Table 2.

Summary of studies on response format

Author (date); Country; Study Design; Measurement properties evaluated	Objective	Sample size (N); Setting; Age; Population	Main findings	QualSyst Score (%)
Baxter (2011) [16]; USA; Quantitative; Prospective; Feasibility, construct validity, responsiveness	To create and validate a pictorial scale with regular incremental levels between scores depicting increasing nausea intensity (BARF scale)	N = 127; Hospital; 7–18 years; Emergency department and surgery	The Spearman correlation coefficient of the first paired BARF and VAS for nausea scores was 0.93. VAS for nausea and BARF scores (P = .20) were significantly higher in patients requiring antiemetic agents and decreased significantly after treatment, while posttreatment pain scores (P = .47) for patients receiving only antiemetic agents did not. All patients understood the pictorial faces scales	68
Benson (2016) [17]; UK; Quantitative; Prospective; Construct validity	To test items, identified through previous qualitative interviews, that might form the basis of a new Malocclusion Impact Questionnaire for young people	N = 184; Hospital; 10–16 years ; Dental outpatients	Using Rasch analysis it was shown that all but one item had disordered thresholds, indicating response categories were not functioning as expected. The original 5-point response scale format was reduced to 3 points	60
Berntson (2001) [18]; Sweden; Quantitative; Cross-sectional; Acceptability, construct validity reliability	To evaluate the concordance between pain assessments made on a VASa and a 4-point verbal descriptor scale and establish scale preference	N = 12; Hospital; 10–18 years; juvenile arthritis	Slight pain on verbal scale corresponded to a wide interval of 7–65 on VAS suggesting VAS was difficult to interpret. Preference was for VAS (69%) but this did not show the most reliable results	68
Borgers (2003) [19]; Netherlands; Quantitative; Prospective; Feasibility, reliability	To investigate the effects of partially labelled response options and vague quantifiers in response stability compared to completely labelled response options and the use of clearly quantified words in children of different ages	N = 91; Home; 8–16 years; Healthy	No effect on stability over time was found when offering vague quantifiers in the response options (p > 0.05). Young children do not benefit from the extra information of completely labelled response options. Offering different types of response option can lead to substantially different structural models	75
Borgers (2004) [20]; Netherlands; Quantitative; Prospective Feasibility, reliability	To examine the effects of negatively formulated questions, number of response options and offering a neutral midpoint as response option question characteristics on the reliability of responses	N = 222; Home; 8–16 years; Healthy;	Negatively formulated questions had no effect on reliability, although children respond consistently differently to negatively formulated questions as opposed to positively formulated ones. Offering about 4 response options is optimal (reliability increased up to 6, more than 7 caused a decrease)	80
Campbell (2011) [21]; Canada; Quantitative; Cross-sectional; Feasibility	To investigate the utility of a VASa to measure peer conflict resolution knowledge in children with language impairment (LI) and typically developing peers (TLD). Are children with varying language status able to express nuances in social knowledge by marking responses along the full VAS	N = 26; School; 9–12 years; Healthy	Those with TLD used the whole VAS; most (83%) with LI relied solely on scale anchors	59
Castarlenas (2013) [22]; Spain; Quantitative; Cross-sectional; Acceptability, construct validity	To assess whether the NRS-11b is a valid tool with 6–8 year old children when presented verbally	N = 126; School; 6–8 years; Healthy	The NRS-11 showed high convergent construct validity with VASa, FPS-Rc and CASd (r = 0.73–0.86), adequate discriminant validity (z = 2.05–5.55) and adequate criterion validity (r = 0.45–0.70). Preference order = CAS > NRS > FPS-R > VAS	75
