Proxy-report in individuals with intellectual disability: A scoping review

Abstract

Background:

Views can be collected from individuals (self-report) or others on their behalf (proxy-report).

Objective:

We aimed to review the literature on methods and statistical approaches used to evaluate observer versus self-report responses from individuals with intellectual disability or Down syndrome.

Methods:

A series of key questions related to statistical approaches and data collection methods were formulated a priori to inform the search strategy and review process. These addressed the topics of self-report in individuals with intellectual disability, including Down syndrome. Using the National Library of Medicine database, PubMed, detailed literature searches were performed. The quality of available evidence was then evaluated, the existing literature was summarised, and knowledge gaps and research needs were identified.

Results:

Fifty relevant original articles were identified which addressed at least one key question. Study details, including: research design, internal validity, external validity, and relevant results are presented. Review of studies of individuals with intellectual disability which used a variety of statistical approaches showed mixed agreement between self-report and proxy-report.

Conclusion:

Few studies identified to-date have used self-report from individuals with Down syndrome, but lessons from the existing intellectual disability literature can guide researchers to incorporate self-report from individuals with Down syndrome in the future.

1 |. INTRODUCTION

General population health is measurable and trackable through patient-reported outcome measures; some patient-reported outcomes may use proxy-report (Cella et al., 2007; Koh et al., 2014). Proxy-report refers to responses provided by a respondent, such as a caregiver or parent, about a member of the population or sample of interest. National and population surveys rely on a single respondent’s answers on behalf of a household (Gallagher & Mechanic, 1996; LaVange et al., 1996). In specific populations, such as adolescents, the topic of proxy-report may be pertinent (Shapira et al., 2021). Children and adolescents are typically considered able to provide reliable responses to surveys by age 8–11 (Erhart et al., 2009; Ravens-Sieberer et al., 2010). Further, proxy report is often used when the target is unable to provide a response, such as in patients with dementia, and proxy answers on another’s behalf (Bangma et al., 2021).

In specific situations using survey for data collection, it is acceptable to use proxy report rather than self-report. To evaluate the validity of a proxy’s report, the reliability of a measure based on proxy report rather than self-report necessitates evaluation not only of degree of association but also degree of agreement (Tinsley & Brown, 2000). Association, the degree to which the ratings of different judges are proportional as expressed as deviation from their means, and agreement, the extent to which different judges tend to make the same judgement about a rated subject, differ. Different statistical tests are used to evaluate association or agreement, with the goal of establishing a level of evidence to support the substitution of a proxy measure for a self-reported measure (McGraw & Wong, 1996; Shrout & Fleiss, 1979).

Ideally, with increasing inclusion of individuals with intellectual disability, outcome measures would include the voices of patients with intellectual disability through self-report. Outcome measures may require modifications when self-report is provided by individuals with intellectual disability compared to peers without intellectual disability. These modifications may be attributable to differences in the four major cognitive processes involved in response formation: question interpretation, information retrieval, judgement formation, response formatting and response editing (Sudman et al., 1996). Research on the use of self-report in the field of intellectual disability exists, although the majority of prior work is focused on the evaluation of the reliability of self-report by adults with intellectual disability, with little focused on adolescents with intellectual disability (Mindham & Espie, 2003; Nicolaidis et al., 2020). When surveying adolescents with intellectual disability considerations might include: the reliability in self-report data due to differences in memory or recall which is part of the information retrieval process of response formation, comprehension or understanding of questions dependent on reading level which is part of the judgement formation, and ability to recall detailed or distant events which is involved in response formatting and editing. These issues pose a threat to the reliability and validity and subsequent inference made based on self-report.

In particular, one syndrome with varying degrees of intellectual disability is Down syndrome. Present in one in 800 infants born in the United States, Down syndrome is associated with a unique medical and developmental profile with a range on intellectual disability, and wide variation in timing of developmental milestones, and attainment of activities of daily living (de Graaf et al., 2015; de Graaf et al., 2017). Individuals with Down syndrome have varied adaptive function and independence, and communication; many individuals with Down syndrome are able to write his / her own name without difficulty (76%) yet a smaller proportion (34%) demonstrate the ability to communicate in writing without difficulty; many understand verbal communication without difficulty (58%) and fewer using verbal communication without difficulty (45%) (Matthews et al., 2018). Proxy-report from parents of 10–12 year old children with Down syndrome surveyed found 53% of the children write at least somewhat well, 32% write at least reasonably well, and 3% very well; in young adulthood (21–30 years of age), around 40% write at least reasonably well, around 60% at least somewhat well, and 8% very well (de Graaf et al., 2018). Using these communication and writing skills, adolescents with Down syndrome have expressed their views about their lives, who they are, and how they look (Skotko et al., 2011). While adults with Down syndrome have reliably rated their health-related quality of life in one study (Graves et al., 2016), many studies have relied on proxy-report from parents of children with Down syndrome (Becker & Dusing, 2010).

As researchers studying Down syndrome, we began this project to ask the question: how have individuals with Down syndrome directly reported their health views? And, how might those views compare to others’ perceptions of health? We aimed to conduct a literature review to identify studies of self-report from adolescents with Down syndrome, and to search the broader literature studying individuals with intellectual disability which might be relevant to research on Down syndrome. Although patient-reported outcomes are commonplace in the general population, we wanted to understand the state of the evidence for individuals with intellectual disability and Down syndrome. The purpose of this review was to identify studies which compare data collection methods used in the evaluation of proxy versus self-report and to identify the statistical approaches that are used to evaluate for agreement between reporters. The overarching steps to be completed in this review were to: (1) formulate key questions a priori; (2) search PubMed to identify original research articles that address the key questions; (3) appraise these articles using United States Preventative Services Task Force (USPTF) criteria for evaluating the internal validity, external validity of individual studies (United States Preventative Services Task Force, 2015); (4) summarise the published literature and identify gaps in the evidence and need for future research.

2 |. METHODS

2.1 |. Key questions

We replicated the methodology from our previous work (Capone et al., 2018; Santoro et al., 2020; Santoro & Steffensen, 2021), and began by formulating a series of key questions related to self-report/proxy-report in individuals with intellectual and developmental disabilities or Down syndrome to answer in the literature review (Capone et al., 2018; Santoro et al., 2020; Santoro & Steffensen, 2021). In addition, throughout our project, we followed all components of the PRISMA Scoping Review guidelines (Tricco et al., 2018) which were sufficient to address our question, rather than a systematic review (Page et al., 2021).

By consensus, the following key questions were formulated by the authors:

1. In published studies, what data collection methods are used to obtain self-report in individuals with intellectual and developmental disabilities?

2. In published studies, what data collection methods are used to obtain self-report in individuals with Down syndrome?

3. What statistical approaches are used to compare proxy-report and self-report in individuals with intellectual and developmental disabilities or Down syndrome?

4. How well do responses compare between proxy-report and self-report in individuals with intellectual and developmental disabilities or Down syndrome?

