Adapting the Bayley Scales of infant and toddler development in Ethiopia: evaluation of reliability and validity

Abstract

BackgroundThere is a need for valid and reliable observational measures of early child development in low‐income and middle‐income country settings.MethodsThe aims of the study were to adapt the Bayley Scales of Infant Development (Bayley III) for a rural Ethiopian setting and evaluate reliability and validity. The study was carried out between January 2008 and January 2009 in the Butajira demographic surveillance site, south central Ethiopia. The Bayley III was adapted to be socioculturally appropriate for a rural Ethiopian context. Nurses and high school graduates were trained in administration of the measure for 10 days. Inter‐rater reliability was evaluated (n = 60). Content, construct and convergent validity was then examined on a population‐based cohort of children at the ages of 30 (n = 440) and 42 months (n = 456). Mokken scale analysis was used to determine the scalability of items in unidimensional, hierarchical sub‐scales. The mean score was compared by age of child and by stunting status (less than −2 z scores below the standard height‐for‐age).ResultsThe intra‐class correlations between raters were above 0.90 for all sub‐scales of the child development measure. Some scale items were not contextually relevant and showed poor scalability. However, the majority of items scaled onto the existing sub‐scales of the international measure to form adequate‐to‐strong hierarchical scales with good internal consistency (Cronbach's α above 0.70 except for gross motor and expressive language sub‐scales). Item‐scale coefficients were good. The mean score of all sub‐scales was significantly higher in the older group of children (33.02 higher total score; P < 0.001) and in the children who were stunted (total Bayley score 2.58 (95% confidence interval 0.07 to 5.10) points lower at 30 months and 3.87 (1.94 to 5.81) points lower at 42 months.ConclusionsAn adapted version of an international, observational measure of child development was found to be reliable, valid and feasible in a rural Ethiopian setting.

Reliability and Validity of an Adapted and Translated Version of the Mullen Scales of Early Learning (AT‐MSEL) in Rural Guatemala

“…Intraclass correlation coefficients indicated excellent interrater reliability across subscales, consistent with those reported in the MSEL manual (Mullen, ), as well as high agreement reported between raters for an adapted version of the MSEL in South African children (Bornman et al, a). Results are also consistent with those reported for other tests of early childhood development adapted and translated for Ethiopian (Hanlon et al, ) and Nepalese (Ranjitkar et al, ) children. Although other previous studies in low resource settings have documented adequate interrater reliability for tests developed for school‐age children (Ruiseñor‐Escudero et al, ), it is encouraging that high interrater reliability is also being achieved for early childhood assessments, as these assessments generally require a higher level of skill and training of test administrators and so are more vulnerable to errors in administration and scoring.…”

“…Intraclass correlation coefficients indicated excellent interrater reliability across subscales, consistent with those reported in the MSEL manual (Mullen, 1995), as well as high agreement reported between raters for an adapted version of the MSEL in South African children (Azari et al, 2017), Ethiopia (Hanlon et al, 2016), and Nepal (Ranjitkar et al, 2018) ranges from unacceptable to excellent and often varies across subscales. This variability may be due to challenges adapting and translating tests to another cultural context while preserving item intent and level of difficulty; however, language subscales in particular showed weaker internal consistency in those studies.…”

“…Internal consistency statistics were also high across AT‐MSEL subscales, demonstrating that the measure performed as expected. In the existent literature, internal consistency of developmental measures developed in high‐income countries and utilized in low resource settings including Iran (Azari et al, ), Ethiopia (Hanlon et al, ), and Nepal (Ranjitkar et al, ) ranges from unacceptable to excellent and often varies across subscales. This variability may be due to challenges adapting and translating tests to another cultural context while preserving item intent and level of difficulty; however, language subscales in particular showed weaker internal consistency in those studies.…”

“…One approach to address this problem is to develop a new test for use within a specific cultural setting. While this can be the preferred method in some circumstances (Gladstone et al, 2010), this approach limits comparison and applicability across settings and can be impractical due to the high cost and the extensive time required (Geisinger, 1994;Hambleton & Patsula, 1999;Hanlon et al, 2016).…”

The reliability and validity of Bayley-III cognitive scale in China's male and female children

“…The Bayley Scales of Infant and Toddler Development-3rd Edition (Bayley-III) is a well-known psychometric instrument, which provides useful information for the early identification of infants who might have developmental problems [1]. Since publication in 2006, the Bayley-III has been used widely in pediatric populations worldwide [2][3][4][5][6][7][8][9][10][11], and numerous researchers have confirmed the strong reliability and validity of the instrument [9][10][11][12][13][14][15]. with five separated subscales accordingly.…”