Chambers (1998) [23]; Canada; Quantitative; Cross-sectional; Feasibility	To examine the potentially biasing impact of neutral or smiling face as a no pain anchor on children’s reports of pain in response to a series of vignettes	N = 100; Childcare centres; 5–12 years; Healthy	Children who use a smile anchored scale had significantly higher pain scores for no pain and pain negative emotions (p < 0.001) and significantly lower faces pain scores for pain/positive vignettes than children who use a neutral anchored face scale (p < 0.001). Faces scales that use smiling anchors may confound affective states with pain ratings	63
Chambers (1999) [24]; Canada; Quantitative; Cross-sectional; Acceptability, feasibility	To examine the potential for bias in children’s’ self-report of pain when using scales with smiling rather than neutral anchors and to establish preference of type of faces scale	N = 75; Hospital; 5–12 years ; Endocrine/diabetes	Scores across scales were highly correlated (r = 0.81–0.93). There was no age or gender interaction effect. Pain was rated significantly higher when scales with a smiling, rather than neutral, anchor were used (p = 0.001). 52.1% of children preferred scales they perceived to be happy and cartoon-like	75
Chambers (2005) [25]; Canada; Quantitative; Cross-sectional; Acceptability, feasibility	To determine whether scales beginning with a smiling rather than neutral “no pain” face would produce higher ratings in the assessment of postoperative pain intensity in children and to compare ratings using different faces. Preference also asked	N = 78; Hospital; 5–13 years ; Post-surgical	Children’s ratings of postoperative pain intensity are influenced by the presence of smiling “no pain” face at the beginning of faces scales, with such scales producing significantly higher ratings than scales with neutral “no pain” faces (p < 0.01). Ratings on the independent CASd measure were more comparable to those provided on faces scales with neutral “no pain” faces. 55.6% preferred Wong Baker faces scale despite it giving the highest pain scores	83
Decruynaere (2009) [26]; Belgium; Quantitative; Cross-sectional; Construct validity, feasibility	To examine with the rating scale model how a sample of healthy children from 4–7 distinguish different faces when rating imaginary painful situations	N = 121/76; Schools and sports centres; 4–7 years ; General	Children performed better on a 3-point faces scale than 6-point scale. Ability improves with age on a 3-point faces scale. 4–5-year-olds could only distinguish 2 response categories	70
Emmott (2017) [27]; Canada; Quantitative; Cross-sectional; Construct validity, feasibility	To evaluate validity and feasibility of 2 simplified pain scales—S-FPS and S-COS in pre-school age children	N = 180; Hospital; 3–6 years ; Venepuncture	The ability to discriminate pain vs no pain was improved with S-FPSd and S-COSf (p = 0.858) compared with FPS-Rc (p = 0.036 with S-FPS and p = 0.022 with C-COS) within 4–6-year olds but not 3-year olds. Quantitative pain rating remains challenging for 3-year-olds	88
Fanciullo (2007) [28]; USA; Quantitative; Cross-sectional; Acceptability, construct validity, feasibility	To determine initial psychometric properties and feasibility of a new Computer Face Scale for measuring children’s pain	N = 54; Hospital; 3–17 years ; Hospitalised in pain/healthy	76% of children from3 years preferred moveable online faces to select their degree of pain over paper and pen static faces. Paired t tests showed significantly more hospitalised children reported pain than non-hospitalised (p < 0.001). Correlation with Wong-Baker faces scale r = -0.72	75
Fritz (1994) [29]; USA; Quantitative; Prospective; Feasibility	To determine whether the use of pictorial anchors improved a VASa designed to assess asthma episodes	N = 77; Summer camp; 8–15 years; Asthma	The mean VAS scores increased by 64% using the pictorial VAS while the mean PEFRsg in the 2 years were almost identical, suggesting that changes on the VAS were not due to differences in pulmonary functioning. For boys, the increase in individual mean VAS score in year 2 using the pictorial VAS was 44%; for girls, the increase in individual mean VAS scores was 112%. Use of a pictorial anchor led to greater usage of the full range of the scale	77