2.2 |. PubMed literature search and data charting process

Literature searches were conducted in November 2020 to January 2021 using the National Library of Medicine (NLM) biomedical literature database PubMed (MEDLINE) (NCBI 1946–2020) to identify original research manuscripts addressing our topic. The National Library of Medicine (NLM) biomedical literature database PubMed was used because it is a large database focused on medicine aligning with our focus on health views. The authors worked with a librarian to identify search terms. We used the Medical Subject Headings (MeSH) (the NLM controlled vocabulary thesaurus for indexing) to capture related entry terminology in our searches. For example, the MeSH term ‘Down syndrome’ included the search entry terms: Downs syndrome, Down’s syndrome, Mongolism, Trisomy 21, Partial Trisomy 21. The MeSH terms ‘intellectual disability’ or ‘Down syndrome’ were combined using the Boolean operator ‘AND’ with the terms ‘proxy’, ‘parent’, ‘guardian’, or ‘self-report’ in the title or abstract and the terms ‘scale’, ‘measure’, or ‘instrument’ in the title or abstract, to capture the unfiltered literature. Then, the limiters ‘Human’, ‘English’ were applied to narrow the scope of the search to filtered literature. Search terms:

Search string with these filters (English): ((proxy OR parent OR guardian OR self report) AND (down syndrome or intellectual disability)) AND (scale or measure or instrument) AND ((y_10[Filter]) AND (english[Filter])). We did not filter or limit searches by publication year. Abstracts were reviewed and included according to their relevance to key questions. Whenever an abstract made mention of any key question (or there was doubt) the full article was procured. Sections 2 and 3 were then reviewed to determine which articles met inclusion or exclusion criteria. Article inclusion criteria included: data addresses at minimum one key question and supporting data is original (not previously published). Exclusion criteria included: data does not address at least one key question, study uses an uninterpretable methodology, case series <5, does not provide supporting data, did not present data specific to Down syndrome or intellectual disability, or study only included proxy-report. For this review, self-report data included respondents’ opinions, attitudes, and self-perceptions.

Using only the PubMed articles meeting inclusion, data to key questions were extracted from the abstract, methods, and results sections and entered into a preformatted Excel data extraction table (DET). The series of steps in review included:

1. Review of abstracts to ID those articles to move to a full text

2. Full text review for inclusion

3. Data abstraction from full texts of articles identified in step 2. The lead author (S.S.) reviewed articles for inclusion, and extracted data for all articles. A second reviewer (M.C.) assessed 10% of final articles, including all of the articles summarised in Table 3, and agreed with the findings from the first reviewer.

TABLE 3.

Statistical analysis and results summary of 14 studies comparing self-report and proxy-report in individuals with intellectual disability identified through literature review

