A Pilot Study Comparing Two Developmental Screening Tools for Use With Homeless Children

Literature review 

Stressors associated with homelessness and poverty affect the psychosocial and physical well-being of children and their families. Negative outcomes associated with homelessness in children often were reported; these outcomes include physical illnesses, developmental/behavioral problems, depression, fear or trauma from violence, and difficulties with school. For example, Ziesemer, Marcoux, and Marwell (1994) studied homeless and low-income housed school children and found that about three quarters were at increased risk for academic failure or for exhibiting behavioral problems. Davey (1998) studied a group of 52 children from 5 to 11 years of age who resided in a shelter in Florida. These children were found to have increased stress and behavioral problems, as well as lower self-image. Other researchers found that older homeless children had mental and behavioral issues as well (e.g., Ceballo and McLoyd, 2002, Davey and Neff, 2001, Menke, 2000).

Effects of homelessness on children's physical growth are less consistent. Miller and Lin (1988) reported a survey (N = 82 families with 157 children aged 17 days to 17 years) at a homeless shelter in the Seattle area. They found that slightly more than one third of the children were obese and that 13% were rated by their parents as in fair or poor health. More than half did not have health insurance and utilized emergency rooms at a higher rate than did average children. Wood and colleagues (1990) similarly reported that 13% of the homeless children in their study were overweight, and, in a 2007 study of homeless children 6 to 19 years of age, Grant and colleagues (2007) found an even higher percentage of obesity (about 33%). In an earlier study, Lewis and Meyers (1989) examined the growth and development status of homeless children entering a shelter in Boston. More than 90% of these children were younger than 5 years. The authors found that weight-for-height, weight-for-age, and height-for-age of the sample were similar to those of national samples of low-income children. Fierman and colleagues (1991) compared the growth of homeless children 3 months to 11 years of age with National Center for Health Statistics (NCHS) standards as well as with growth of age-matched children from low-income households. They found homeless children to have significantly shorter stature when compared with housed children and with NCHS standards. Two factors associated with lower-height percentiles were larger families and single mothers. However, duration of homelessness was not associated with decreased height or weight-height among homeless children.

Developmental screening is critically needed to help identify children with developmental problems early in life, particularly for homeless children who are at increased risk for compromised health and safety, as well as learning and developmental delays. According to studies completed a decade ago, an estimated 12% to 16% of children younger than 18 years in the United States have developmental or behavioral disorders (Boyle et al., 1994, Boyle et al., 1996). A 2006 report (using 2000 data pertaining to 8-year-old children in Atlanta) showed that although prevalence of developmental disabilities (such as mental retardation) was slightly lower (3% decrease) compared with 1996 data, male and Black children still had the highest prevalence (Bhasin, Brocksen, Avchen, & Van Naarden Braun, 2006). Sadly, national statistics indicate that only about one third of children with developmental or behavioral problems are recognized as having problems before entering kindergarten (Glascoe & Macias, 2003).

Routine developmental screening of homeless children can be a daunting task because of multiple stressors faced by homeless children and their families. Poverty (Duncan & Brooks-Dunn, 2000), lack of health care access/inconvenience (Riemer, Van Cleve, & Galbraith, 1995), parenting stress (Petterson & Albers, 2001), and ineffective parenting practices (Koblinsky, Morgan, & Anderson, 1997) are associated with negative developmental outcomes in homeless children. Parent-completed screening measures, however, have been used successfully with specific groups of children already flagged as developmentally delayed or as high-risk infants (e.g., extremely premature infants) (Klamer et al., 2005, Plomgaard et al., 2006, Skellern et al., 2001). Despite this, homeless children have not been the focus of previous studies, and in a survey of 102 agency directors of homeless shelters, Hicks-Coolick, Burnside-Eaton, and Peters (2003) reported that only 19% of children at shelters received developmental assessment. The transient nature of their lives permits no real continuity of care. Thus, a first step toward identifying these at-risk children is exploration regarding whether a parent-completed developmental measure would yield an accurate assessment in comparison with a professionally conducted developmental screening.

