A Parent-Response Screening Inventory for Fragile X Syndrome: Development & Testing with an International Sample

Abstract

Fragile X syndrome, caused by a mutation in the FMR1 gene, is the most commonly inherited form of intellectual disability in children. Because the physical and early behavioral signs of Fragile X are often subtle, parents are often in the best position to advance early recognition and treatment. The Biopsychosocial Screening Inventory for Fragile X (BIPSSI-FX) was designed as an early detection parent-response inventory. A mixed-methods exploratory study of 886 caregivers, recruited from 22 countries across 5 continents, of children aged 1 through 18 years was used to refine a reliable and valid instrument.

Fagile X syndrome (FXS [MIM 300624]), the most commonly inherited cause of intellectual, learning, and psychoneurological disabilities in humans, is caused by a mutation in the Fragile X Mental Retardation 1 gene (FMR1 [MIM309550]). This gene encodes for the Fragile X mental retardation protein (FMRP), which is essential for typical brain function (Chiurazzi, Neri, & Oostra, 2003). Fragile X syndrome is inherited as an X-linked dominant disorder, meaning the gene responsible for the disease is located on the X chromosome and a single faulty (mutated) form of the gene can produce an effect.

The majority of mutations in the FMR1 gene are due to expanded trinucleotide repeats (CGG) located in the 5’ untranslated region of the gene. Repeat expansions are mutations where short deoxyribonucleic acid (DNA) sequences are duplicated a number of times in a row. A trinucleotide repeat is made up of triplet base-pair (nucleotide) sequences that exceed the normal, stable threshold for a healthy gene (see Figure 1). These expansions result in fragile sites within the section of a gene controlling accurate protein translation and suitable protein production. Improper protein production (i.e, overproduction or underproduction) underlies the manifestation of several human diseases.

Figure 1.

Triplet-repeat-mediated pathological mechanism for Fragile X syndrome. Image reproduced from the The U.S. National Library of Medicine biomedical collection holdings.

Subject to the number of repeats, individuals with a mutation in the FMR1 gene may be diagnosed to one of the following status categories: unaffected, intermediate, premutation, or full mutation. When the repeat region expands to over 200 duplicates, the region becomes hyper-methylated. In genetics, DNA methylation is a biochemical modification of a DNA strand through the addition of a methyl group. This abnormality in methylation consequently “switches off” the FMR1 gene and thus FMRP production. Children diagnosed as having a full mutation status completely lack FMRP and exhibit a complement of cognitive and psychosocial difficulties. Those who carry the premutation status (i.e., between 55–200 unmethylated repeats) typically do not exhibit symptomatology of FXS during childhood because FMRP is being produced.

Individuals with a premutation status are at increased risk for the development of late-onset premutation-associated disorders (PAD). Specifically, fragile X associated primary ovarian insufficiency (FXPOI [MIM 311360]) and fragile X associated tremor-ataxia syndrome (FXTAS [MIM 300623]). Researchers suggest that these conditions may possibly perform as an indicator of mutation shift from that of premutation to full mutation status within a single generation. Because it generally takes several generations for a mutation to spread, this rapid mutation shift is considered a unique aspect of FXS. However, this unestablished concept is an area of ongoing active research

In agreement with the hallmarks of X-linked dominant inheritance, females have a higher FXS prevalence rate than males. Currently, FXS has an estimated frequency of 1/4000 males and 1/7000 females worldwide (Willemsen, Levenga, & Oostra, 2011). Males typically demonstrate more severe physical, cognitive and behavioral disability than females. Clinical features of an FXS full mutation are identifiable in infants as early as 9 months of age (Roberts, Hatton, Long, Anello, & Colombo, 2012; Visootsak, Warren, Anido, & Graham, 2005). These include developmental delays, learning disabilities, prominent facial features, and such social-behavioral problems as impulsivity and attention deficit.

Early Detection of At-risk Children

Prompt detection and intervention greatly enhance quality of life for children with FXS (Hagerman et al., 2009; Lisik & Sieron, 2010). Yet, early detection of FXS-specific signs and symptoms by healthcare professionals is challenging. Early recognition of at-risk children who require further FXS molecular diagnostic testing is based, primarily, on trend data generated from screenings of developmental progression during pediatric well-visits. The average age of an FXS confirmatory diagnosis via molecular testing is approximately 37 months for boys and 42 months for girls. Data suggest that the variability associated with delayed FXS diagnosis can be up to 96 months post non-attainment of the first developmental milestone (Bailey, Raspa, Bishop, & Holiday, 2009; Centers for Disease Control and Prevention (CDC), 2002) This significant time lapse suggests that barriers exist to the early detection, diagnosis and subsequent treatment of FXS.

