persistent manner
b152: Emotional functions. Specific mental functions related to the feeling and affective components of the processes of the mind
Activity (d)d550: Eating. Carrying out the coordinated tasks and actions of eating food that has been served, bringing it to the mouth and consuming it in culturally acceptable ways, cutting or breaking food into pieces, opening bottles and cans, using eating implements, having meals, feasting, or dining
d560: Drinking. Taking hold of a drink, bringing it to the mouth, and consuming the drink in culturally acceptable ways, mixing, stirring and pouring liquids for drinking, opening bottles and cans, drinking through a straw or drinking running water such as from a tap or a spring; feeding from the breast
d660: Assisting others with their learning, communicating, self-care, movement, within the house or outside; including self-care, movement, nutrition and health
Participation (d)d9205: Socializing. Engaging in informal or casual gatherings with others, such as visiting friends or relatives or meeting informally in public places
Environmental factors (e)e110: Products or substances for personal consumption. Any natural or human-made object or substance gathered, processed, or manufactured for ingestion
e1151: Assistive products and technology for personal use in daily living
e340: Personal care providers and personal assistants
e410: Individual attitudes of immediate family members
e5800: Health services and programmes at a local, community, regional, state, or national level, aimed at delivering interventions to individuals for their physical, psychological, and social well-being

Content validity was considered satisfactory if the purpose of assessment, the target population, and the concepts being measured were clearly identified; content should have been identified with input from the target population as well as experts and investigators. Construct validity was assessed through hypothesis testing with related instruments, using a priori estimations of the direction and magnitude of statistical association.[27] Measures of reliability were examined with reference to kappa[25, 31] and intraclass correlation coefficients (ICCs).[28] Reliability was considered satisfactory for population-based research if the ICC (or weighted kappa) was at least 0.7 in a sample size of at least 50 patients.[27]

Results

The search identified 6299 references after duplicates were removed. After screening, 722 full-text papers were obtained, of which 464 papers were excluded. In total, 254 papers detailed clinical, instrumental, and patient- or proxy-reported measures. From this group, 23 papers describing 15 ordinal-scale measures used to classify the eating and drinking ability of people with CP were identified (Fig. 1). Agreement between reviewers was 98% (kappa=0.95).

Figure 1.

The process of identification of ordinal scales used to classify eating and drinking abilities of people with cerebral palsy.

Overall, 13 of the 15 measures were developed as clinical assessments by health professionals;[3, 6, 8, 13, 20, 32-39] nine of the measures made use of information gathered in interviews or questionnaires from parents of children with CP or developmental disabilities.[3, 6, 20, 33, 35, 37, 38, 40, 41] One study[42] made use of a single item taken from the Pediatric Evaluation of Disability Inventory (PEDI).[41] One scale was developed for use in the context of a CP surveillance programme.[40] Two measures were used in conjunction with videofluoroscopic examinations of swallowing.[8, 39]

All measures were developed for use with children ranging in age from 5 months to 18 years. All participants reported eating and drinking performance, that is what a person actually does in their current environment, rather than capacity, which is their highest level of functioning. The measures reported the range of ability utilizing three to six different categories. The age ranges and constructs covered by each of the measures are provided in Table 2.

Table 2. Ordinal measures used to classify eating and drinking performance for people with cerebral palsy: content, structure, validity, and reliability
Scale/authorScaleUserSampling frameICF categoryContent validityReliability
  1. ICF, International Classification of Functioning, Disability and Health; CP, cerebral palsy; DSS, Dysphagia Severity Score; GMFCS, Gross Motor Function Classification System; PEDI, Pediatric Evaluation of Disability Inventory.