	Data sources measures	Test	Result of comparison between self-report and proxy-report
11 studies with analyses comparing report
Balboni et al. (2013)	Agreement between self-report and report of others.	Association: N/A	-
	Personal Outcomes Score (POS) 48 items, RC = 3 points	Agreement: Intra class correlation (ICC) with two-way random effects model used for all the comparisons investigated.	Between self-report and third-party-point-of-view, ICC coefficients were poor or fair not only for the Emotional and Physical Well-being domains and the Interpersonal Relationship domain, but also for the Well-being factor, the Self-determination domain, the Social Inclusion domain and the Social Participation factor. Comparing two types of proxy to each other: All the ICC coefficients are statistically significant (p < .001) and all the differences between the scores obtained by the pairs of caregivers are not statistically significant.
		Additionally, t-test for matched samples and here statistically significant differences are found, Cohens d for matched samples reported
Benavidez and Matson (1993)	Differences between students with intellectual disability and ‘intellectually average’ students.	Association: Pearson correlation, between and within scales and informants.	For students with intellectual disability, self-report and informant versions of the CD1 and the RCDS were significantly correlated with each other.
	Children’s Depression Inventory (CDI), 27 items, RC = 4 point	Agreement: Fishers transformation for correlation coefficients.	Differences between these correlations were tested again using Fisher’s transformation of r at an alpha level of 0.05. According to these analyses, none of the correlations were significantly different (BID: 2 = 0.980; CDI: 2 = −1.605; RCDS: 2 = 0.979). With the exception of this particular correlation, there were no significant differences in magnitude of correlations of the other self-report and informant-report scales.
	Bellevue Index of Depression (BID), 40 items, RC = 5 points	ANOVA for differences in means between students on depression scales, with Bonferroni correction
	Weschler Intelligence Scale for Children-Revised (WISC-R)
Douma et al. (2006)	Adolescents with intellectual disability grouped as lower or higher IQ with 70 IQ Less than 70 or IQ greater than or equal to 70, compared to their parents.	Association: N/A	-
	Youth Self-Report (YSR), 119 items with 3-point RC and parent report using the CBCL, a 120 item measure with 3-point response scale. The YSL is considered an adapted form of the CBCL.	Agreement: Parent-adolescent self-report score differences using ICC two-way mixed effects to test for agreement	ICC between parent and self-report demonstrate poor reliability (<0.50) although agreement is moderate between parent and self-report within the higher IQ age group on 2 of the 10 scales: Anxious/depressed ICC = 0.50; Aggressive behaviour ICC = 0.53
		Fishers Z used to test for difference in ICC between groups.
Fisher et al. (2014)	Comparison of self-report of young adults with Williams syndrome to parent report, as well as observed behaviour corresponding to self and parent count measure.	Non-parametric statistics due to small sample size. Association: Mann–Whitney U test and Spearman’s correlation to compare self-report to observed behaviour as well parental report to observed behaviour	Although self-report ratings were consistent across measures, these measures did not correspond to the individuals’ actual behaviours during the community observations.
			Parents did more accurately predict their child’s real-life social approach behaviours.
	Self-Report Approachability Scale for young adults, 8 items with dichotomous Yes/No response as well as a count measure (0–8). of number.	Agreement: Cohen’s kappa for comparison of self and parent report	Parental reports did not correspond to their child’s self-report measures
	Patent Report Approachability Scale, 8 items, with 5-point response category as well as a count measure (0–8).
Knüppel et al. (2018)	Adolescents with diagnosis of Autism Spectrum Disorder (ASD) self-report and parental proxy report, with parents reporting as they thought their child would.	Association: Pearson’s correlation compared between mean INICO-FEAPS scores	Association between self and proxy report based on total INICO-FEAPS is higher (r = 0.64) than correlations of any of the individual 10 domain scores, with r ranging from 0.35 to 0.61
	INICO-FEAPS scale, 72 items across 8 domains, 4-point Likert type response categories.	Agreement: N/A	-
Mileviciute and Hartley (2015)	Adults with mild ID, with and without diagnosis of depression, to proxy report of staff.	Association: N/A	-
	Self-report using: Self-Reported Depression Questionnaire (SRDQ), 32 items, 3-point response categories	Agreement: One-way repeated measure ANOVA Paired sample t-tests with Bonferroni correction, using Ratio frequency score (domain total/highest possible domain total) across SDRDS, GDS and CGDS.	Adults with mild intellectual disability self-reported a higher frequency of affective and cognitive depressive symptoms than staff reported on the informant measure. Staff reported a higher frequency of somatic symptoms than adults with mild intellectual disability on one of the self-reported questionnaires (GDS) and a similar frequency on the other self-reported questionnaire (SRDQ). Important differences were found in the types of depressive symptoms based on their IQ, age and presence of a co-occurring psychiatric disorder
	Glasgow Depression Scale for people with a Learning Disability (GDS), 20 items, 3-point response categories.
	Proxy report using: Glasgow Depression Scale-Caregiver Supplement (CGDS), 16 items, 3-point response categories.
	Scale items were mapped to one of 3 domains: Affective, Cognitive and Somatic
Penketh et al. (2014)	Comparison of adults with intellectual disability self-report of SRAAS to staff proxy report of MAST, thwo different measures.	Association: Correlation between MAST proxy report score and self-report SRAAS score	There was a significant positive correlation between the MAST total scores and self-report comfort-seeking behaviour as measured by the SRAAS. In addition, a large effect size was reported and the MAST and the SRAAS shared 25.4% of the variance.
			As MAST scores increased so did SRAAS total frequency scores, therefore indicating that secure attachment style was significantly positively correlated with higher total frequency scores on the SRAAS.
	Self report of the Self-report Assessment of Attachment Security (SRAAS), number of items not reported, 6-point response categories	Agreement: N/A	-
	Proxy report of Manchester Attachment Scale-Third party observational measure (MAST), 16 items, 4-point response categories.
Schützwohl et al. (2018)	Comparisons of CANDID scores from self-report and two proxy types: staff and caregivers.	Association: G coefficient for inter-rater agreement.	Ratings of carers from the living environment showed poor agreement, as indicated by a G coefficient lower than 0.40, arose for the area ‘self-care’ (G = 0.32), ‘social relationships’ (G = 0.13) and ‘exploitation risk’ (G = 0.06).
	Camberwell Assessment of Needs for Adults with Developmental and Intellectual Disabilities (CANDID), number of items not reported although 25 ‘areas’ reported on, 3-point response categories.	Agreement: Paired t-test with Bonferroni correction. McNemar chi-square with Yates continuity correction compared the likelihood of discordance of ratings across self and proxy report.	Among self-ratings and ratings from carers from the living environment, absolute agreement was good for mean number of total needs (ICC = 0.63, 95% CI 0.27–0.79) and for mean number of met needs (ICC = 0.61, 95% CI 0.18–0.78). It was poor for mean number of unmet needs (ICC = 0.17, 95% CI −0.10–0.38). Similarly, absolute agreement between self-ratings and ratings from staff from sheltered workshops was good for mean number of total needs (ICC = 0.75, 95% CI 0.67–0. 81) and for mean number of met needs (ICC = 0.78, 95% CI 0.71–0.84), but poor for mean number of unmet needs (ICC = 0.27, 95% CI 0.02–0.45). agreement between self-ratings of adults with intellectual disability and proxy ratings was found to be generally moderate to almost perfect.
		Total agreement assessed by two-way random model ICC.
Watson et al. (1988)	Adults with developmental disability compared to proxy report of individuals support staff	Association: Spearman’s correlation	Within the community group, agreement between the subjects and the raters reached significance on only the somatoform disorder subscale rs < 0.05). For the institution group, there were significant correlations on the schizophrenia, affective disorder, and psychosexual disorder subscales, as well as for the total score.
	Psychopathology Instrument for Mentally Retarded Adults (PIMRA), 57 items, dichotomous response category.	Agreement: N/A	-
	PIMRA scored for
Wilson et al. (2016)	Comparison of child’s self-report to parent’s proxy report using the Paediatric Quality of Life Inventory (Peds QL4.0), a measure of quality of life, for children with Prader Willi Syndrome	Association: N/A	-
	Peds QL4.0, 23 items across 4 domains, response categories not noted.	Agreement: Two-way mixed ICC to compare consistency of self and proxy report.	The Bland Altman Plots show a high degree of variability inparent and child reports with wide difference values.
		Bland–Altman plots of differences between self-and proxy	ICC between the parent and child report of the quality subscales were very low, ranging from 0.05 to 0.24, with no statistically significant difference from zero.
			Bland–Altman plots indicate wide differences between self- and proxy report, and are suggestive of bias towards higher (positive) self (child) ratings in domains of physical, social, psychological and total quality of life.
Schwartz and Rabinovitz (2003)	Comparison of self-report by verbal adults with intellectual disability to proxy report provided by parents and by staff members	Association: correlation was used to evaluate association between self and proxy reports.	Correlations were statistically significant with self-staff proxy correlation r = 0.48 and self-parent proxy r = 0.43. Proxy staff reported highest LS score and self-reported lowest LS scores.
		GLM and paired t-tests were used to evaluate differences in mean scored between self and proxy reports.	Difference in mean LS scores of self-compared to proxy staff were found, with staff reporting higher LS scores than individual self-report LS scores. No statistically significant difference were found between self and parent report of LS.
	Lifestyle Satisfaction Scale, 14 items, 5-point response categories.	Agreement: N/A	-
Three additional studies
Gordon et al. (2007)	Zung depression scale	Analysis not specified	Authors report ‘The correlations between ratings of depression by key-workers were high and key-worker ratings correlated significantly with self-rating. The depression scores were found to be temporally stable.
			Conclusions. The findings suggest that proxy raters appear to be able to make reliable and valid judgements about other people’s self-reported levels of depression’.
Perry and Felce (2002)	Structured interview about choice	Analysis to compare not described in methods	Authors report: ‘Compared to two proxy measures; good correlation with one measure, poor correlation with other measure’
Raluy-Callado et al. (2013)	Hunter Syndrome-Functional Outcomes for Clinical Understanding Scale (HS-FOCUS) questionnaire	Not described in Methods	Authors report: ‘For this subgroup of patients above 12 years old, parent and patient self-reported scores were compared and parent’s ratings suggested a slightly higher level of dysfunction and distress than those of patients. The correlation coefficient between patient self-reported and parent overall scores was found to be positive ranging from 0.69 in HUI3 to 0.81 in CHAQ’.

See Figure 1 for an overview of the identification, inclusion and exclusion process and Data S1 for full details of extracted data. Fields for publication year, population details (number of subjects studied, age of sample, location, source of subjects and demographics, and research methodology were included in the DET. Data related to subject age, topic of study, and methods of obtaining self-report were collected and are summarised in section 3.

FIGURE 1.

PRISMA diagram of articles identified, screened, eligible, and included in review of self-report in individuals with intellectual disability and Down syndrome. Research design hierarchy in accordance with United States preventative services task force methods.

