Participants
Data from participants in the RTT Natural History Study (Study #5211), part of the Rare Disease Clinical Research Network, were used for this secondary data analysis study. Participants were enrolled at 14 sites across the United States from 2014 to 2021: University of Alabama-Birmingham, Baylor College of Medicine, Boston Children’s Hospital, University of California San Diego, Children’s Hospital Colorado/University of Colorado School of Medicine, Children’s Hospital of Philadelphia, Cincinnati Children’s Hospital, Gillette Children’s Hospital, Greenwood Genetic Center, Oakland Children’s Hospital, Rush University, St. Louis Children’s Hospital, Vanderbilt University Medical Center, and Washington University School of Medicine. Individuals who were diagnosed with classic RTT (Neul, 2010) by an experienced child neurologist, geneticist, or developmental pediatrician were included in the analyses.
From 2014 until September 2021, there were 649 participants with classic RTT. For this analysis, we used the baseline visit data. We randomly selected two-thirds of the participants for the development sample (n = 425) which was used to identify the factor structure and the remaining one-third were used as a validation sample (n = 224). Table 1 presents the demographic data for both samples. Over 99% of participants were females. The average age was 14.6 years (10.2 SD). The majority of participants were white and non-Hispanic. Similar to other reports in RTT (Cuddapah et al., 2014; Neul et al., 2008), participants had a variety of mutation types, with R168X, R306C, T158M, Carboxy-Terminal Truncations (CTT), and a group of other point mutations (i.e., all other point mutations besides the 8 common recurrent MECP2 mutations) being the most common.
Measures
Interval History Form. The Interval History form is a caregiver-completed measure used to gather a variety of information about individuals with RTT, including living arrangement, overall functioning and symptom severity, classroom type, therapy use, and socioeconomic indicators. For these analyses, we focused on items that assessed RTT symptoms across six core domains: (1) Hand skills; (2) Communication; (3) Sitting, standing, and walking;(4) RTT behaviors (e.g., hand wringing, hyperventilation, teeth grinding); (5) Mood and abnormal behaviors; and (6) co-occurring conditions (e.g., pain tolerance, constipation, drooling). Of note, items that assessed improvement in symptoms over the prior 6 months were not included in the analyses.
For each item across the six domains, parents were asked to assess symptoms exhibited in the previous 6 months. Many items use a 5-point Likert-type response scale (e.g., Never, Occasionally, Frequently, Very Frequently, Constantly). Others assess the presence of a behavior (e.g., “My child has demonstrated the following hand skills”). Given the Interval History form was not originally designed as an outcome measure, several items needed to be recoded prior to the analyses.
First, all items on the Interval History Form were rescaled such that a higher score indicated more severe RTT symptomology. In addition, several items from the Hand skills, Communication, and Gross motor domains needed to be re-coded prior to conducting the analyses. A single item on the Hand skills domain asked parents to rate whether their child demonstrated a number of different hand skills. For each, parents indicated whether it was present and if the child could do the skill alone or hand-over-hand only (i.e., with assistance from someone else). We ranked the hand skills by difficulty based on clinician input. We then converted these skills into a 21-point scale by combining ability to perform with type of skill. The highest severity score of 20 was assigned to individuals with no hand skills whereas the lowest severity score of 0 was for those who could use a pincer grasp with finger alone. For the communication domain, we re-scaled the item assessing the child’s spoken language skills to a 5-point Likert-type scale. Similar to the Hand skills item, we ordered the language skills by difficulty based on clinician input and assigned each a score, with the higher score indicating higher levels of severity. A score of 5 was used when an individual had no spoken language or sounds, except crying or screaming and a score of 0 was used when the individual spoke normally or used complete sentences. An overall walking variable was created using ability to walk and walking distance, from 0 (independently walking >100 yards) to 8 (inability to walk). A sitting variable was created based on ability to sit and level of assistance needed, from 0 (Sitting without help for a long time [greater than 5 minutes]) to 6 (can not sit alone). A similar variable was created for standing. A total score was calculated based on the results of the psychometric analyses.
Clinical Severity Scale (CSS). The CSS (Neul et al., 2008) is a clinician-reported measure that was developed to monitor and assess the severity of clinical features in RTT. The 13-item scale assesses common clinical features in RTT, including age of regression, age of stereotypy onset, growth deceleration, sitting and walking ability, hand use, scoliosis, verbal and non-verbal communication, respiratory function, autonomic symptoms, and seizures. Each item is rated on an item-specific 5- or 6-point ordinal scale, with some scores ranging from 0 to 4 or 0 to 5. A total score is calculated, with higher scores indicating greater clinical severity.