Gharaibeh (2002) [30]; Jordan; Quantitative; Cross-sectional; Acceptability, construct validity, reliability	To test the reliability and cultural validity of the following three pain assessment scales: Faces Scale, the Word Description Scale, and the Poker Chip Scale	N = 95; Hospital; 3–14 years; Venepuncture	55.8% of children preferred the Poker Chip Tool to the Faces Scale and the Word Descriptive Scale. There was significant convergent validity (p < 0.01) and test–retest reliability (p < 0.01) between the three scales	60
Goodenough (1997) [31]; Australia; Quantitative; Cross sectional; Construct validity, feasibility	To compare the utility of the Faces Pain Scale with three other self-report measures (VASa, Poker Chip, VRSh) of pain severity. These four scales were compared and contrasted in terms of the facility of application and comprehension by young children and their relative response frequency distributions	N = 50; Hospital; 4–7 years; Immunisation	Scores on all 4 scales correlated well (r > 0.7). The scales seemed to be measuring the same construct of pain. The faces scale was well understood. 12% had difficulty with the Visual analogue toy scale. The faces pain scale was skewed to low possibly because there are too many response options for the age group causing them to choose the extreme options	60
Gulur (2009) [58]; USA; Quantitative; Prospective; Acceptability, construct validity, reliability	1) to determine whether children understood the link between the facial expressions of smiling and frowning and the subjective feelings of happiness and pain/hurt. 2) to determine whether children understood that relative degrees of smiling or frowning were linked to relative degrees of happiness and pain/ hurt. 3) to determine the concurrent validity of the Computer Face Scale with the Wong-Baker Faces Scale. 4) to determine the test–retest reliability of the Computer Face Scale	N = 79/50; Hospital; 3–17 years; Study 1 Surgical; Study 2 general inpatients	The computerised scale showed concurrent validity with Wong-Baker faces (r = -0.68). 15-min test–retest reliability r = 0.77. 77% preferred the computerised faces scale. Participants were able to use both scales	45
Gupta (2016) [96]; USA; Qualitative; Cross-sectional; Acceptability, feasibility	To evaluate comprehension and usability of a modified electronic version of the electronic version of the FPS-Rc for children aged 4–17 years with sickle cell disease	N = 22; Unclear; 4–17 years; Sickle cell	Children age 4–6 years were generally unable to demonstrate understanding of the FPS-R and its response scale. Children > 7 years understood the scale and could complete it electronically. Those aged 7–8 years often needed parental assistance	55
Hicks (2001) [32]; Canada; Quantitative; Cross-sectional; Construct validity	1) to revise the FPSi from 7 faces to 6 to make scores comparable to other measures (0–5 or 0–10). 2) to evaluate the validity of the revised version	N = 76/45; Ear piercing/Hospital; 4–12 years; Healthy/hospital	The validity of the revised scale is supported by a strong positive correlation (r = 0.93) with a VASa measure in healthy children aged 5–12 years. In hospitalised children the revised scale correlated with the VAS (r = 0.92) and CASd (r = 0.84)	60
Hunter (2000) [33]; Australia; Quantitative; Cross-sectional, Feasibility	To further investigate the psychometric properties of the faces pain scale	N = 135; School; 3.5–6.5 years; Healthy	All children were capable of making meaningful discriminations. Children had difficulties with the middle of the scale suggesting that it formed an acceptable series but could not be considered an interval scale. The scale is best reserved for school age children	50
Irwin (2009) [95]; USA; Qualitative; Cross-sectional; Feasibility	To conduct cognitive interviews with children and adolescents to gain feedback on items measuring physical functioning, emotional health, social health, fatigue, pain and asthma specific symptoms for PROMIS item bank	N = 77; Hospital/community; 8–17 years; Healthy/asthma	Response options were understood by the majority of participants (up to 5 options). Children could clearly identify variable levels of functioning. Younger children misunderstood the word difficulty, so it was changed to trouble	65
