Cross-Cultural Validity: Canadian Norm Values of the Alberta Infant Motor Scale Evaluated for Dutch Infants

Abstract

Purpose: 

To examine whether the Canadian normative values of the Alberta Infant Motor Scale (AIMS) are appropriate for Dutch infants.

Method: 

In a cross-sectional study, 499 infants developing typically (0.5-19 months) were assessed using the AIMS home video method. The scaling method was used for calculating item locations of the Dutch sample, and Welch test to compare Canadian and Dutch raw scores.

Results: The 

AIMS items (45 of 58) met the criterion for stable regression to calculate item locations of the Dutch data set and compare these with the Canadian data set. Dutch infants passed 42 of 45 items at an older age. Most monthly age groups of Dutch infants had lower mean AIMS scores.

Conclusion: 

The Canadian norms are not appropriate for the Dutch study sample. Dutch infants appear to develop in a similar sequence but at a slower rate. This has implications regarding the clinical use of the AIMS in the Netherlands.

In Brief

To examine if the Canadian normative values of the Alberta Infant Motor Scale are appropriate for Dutch infants.

INTRODUCTION

The Alberta Infant Motor Scale (AIMS) is a frequently used observational instrument to measure infant gross motor development from birth until independent walking (0-19 months). The AIMS discriminates typically from atypical motor development and was originally developed and norm referenced with 2202 infants in Alberta, Canada in 1994.¹ The neuromaturational theory provides a framework for the infant's sequential motor development assessed with the AIMS. Variability in rate of acquiring motor skills is better explained by the dynamic systems theory, where infant development is considered in its context.² Because of these contextual influences, the question arose whether policy (eg, back to sleep campaign) and changes in ethnic diversity of the Canadian population would have influenced infant motor development.³ The reevaluation study published in 2014 provided evidence that the established norms for the Canadian infants remained valid.³

Mendonca et al⁴ investigated the cross-cultural validity of standardized motor developmental screening and assessment tools to evaluate motor development of children up to 2 years of age. They concluded that standardized motor developmental assessments have limited validity in cultures other than the culture in which the normative sample was established. The use of culturally specific assessment tools might have consequences for clinical use, in terms of under- or overreferral for services.

Several studies examined the validity of the AIMS normative values in other countries with ambiguous results.^5–7 Syrengelas et al⁵ found no difference between Canadian and Greek infants in a sample of 1068 full-term born Greek infants. A cross-cultural analysis of motor development of Brazilian, Greek, and Canadian infants assessed with the AIMS supported that Brazilian infants develop more slowly than the Greek and Canadian infants.⁶ A Dutch pilot study by Fleuren et al⁸ reported a similar trend. Dutch infants develop more slowly than Canadian infants assessed with the AIMS.

The results of these reports are accompanied by the voiced concern of pediatric physical therapists in the Netherlands. Dutch infants are more often identified with a developmental delay according to the Canadian AIMS norms than they would identify on the basis of clinical views and expertise. The contextual influence of culture on normative values remains a discussion.⁹ The objective of this study was to examine whether the currently used original Canadian AIMS norms are appropriate for infants in the Netherlands.

METHODS

This study was a cross-sectional descriptive study, comparable to the reevaluation study of the Canadian norm values of 2014.³ The Research Ethics Committee of University Medical Center Utrecht approved the study (protocol no.15-029C).

Participants

As in the Canadian reevaluation of the AIMS norms, the target was to include a minimum of 450 Dutch infants, with age groups of two weeks to 19 months (see Supplemental Digital Content Appendix 1, available at: https://links.lww.com/PPT/A268). This spread proportional to the original AIMS distribution. Because the AIMS is developed for infants developing typically and is not appropriate for infants with a syndrome or neurological disorders, infants with pathology or a disorder known to influence gross motor development were excluded. The sample should comprise approximately 8% premature born infants (gestational age less than 37 weeks) and 10% infants from non-Western origin, which is in accordance with the population composition in the Netherlands.¹⁰ Confirming to Statistics Netherlands (www.cbs.nl), non-Western origin is defined as an infant either having at least 1 parent (first-generation) or grandparent (second-generation) from non-Western origins.¹¹