Da Graca Andrada et al.[40] – Portuguese survey of CP in Europe scaleFive-level ordinal scale (I–V) with increasing limitations to performance for each construct within each level. Ability to bring food and drink to the mouth, ability to chew and swallow, and time taken for meals are assumed to all be related; tube feeding occurs in level VSpeech and language therapists, healthcare professionals, and parentsChildren with CP; n=30; spastic bilateral, spastic unilateral, dyskinetic, ataxic, unspecified severity; age 4–7y; on Portuguese CP surveillance registerTotal=6: b440 safety; b510 oral skills; b5105 swallowing; b539 tube feeding; d550/d560 assistance required; e110 food textures; unspecified ‘duration’None reportedn=30 children with CP; kappa=0.43 healthcare professionals vs parents; kappa=0.56 speech and language therapists vs parents; kappa=0.61 speech and language therapists vs healthcare professionals
Calis et al.[32] – DSSFour levels linked to DDS[46] part 2 scores: (1) no dysphagia; (2) mild dysphagia; (3) moderate to severe dysphagia; and (4) profound dysphagia (nil by mouth)Clinician/researcher qualified to use DDSIndividuals with CP; n=166; GMFCS level IV or V; age range 2y 1mo–19y 1mo, mean age 9y 4mo; attending specialized day care, residential, and educational centres in Western region of the NetherlandsTotal=8: b126 awareness; b440 aspiration; b510 loss of food; b510 oral skills; b510 safety of oral feeding; b5105 swallowing; b539 gastro-oesophageal signs; b539 tube feedingNone reportedNone reported
Dahlseng et al.[20] – Norwegian survey of CP in Europe scaleFive-point ordinal scale (1) independent; (2) in need of some assistance; (3) totally dependent upon assistance not tube feeding; (4) partly tube fed; and (5) mainly tube fed. Levels 1 and 2 are children with no ‘feeding problems’; levels 3–5 are children with ‘feeding problems’Paediatric neurologist with parents using registration formChildren with CP; n=661; GMFCS levels I–V; age range 3y 8mo – 10y 5mo; on Norwegian CP surveillance registerTotal=2: b539 tube feeding; d550/d560 assistance requiredNone reportedNone reported
Erkin et al.[33]Four levels defined by food textures managed: normal, mild dysfunction, moderate dysfunction, or severe dysfunction (tube fed)Clinician/researcher from maternal interviewChildren with CP; n=120; GMFCS level I to V; age range 2–18y; attending Ankara Physical Medicine and Rehabilitation Education and Research HospitalTotal=5: s320 structure of the mouth; b510 ingestion ‘feeding difficulty’; b5105 swallowing; b539 tube feeding; e110 food textures and fluid consistenciesNone reportedNone reported
Fung et al.[3] – North American growth project questionnaireFour levels defined by food textures managed: normal, mild, moderate, or severe feeding dysfunctionClinician/researcher from parent questionnaireChildren with CP; n=230; GMFCS level III to V; age range 5–14y; recruited from six centres in the USA and CanadaTotal=6: b440 respiratory illness; b450 coughing, safety; b510 ingestion oral skills, ‘feeding dysfunction’; b5105 swallowing; b539 tube feeding; e110 food textures and fluid consistenciesNone reportedNone reported
Gisel and Alphonce[34] Classification SystemThree levels defined by standard deviation from chewing norms based on length of time between specified food textures entering mouth and first swallow: mild, moderate, or severeClinician/researcherIndividuals with CP; n=100; range of severity of CP; age range 2–16y; from Montreal areaTotal=3: b510 duration of ingestion; b5105 swallowing; e110 food texturesChewing cycle curves developed for typical population in laboratory (n=103, age range 2–8y)None reported
Haley et al.[41] –PEDIFour levels defined by food textures managed: (1) pureed/blended; (2) ground/lumpy; (3) cut up/chunky; and (4) all food texturesClinician/researcher from parent reportChildren with CP; n=170; GMFCS level I–V; age range 1y 6mo–3y; recruited across Queensland and Victoria in AustraliaTotal=2: b510 ingestion; e110 food texturesItem A in self-care domain of PEDI; concurrent validity reported for whole instrument not individual itemReliability reported for Norwegian version of whole instrument, not individual item