2.3 |. Evidence ratings by condition

Included articles were critically appraised by a single reviewer to determine each study’s research design (prospective, cross-sectional or longitudinal), sampling mode, mode of administration, subject recruitment, total number of subjects, use of control subjects, and the extent of internal validity and external validity of study design and study measures. We used the United States Preventative Services Task Force (USPTF) criteria for assessing articles based on experience with this methodology from our previous work (Capone et al., 2018; Santoro et al., 2020; Santoro & Steffensen, 2021). The USPSTF is a panel of experts that systematically reviews the evidence of effectiveness and develops recommendations. The research design hierarchy is based on individual study design which guides the definition of ratings of internal and external validity. The evaluation of internal validity considers study design factors such as ascertainment and selection bias, test procedures and consideration of confounding variables. For example, the internal validity of a cohort study is rated as good if it ‘Meets all criteria: comparable groups are assembled initially and maintained throughout the study (follow-up greater than or equal to 80%); reliable and valid measurement instruments are used and applied equally to all groups; interventions are spelled out clearly; all important outcomes are considered; and appropriate attention to confounders in analysis. In addition, intention-to-treat analysis is used for RCTs’ (USPSTF Procedure Manual [2015], p. 70 (United States Preventative Services Task Force, 2015)). External validity considers the generalizability of findings to a broader (more representative) population (United States Preventative Services Task Force, 2015). We used the criteria on research design hierarchy and the rating system used for scoring internal and external validity in the USPSTF report, Appendix VI and VII (United States Preventative Services Task Force, 2015).

2.4. Synthesis of results

See Table 1, Table S1, and Figure 1 for the summary of evidence rating of the articles included in our review, and the Table S2 for rating of each individual article. Tables were created summarising description information and descriptive statistics from identified studies, including: counts, frequencies, and ranges.

TABLE 1.

Characteristics of the 50 articles identified through literature review, categorised by topic

	Publications (N) dates	Diagnosis	Age range	Source of subjects	Methods	Research design hierarchy*	Internal validity category	External validity category
Self-report and proxy-report	(18) 1988–2019	Intellectual disability (ID), learning disabilities, autism spectrum disorder, other genetic syndromes with intellectual disability (Williams syndrome, Prader-Willi syndrome, Mucopolysaccharidosis type II)	11–96 years	Facilities for individuals with disabilities, schools, database, summer camp	Prospective survey, interview	II-1 (2), II-2 (13), III (3)	Good (12), Fair (6)	Fair (11), Good (7)
Self-report only	(25) 2002–2019	Intellectual and developmental disability, autism spectrum disorder, Down syndrome	6–76 years	County offices of disability services, work in disability-specific setting, attend a day programme, registries, research networks, schools, clinics and hospitals	Prospective survey, cognitive interviewing, administered neuropsych-ological testing	I (1), II-2 (9), II-3 (1), III (14)	Good (18), Fair (7),	Fair (19), Good (6)
Self-reported pain perception	(5) 1998–2017	Intellectual disability, Down syndrome	19–71 years	Care centres for intellectual disability	Prospective survey, observational	II-2 (4), III (1)	Fair (4), Good (1)	Fair (3), Good (2)
Self-report compared to control	(1) 2012	Intellectual disability	18–47 years	Organisations for adults with intellectual disability	Instrument development and survey	II-2	Good	Good
Comparisons between group of: self-report, supported, and proxy interview	(1) 2017	Intellectual disability	40+ years	National intellectual disability database	Prospective interview	II-2	Good	Good

Note: II-1: Well-designed controlled trial without randomisation. II-2: Well-designed cohort or case–control analytic study. II-3: Multiple time series with or without the intervention; dramatic results from uncontrolled experiments. III: Opinions of respected authorities, based on clinical experience; descriptive studies or case reports; reports of expert committees.

^aI: Properly conducted randomised controlled trial (RCT).

The authors grouped the specific topics of identified studies into the WHO definition of health categories of physical, mental, and social well-being, or other (WHO, 2018) based on our focus on health status, health related behaviours, and health-related quality of life. Based on our review of the identified studies, the authors proposed and agreed upon research gaps in need of future research and research situations which might favour self-report over proxy-report, or proxy-report over self-report.

3 |. RESULTS

3.1 |. Original literature review

We identified 50 articles which fit our inclusion criteria of having original data, answering a key question about self-report in individuals with Down syndrome or intellectual disability, in humans, and reporting on more than five cases (Figure 1, Table S1). These studies were published from 1988–2019, and included participants age 6–96 years overall. The data collection methods included: survey, interview, administered testing, and instrument development. All 50 studies included self-report information for a cohort, of which some also had proxy-report information [N = 18], and two studies compared responses from self-report group to a control group [1], or between groups [1] (Table 1). In addition to the bracketed number of studies, please see the Table S2 for specific details on each of the 50 studies.

Data extracted from original literature review which addressed one of our four key questions are presented below.

3.1.1 |. In published studies, what data collection methods are used to obtain self-report in individuals with intellectual and developmental disabilities?

Forty-six articles evaluated self-report in various groups of individuals with intellectual and developmental disabilities and answered this key question. Of these 46, 40 articles studied participants with intellectual disability in general. These 40 studies were published from 2002 to 2019. Self-report was collected from participants with intellectual disability as young as 6 years of age in one study; from adolescents with intellectual disability in seven studies, and adults with intellectual disability were the focus in 22 studies. Modes of administration and sampling modes in these 40 studies were diverse including: cognitive interview [2], focus group [1], interviewer-administered survey [24] of which two used a computer-assisted programme [1] or interviewer-administered modifications [1], self-administered survey [14] with paper instrument [12], caregiver support or assisted completion [2], with visuals added [2], and with a pretest procedure to determine who could use a three-point scale response option [1]. The topics of self-report were varied among physical well-being (Aznar et al., 2012; Balboni et al., 2013; Carbó-Carreté et al., 2019; Pérez-Cruzado & Cuesta-Vargas, 2013; Schützwohl et al., 2018; Schwartz & Rabinovitz, 2003; Verdugo-Alonso et al., 2017; Wigham et al., 2011; Wigham et al., 2014), mental well-being (Benavidez & Matson, 1993; Bond et al., 2019; Dagnan et al., 2008; Esbensen et al., 2005; Gordon et al., 2007; Hall et al., 2014; Haynes et al., 2013; Hermans et al., 2012; Kellett et al., 2015; Masi et al., 2002; Mileviciute & Hartley, 2015; Penketh et al., 2014; Stancliffe et al., 2014; Watson et al., 1988), and social well-being (Benromano et al., 2017; Bromley et al., 1998; Janeslätt et al., 2019; Johnson et al., 2014; Kellett et al., 2005; Kellett et al., 2015; Kramer et al., 2009; Liljenquist et al., 2019; Miller & Chan, 2008; Mumbardó-Adam et al., 2018; O’Donovan et al., 2017; Pérez-Cruzado & Cuesta-Vargas, 2013; Perry & Felce, 2002; Turnpenny et al., 2018; Vlot-van Anrooij et al., 2018; Walsh et al., 2018; Watkins et al., 2006) (Table 2).

TABLE 2.