Impediments to reliable routine developmental screening, referral for molecular testing and timely diagnosis of FXS in children include but are not limited to deficient health care access and utilization among vulnerable populations, time constraints in the practice area, and availability of practical and psychometrically valid FXS-specific measures (Chung, Lee, Morrison, & Schuster, 2006; Dosreis, Weiner, Johnson, & Newschaffer, 2006; Kemper & Bailey, 2009; Mackrides & Ryherd, 2011). Bearing these limitations in mind, and considering that the amount of time mothers and fathers spend caring for their child is often substantial during the early years of development, parents are often more likely to notice developmental problems or abnormal behaviors which may go unrecognized by others (Bailey, Skinner, & Sparkman, 2003; Squires, Bricker, Heo, & Twombly,, 2001; Theeranate & Chuengchitraks, 2005). Consequently, parent as first-line screeners may be an invaluable resource in assisting pediatric healthcare professionals to overcome impediments to identifying at-risk individuals in need of further evaluation for FXS.

In addition to real-time screening and early recognition of FXS-specific characteristics, determinants of familial risk may also inform early diagnosis of FXS among children. A premutation or carrier status in the birth parent or biological caregiver, even in the presence of subtle developmental and neuropsychological deficits in the child, may function as a supplementary indicator for further testing. No research has specifically addressed the effectiveness of familial history information, obtained in the general pediatric setting, as a complement to risk-assessment calculations in screening for FXS during early childhood. However, associations have been noted between family history screening and the subsequent diagnosis of monogenic and polygenic heritable diseases (Dolan & Moore, 2007; Hariri, Yoon, Moonesinghe, Valdez, & Khoury, 2006; Heald, Edelman, & Eng, 2012). These findings support the viability of incorporating a familial risk component into a screening measure for FXS.

Currently, no reliable and valid parental screening measure exists specifically for use with children suspected of having FXS. In response to this need, we developed a new parent-response instrument, the Biopsychosocial Screening Inventory for Fragile X (BIPSSI-FX). The profile is designed to elicit information regarding FXS-specific psychological and behavioral characteristics commonly exhibited by children with an FMRI full mutation. The purpose of this study is to report psychometric properties of the BIPSSI-FX using a large international sample of children. Specifically, we evaluated the instrument’s internal consistency reliability, construct validity, and factor structure.

Materials and Methods

Conceptual Framework

An adaptation of Gottlieb’s Probabilistic Epigenetic Biopsychosocial Conceptual Framework of Development was used to guide creation of the instrument in this study (Gottlieb, Walshten, & Lickliter, 1998). A key characteristic of Gottlieb’s meta-theoretical model is the perspective of individual growth and development via bidirectional influences of environmental, behavioral, neural, and genetic factors. The framework presents the premise that when comparing individual performance on motor, cognitive, behavioral and/or developmental milestone tests against normative indices, it is possible to differentiate between different types of genetic syndromes involving intellectual and developmental disabilities (e.g., FXS vs. Down’s syndrome). Specifically, as a child matures, we are able to contrast specified developmental, physical, social-behavioral, and cognitive outcomes with standardized or expected patterns. In this manner we can ascertain difficulties in functional areas, unique to FXS, warranting further evaluation. Furthermore, we integrated the concept of parents as first-line screeners for detection of developmental delay and social-emotional symptoms often associated with FXS among children (see Figure 2).

Figure 2.

Johnson’s Biopsychosocial Spiritual Framework for Human Development, adapted from Gottlieb (2002)

BIPSSI-FX Item Generation

Phase 1.

The initial 50 items for the multi-dimensional BIPSSI-FX were developed based on concepts of interest and a comprehensive review of the psychological and pervasive developmental disorders (PDD) literature.

Phase 2.

After the items were generated and grouped theoretically into five subscales, face and content validity were addressed using FXS clinical specialists, general pediatric clinical specialists and experts in instrument development. Initial verification of the instrument’s potential viability via internal consistency estimation, test-retest reliability, and concurrent validity were evaluated through pilot testing using a small African American cohort (n = 16) (Powell-Young, Sherman, Brossman, & Johnson, 2012). Qualitative feedback from participants was used to modify, increase clarity and readability, and add or eliminate questions before administering the questionnaire to an international sample.

Phase 3.

Using a 200-subject subset of the current sample, (i.e., the first 200 questionnaires that were completed and submitted), we performed a series of exploratory analyses to re-evaluate items for final profile inclusion. These items were factored to examine the latent structure represented by the item pool. Based on these and pilot data analyses, a total of 24 profile items were removed. These deletions resulted in the current version of the BIPSSI-FX which consists of 26 items within four subscales.