Joffer (2016) [34]; Sweden; Qualitative; Cross-sectional; Feasibility	To explore how adolescents interpret and reason when answering a question on self-rated health	N = 58; School; 12–18 years; Healthy	Participants’ understandings of the response alternative “Neither good, nor bad” varied. Some regarded it as normal and “in the middle”, some as a negative alternative, and others as a passive state. The five response options all demonstrated differences in self-rated health	60
Jung (2018) [35]; Korea; Quantitative; Prospective; Construct validity	To develop and validate the “Pain Block” concrete ordinal scale for 4- to 7-year-old children. Psychometric properties were compared with the FPS-Rc	N = 163; Hospital; 4–7 years; Emergency dept	Agreement between the 2 pain scales was acceptable, with 95% of the values within the predetermined limit (r = 0.82). The pain scores for both pain scales were significantly decreased when analgesics or pain-relieving procedures were administered (p < 0.001). The Pain Block pain scale could be used to assess pain in 4- to 7-year-old children capable of understanding and counting up to the number 5, even if they do not understand the FPS-R pain scale	68
Keck (1996) [36]; USA; Quantitative; Prospective; Acceptability, construct validity, reliability	To investigate the Faces and modified Word Descriptor Scale for concurrent validity, discriminant validity and test retest reliability	N = 118; Hospital; 3–18 years; Haematology and oncology; venepuncture	Both the word descriptor and faces scales demonstrated discriminant validity (p < 0.001 for scores before and after painful procedure) and concurrent validity (r > 0.71) and test–retest reliability (faces r = 0.9 and word scale r = 0.92). All children understood the scales. The majority of children in all age groups preferred the faces scale (65.1%)	50
Klassen (2015) [60]; Canada; Mixed methods; Cross-sectional; Acceptability, construct validity, feasibility, reliability,	1) to conduct individual cognitive interviews with adolescents age 12–18 with different health conditions to obtain their feedback on the instructions, response options and items of a transition questionnaire (Transition-Q) with a 5-point Likert response option and to identify any missing content and to revise the scale as necessary. 2) conduct a large-scale field test to examine reliability and validity	N = 32/37; Hospital; 12–18 years; Chronic conditions	Item response option thresholds weren’t ordered for 13 of 18 items. Items were rescored in to 3 response options. 14 participants did not like the agree/disagree response format. It was changed to frequency (never, sometimes, often and always). This was preferred by 8/9 in the second round. Cronbach’s ⍺ = 0.85. Test–retest reliability = 0.9	90 (quant) 55 (qual)
Lawford (2001) [37]; UK; Quantitative; Cross-sectional; Feasibility, reliability	To provide an empirical basis for selecting the response format of a QOL measure for 3–8-year olds (4 point Likert scale vs 4 coloured circles)	N = 28; Nursery school; 4–5 years; Healthy	The Likert scale took significantly longer to complete (p < 0.005). The coloured circle format had higher internal consistency than the Likert scale (⍺ = 0.7 vs 0.48)	65
Leske (2015) [38]; USA; Quantitative; Cross-sectional; Construct validity	To use Rasch analysis to refine the Intermittent Exotropia Questionnaire, removing items that do not contribute meaningful information and ensure response options are properly interpreted	N = 575; Eye clinics; 8–17 years; Intermittent exotropia	Performance of the child and adult versions were enhanced by reducing the number of response options from 5 to 3	80
Locker (2007) [39]; Canada; Quantitative; Cross-sectional; Construct validity	To assess the performance of negatively and positively worded items in questionnaires to measure child perceptions of child oral health-related quality of life	N = 91; Dental clinics; 10–14 years; Dental/oro-facial	Positively worded items elicited significantly more ‘don’t know’ responses and missing values. The performance of positively worded items was unsatisfactory	85