Topic of the instrument studied in 50 articles identified by literature review, grouped into physical, mental, and social well-being or other

	Respondent	Physical well-being topics	Mental well-being topics	Social well-being topics	Other
Self-report data	Individuals with Down syndrome	Perceptions of pain (de Knegt et al., 2013; de Knegt et al., 2017a; de Knegt et al., 2017b)			Views about participation in research (Cooper et al., 2016)
	Individuals with intellectual and developmental disabilities	Behaviour (Kellett et al., 2015), everyday activities (Kramer et al., 2009), physical activity (Johnson et al., 2014; Walsh et al., 2018) were reported. Health topics including: health behaviours and health status (Vlot-van Anrooij et al., 2018), concerns related to epilepsy (Watkins et al., 2006), and perceptions of pain (Benromano et al., 2017; Bromley et al., 1998)	Quality of life (Balboni et al., 2013; Carbó-Carreté et al., 2019; Pérez-Cruzado & Cuesta-Vargas, 2013; Verdugo-Alonso et al., 2017), met/unmet needs (Schützwohl et al., 2018), life satisfaction (Schwartz & Rabinovitz, 2003), and adverse events and trauma (Wigham et al., 2011; Wigham et al., 2014). Mental health (Douma et al., 2006; Haynes et al., 2013; Hermans et al., 2013) concerns were self-reported including: depression (Benavidez & Matson, 1993; Esbensen et al., 2005; Gordon et al., 2007; Hall et al., 2014; Hermans et al., 2012; Kellett et al., 2015) and depressive symptoms (Mileviciute & Hartley, 2015), life events (Bond et al., 2019), depression and anxiety (Dagnan et al., 2008; Hall et al., 2014; Kellett et al., 2015; Masi et al., 2002), coping with negative emotions (Penketh et al., 2014), loneliness (Stancliffe et al., 2014), and psychopathology (Watson et al., 1988).	Aspects of independence such as autonomy (Janeslätt et al., 2019), choice (O’Donovan et al., 2017) (Perry & Felce, 2002), life satisfaction (Miller & Chan, 2008), self-determination (Mumbardó-Adam et al., 2018), dependence, and self-efficacy (Pérez-Cruzado & Cuesta-Vargas, 2013), social care skills (Turnpenny et al., 2018) were reported, as well as social topics: social support (Pérez-Cruzado & Cuesta-Vargas, 2013), Interpersonal problems (Kellett et al., 2005), social experiences (Liljenquist et al., 2019).	Views on exercising rights compared to control group of college students (Aznar et al., 2012)
	Individuals with autism and intellectual disability		Quality of life (Knüppel et al., 2018) bullying and happiness (Lung et al., 2019)
	Individuals with genetic syndrome with ID		Quality of life (Raluy-Callado et al., 2013; Wilson et al., 2016) anxiety (Dodd et al., 2009)	Social behaviour (Fisher et al., 2014)
With Proxy-report to compare	individuals with intellectual and developmental disabilities or Down syndrome		Quality of life (Balboni et al., 2013; Schwartz & Rabinovitz, 2003; Verdugo-Alonso et al., 2017), mental health and related topics (Benavidez & Matson, 1993; Gordon et al., 2007; Mileviciute & Hartley, 2015; Watson et al., 1988)	Topics related to independence (Penketh et al., 2014; Perry & Felce, 2002; Schützwohl et al., 2018)
	Individuals with other genetic syndromes with ID		Quality of life (Raluy-Callado et al., 2013; Wilson et al., 2016)	Social behaviours (Fisher et al., 2014)

Note: See article References for associated citations.

Two of the 46 articles focused on participants with autism and intellectual disability. Quality of life was assessed through paper survey with written and visual instructions (Knüppel et al., 2018). Parents could assist their son/daughter in completing the quality of life questionnaire, but it was emphasised that the individuals with autism spectrum disorder should decide how to rate the items themselves (Knüppel et al., 2018). Happiness was assessed through self-report questionnaire (Lung et al., 2019).

Four of the 46 articles focused on participants with a specific genetic syndrome with intellectual disability. Interview (Fisher et al., 2014), and self-administered survey (Raluy-Callado et al., 2013; Wilson et al., 2016) were used. In one study, paper instruments were modified: child measures were chosen to ensure that the items were appropriate for the cognitive level of the participants with Williams syndrome and to enable collection of parent report data (Dodd et al., 2009). For participants who were aged over 18 years, items that referred to school the word ‘school’ were replaced with ‘work’ and items that referred to kids, the word ‘kids’ was replaced with ‘people’ to ensure that the item content was appropriate for all participants (Dodd et al., 2009).

3.1.2 |. In published studies, what data collection methods are used to obtain self-report in individuals with Down syndrome?

Four articles answered this key question; these four studies included adults with Down syndrome with youngest participant age of 19 years. The topic of study included views about participation in research (Cooper et al., 2016) and perceptions of pain (de Knegt et al., 2013; de Knegt et al., 2017a; de Knegt et al., 2017b) (Table 2). In one study, semi-structured interviews with the participant with Down syndrome were used as one component of RCT for Alzheimer’s disease (Cooper et al., 2016). In three studies, neuropsychological testing and interview were conducted to evaluate various pain scales (pictures of faces, visual analog, pyramid scale) (de Knegt et al., 2013; de Knegt et al., 2017a; de Knegt et al., 2017b).

3.1.3 |. What statistical approaches are used to compare proxy-report and self-report in individuals with intellectual and developmental disabilities or Down syndrome?

Of the 18 articles with self-report and proxy-report information presented, three did not compare responses between proxy and self-report (Bond et al., 2019; Dodd et al., 2009; Miller & Chan, 2008), and in one article, the study aimed to evaluate a scales psychometric properties in a different population and comparing response between self and proxy was not even a secondary objective (Verdugo-Alonso et al., 2017). Fourteen articles presented data comparing responses from self-report and proxy-report, of these, three did not clearly outline their statistical analysis in section 2 (Gordon et al., 2007; Perry & Felce, 2002; Raluy-Callado et al., 2013), and 11 articles presented a clear methodology of the statistical analysis performed to answer this key question (Balboni et al., 2013; Benavidez & Matson, 1993; Douma et al., 2006; Fisher et al., 2014; Knüppel et al., 2018; Mileviciute & Hartley, 2015; Penketh et al., 2014; Schützwohl et al., 2018; Schwartz & Rabinovitz, 2003; Watson et al., 1988; Wilson et al., 2016) (Table 3). Data presented from several of the articles was collected primarily for the evaluation of the validity of a novel measure, not to compare responses between reporters (Douma et al., 2006; Penketh et al., 2014; Wilson et al., 2016). While establishment of an instrument’s validity is necessary for inference, the psychometrics used are generally not sufficient to establish intra-rater reliability between raters. Relative to the type of evaluation of reliability of proxy versus self-report, three articles evaluate association only (Knüppel et al., 2018; Penketh et al., 2014; Watson et al., 1988). In one of these articles, the target population is adolescents with autism spectrum disorder; there is no measurement or control of levels of intellectual disability as a covariable. The remaining eight articles all evaluate agreement, primarily using t-tests, Cohen’s d, Cohen’s kappa, or Bland–Altman, (Table 3). Interclass correlation coefficient (ICC), considered the most rigorous statistical test was employed by only four articles (Balboni et al., 2013; Douma et al., 2006; Schützwohl et al., 2018; Wilson et al., 2016). Each article presented the appropriate information, including model form, rater type (Shrout & Fleiss, 1979), as well as type of agreement, as recommended by McGraw and Wong (McGraw & Wong, 1996).

3.1.4. How well do responses compare between proxy-report and self-report in individuals with intellectual and developmental disabilities or Down syndrome?