Methods 

This prospective, comparative pilot study was conducted at the ACCESS shelter located in an urban city in northeast Ohio. A convenience sample of 20 homeless mothers and their children (n = 21, 4 to 60 months of age) were recruited. Inclusion criteria for the mothers were as follows: English speaking, aged 18 years or older, residing in the ACCESS shelter, not under the influence of drugs or alcohol during shelter stay, not presently victims of domestic violence, and not directly released from a mental facility. These criteria, required by the ACCESS shelter for admission (DiMarco, 2000, DiMarco, 2007), were in place while the study was conducted. Only mothers whose children were between 4 and 60 months of age were included because of age limits set by a measure used, the Ages and Stages Questionnaires (ASQ) (4 to 60 months). Children who were chronically ill or who had a known history of developmental delay were excluded. To minimize the time commitment from mothers and children, no more than two eligible children per mother were invited to participate. At completion of the study, each mother received $10 cash per child, and each child received a small gift such as book, stuffed animal, or crayons.

Instruments 

The Denver Developmental Screening Test (Denver II) (Frankenburg, Dodds, Archer, Shapiro, & Bresnick, 1992b) is a widely used, readily administered screening tool for early identification of developmental delays in children from birth to 6 years of age. It covers four areas of development: (a) personal/social, (b) fine-motor/adaptive, (c) gross motor, and (d) language development. Delay in a specific area is suspected when a child fails an item that 90% of same-aged children pass. A child receives a “caution” when failing an item that 75% to 90% of same-aged children pass. Each child can receive an overall rating of normal (“no delays and a maximum of one caution”), suspect (“2 or more cautions and/or 1 or more delays”), or un-testable (“if refused one or more items completely to the left of the age line or more than one item intersected by age line in the 75% to 90% area”) (Frankenburg et al., 1992a, p. 13). Interrater reliability for the Denver II was between 93.8% and 97.8%. Seven to 10-day test-retest reliability showed 23% of the items were deemed fair to good and 59% excellent (Frankenburg et al., 1992b). Validity was established “by the precision with which the ages corresponding to 25%, 50%, 75%, and 90% passing for each item and subgroup have been determined (Frankenburg et al., 1992b, p. 94).” Construct validity was not performed for the reason that “the Denver II is not a test of some hypothetical construct (e.g., intelligence or physical dexterity), it simply defines the ages at which children accomplish a broad variety of specific tasks (p. 94).”

The ASQ, a parent-completed measure, also was used to measure child development. The ASQ consists of 19 questionnaires for children from 4 to 60 months of age. The ASQ screens for five areas: (a) personal-social, (b) fine motor, (c) gross motor, (d) communication, and (e) problem solving. At the end of each questionnaire, seven additional questions relate to parents' overall perception of and concerns about their children's development. Written at a fourth- to sixth-grade level (Bricker & Squires, 1999), the ASQ is feasible for use in community settings (Rydz et al., 2006). Specific cut-off points for each age were used for identification of children at risk for developmental delay. When compared with standardized tests appropriate for age and tasks (e.g., Bayley Scales of Infant Development, the Revised Gessell Developmental Examination, the Stanford-Binet Intelligence Scale, the McCarthy Scales of Children's Abilities, and the Battelle Developmental Inventory), the sensitivity and specificity of the ASQ is 51% to 92% and 81% to 100%, respectively (Squires, Bricker, & Potter, 1999). Sample items for a 4-month-old from the Denver II and the ASQ are listed in Table 1.

Table 1. Selected sample items from Denver II and ASQ for a 4-month-old child

	Domain	Denver II	ASQ
	Personal-social	Smile spontaneously∗	In front of a large mirror, smile or coo at himself or herself
		Regard own hands	Watch own hands
		Work for toy	While in sitting position, reach for a toy on a table close by, even though his or her hand may not touch it†
	Language	Make squeal sounds	Make high-pitched squeal sounds
		Laugh∗	Laugh
		OOO/AAH∗	Make sound when looking for toys or people
	Fine motor/adaptive	Follow an object 180°	Follow a toy from side to side in front of him or her
		Regard raisin	Sitting position, look at a toy placed on the table or floor in front of him or her
		Hand together	Bring hands together over chest, touching fingers‡
		Grasp rattle∗	Grab or scratch at his or her clothes
	Gross motor	Sit up with head steady∗	Hold his or her head steady when held in a sitting position
		Head up 90°∗	On tummy, hold head straight up, looking around
		Bear weight on legs	—

ASQ, Ages and Stages Questionnaires. Adapted from Bricker, D. & Squires, J. (1999). Ages & Stages Questionnaires® (ASQ): A parent-completed, child-monitoring system. (2nd ed.) Baltimore, MD: Paul H. Brookes; adapted by permission; http://www.brookespublishing.com/store/books/squires-asq/index.htm; and Frankenburg, W. K., Dodds, J., Archer, P., Bresnick, B., Maschka P., Edelman, N., & Shapiro, H. (1992a). The Denver-II: Training manual. Denver, CO: Denver Developmental Materials; adapted with permission.