Hung et al.[35]Three levels defined by level of assistance: (1) totally dependent upon caretaker; (2) partially dependent (some help required); and (3) totally independent in feedingClinician/researcher from parent reportChildren with CP; n=75; age range 5mo–10y; quadriplegia vs hemi- vs diplegia, wheelchair bound vs community walker; attending rehabilitation centre, Chang Gung Memorial Hospital, Kaohsiung TaiwanTotal=1: d550/d560 level of assistanceNone reportedNone reported
Morton et al.[8]Three levels defined by respiratory illness: (1) no respiratory infection; (2) minor respiratory infections requiring one course of antibiotics; and (3) recurrent respiratory tract infections with two or more courses of antibioticsClinician/researcherChildren with CP; n=26; spastic quadriplegia, unspecified severity; age range 7mo – 16y; attending Derbyshire Children's HospitalTotal=1: b440 respiratory illnessNone reportedNone reported
Reilly et al.[36]Four levels defined: (1) no apparent feeding problem; (2) mild swallowing /feeding difficulty; (3) moderate swallowing or feeding difficulty; and (4) severe swallowing and feeding problemsClinician/researcherChildren with CP; n=30; spastic or athetoid, unspecified severity; age range 4y 4mo –17y 11mo; attending one special school for children with motor disabilityTotal=3: b510 unspecified ‘feeding difficulty’; b5105 swallowing; e110 food textures and fluid consistencyNone reportedNone reported
Reilly et al.[13]Four categories of oral motor dysfunction: none, mild, moderate, and severeClinician/researcherChildren with CP; n=49; range of severity and type of CP; age range 1–6y; from two inner London district health authoritiesTotal=1: b510 ingestion; unspecified ‘feeding dysfunction’None reportedNone reported
Selley et al.[37] – Feeding Difficulty Symptom ScoreFive groups defined: (1) general mealtime difficulties; (2) saliva control problem only; (3) swallowing difficulty; (4) fear of choking or coughing, with or without a saliva control problem; and (5) both swallowing difficulty and fear of choking or coughingClinician/researcher from parent reportChildren with CP; n=117; range of severity and type of CP; age range 1–18y, mean age 8y 7mo; attending Feeding and Swallowing Advisory Centre at Vranch House Special School, ExeterTotal=5: b153 fear of choking; b440 choking; b450 coughing; b510 saliva control; b5105 swallowing; unspecified ‘general mealtime difficulties’None reportedNone reported
Sheppard[38] –Dysphagia Management Staging ScaleFive levels: (1) no symptoms; (2) mild; (3) moderate; (4) severe; and (5) profound swallowing or feeding disorder. Ability to bring food and drink to the mouth is considered separately; number of management strategies used (linked to DDS assessment) determines severity ratingClinician/researcher qualified to use DDS;[46] information taken from parent questionnaireNone reportedTotal=15: b126, b130, and b152 mealtime behaviours; b535 gastro-oesophageal signs; b539 tube feeding; b440 respiratory function; b450 safety; b510 ingestion oral skills; b510 saliva control; b5105 swallowing; d660 nutrition/hydration adequacy; e1151 postural management strategies; e110 food textures and fluid consistencies; e340 special techniques; e5800 management strategiesNone reportedNone reported
Sullivan et al.[6] – Oxford feeding studySix levels defined: (1) always needs help; (2) some difficulty needs help; (3) can feed but slow and messy – help given; (4) tube fed or not fed by mouth; (5) some difficulty, no help; and (6) self-feedsClinician/researcher from parent questionnaireChildren with neurological impairment; n=100; mild, moderate, severe CP; age range 4–13y, mean age 9y SD 2y 5mo; from four counties of UK regionTotal=2: b539 tube feeding; d550/d560 assistance required, duration, and messiness associated with self-feedingNone reportedNone reported
Zerilli et al.[39]Three groups defined: (1) minimal or no risk of aspiration; (2) moderate risk of aspiration; and (3) no oral feeding owing to excessive risk of aspirationClinician/researcherChildren with CP; n=11; unspecified severity or type of CP; age range 11mo–13y; attending C.S. Mott Children's Hospital, Michigan, USATotal=3: b440 aspiration; b5105 safety; e5800 special techniquesNone reportedNone reported