In total, 14 articles compared the information and answered this key question. In the 14 articles, the participants had intellectual disability in general [11], or syndromic intellectual disability [3]. In the 14 articles, the participants were adolescents [4] or adults [10]; the youngest ages when self-report information was compared to proxy-report was 8 years (Wilson et al., 2016), and 11–12 years (Benavidez & Matson, 1993; Douma et al., 2006; Raluy-Callado et al., 2013). Self-report information was elicited from interview (self-structured, with modifications), verbally-administered survey, and self-report paper survey in these 15 studies.

In these 14 articles, the extent of agreement between proxy-report and self-report varied with eight showing similarities in response (Table 3). Two articles showed agreement between responses from individuals with intellectual disability and their proxy (Balboni et al., 2013; Benavidez & Matson, 1993); four articles showed association between responses (Benavidez & Matson, 1993; Knüppel et al., 2018; Penketh et al., 2014; Schwartz & Rabinovitz, 2003), and two showed similarities in responses but did not clarify if association or agreement were analysed (Gordon et al., 2007; Raluy-Callado et al., 2013).

Some studies showed mixed results (Perry & Felce, 2002; Schützwohl et al., 2018; Watson et al., 1988). For example, in one study of self-report and caregivers perspective on needs, agreement between self-ratings of adults with intellectual disability and proxy ratings was found to be generally moderate to almost perfect (Schützwohl et al., 2018). However, specific topics showed poor agreement, as indicated by a G coefficient lower than 0.40, including: ‘self-care’ (G = 0.32), ‘social relationships’ (G = 0.13) and ‘exploitation risk’ (G = 0.06) (Schützwohl et al., 2018). Specific aspects of scales showed varying agreement; one study found stronger relationships between personal development and rights and between rights and self-determination, but smaller correlation coefficients for self-determination and physical well-being (Verdugo-Alonso et al., 2017). Survey modifications were used, including asking proxy to complete ‘as if you were the person with intellectual disability’; in this strategy, all the correlation coefficients are statistically significant (p < .001) and all the differences between the scores obtained by the pairs of caregivers were not statistically significant, except for the ‘Right’ domain (t[42] = 2.09, p = .05; Cohen’s d = 0.31) (Balboni et al., 2013). Balboni et al. stated that ‘both self-report and report of others assessment strategies are necessary and that estimation of the client’s point of view may be a valid and reliable substitute of self-report when clients are not able to answer’ (Balboni et al., 2013).

Eight articles showed differences between proxy- and self-report (Douma et al., 2006; Mileviciute & Hartley, 2015; Perry & Felce, 2002; Watson et al., 1988). Notably, in Perry et al., the self-report scales could not be used properly by two thirds of the sample (Perry & Felce, 2002). One study found general disagreement, but some areas of agreement on: the somatoform disorder subscale in the community group, and the schizophrenia, affective disorder, and psychosexual disorder subscales as well as for the total score in the institution group (Watson et al., 1988). Mileviciute et al. suggest that ‘Informant questionnaires offer valuable information, but assessment should include self-reported questionnaires as these questionnaires add unique information about internalized experiences (affective and cognitive symptoms) of adults with mild intellectual disability that may not be apparent to caregivers’ (Mileviciute & Hartley, 2015) Anothe r benefit of self-report is accuracy; proxy of adults with intellectual disability understated quality of life when compared to self-report (Vlot-van Anrooij et al., 2018). In those with intellectual disability and autism spectrum disorder, significant differences were found for two domains, social inclusion and interpersonal relationships, with self-reports having higher scores compared to proxy-reports (social inclusion mean scores: 79.00 versus 75.62, respectively, p < .001; interpersonal relationships mean scores: 71.88 versus 66.02, respectively, p < .001) (Knüppel et al., 2018).

No studies were identified which compared responses been compared proxy-report and self-report in individuals with Down syndrome.

Author-proposed research gaps (Table 4), and research scenarios for proxy- or self-report are presented (Table 5).

TABLE 4.

Research gaps identified by authors based on literature review, and corresponding future research needs

Topic	Research gap and needs
Population	Study of specific groups of individuals with intellectual disability, such as specific genetic syndromes, and comparison to existing literature for intellectual disability
	Creation of a standard set of demographic information to collect in individuals with intellectual disability when presenting self-report data, such as IQ or level of intellectual disability
	Compare developmental levels and level of intellectual disability to response rate to determine to what extent this impacts rate of self-report
Methods	Comparison of survey methods in individuals with intellectual disability, such as between interviewer-administered survey and self-administered survey
	Comparison of self-report views to topics, are there specific topics which are more amenable to self-report
	Creation of universal tools for self-report in individuals with intellectual disability, such as a validated, appropriate consent language,
	Develop an approach to avoid ‘yea-ing’/acquiescence bias in individuals with intellectual disability
	Develop visual modifications for individuals with Down syndrome
	Determine a standard approach to statistical analysis comparing responses from two reporters; compare results with existing approaches
Inclusion	Consider guiding principle of defaulting to self-report over proxy-report; accept this inclusive approach across field
	Develop criteria for times in which self-report is not preferred/when proxy-report is acceptable
	Add study of self-report to research which asks proxy-report views

TABLE 5.

Proposed research considerations for proxy- or self-report in individuals with intellectual disability or Down syndrome using the cognitive model of survey response by authors based on literature review

	Self-report > proxy-report		Proxy-report > self-report
	Considerations	Research example	Considerations	Research example
Question interpretation	Reading level of individual with intellectual disability or Down syndrome Receptive language skills of individuals with intellectual disability or Down syndrome for interviewer-administered Cognitive ability of individual with intellectual disability or Down syndrome to understand abstract concepts Developmental stage may be very literal Cognitive ability of individual with intellectual disability or Down syndrome to understand complex questions	Concrete information, with limited interpretation, such as: reporting if they are currently feeling happy	Proxy interpretation of question: as they respond versus as they think the individual with intellectual disability or Down syndrome would respond Proxy is a secondary source for information about individual with intellectual disability or Down syndrome	Complex or abstract concepts, such as: reporting one’s overall temperament or personality, or academic progress over the past year
Information retrieval	Detail of information recalled: e.g., individuals with intellectual disability or Down syndrome may recall general events rather than specific historic dates Accuracy required in information recalled: e.g. individuals with intellectual disability or Down syndrome may recall events as a ‘recent’ memory which occurred years ago Immediacy of recall	Current information, such as: reporting if they attended school that day	More detailed information may be recalled by proxy Factual information may be more accurately recalled by proxy Distant events or recall of an experience over the past year may be preferred for proxy	Factual, detailed, specific information over a long time, such as: recalling detailed medical history
Judgement formation	Cognitive ability of individual with intellectual disability or Down syndrome may impact judgement about abstract concepts Cognitive ability of individual with intellectual disability or Down syndrome to perform mental calculations Subjective information may be preferred to be directly from individuals with intellectual disability	Concrete factual information, such as: Orientation in the Mini-Mental Status Exam	Abstract concepts may be preferred for proxy Mental calculations may be more accurate by proxy	Calculations over time, such as: Estimating the number of times one has been to the doctor in the last year
Response formatting	Reading level of individual with intellectual disability or Down syndrome Receptive language skills of individuals with intellectual disability or Down syndrome for interviewer-administered Cognitive ability of individual with intellectual disability or Down syndrome to understand response options including relative ranking of options Expressive language skills of individuals with intellectual disability or Down syndrome to give a verbal response	Responses with associated visuals or well-known symbols, such as: Responses of ‘thumbs-up’ or ‘thumbs-down’	Multiple response options may be better for proxy Response options with complex language, with complex concepts, or with abstract concepts	Responses with text and multiple options to compare, such as: A 5-point Likert scale of (1) Strongly disagree; (2) Disagree; (3) Neither agree nor disagree; (4) Agree; (5) Strongly agree.
Response editing	Reading level of individual with intellectual disability or Down syndrome Writing level of individual with intellectual disability or Down syndrome if self-administered paper survey Expressive language skills of individuals with intellectual disability or Down syndrome to give a verbal response Stamina to complete task-oriented evaluations Anxiety provoked by an ‘I don’t know’ response, or a response which is not socially desirable may be greater May be more likely to ‘yea-say’	Responses with concrete answers, such as: ‘yes’, ‘no’, or ‘skip’.	May be less uncomfortable with responses of ‘I Don’t Know’ Unclear if proxy would be better for socially-undesirable responses May be less uncomfortable disagreeing with a researcher	Responses which as for constructive feedback, such as: What could we improve in this survey?