Under the umbrella of a long-term research project, GODIVA, data were collected in multiple studies. GODIVA is an acronym for Gross mOtor Development of Infants using home Video registration with the Alberta Infant Motor Scale. The GODIVA project commenced in 2013 and is ongoing. The GODIVA project comprises 5 studies, including the current study. Two studies, namely, the AIMS home video validation study¹² and a pilot longitudinal study on gross motor development of infants between 1.5 and 15.5 months, are completed. The pilot longitudinal study explored the feasibility of the home video method for parents in longitudinal research design.¹³ These studies generated cross-sectional data for this study. In addition, participants were recruited specifically for this study between January 2014 and April 2018. For all GODIVA studies, recruitment of infants was conducted via social media, word of mouth, flyers, and posters in Well Baby Clinics. After showing interest to participate, parents received official information and video instructions together with the informed consent form. Parents were asked for demographic information.

Measurements

The AIMS is a norm-referenced observational instrument that measures infant gross motor development (0-19 months) and has good psychometric properties.¹ The infant is observed in 4 different postures (supine, prone, sitting, and standing). The test items are based on spontaneous movements of the infant. The instrument consists of 58 items, which can be scored as “observed,” “not observed,” or “mastered.” The observed items together represent the motor repertoire of the infant.¹

Test Procedure

Parents used the AIMS home video method. This method is a validated and reliable way of assessing the AIMS with an intra-class correlation agreement between live and video assessment of 0.99 and standard error of measurement of 1.44 items.¹² Parents are asked to film their child in the 4 AIMS positions with guidance from an instructional film and instruction card. Using this method ensures that the infant is in a typical environment and filmed at a convenient time for both parents and child. Parents uploaded the home videos in a secure digital environment or saved videos on a secured USB stick. The researchers scored the AIMS from the video recordings and parents received feedback via e-mail.

More than 90% of the videos were assessed by 2 trained researchers. The agreement on item level between the 2 observers on 8 infants was 97.8%. Consensus meetings were organized frequently to discuss disputable items. In total, 4 other trained pediatric physical therapists assessed the remaining 10% of AIMS home videos. Their training consisted of 2, 4-hour training sessions during which infants were scored from videos. At the end of the training, 2 videos were scored passing was item agreement of 80% with the consensus score of 3 experienced researchers.

Data Analysis

The participant demographic means and standard deviations (SD), medians with interquartile ranges, or counts with percentage are shown in the Table. The alpha level was 5% and data were analyzed using the statistical package SPSS 24.0.

TABLE - Distribution of Monthly Age Groups of the Original Canadian and the Dutch Sample

Age Group, mo	n, NL-CAN	Boys/Girls	Mean (SD) Total Dutch AIMS Scores	Mean (SD) Total Canadian AIMS Scores	Difference CAN-NL	P (Welch Test)
0-1	10/22	6/4	5.6 (1.17)	4.5 (1.37)	−1.1	.03a
1-2	13/56	9/4	6.7 (1.49)	7.3 (1.96)	0.6	.23
2-3	21/118	9/12	8.6 (1.69)	9.8 (2.42)	1.2	.01a
3-4	30/90	14/16	12.3 (2.55)	12.6 (3.29)	0.3	.61
4-5	27/122	21/6	14.9 (2.66)	17.9 (4.15)	3.0	>.001a
5-6	44/189	20/24	18.5 (4.22)	23.2 (4.75)	4.7	>.001a
6-7	41/225	22/19	22.4 (3.40)	28.3 (5.50)	5.9	>.001a
7-8	44/222	22/22	28.8 (7.43)	32.3 (6.85)	3.5	.01a
8-9	42/220	24/18	31.9 (7.85)	39.8 (8.69)	7.9	>.001a
9-10	44/189	22/22	37.0 (8.83)	45.5 (7.47)	8.5	>.001a
10-11	33/155	21/12	43.0 (7.72)	49.3 (5.92)	6.3	>.001a
11-12	31/155	15/16	44.5 (8.37)	51.3 (7.11)	6.8	>.001a
12-13	29/124	15/14	50.2 (6.55)	54.6 (4.52)	4.4	>.001a
13-14	20/86	13/7	51.5 (4.86)	55.6 (5.01)	4.1	>.001a
14-15	21/61	10/11	51.8 (4.98)	56.9 (1.97)	5.1	>.001a
15-16	18/40	9/9	56.4 (1.94)	57.8 (0.45)	1.4	.01a
16-17	13/49	5/8	54.5 (4.50)	57.8 (0.55)	3.3	.02a
17-18	9/49	4/5	57.3 (2.00)	57.9 (0.35)	0.6	.40
18-19	9/30	2/7	56.7 (2.50)	57.7 (0.64)	1.0	.27
Total	499/2202	263/236

Abbreviations: AIMS, Alberta Infant Motor Scale; CAN-NL, Canadian-Netherlands; NL-CAN, Netherlands-Canadian.