Luffy (2003) [57]; USA; Quantitative; Cross-sectional; Acceptability, construct validity, reliability	To compare the validity, reliability and preference of pain intensity measurement tools—the African American Oucher scale, Wong-Baker Faces scale and VASa	N = 100; Outpatient clinics; 3–18 years; Sickle cell	Faces and African American Oucher are valid (no significant difference in scores between Oucher and Wong-Baker faces) and reliable (test–retest p < 0.005) tools for measuring pain in children. The VAS was not. 56% preferred the faces scale	50
Maïano (2009) [40]; France; Quantitative; Cross-sectional; Construct validity, reliability	To test the factor validity and reliability of 2 versions (graphical scale vs Likert scale) of the Very Short Form of the Physical Self-Inventory (PSI-VSF), with a sample of adolescents with mild to moderate intellectual disability	N = 342; School; 12–18 years; Learning difficulties	Both versions showed good structural validity, with the graphical version being superior. The graphical faces scale version had higher internal consistency (⍺ = 0.7–0.74 vs 0.65–0.67) than the Likert scale	80
McGrath (1996) [55]; Canada; Quantitative; Cross-sectional; Construct validity, feasibility	To determine the validity of the CASd as a pain measure for children by evaluating the psychometric properties of the scale and comparing them to the properties of the VASa	N = 104; 5–16 years; Routine check-up/pain clinics; Healthy/recurrent headache	There was no significant difference in pain scores between the VAS and CAS for the same event. Higher mean scores were reported for severe tissue damage injuries such as broken bones than for minor bruises. 87% found the CAS very easy to use whereas 22% found the VAS easy to use	70
Miro (2004) [41]; Spain; Quantitative; Prospective; Acceptability, construct validity, reliability	1) determine the initial psychometric properties of the Catalan version of the FPS-Rc 2) compare patients’ opinion of the FPS-R with the CASd	N = 371; Hospital/school; 7–15 years; Hospitalised/healthy	Correlations between the FPS-R and CAS ranged from r = 0.83–0.9. Relationship between pain and affective state r = 0.32. Test–retest ranged from r = 0.26–0.7. The proportion of children that preferred the FPS-R was significantly higher than the proportion that preferred the CAS (66–68%)	46
Morley (2014) [42]; Canada; Qualitative—cognitive interviews; Cross-sectional; Feasibility, acceptability	To cognitively test the Pediatric Advanced Care Quality of Life Scale (PAC-QoL) to establish whether the items and response options were understood	N = 34; Hospital; 8–18 years; Oncology;	Response scale was accurately interpreted in 88–93% of cases. When participants had trouble distinguishing between responses it involved options in the middle of the 4-point scale (sometimes and often)	65
O’Sullivan (2014) [43]; Canada; Qualitative; Cross-sectional; Feasibility	To evaluate and refine a new instrument for cancer symptom screening (SSPedi), including evaluating understanding of the response scale	N = 30; Hospital; 8–18 years; Oncology	Response options (5-point Likert) were understood by 90% of children	60
Ogden (2008) [44]; UK; Mixed methods; Cross-sectional; Acceptability, feasibility	To identify changes needed to adapt the IMPACT instrument for use in British children with inflammatory bowel disease and to see whether children preferred the Likert scale or the VASa	N = 20; Outpatients; 8–16 years; Gastroenterology	Participants distinguished between the responses in the Likert scale and related their answers to the response options proficiently. Some children only guessed that ‘moderate’ meant ‘in the middle’ because of its position in the scale (5 point). 75% preferred the Likert scale to the VAS as it was easier and quicker to complete (p < 0.01)	55 quant 45 qual