Design, Administration and Scoring

Subscale 1, Developmental Milestones (5 items), provides information on attainment of adaptive, motor, language, and social maturation according to well-documented standardized norms. Subscale 2, Social-Behavioral (11 items), assesses a spectrum of social-emotional-behavioral limitations associated with FXS. Subscale 3, Cognitive (4 items), identifies features related to language and intellectual impairment. Subscale 4, Biological Parent (6 items), evaluates traits associated with FX premutation carrier status of the suspect child’s biological respondent. We consider inclusion of a biological parent subscale a unique aspect of the BIPSSI-FX because items within this subscale assist in the appraisal of premutation status among first (e.g., parents, siblings) or second-degree (e.g., grandparents, aunt) blood related caregivers of FXS suspect children. Integration of this component into the measure serves a two-fold purpose. As part of the overall profile, scores generated with this subscale provide a more comprehensive picture of FXS risk.

Individually, healthcare professionals may use this measurement index as an alert indicator for potential carriers of the FMR1 premutation. Although a premutation status is rarely associated with mental deficits, these expansions are unstable and prone to an FMR1 full-mutation expansion conversion within a single generation (Levesque et al., 2009). Each profile item is rated on a Likert-type scale from 0 to 2. Point-assigned response options reflect the degree of subscale achievement-functional limitation. Scoring is based on summated means calculated for each domain. The higher the mean subscale score, the greater the functional deficit in that area. Specifics related to BIPSSI-FX scoring and interpretation are detailed elsewhere (Powell-Young, Sherman, Brossman, & Johnson, 2012).

Sample and Setting

This study featured a comparative, exploratory, field-study design. In contrast to the phase 2 pilot study in which administration of the survey was by traditional paper-pencil method, the BIPSSI-FX was administered as a web questionnaire accessed on the OSU/ Langston University server. A computer science and information technology professor and department head built and maintained the data collection website. Internet data collection methodology was employed in an effort to increase both sample size and diversity. Using internet listserv groups and organizations that support child health and families of children with cognitive challenges such as the FRAXA Research Foundation, National Fragile X Foundation and Autism Network International, a convenience sample of N = 886 biological and surrogate caregivers of children aged 12 months through 18 years of age volunteered to participate in this study. To be included in the study, caregivers had to be adults (> 18 years of age) and able to read and understand either the English or Spanish version of the BIPSSI-FX. As we previously alluded, of the total subjects recruited, 200 were used for initial refinement and validation. Results from analyses performed on the remaining sample (n=686) are reported within this article.

Measures

Demographic Survey.

We collected general demographics including ancestry, gender, age, familial relationship, geography, education, and annual household income.

Diagnosis Information Form (DIF).

This document was used to facilitate grouping of the children into FMR1 full mutation group for evaluation of convergent validity. To be categorized as having a FMR1 full mutation, respondents must have reported that an FXS diagnosis was confirmed via genetic testing.

Procedures

Before the study began, we obtained approval from the Oklahoma State University IRB and various regional, national, and international listserv masters. Subsequently a notice was posted to recruit primary caregivers of children diagnosed with a developmental disability and caregivers of typically developing children. A University website, specifically designed for the study, provided a mechanism by which the primary caregivers could complete online questionnaires and return the encrypted, automated surveys to an access database at the server hosting the website. Anonymity was protected in that no identifying data were collected, including respondent IP addresses.

Each prospective volunteer completed and submitted the BIPSSI-FX, a Primary Caregiver Information Profile, a Diagnosis Information form, and the How to Participate Information sheet on which they acknowledged a statement of informed consent. Instructions for completing the study were clearly delineated on the recruitment website. A reminder was integrated into the biological parent section alerting non-biological caregivers to refrain from completing that section.

Data Analyses

All analyses were conducted using the Statistical Package for the Social Sciences (SPSS^®, Version 17). Prior to analyses, all variables were edited separately for accuracy, completion, and credible values. Frequencies, central tendency measures, and measures of variability were used to generate descriptive summarizations. Cronbach’s alpha coefficient estimates were used to evaluate internal consistency reliability. A threshold criterion of 0.70 was considered acceptable to demonstrate reliability (Polit & Beck, 2006). Respondents who self-identified as a non-biological caregiver were not included in the reliability analysis conducted for the Biological Parent subscale. Pearson’s point biserial (r_pb) assessed the construct validity, more specifically, convergent validity of the BIPSSI-FX subscale scores by diagnostic grouping. To demonstrate convergent validity, we expected moderate (>.30) to high (>.50) correlations (Cohen, 1988) in a positive direction between each subscale score and the total profile score with a confirmed genetic diagnosis of FXS.