Seven of the measures included information about whether a child is fed via a tube.[3, 6, 20, 32, 33, 38, 40] Seven measures included information about food texture or fluid consistency managed by the child.[3, 33, 34, 36, 38, 40, 41] Seven measures included details about swallowing[3, 32, 33, 36-38, 40] and five scales included information about the oral skills required to bite and chew food.[3, 32, 36, 38, 40] Five scales reported ‘feeding dysfunction’ or ‘difficulties’, although these were not defined.[3, 13, 33, 36, 37] The safety of oral feeding was included in six scales[3, 32, 37-40] with aspiration being specifically noted by two measures.[8, 39] Respiratory function or respiratory illness was included in three measures.[3, 8, 38] Four measures included details about the level of assistance required by the child to eat or drink.[6, 20, 35, 40] Three measures made use of the time taken or the duration in order to define the categories assigned to a child;[6, 34, 40] two of these measures used duration of mealtimes as an indication of severity.[6, 40]

The number of constructs assessed to assign a classification category in the different measures ranged from 1, such as the level of assistance required[35] or number of chest infections,[8] to 15.[38] The content assessed by the included measures were coded with reference to the ICF[30] (Table 1). Table 3 shows the frequency of use of ICF categories across all identified measures. Six of the identified measures included more than one category within each level, creating the potential to assign a child to more than one category.[3, 6, 20, 33, 38, 40] For example, children able to swallow safely and self-feed but who also received some of their nutrition by tube would not be easily categorized. One measure identified the ability to feed oneself and the ability to bite, chew, and swallow safely as distinct skills;[38] one measure combined the level of assistance required with swallowing safety, assuming close correspondence of these constructs with increasing severity.[40] The terms ‘mild’, ‘moderate’, and ‘severe’ were used by eight different scales, each with a different definition.[3, 13, 32-34, 36, 38, 39]

Table 3. Frequency of International Classification of Functioning, Disability and Health (ICF) categories associated with constructs of identified ordinal scales
ICF categoriesTotal
Body structure
s320: Structure of mouth1[33]
Body function
b510: Ingestion functions10[3, 13, 32-34, 36-38, 40, 41]
b5105: Swallowing9[3, 32-34, 36-40]
b535: Sensations associated with digestive system1[38]
b539: Functions related to the digestive system, including feeding tube7[3, 6, 20, 32, 33, 38, 40]
b440: Respiration functions, including aspiration6[3, 8, 32, 38-40]
b450: Additional respiratory functions, including coughing3[3, 37, 38]
b126: Temperament and personality functions, including attention and awareness2[32, 38]
b130: Energy and drive functions, including appetite1[38]
b152: Emotional functions2[37, 38]
Activity
d550: Eating; and d560: Drinking4[6, 20, 35, 40]
d660: Assisting others including nutrition and health1[38]
Participation
d9205: Socializing0
Environment
e110: Products or substances for personal consumption including food textures and fluid consistency7[3, 33, 34, 36, 38, 40, 41]
e1151: Assistive products and technology for personal use in daily living1[38]
e340: Personal care providers and personal assistants1[38]
e410: Individual attitudes of immediate family members0
e5800: Health services including delivering interventions2[38, 39]