Okanda (2010) [45]; Japan; Quantitative; Cross-sectional; Feasibility	To investigate whether 3–6-year-old children exhibit a ‘yes’ bias to various yes–no questions and whether their knowledge status affects the production of a yes bias	N = 135; Kindergarten/ nursery; 3–6 years; Healthy	3-year-olds had a strong tendency to exhibit a yes bias to both preference-object and knowledge object yes–no questions (even though they know the answer, p < 0.01). 4-year-olds could appropriately answer preference questions but showed a yes bias to knowledge questions (p < 0.1). 5- and 6-year-olds did not show a response bias to yes questions but showed a weak tendency to say yes to knowledge questions regarding familiar objects	55
Ortqvist (2012) [46]; USA; Qualitative; Cross-sectional; Feasibility	To examine how well the Knee Injury Osteoarthritis Outcome Score for Children (KOOS-Child) is understood	N = 34; Outpatient clinics; 10–16 years; Knee injury	Most children understood how to use a 5-point Likert response scale. The response option ‘moderate’ was persistently perceived as confusing. Most could interpret the meaning of the word by its location in the scale but could not define the word and suggested replacing it with the word ‘some’	70
Pagé (2012) [56]; Canada; Quantitative; Prospective; Acceptability, construct validity, feasibility	To evaluate the convergent and discriminant validity of the NRSb for pain intensity and unpleasantness in children after surgery	N = 83/69; Hospital; 8–18 years; Orthopaedic/general surgery	The NRS correlated highly with the VRSh and FPS-Rc (p < 0.001). Scores were significantly higher at 48–72 h post-surgery than at 2 weeks (p < 0.001). Children found the faces scale easiest to use (51%). The VRS was least liked (13%) and hardest to use	82
Rebok (2001) [92]; USA; Qualitative—cognitive interviews; Cross-sectional; Acceptability, feasibility	(1) to determine whether children can answer health survey items. (2) to test the feasibility of a pictorial questionnaire format using cartoon drawings of a child. (3) to examine several types and numbers of response formats to see which are preferred and most easily understood. (4) to test children’s understanding of specific concepts of health and wording of different response formats	N = 114; School/kindergarten; 5–11 years; Healthy	74% preferred circle responses to VASa, with 68% preferring graduated circles. 74% preferred 4 rather than 3 circles. 100% preferred a horizontal presentation. Younger children gave a significantly higher number of extreme responses. Younger children effectively reduced a 5-point response format to 3 points by using only the middle and extremes. 67% preferred the 5-point response format (rather than 4 point)	70
Shields (2003) [47]; USA; Quantitative; Cross-sectional; Feasibility	To identify demographic and cognitive variables that would maximise the accuracy of predicting children’s abilities to use a VASa	N = 40; Kindergarten; 5–7 years; Healthy	Only 42% of participants could use a VAS. Cognitive ability (IQ ≥ 100) combined with chronological age (≥ 5.6 years) was the best predictor of accurate use	80
Shields (2005) [48]; USA; Quantitative; Cross-sectional; Feasibility	To determine whether age, combined with estimated IQ, is an accurate predictor of a child’s successful use of a VASa in a non-clinical situation vs an acute, clinically emergent situation	N = 104; Hospital; 5–11 years; Healthy/lacerations	Estimated IQ and the ability to do a seriation task were the best predictors of 5–6-year-olds ability to accurately use the VAS (p < 0.001). Estimated IQ was not as important as chronological age and ability to perform a seriation task in those 7 years and over	83
Stanford (2006) [49]; Canada; Quantitative; Cross-sectional; Feasibility	To examine variations in 3- to 6-year-old children’s ability to accurately use a common self-report scale to rate pain in hypothetical vignettes (faces pain scale revised)	N = 112; Community; 3–6 years; Healthy	5- and 6-year-old children were significantly more accurate (40% errors) in their use of the FPS-Rc in response to the vignettes than 4-year-old children, who in turn were significantly more accurate than 3-year-old children (60% errors). Over half of 6-year-olds demonstrated difficulty using the FPS-R in response to the vignettes. Child age was the only significant predictor of children’s ability to use the scale in response to the vignettes (p < 0.001). The ability to use the scale improved with age	65