4 |. DISCUSSION

Through literature review using MeSH terms in PubMed, we identified 50 articles which provided original data about self-report to answer one of our key questions on responses from individuals with intellectual and developmental disabilities, responses from individuals with Down syndrome, and comparisons to proxy-report. In response to our four key questions, we found that:

1. Statistical analysis methodology comparing proxy- and self-report was described in 11 articles with a variety of approaches used.

2. Fourteen articles compared report from individuals with intellectual disability to proxy-report; some studies found agreement, while others found disagreement. No studies compared proxy-report to self-report in individuals with Down syndrome.

3. Adults with Down syndrome reported views on research or pain perception in four studies and all used an interviewer-administered mode of administration; there were limited studies using self-report from individuals with Down syndrome.

4. Individuals with intellectual disability as young as age 6 years provided self-report data in 46 studies on topics ranging from mental health, to independence, to social issues, to health; data was obtained using interviewer-administered modes and self-administered modes with varying methods of assistance and support.

In conducting this literature review, we answered our key questions, yet there are limitations in the existing literature for individuals with intellectual disability, particularly regarding statistical comparisons between self- and proxy-report in individuals with intellectual disability. We only identified 11 articles which described the statistical approach for comparing responses clearly; among these, no consistent, standard statistical approach emerged. This may be due to study differences, as the statistical approach used is often dictated by the form of data (nominal, ordinal, continuous as well as source and number of raters), with the common goal of establishing a level of evidence to support the substitution of a proxy measure for a self-reported measure (McGraw & Wong, 1996; Shrout & Fleiss, 1979). However, in some studies, comparing self- and proxy-report was not the primary aim of the study and the statistical methods used for the primary aim (validation) may not be adequate for comparing reporters (Verdugo-Alonso et al., 2017). In evaluating the studies comparing responses between self-report and proxy-report, we found that the methods in some manuscripts were confusing; it was difficult for us to understand what statistical approaches were used, and it was unclear what responses were being compared. In one study, we found that they reported ‘parent-adolescent agreement, and indications of construct validity were about similar in all samples’ and described ICCs as ‘not significantly different between samples’, however, our review of their results shows reporting of ICC values less than 0.50 consistent with poor reliability, inconsistent with the authors’ conclusions (Douma et al., 2006). Comparisons often used total scores, or total subscale scores between self- and proxy-report, rather than assessing responses to individual items.

In assessing the statistical methods used, we reviewed the various statistical approaches which can be used to compare responses. To establish the reliability of a measure based on proxy-report rather than self-report, a measure of agreement, not only association, is needed (Tinsley & Brown, 2000). Association represents the degree to which the ratings of different judges are proportional as expressed as deviation from their means, and may be determined by correlation, such Pearson correlation coefficient, gamma Spearman’s rank order, Somer’s d-statistics, Kendall’s tau-b, or lamda, dependent of the level of measurement. Alternatively, agreement represents the extent to which different judges tend to make the same judgement about a rated subject, and is assessed with t-tests, Cohen’s kappa, or Bland–Altman plots. And, ICC reflects both degree of correlation as well as agreement between measures. Although individual views on preferred approaches may differ (McHugh, 2012), there are standard thresholds of acceptability for statistical evaluations, including: G coefficient (0.7–0.8 = acceptable, and 0.8–0.9 = good) (Cronbach et al., 1963; Shavelson et al., 1992), Cohen’s kappa (Cohen, 1960), Pearson correlation (Small = 0.1 to −0.3 or 0.1 to −0.3, Medium = 0.3 to 0.5 or − 0.3 to −0.5, and Large = 0.5 to 1.0 or −0.5 to −1.0) (Cohen, 1992; Nunnally, 1978) and the ICC (McGraw & Wong, 1996; Shrout & Fleiss, 1979).

Additionally, the studies of self-report from individuals with Down syndrome in the literature is limited; three of four were on the same topic, by the same research team. We found no studies comparing proxy- and self-report of individuals with Down syndrome. Future study would be useful to: compare proxy-and self-report in Down syndrome, and determine if the findings of studies in individuals with intellectual disability can be applied to individuals with Down syndrome. Evaluating the broader intellectual disabilities research on self-report, we found that most studies were prospective surveys, at a single site, and involved a single cohort. In our quality appraisal of the identified studies, the internal validity for studies was more often good (N = 33) rather than fair (N = 17), while the external validity was often fair (N = 33) rather than good (N = 17); larger, multi-site studies which are more generalizable would expand the existing literature. The specific instrument used, the topic studied, and the mode of administration varied; it is difficult to draw general conclusions on the quality of this evidence as a whole given the heterogeneity in the studies we identified. Reflecting on this, we proposed a dozen gaps in the research for future research.

We identified 14 articles which compared report from individuals with intellectual disability to proxy-report; no studies compared proxy-report to self-report in individuals with Down syndrome. Some of those that compared report found agreement, while others found disagreement. Existing studies in the general population to-date have shown differences between proxy-and self-report. For example, Erhart et al. found that adolescents’ and parents’ responses differed regarding adolescent quality-of-life; parents reported higher scores for total score and physical, self-esteem and school subscales, while adolescents self-reported higher scores for the psychological and family subscales (Erhart et al., 2009). Jokovic et al. suggested that proxy ratings should be considered carefully as a potential substitute for self-report ratings, but acknowledged that information provided by parents is useful even if incomplete (Jokovic et al., 2004). And, on average, children rated significantly lower health-related quality of life than their parents on the physical complaints, motor functioning, autonomy, cognitive functioning and positive emotions scales (paired t-test: p < .05) (Theunissen et al., 1998). Theunissen et al. concluded that parent reports cannot be substituted for child reports in the general population, yet both the child and parent reports proved to be valid with child scores appearing to be less extreme than parent scores (Theunissen et al., 1998). Further, in adolescents with autism spectrum disorders and intelligence in the average range (IQ > 70), correlations were higher between parent proxy reports when asked to answer as they believe their child would and adolescent self-reports than between standard parent reports and adolescent self-reports suggesting that parents have different opinions about their children’s quality of life than their children do (Sheldrick et al., 2012).