The available evidence of the validity and reliability of the different scales is summarized in Table 2. The scale from the North American Growth Project[3] was derived from acknowledged expert sources.[13, 43] The Gisel and Alphonce Classification system[34] explicitly states that the source of content for the ordinal scale is laboratory-based studies. Eating curves were developed for the typically developing population based on the time taken to swallow a specified quantity of three different food textures;[44, 45] the categories of the ordinal scale were based on standard deviations. The validity and reliability of the single item used by Weir et al.[42] taken from the PEDI (self-care domain item A)[41] have not been reported. The content of the Dysphagia Severity Scale[32] was derived from the Dysphagia Disorders Survey,[46] a detailed measure of oropharyngeal dysphagia; correlations between the two instruments were low to moderate (r=−0.18 to 0.46). In the case of the remaining 11 measures, content validity was not reported.[6, 8, 13, 20, 33, 35-40] For only one scale[40] were measures of reliability when used by different observers reported (n=30; kappa values=0.43 – 0.61).

Eleven of these measures were found only in peer-reviewed publications by the research groups who developed the scales. Two[6, 34] of the remaining four scales are cited once each by research teams other than the original developers.[47, 48] Use of the Dysphagia Management Staging Scale[38] is restricted to clinicians who have attended certification workshops; the PEDI[41] has been used widely in research, but more specifically to examine eating and drinking ability of individuals with CP in only four papers.[42, 49-51]

Discussion

Fifteen ordinal scales used to classify the eating and drinking ability of people with CP were identified in this systematic review. Eight of the scales utilized the terms mild, moderate, and severe; however, each measure defined these terms in different ways. The terms ‘feeding problem’ or ‘feeding dysfunction’ also lacked precise definition, and might refer to limitations in the ability to bring food and drink to the mouth, or limitations in the ability to bite, chew, and swallow. There appears to be a dearth of evidence to support the validity and reliability of any of the published scales.

Attempts to provide simple objective measures of the severity of limitations to eating and drinking, such as mealtime duration, have been challenged because of the multidimensional nature of the activity.[52, 53] Some of the included measures use no more than two constructs, such as time taken, food texture, or the ability to self-feed, to clearly define the distinct categories of the ordinal scale. An individual's ability to bite, chew, move food and fluid in the mouth, and swallow will impact on the food textures and fluid consistencies that can be managed and the time taken to eat. By limiting the number of constructs used, scales suitable for use in population studies have been positively applied.[8, 20, 34, 35] However, significant information related to the safety of eating and drinking is omitted from these scales. It may be possible to categorize eating and drinking ability by food textures that can be ‘managed’ by an individual with CP, although the definition of ‘managed’ needs clarification: someone may ‘manage’ chopped food with occasional episodes of choking requiring intervention from others, whilst another individual may eat the same food textures with minimal risk of choking; someone may be able to eat a roughly mashed diet and drink thin fluids but experience regular respiratory illnesses associated with primary aspiration, while someone else may be able to ingest the same diet without respiratory compromise.

When different constructs are combined within distinct categories of an ordinal scale, it is not always possible to make clear distinctions between levels. For example, when the need for assistance in bringing food and drink to the mouth is categorized together with the oral skills required to bite, chew, and swallow, severely limited performance in one construct will mask the classification of relatively unaffected performance in another area. People with CP may require assistance to bring food and drink to the mouth but have the oral skills required to bite, chew, and swallow a full range of foods. The reverse can also be observed, most notably in the case of people with Worster-Drought syndrome, in whom the oral skills required to bite, chew, and swallow safely are limited but who suffer no limitations in bringing food and fluid to the mouth.[15] The relationship between eating and drinking ability and other aspects of function such as gross motor function or hand-to-mouth function cannot be clarified when these functions are combined in the same ordinal scale. In the same way, whilst eating and drinking ability will have an impact upon intake of food and fluid, it may not be helpful to combine eating and drinking ability with the ability to meet nutrition and hydration needs. The nutritional and hydrational requirements of one person with CP will be different from another's, even though they may have similar eating and drinking abilities. Another assumption present in some of the scales is that there will always be an association between the presence of a feeding tube and the greatest limitations to eating and drinking ability.[20, 32, 33, 40]