Revised Motor-Behavior Assessment Scale (R-MBA). The R-MBA is a 24-item scale completed by clinicians (Raspa et al., 2020). The measure assesses RTT symptoms across five domains: (a) Motor dysfunction, (b) functional skills, (c) social skills, (d) aberrant behavior, and (e) Rett-specific symptoms. Each item is rated on a 5-point ordinal scale, with scores ranging from 0 to 4. Two items (mouthing hands/objects and stereotypic hand behaviors) are combined to improve item distribution and then rescaled to a 5-point scale (0 =0, 1-2, =1, 3-4 = 2, 5-6 = 3, 7-8 = 4). A total score is calculated by summing all items. Higher scores indicate more severe functioning.
Clinical Global Impression Severity Scale (CGI-S). The CGI-S is a single-item measure completed by clinicians which assesses severity of illness based on current functioning. The CGI-S uses a 7-point Likert type scale (1=Normal, not at all ill; 2=Borderline ill; 3=Mildly ill; 4=Moderately ill; 5=Markedly ill; 6=Severely ill; 7=Extremely ill); however, clinicians were provided a RTT-specific version which included descriptions of functioning across seven domains (Language/communication, Ambulation, Hand use, Social, Autonomic, Seizures, Attentiveness) which served as clinical anchors for each response option (Neul et al, 2015).
Child Health Questionnaire (CHQ). The parent-reported CHQ was used as a measure of health-related quality of life in children and adolescents (Landgraf et al., 1996). Parents rated their child’s health over the past 4 weeks using the 50-item form, which uses Likert-type scaling. Items assess a variety of health constructs, such as physical functioning, bodily pain, behavior, self-esteem, and mental health. Factors are converted to standardized Z-score and then combined into two component scores, Physical health (CHQ-PH) and Psychosocial health (CHQ-PS), which are each transformed to a standardized distribution (mean 50, SD 10), with higher scores indicating better or more positive health states. The CHQ has been used previously in the RTT Natural History study to examine quality of life (Buchanen et al., 2019; Lane et al. 2011)
Scales measure Physi-
cal Functioning, Role/Social-Emotional/Behavioral,
Role/Social-Physical, Bodily Pain, Behavior, Mental
Health, Self-Esteem, General Health, Parental
Impact-Time, Parental Impact-Emotional, Family
Activities, Family Cohesion, and Change in Health
Statistical Analyses
Descriptive analyses were conducted using SAS v7.15 to examine item-level means, standard deviations, skewness, and kurtosis on the Interval History form. Next, we conducted psychometric analyses using a split-sample design. A two-thirds random sample of the participants (n = 425), referred to as the development sample, was used to conduct exploratory factor analyses and the remaining one-third of participants (n = 224), the validation sample, was used for a confirmatory factor analysis. The development sample was used to explore the most appropriate factor structure for the items. Using SAS PROC FACTOR, we examined a series of possible factor structures. We used an oblique rotation method to allow for correlations between the factors. We used several criteria to determine the best factor solution, including the pattern of factor loadings (i.e., demonstration of simple structure), size of the factor loadings (above 0.40), the percentage of variance accounted for by each factor, correlation (for possible redundancy), and clinical relevance. Next, the validation sample was used to examine the consistency of the final factor structure. In Mplus (Muthén & Muthén, 1998–2017), we used weighted least square mean and variance adjusted (WLSMV) estimation to examine the categorical data. Model fit was assessed based on three fit indices: Tucker-Lewis index (TLI), comparative fit index (CFI), and root mean square error of approximation (RMSEA). The TLI and CFI are relative fit indices that compare the final model to the null or baseline model (i.e., a model with the worst fit). The RMSEA is an absolute fit index that measures how perfect a fit our model is. Preferred values of the TLI and CFI are 0.90 or greater and less than 0.08 for the RMSEA (Schreiber, Nora ,Stage, Barlow, & King, 2006).
After the factor structure was established, reliability and validity were examined using the full sample. Inter-class correlations were computed to examine the relationship between the subscales. Next, we calculated Cronbach’s alpha for each subscale and the total score on the RCASS as an indicator of reliability. We examined the total score and Cronbach’s alpha by mutation type and age as a second measure of internal consistency. ANOVAs were conducted to examine differences in mean total scores by mutation type and age group as an indicator of construct validity. Generalized linear regression analyses were conducted to determine whether existing Rett measures, including the CSS and R-MBA, and other validated measures (i.e., CGI-S and CHQ) were associated with the total and subscale scores on the RCASS.