^aSignificant when α < .05.

Data from the original Canadian normative data set were compared with Dutch data using the scaling method.³ An estimate of the age at which 50% of the infants would pass an item is referred to as item location. For each item of the AIMS, logistic regression was used to calculate the item location. The second step was to compare those items of the Canadian sample with the Dutch sample, which had a proportion of infants passing the item between 0.10 and 0.90 in both samples. Of the proportionally valid items, the means and SD of the item locations were calculated in the third step. The last step in the scaling method was to plot the item location for each item of Dutch infants against the Canadian infants and a regression model fitted with intercept of zero (Figure).

Fig.: Difference of the item location for the Dutch infants compared with the Canadian infants on the mean age (weeks) scored per item.

If the order of the items does not change, a curvilinear transformation of the original scale is necessary to generate new tables. To objectify this, the mean total AIMS score was compared between Canadian and Dutch infants with the Welch t test.

RESULTS

A total of 499 infants participated of whom 263 were boys (53%). Thirty-eight infants (7.6%) were preterm with a mean gestational age of 31.2 weeks (SD = 3.5) and 27 with gestational age of less than 34 weeks and 11 between 34 and 37 weeks. Infants' mean test age was 38.2 weeks (SD = 18.9) in comparison with 37.4 weeks (SD = 17.6) in the original Canadian sample (P = .39). Information on ethnic background was available for 412 infants (83%). Of these 412 participants, 5.8% were of non-Western origin, originating from very different countries, for example, Turkey, Eritrea, Israel, Suriname, and Indonesia. The ages of parents were available for 170 mothers (34%) and 153 fathers (31%), with a modal age of 30 to 35 years (43% mothers and 39% fathers). Education was for 234 mothers (47%) and 229 fathers (46%) of whom almost 80% were highly educated.

Thirteen of the 58 items were removed from analysis because of homogeneous scoring or because the proportion of infants passing the item was below 0.10 or above 0.90 in 1 or both data sets. The remaining 45 items were included in the analysis. In 42 items, Canadian infants had a younger mean age of passing. The biggest difference was 14.42 weeks for early stepping (item stand 12): Canadian infants passed this item at a mean age of 50.99 weeks (11.8 months), and Dutch infants 65.41 weeks (15.1 months). Items Supine 5 (hands to knees) and Supine 8 (rolling to supine position without rotation) were passed at a younger mean age by Dutch infants, with 2 days and 2 weeks of difference, respectively. Items Sit 10 (sitting to prone) and Prone 17 (reciprocal creeping 1) were observed at the same mean age, with −0.07 and 0.01 difference.

When item locations of the 45 items are plotted (Figure), a linear line X_Dutch = −7.42 + 1.21 × X_Canada was fitted with a proportion explained variance (R²) of 92.7%. If it is postulated that the intercept is zero, then the linear regression model is X_Dutch = 1.18 × X_Canada, with a 7% higher proportion explained variance (R² = 99.2%). Most differences between the Dutch and Canadian infants were significant, except for infants at 3 to 4 months and 17 to 19 months.

Compared with the Canadian centile scores, 74% of the Dutch infants scored below the 50th centile of which 16% of the infants scored below the 5th centile. According to the Canadian normative values, these infants would have been classified as having a motor delay.

DISCUSSION

This is the first study in which the item locations on the AIMS are calculated and compared with the Canadian item locations with the purpose to examine whether the currently used original Canadian AIMS norms are appropriate for Dutch infants. The mean age at which 50% of the infants pass an AIMS item is at a later age than the Canadian norm group. Regardless of differences in rate between the Dutch and Canadian infants, the sequence of the AIMS items was the same. Infants in the Netherlands attain motor milestones in the same developmental order as the Canadian infants.