The authors of the measures take different viewpoints on the question of who is best placed to report a child's eating and drinking ability across a range of foods and environmental settings: six measures[8, 13, 32, 34, 36, 39] are based on information available to healthcare professionals or investigators, whilst the remaining nine measures make use of information from parents; only two have the potential to be used by both parents and healthcare professionals.[40, 41] The most comprehensive measure is derived from a clinical assessment tool for use only by clinicians who have attended certification workshops.[38] Disagreements between parent judgements about aspects of their children's eating and drinking ability and judgements made by healthcare professionals have been explicitly identified at the level of mealtime duration, eating and drinking difficulty, and food texture safety.13,32 Concerns about the potential discrepancy between parents' judgements about their children's eating and drinking ability expressed through questionnaires and judgements made by specially trained healthcare professionals have been acknowledged.[3, 5] Categorization of eating and drinking ability on the basis of questionnaire data collected from parents was successfully applied in two large-scale population studies.[3, 6] None of the identified measures allow for a direct comparison between ratings made by people themselves or by parents and professionals.

Direct observation of eating and drinking is limited because much activity takes place within the oral cavity, pharynx, and larynx, out of view. Inferences about eating and drinking ability can be drawn from observation of subtle clinical signs.[54] Potentially harmful limitations to eating and drinking leading to aspiration of food and fluid into the lungs, visible through instrumental means such as videofluoroscopic examination of swallowing, have been documented.[1, 9, 55] ‘Silent aspiration’, aspiration that takes place without the usual outward signs such as coughing, has also been documented.[8-10] Only six of the scales refer to aspiration or the consequences of aspiration presenting as respiratory illness;[3, 8, 32, 38-40] two of these scales have been developed for use in the context of videofluoroscopy.[8, 39]

There are limitations to this systematic review in that only those scales that had been produced or translated into English were included. Evidence of measures of reliability and validity of the included scales was considered only if it had been published in peer-reviewed studies. Scales used within CP surveillance registers[22] were included only if published. Scales assessing emotional and behavioural disturbances to eating and drinking function occurring in the paediatric population were not included. There may be unpublished data regarding validity and reliability of scales, and contact with the developers of the Dysphagia Disorders Survey[46] suggests that this may be the case. However, we preferred to include data only from peer-reviewed publications so that an appraisal of the methodological quality of those studies could be examined. We considered using the COSMIN checklist to appraise methodological quality of studies examining validity and reliability,[56] but data emerging in the review were too limited to warrant the approach.

Conclusion

None of the scales identified in this review can be considered as valid and reliable ordinal classification systems of eating and drinking ability for people with CP. This systematic review supports the development of a new system to classify eating and drinking through rigorous and robust methods, as have been applied to other classification systems.[16-19] The Gross Motor Function Classification System and other functional classification systems have had a profound impact upon research and clinical practice.[16-18] A valid and reliable classification system of eating and drinking ability would enable more rigorous investigation of prevalence and an exploration of associations between limitations to eating and drinking ability and other health-related concerns. Development of classification systems and measures should include consultation with potential users to ensure content validity. The adoption of such a system by clinical and research communities would be enhanced by attention being paid to levels of consensus about its content and the reliability of its use. A classification system of eating and drinking for use by parents and healthcare professionals would have the potential to facilitate working in partnership[57] and facilitate more robust clinical and population-based research.

Acknowledgements

This article presents independent research commissioned by the National Institute for Health Research (NIHR) under the Research for Patient Benefit programme. The views expressed in this publication are those of the authors and not necessarily those of the NHS, the NIHR, or the Department of Health. This systematic review and protocol was registered on the Prospero database (http://www.crd.york.ac.uk/NIHR_PROSPERO registration no. CRD42013003701). Particular thanks are given to Sarah Ford, Sarah Butler, and staff at the Library and Knowledge Service, Brighton and Sussex University Hospitals NHS Trust.

Ancillary