For researchers considering self-report, and clinicians interested in patient-reported outcomes, we identified take-home lessons in the research to-date. First, the methods used to obtain self-report from individuals with Down syndrome and intellectual and developmental disabilities more often used interviewer-administration, though self-administered survey was possible, with provision of various forms of support or assistance. Although possible, it was uncommon for individuals with intellectual and developmental disabilities or Down syndrome to independently complete a self-administered paper instrument. In addition, modifications to instruments were frequent, and included: the addition of visual aids, a decreased number of response options, and the use of screener questions to identify participants who should complete self-report. The topics of self-report were more often to be focused on mental well-being and social well-being, and less often focused on physical well-being. When comparing self-report to proxy-report, topics related to mental well-being showed either concordance (quality of life, life satisfaction, depression, and negative emotions) or discordance (mental health, depression, psychopathology, and choice) in individual studies. In the studies we reviewed, we did not find the topic of the instrument to show a pattern in similarity in responses from self and proxy (Table 2).

When choosing the respondent for a survey, many factors are likely to be considered. Differences in cognitive processing could impact the extent to which individuals with intellectual and developmental disabilities or Down syndrome are able to participate in self-report research, or complete self-report patient-reported outcome measures. The mode of administration is an important consideration for collecting self-report data, and interviewer-administration can address reading level, writing ability, and modifications such as allowing for clarification of questions can enable judgement formation. Some of the modifications to paper instruments, such as use of visual aids, can improve survey response (Petersen, 2008), possibly through enabling judgement formation, and assisting with response formatting and editing. The approaches used in these studies addressed some aspects of cognitive processing and allowed researchers to obtain self-report data despite variable intellectual disability.

Based on the limited number of studies comparing self-report and proxy-report, we did not feel that either self-report or proxy-report were clearly preferable. And, there may be situations in which one is preferred over the other, or in which both self-report and proxy-report are obtained synchronously. When considering the use of self-report in research of individuals with intellectual and developmental disabilities or Down syndrome in survey research, there may be times when self-report is ideal. A literature-informed study concluded that with appropriate supports, many individuals with intellectual disability can respond to mental health questions. And there is the need to engage individuals with intellectual disability to provide first-hand information about their health and well-being (Havercamp et al., 2022).

Using the steps in cognitive processing (Sudman et al., 1996), we propose specific instances when self-report from individuals with intellectual and developmental disabilities or Down syndrome may be preferred to proxy-report, or vice versa (Table 5). We hoped to offer instances and examples to guide researchers to begin to ask self-report more often from individuals with Down syndrome. When possible, for example if questions are straight-forward, the question aligns with their reading level, response options can be limited, and wording aligns with their reading comprehension, self-report from individuals with intellectual and developmental disabilities is preferred. Self-report for individuals with Down syndrome may benefit from visual options given their strengths in visual learning (Kumin, 2012). The use of visual support provides static information that allows longer processing time, simplifies the message by breaking it down into the core components, and can provide a link between verbal and visual information (Down syndrome Visual Supports Guide, 2021). Alternatively, proxy-report may be needed if sample size is small due to higher reliability from proxy (Erhart et al., 2009), or to provide information to complement self-report (Eiser, 2001; Eiser & Morse, 2001).

In the studies, we identified which compared proxy-report to self-report, there was inconsistent agreement: some with association/agreement, some without, and others with mixed findings. Future research could evaluate: the accuracies of various statistical approaches in comparing responses, the extent of agreement from respondents on specific items, the extent to which disagreement in an item contributes to disagreement in total scores, and our proposed considerations of specific instances which may be more amenable to be reported by proxy over self, or by self over proxy.

This literature review is limited by the information presented and published in the existing medical literature. The 50 articles were identified through searching with MeSH terms and the PubMed database; relevant articles may have been missed and have identified such articles (Graves et al., 2016; Krahn et al., 2014; Skotko et al., 2011; Santoro et al., 2021). Our search string may have missed key citations by not using wildcards. Additionally, it is possible that other relevant articles may not be indexed in PubMed; we are aware of studies in individuals with intellectual disability (Claes et al., 2012; Scott & Havercamp, 2018), at least one such study of adults with Down syndrome (Havercamp et al., 2017) and one study of adolescents with Down syndrome (Robles-Bello et al., 2020). Article screening in the first stage and quality appraisal of studies were done by a single researcher, but with input from the other co-authors. This is not ideal, or standard practice, but the best option for consistency in screening and appraising. The studies we identified varied in population (age, diagnosis, source) and methods (type, administration, modifications) complicating our ability to combine results of studies and draw broad conclusions. To address this, we present the original data from these studies and our full review data which may be of use to other researchers focused on this area (Table S2). There are likely additional features of the identified studies which we did not study, but which could be studied in the future. For example, we did not review the details of Institutional Review Board approval or consent procedure in the studies we identified, and there are likely moral-ethical deliberations for choosing either method of report. In addition, there are limited relevant studies which compared self-report to proxy-report, demonstrating an important research gap, and no consistent or ‘gold standard’ approach was identified among those studies which compared responses which could reflect different opinions in the field. Specific features of individual instruments, such as evaluation of item structure and response options, on agreement could also be studied. Future research should validate or adapt studies of self-report in individuals with intellectual and developmental disabilities to Down syndrome, autism with intellectual disability, and other genetic syndromes with intellectual disability. Ideally, when research studying individuals with intellectual disability is reported, it would be beneficial for the authors to present cohort data (for example, if there is a subgroup with Down syndrome) in the Data S1 to harmonise research data allowing other researchers to study specific cohorts.

As has been shown for individuals with disability, the source of information can have important implications on responses to surveys and patient-reported outcomes. It is essential to closely consider the reliability and validity of proxy-report data given factors, such as biases in health care (Iezzoni et al., 2021; Agaronnik et al., 2019), when acting on results. Collecting self-report data is a necessary step to compare self-report responses to proxy-report, and then identify strengths and limitations of each. Future studies could address the gaps in the literature which we have proposed; including increasing the number of studies of self-report in individuals with Down syndrome, and beginning to study self-report patient-reported outcomes in individuals with intellectual disability.

5 |. CONCLUSION

Review of studies of individuals with intellectual disability, which used a variety of statistical approaches, showed mixed agreement between self-report and proxy-report. Limited studies to-date have used self-report from individuals with Down syndrome, but lessons from the existing intellectual disability literature can guide researchers to incorporate self-report from individuals with Down syndrome in future research.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study were derived from the following resources available in the public domain: PubMED at https://www.ncbi.nlm.nih.gov/pubmed/ and the full data from our review is listed in the Data S1.