Fifty percent of the Canadian infants passed almost all items at an earlier age. Only 3 items are reached earlier by Dutch infants. Item Prone 17 (reciprocal creeping) was passed at the exact average age by the Dutch and Canadian infants. In the original AIMS norms, reciprocal creeping with rotation (item Prone 21) was passed at an earlier age (41.49 weeks) than reciprocal creeping without rotation (Prone 17; 43.56 weeks), while in the Dutch sample, infants passed item Prone 21 at 51.19 weeks and Prone 17 at 43.57 weeks, which is a more logical order. Item Prone 21 was removed in the Canadian reevaluation study, because of a large difference in mean age of passing this item. Darrah et al³ had concerns that this item was scored incorrectly. However, it is not clear whether the incorrect scoring of the item occurred in the original study or in the reevaluation study.

The findings of our study extend those of a recent study on the Bayley Scales of Infant and Toddler Development (Bayley-III), another measurement tool for infant development.¹⁴ Steenis et al identified significant differences in the age at which Dutch infants reached gross motor milestones compared with American infants aged 3.5 to 25.5 months.¹⁴ The study confirmed that the item sequence was adequate for the assessment of Dutch children, but more Dutch infants would have been identified with a developmental delay using the American norms of the Bayley-III. Therefore, new Dutch norms were developed.¹⁵

Because the AIMS is never interpreted on an individual item level, it is important to review the AIMS total scores. For almost all monthly age groups, total AIMS scores differ significantly, except for 3 age groups at the borders of the test age. This can be explained by little variances in scores due to a bottom and ceiling effect of the AIMS. According to Darrah et al, the AIMS is considered most sensitive in the period from 4 to 12 months.³

The findings in our study of 74% of the infants scoring below the 50th centile are in accordance with the results of the pilot studies of Fleuren et al⁸ and De Kegel et al.¹⁶ Fleuren et al found that 75% of Dutch infants aged 0 to 12 months score below the 50th percentile of the Canadian norms. De Kegel et al showed that 64.8% of the Flemish infants score below the 50th percentile.

Strengths of this study are the large sample with an adequate representation of premature born infants and the method used, being comparable with the Canadian reevaluation study to calculate the item locations. The calculated item locations provide a better understanding of the sequence of gross motor development and the differences between the Canadian and Dutch infants. The results of our study support the therapists' thoughts about the differences they experienced in clinical practice regarding the AIMS scores. Another strength in this study was the use of the home video method. The method provided the opportunity for the researchers to discuss difficult items and doing so, increased the reliability of the assessments. This was confirmed by the high agreement (κ = 0.98) among the researchers.

There are some limitations including the small number of infants in the age groups of 0 to 3 and 17 to 19 months. Although this does not affect the item location calculations, it does affect the mean AIMS scores of the age groups. Despite these small groups, the course of the centile scores is comparable with that of the original sample. The AIMS has a floor effect in the earlier ages and a ceiling effect at later ages; therefore the effect of fewer infants in these age groups is less.

Second, it is debatable whether the sample is a perfect representation of the Dutch population due to relatively large proportion of missing demographic data. Approximately 80% of more than 200 parents were highly educated. This is in accordance with the Canadian reevaluation study sample, where 90% of the parents had a post–secondary education, 9% had finished high school, and less than 1% had an education less than high school.

There are inconsistent results in the association between higher maternal education and motor outcome.^17–22 Because the study sample of the Canadian reevaluation also consisted of approximately 90% highly educated mothers and the original norm reference study did not ask for maternal education, the data are comparable. The sample does not count a representative percentage of infants of non-Western origin. The researchers made considerable effort to recruit non-Western parents for the study, with limited results. Language problems and restraints for filming their child were some reasons for not participating in the study. The results of our study should therefore be interpreted with care.

Despite these limitations, this study supports relevant differences in developmental rate between the Canadian and Dutch infants according to the AIMS. Future research should therefore explore normative values of countries that frequently use the AIMS in clinical practice. In addition, the results of this study raise the question: Which factors contribute to infant gross motor development and the considerable difference in gross motor developmental rate of infants living in different parts of the world?