Staphorst (2017) [50]; Netherlands; Mixed methods; Cross-sectional; Acceptability, construct validity feasibility	To develop a generic, short and child-friendly instrument: the DISCO-RC questionnaire (DISCOmfort in Research with Children)	N = 46; Outpatients; 6–18 years; Unclear	Children preferred a 5-point Likert scale as a response option. The 5-point Likert scale coloured numeric VASa and simple VAS were strongly correlated (r = 0.76 – 0.99)	60 (quant) 65(qual)
Tesler (1991) [51]; USA; Quantitative; Cross sectional; Acceptability, construct validity, reliability, responsiveness	A program of studies designed to select and test a pain intensity scale for inclusion in a multidimensional pain assessment tool for children, focusing on determining each scale’s reliability, validity, ease of use and preference.5 scales were tested: a word graphic scale. VASa, graded graphic rating scale, 0–10 magnitude estimation scale and CASd	N = 1223; Hospital, outpatient, school; 8–18 years; Acute/healthy/chronic illness	Convergent validity for the 5 scales was supported (r = 0.66–0.84). The word graphic rating scale (Likert) was preferred by 47% of sick children. When used in a multidimensional pain assessment tool it showed test–retest reliability (r = 0.68–0.97) also showed sensitivity to change (p = 0.002)	65
Tomlinson (2019) [93]; Canada; Qualitative; Cross-sectional; Feasibility	To develop a new self-report symptom screening tool for children receiving cancer treatments who are 4–7 years of age (mini-SSPedi), based on SSPedi	N = 100; Hospital; 4–7 years; Oncology	Dichotomous response scale (yes/no) was understood by all participants. 80% understood the Wong-Baker faces, 70% understood FPS-Rc and 65% understood the pieces of hurt scales	60
van Laerhoven[59] (2004); Netherlands; Quantitative; Cross-sectional; Acceptability, feasibility	To examine which response options children prefer and which they find easiest to use (VASa vs Likert). To examine the relative reliability of the different response options	N = 122; Outpatients; 6–12 years; Not specified	Children preferred the Likert scale. They considered the Likert scale easiest to fill out. Results of the different response options correlated strongly with each other (r = 0.67– 0.90)	59
von Baeyer (2013) [52]; Canada; Quantitative; Cross sectional; Feasibility	To evaluate a binary question followed by simple response options for pain assessment in young children (FPS-R)	N = 184; Preschool/day care; 3–5 years; Healthy	3- and 4-year-olds performed significantly better using the simplified task than the FPS-Rc (p < 0.001). The simplified pain task made no difference to the 5-year olds who had almost identical mean scores using both methods. Response bias is common in children under 5	68
Vreeman (2014) [94]; Kenya; Qualitative—cognitive interviews; Cross-sectional; Acceptability, feasibility	To improve the understandability of paediatric antiretroviral adherence measurement items through cognitive interviewing with paediatric caregivers and HIV-infected adolescents	N = 10; HIV clinic; 13–18 years; HIV	Participants inconsistently quantified the differences between 4-point Likert response options. Visual analogue scales and the addition of a response option to give 5-points yielded more divergence and were considered hard to understand. It was suggested that VASa would require pictorial cues to orientate the participant to scale meaning	70
Watson (2006) [53]; USA; Quantitative; Cross-sectional; Feasibility	To evaluate the psychometric properties of the fruit and vegetable self-efficacy (FVSEQ) questionnaire	N = 1477; School; 9–10 years; General	Item response modelling showed that the 5-point response scale was not fully utilised	86
West (1994) [54]; USA; Quantitative; Cross-sectional; Feasibility, construct and convergent validity	To identify a clinically feasible and accurate method of measuring pain intensity in paediatric oncology patients in the ITU (FPS and Poker chip)	N = 30; Intensive care; 5–13 years; Oncology	Pain rating scales on the two tools were correlated (faces pain scale and Poker Chip, r = 0.67). 91.6% preferred the faces pain scale to the poker chip tool	50