Internal structure of the scale for adequacy and stability was examined using factor analysis and systematic cluster analysis which included maximum likelihood extraction followed by direct oblimin (oblique) rotation with Kaiser normalization. A Bartlett’s test of sphericity with a significance of <. 001 and a Kaiser-Meyer-Olkin measure of sampling adequacy ≥ 0.6 were used to determine data suitability for CFA (Kaiser, 1974). Both the sample adequacy and use of factor analysis on the data were supported (KMO =.734). Utilizing criteria established by Comrey and Lee (1992), the sample size was considered “very good”. Post rotation, an item was added only if it had an absolute factor loading of 0.30 or greater and it did not cross-load highly (> 0.35) on any other factor. Statistical significance was assumed for p < .05.

Results

Description of Respondents

Mean age of the caregiver respondents was 37.8 ± 8 years. Approximately 88% of the primary caregivers were first- or second-degree biological kin (i.e., mother, father, grandmother, aunt, uncle). The largest proportion (55%) of participants had an annual household income of greater than $50,000. More than 90% of the sample had some college education. Ancestral self-classifications for the sample were: European, 83%; Latino-Hispanic, 4%; Asian-Pacific Islander, 4%; African, 3%; Native American, 1%, and biracial-other, 5%. Mean subscale scores for the BIPSSI-FX with this sample ranged from 0.81 to 1.39. Table 1 provides a summary of subscale scores.

Table 1.

Profile scores, reliability estimates, and convergent coefficients for the BIPSSI-FX (n=686).

BIPSSI-FX Subscales
	Score (±SD)	Internal Reliability α	Convergent Coefficients R
			Fragile X Dx
Developmental Milestones	.92(.49)	.75	.16*
Social-Behavioral	.81(.40)	.85	.18*
Cognitive	1.39(.49)	.77	.16*
Biological Parent	.47(.32)	.58	.00
Total Scale	.83(.24)	.80	.20*

Note:

^*p <.05

Dx = diagnosis

Internal Consistency Estimation

Reliability estimates for the subscales ranged from a low of 0.58 to a high of 0.85 (See Table 1). Total scale Cronbach’s alpha coefficient was 0.80. Internal reliabilities for the developmental, social-behavioral, and cognitive subscales were acceptable ranging from a low of 0.58 for the biological parent subscale to a high of 0.85 for the social-behavioral subscale. These estimations were similar to those values generated with the African American pilot sample. The biological parent subscale, however, did not achieve the criterion of 0.70.

Convergent Validity

Correlation coefficients between the BIPSSI-FX subscale scores and the BIPSSI-FX total profile score with a positive FXS diagnosis via genetic analysis were compared. The highest coefficient generated (i.e., 0.20) did not meet minimum criterion of 0.30. Nevertheless, these findings showed some evidence of convergent validity. Except for of the biological parent subscale, coefficients were positive (see Table 1). Furthermore, children diagnosed with FXS had significantly higher BIPSSI-FX respondent scores when compared with the respondent scores of those who did not have a confirmed diagnosis of FXS.

Factor Structure

Maximum likelihood analysis with data from this sample supports the hypothesized four-factor model. Item loadings, eigenvalues, and percent of variance are presented in Table 2. The four factors accounted for approximately 45% of the variance. The first two factors represented the highest proportions of the variance (31%). All of the designated items for the social-behavioral, developmental milestones and cognitive subscales loaded as Factors 1, 2 and 3, respectively. Six of the seven items that form the biological parent subscale loaded as Factor 4 (See Table 2).

Table 2.

Factor structure, item content and item loadings of the BIPSSI-FX by confirmatory factor analysis (n = 686)

Item content and subscale loadings	1	2	3	4
Factor 1: Social-Behavioral
Emotional problems at home-school	.79
Emotional problems at home-school	.71
Problem with nervousness-anxiety	.65
Therapy for behavioral-emotional problems	.62
Medication for emotional-behavioral problems	.61
Exhibits sad or depressed mood	.56
Difficulty attention span-focus	.54
Self-inflicted injury	.48
Repetitive body movement	.47
Repetitive speech	.45
Difficulty falling or staying asleep	.41
Factor 2: Developmental Milestones
Age sit without support		.71
Age walking without assistance		.68
Age said first word		.67
Age first wave bye-bye		.51
Age first respond to name		.47
Factor 3: Cognitive
Speech problems or therapy			.75
Attend special education classes			.67
Language problems or therapy			.66
Told child has learning disability			.49
Factor 4: Biological parent
Depression in mother before becoming parent				.60
Depression in father before becoming parent				.55
History of developmental disabilities moms family				.44
History of alcohol or drug use mom or dad				.43
Females in family with cysts on ovaries or hysterectomy				.32
Problems with math mom or dad				.32
Eigenvalues	5.0	3.4	1.8	1.6
% of Variance	18.6	12.6	6.9	6.0