∗ These skills should be acquired before 4 months of age. † This item is listed under Fine motor category in the ASQ. ‡ This item is listed under Gross Motor category in the ASQ.

During the assessment, each child's height was measured with a Stadiometer or an infantometer made by Seca (Model 206 bodymeter measuring tape, with wall stop and 210 length measuring mat). Weight was obtained from scales made by Health-O-Meter (Model 402 series and 1522KL). Weight-for-height for children younger than 36 months then was calculated. Interpretation of weight-for-height percentile is as follows: underweight, <5%; healthy weight, 5% to <95%; and overweight, ≥95% (CDC, 2008). Body mass index (BMI) for children 2 years and older was calculated using the CDC online BMI calculator for children and teens. BMI-for-age weight status categories and the corresponding percentiles are as follows: underweight, <5%; healthy weight, 5% to 85%; at risk for overweight, 85% to <95%; and overweight, ≥95% (CDC, 2008). For children between 2 and 3 years of age, either BMI or weight-for-height was calculated, depending on how height was obtained. If obtained with a standing scale, then a BMI score would be calculated. If height could only be obtained in a lying-down position, weight-for-height would be calculated. In this study, height for all children 2 years or older was obtained with a standing scale, and their BMI scores were calculated.

Results 

A total of 20 mothers and 21 children participated in the study. Two of the children were from the same family. SPSS 15 was used for data analysis. A majority of mothers were African American. The ethnicity reported was consistent with the shelter's population. The average maternal age was 28.5 years (range: 18 to 45 years), the average maternal education was 11.7 years (range: 10 to 13.5 years), and the average number of times they were homeless was 1.7 (range: 1 to 6). The average age of children was 30.5 months (range: 4 to 58 months). Eight of the children were younger than 2 years. Other demographic information is presented in Table 2.

Table 2. Maternal and infant demographics

		Mother (N = 20)		Child (N = 21)
		n	%	n	%
	Race
	Black	18	90	20	95.2
	White	1	5	1	4.8
	Other	1	5
	Sex
	Male			8	38.1
	Female			13	61.9
	Marital status
	Never married	17	85
	Separated	2	10
	Divorced	1	5
	Employment
	Full time	4	20
	Part time	2	10
	Unemployed	12	60
	Student	2	10
	Income
	None	4	20
	Employment	6	30
	AFDC/TANF	4	20
	Social security	2	10
	Other	4	20
	Insurance
	None	2	10
	HMO private	1	5
	Medicaid	9	45
	Medicaid HMO	6	30
	Temporary lost	2	10

AFDC/TANF, Aid to Families with Dependent Children/Temporary Assistance for Needy Families; HMO, health maintenance organization.

Information on weight and height was collected from all children. A weight-for-height percentile was calculated for eight children younger than 2 years. One 5-month-old infant was underweight and two (14.5 and 21 months of age) were overweight. BMI was calculated for the remaining 13 children ages 2 years and older. The majority (n = 9, 62.3%) had healthy weights. One child was underweight, two were at risk of becoming overweight, and one was overweight. No children were identified with short stature based on the CDC length-for-age. Overall, 14 of 21 children (67%) had healthy weight, three (14.3%) were overweight, two (9.5%) were at risk for becoming overweight, and two (9.5%) were underweight.

Health care providers used the Denver II instrument to assess children for possible delay. No delay was found in the areas of psychosocial and gross motor development; however, nine of the 21 children (42.9%) received suspect results in the area of language development. In contrast, the ASQ revealed three areas of concern identified by mothers: fine motor (n = 7), communication (n = 4), and problem solving (n = 4). The comparison of screening results between health care providers and mothers revealed that three children (14.3%) were identified by both screening tools to be at risk for language/communication delay and one child (4.8%) was at risk for fine-motor delay. Detailed information about Denver II and ASQ results are presented in Table 3. The percentage agreement between these two tools was strongest in gross motor (95%) and personal social development (95%) but weakest in language development (67%) (Table 4).