Discussion

The purpose of this study was to identify, evaluate and report psychometric properties of the BIPSSI-FX using a large international sample. Measurement performance with the current study demonstrates favorable psychometrics. With the exception of the biological parent subscale, internal consistency estimates support reliability of the BIPSSI-FX. Convergent validity between the subscale and profile scores with an FXS diagnosis was low. Findings from CFA demonstrate a coherent four-factor structure. A major point is the measurement performance of the biological parent subscale. The items within this subscale clustered, and appear to measure a single construct. Yet, the reliability estimation and convergent coefficient values generated with this sample were weak at 0.58 and 0.00, respectively. Examination of the correlation matrix revealed that item elimination would not have appreciably improved the reliability estimation. Although it appears that the subscale measures the construct of interest, the reliable capture of this information is questionable.

Possible reasons for the low reliability estimate (i.e., source of measurement error) generated for this subscale include language translation and differences in cultural and ethno-cultural behaviors. Globally, English is either spoken as a second language or not at all. Among individuals outside the U.S. who profess language proficiency, aspects such as word phrasing may compromise translation of the source language text. Conceptualization of the instrument’s response categories may encumber the participant’s understanding of profile items and influence the accompanying response.

Cross cultural psychologists suggest that differences in human behavior characteristics like parent-child socialization, expressions of social and emotional development, personality traits, and cognitive abilities can influence measurement equivalence (de Klerk, 2008). For example, two international participants commented in the open-ended sections of the survey that cultural variants made some sections of the surveys challenging to decipher. Their comments indicated that some respondents may have misinterpreted the directions needed to accurately complete the research instrument. Item adjustments may be warranted to obtain equivalent or comparable measures across cultures.

Additional obstacles specific to this study were the using the internet to collect data, transitory personal states, and lack of instrument clarity. Internet use limits the researcher’s ability to control for such impact conditions as environmental distraction, problems with the computer, or difficulties with test questions (Harris & Dersch, 1999). Transitory personal states such as the caregiver’s fatigue, hunger, anxiety, and fluctuations in mood have influenced the manner in which the primary caregivers respond on self-report measures with caregivers of children who have developmental delays or disabilities demonstrating increased vulnerability to these temporary states (Perske, 1981; Schilmoeller & Baranowski, 1998).

To further develop and refine the BIPSSI-FX, we plan to collect data on more culturally diverse children and conduct discriminate function analysis to determine the BIPSSI-FX items that best predict the pre- and full-mutations. The protocol related to future data collection will require both the respondents and their children to submit a biological sample for mutation analysis. In addition, the Biological Parent subscale will be expanded to include Fragile X Associated Tremors and Ataxia Syndrome (FXTAS), which is believed to have potential for the bidirectional identification of children with FXS and adults with the FMR1 premutation (Berry-Kravis et al., 2007; Berry-Kravis, Goetz et al., 2007; Berry-Kravis, Knox, & Hervey, 2011; Hagerman et al., 2009).

Study Limitations

Of particular importance was the inability to control for the intellectual capacity of the caregiver respondents. Lack of fluency in English or Spanish may limit one’s ability to read and interpret the test items. Moreover, the instrument has a potential cultural bias, particularly with the social-behavioral related items, in that it was created principally by an African American female. Input from several non-African American researchers counterbalanced this limitation. Although international data may help to overcome the limitations of language and cultural bias, further research is needed before cross-cultural generalization can be made. Another potential limitation is associated with self-report bias (i.e., telescoping, exaggeration and selective memory) and error variance in social assessments.

Conclusion

Statistical evidence supports the BIPSSI-FX as a valid instrument for identifying children at risk for FXS. Although the BIPSSI-FX cannot replace essential, standardized, early assessment tests, it provides a medium for parents as frontline screeners to articulate and quantify their concerns about their children’s development to professionals. This tool and the associated discussions it prompts between parents and health-care professionals may facilitate diagnoses earlier in a child’s life than is the norm presently. Using the BIPSSI-FX to augment comprehensive assessment of infants and toddlers may facilitate informed decisions of whether a child should be tested for the FMR1 gene mutation.