Table 3. Denver II and ASQ results (N = 21)

		Denver II		ASQ
		Pass	Suspect	Pass	Fail
	Personal social	21	0	20	1
	Gross motor	21	0	20	1
	Language	12	9	17	4
	Fine motor/adaptive	20	1	14	7
	Problem solving			17	4

ASQ, Ages and Stages Questionnaires. Adapted from Bricker, D. & Squires, J. (1999). Ages & Stages Questionnaires® (ASQ): A parent-completed, child-monitoring system. (2nd ed.) Baltimore, MD: Paul H. Brookes; adapted by permission; http://www.brookespublishing.com/store/books/squires-asq/index.htm; and Frankenburg, W. K., Dodds, J., Archer, P., Bresnick, B., Maschka P., Edelman, N., & Shapiro, H. (1992a). The Denver-II: Training manual. Denver, CO: Denver Developmental Materials; adapted with permission.

Table 4. Percent agreement between Denver II and ASQ

			ASQ
	Denver II		Fail	Pass	%
			Personal social
	Personal social	Fail	0	0	95
		Pass	1	20
			Fine motor
	Fine Motor	Suspect	1	0	71
		Pass	6	14
			Gross motor
	Gross Motor	Fail	0	0	95
		Pass	1	20
			Communication
	Language	Suspect	3	6	67
		Pass	1	11

ASQ, Ages and Stages Questionnaires. Adapted from Bricker, D. & Squires, J. (1999). Ages & Stages Questionnaires® (ASQ): A parent-completed, child-monitoring system. (2nd ed.) Baltimore, MD: Paul H. Brookes; adapted by permission; http://www.brookespublishing.com/store/books/squires-asq/index.htm; and Frankenburg, W. K., Dodds, J., Archer, P., Bresnick, B., Maschka P., Edelman, N., & Shapiro, H. (1992a). The Denver-II: Training manual. Denver, CO: Denver Developmental Materials; adapted with permission.

Discussion 

Early identification of developmental disabilities and early interventions may have a substantial impact on financial, educational, and societal costs in the future. Certainly, children with developmental delays who get early help are in a better position to experience personal benefits, particularly in areas so critical for learning and developing as social beings. Results of this pilot study indicate that homeless children are in need of monitoring for their growth and development. Several findings warrant discussion.

Second, homeless children may be at particular risk for lagging behind in their language development. Whitman et al. (1990) found that in a group of homeless children, language delay was the major developmental issue. Results of this study provide support for these findings. Results from the Denver II screening revealed that about 40% of the children had suspect results in language development. Although mothers in this study identified only four children (19%) with possible language delay, it is important to note these mothers were highly familiar with their children's speech patterns in ways not clear to a stranger. The higher percentage of children identified with suspect language development through the Denver II screening also could reflect that these children were not as talkative with providers as they were with their mothers. Despite dissimilar results, however, language development, particularly among homeless children, remains an important area for health care providers to closely monitor.

Third, overweight issues were present in this sample of children. One out of four were either at risk for becoming overweight or already were overweight. This finding is consistent with findings by Grant et al., 2007, Miller and Lin, 1988, and Wood et al. (1990) that 13% to 35% of homeless children were obese. A recent study (Schwarz, Garrett, Hampsey, & Thompson, 2007) from Baltimore also showed a higher rate of homeless children at risk for becoming overweight or already overweight, although children younger than 7 years old were found to have a healthy weight. This was not the case in our study, which showed that a much higher percentage of children younger than 5 years were at risk for becoming overweight or were already overweight.

In our pilot study experience with this population, we found that the ASQ was accepted by homeless mothers and that it is a quick method for initial screening. We also learned that homeless mothers not only expressed interest and concern about their children's development but were open to suggestions about child stimulation and ways to improve development. We did not collect information about children's past growth patterns, caregivers' BMI status, and daily caloric intake of these children, so possible relationships between these data and study findings were not explored. Other study limitations included a small sample size, selection bias, and the cross-section design.

Our findings support the 2001 and 2006 AAP policy statements regarding the importance of developmental screening for monitoring development of young children. Although more studies with larger sample sizes are needed to determine usefulness of parent-completed measures for detecting developmental risks in all children, developmental screening is particularly critical for homeless children because of the stressors created by homelessness and poverty. Health care providers are encouraged to solicit parental input about child development whenever possible. This step is particularly important for nurses and nurse practitioners, who are in a unique position to assess the applicability of such instruments and to provide critically needed interventions for these children.