DySMA – an Instrument to Monitor Swallowing Function in Children with Spinal Muscular Atrophy ages 0 to 24 Months: Development, Consensus, and Pilot Testing

Step 1: Rapid Review and preparation

As a first step, we searched for original articles reporting clinical items for swallowing in SMA type 1 children (or with predominantly 2 SMN2 copies) younger than 24 months. These could be either clinical instruments, parts of clinical instruments, or single items reported as outcomes. The exact inclusion and exclusion criteria are listed in Appendix 1.

The Medline and CINAHL databases were searched from inception to August 2022 for original research published in English using medical subject headings (MeSH) and key terms from “infant”, “spinal muscular atrophy”, and “swallowing disorder”. This search resulted in 229 publications. Additionally, six records were identified through other sources. After removing duplicates, 223 records were screened, and finally, six full texts were included in the qualitative synthesis. In the case of multiple studies by the same authors using the same instrument, only one publication was included in the analysis (see Appendix 1 for the PRISMA flow diagram of the reviewing process [18]).

We developed an initial item pool based on the items identified in the rapid review and preliminary studies of the working group [11, 19]. The exact wording of the items and categories for bundling items were determined in formulation session 1 (FS1). Two SLPs, one physician (MD), and two psychologists from the working group attended the meeting, and the initial item pool was generated.

Step 2: Modified Delphi method

A multi-stage modified Delphi process consisted of several formulation sessions, online group interviews, and an anonymous online survey (see Fig. 1).

Interviews

The experts for the interviews were recruited in Germany via mailing lists of pediatric dysphagia networks and by e-mail using the snowball system. The goal was to recruit ten participants. Due to the serious lack of experience with this patient group, the search was adapted and expanded to include experts who had no explicit experience with SMA but with infant dysphagia and extended to Switzerland.

Eligibility was also conditional on being an SLP (or equivalent) and having self-reported expertise in dysphagia in children younger than two years.

We conducted semi-standardized interviews via the Zoom video platform using an interview guide. All participants provided written informed consent. The interviews were audiotaped.

Two SLPs with expertise in both the construct (swallowing development in children) and the target population (young children with SMA) conducted the interviews. Neither was eligible to vote.

The interviews aimed to discuss and reach a consensus on the relevance of the preformulated categories and items for the construct and the target group. In addition, the participants could add items that had not been previously recorded. Approval or rejection was nominally scaled (yes/no/not sure).

Qualitative data, such as discussions or additional suggested items, were transcribed with simple coding and included in the following formulation session (FS2). Categories and items were excluded, added, or reformulated, resulting in a new pool.

Online Survey

In the following round, an online survey was created using Lime Survey 5.x, hosted at Hamburg University (https://www.rrz.uni-hamburg.de/services/software/alphabetisch/limesurvey\!.\!html). The questionnaire aimed to generate consensus on the categories and items of the DySMA in terms of their relevance to the planned instrument using a 5-point Likert scale (1 = disagree; 5 = fully agree). In addition, the wording was assessed on a nominal scale (yes/no), including an open text box. Demographic data on occupation, work location, and years of experience were collected. An SLP and an MD pretested the function and completeness of the online survey.

SLPs, PTs, and NPs with experience diagnosing children with SMA (<24 months) were invited to participate in the survey. Only members of these professional groups with relevant professional experience could participate. Participants were recruited via e-mails to neuromuscular centers treating SMA patients and professional association mailing lists. Participation was anonymous. It was not known in advance how many potential participants with appropriate qualifications there were. According to COSMIN criteria [17], a minimum of 30 participants would have been desirable. The participants from the previous interviews were all invited to participate again, so the questionnaire also served as a summary and feedback on the results of the interviews.

After the final analysis of the online questionnaire, another formulation session (FS3) was conducted, and the framework of the tool, including the scoring, was determined. Practicability was discussed with two PTs and two NPs who had tested the DySMA in their clinical practice with the target group (cognitive debriefing). They provided feedback on the feasibility of individual items and the practicality of the instrument, resulting in another adjustment to the questionnaire.

Consensus definition and data analysis

All type of data was tabulated in Excel (2016, Microsoft) and exported for further analysis to SPSS (Version 27, IBM). Demographic data, frequencies, and percentages were analyzed descriptively.

The consensus threshold for nominal scaled voting was set at≥75% [20]. Concerning the 5-point Likert scale, a predefined threshold of 1.25 of the interquartile range (IQR) was set for consensus, corresponding to 75% [21].

Step 3: Pilot testing of the DySMA

The pilot trial was conducted in a single neuropediatric center treating SMA by two SLPs with three groups of children. Group I included eight healthy controls from the ongoing DySMA norm study. The children were recruited between April and June 2023 through playgroups, maternity clinics, and daycare centers in Hamburg, Germany. Inclusion criteria were birth after 37 weeks of gestation, unremarkable development, no known chronic illnesses according to routine pediatric check-ups, and an apparent absence of infection on the examination day. Group II is part of the ongoing DySMA study and includes six infants with SMA who received early and clinically pre-symptomatic therapy. Group III consists of six infants and toddlers with symptomatic SMA who received treatment. These are four children from the DySMA study and two from the CHILDYS-RD study, for which data have been previously published [11]. Inclusion criteria were genetically proven 5q-associated spinal muscular atrophy with two SMN2 copies and therapy with one of the three currently approved treatments. All parents provided written informed consent.

Due to different study designs, the measurement time points varied: children from the DySMA study had up to five data collection points, children from the CHILDYS-RD study had two, and healthy controls had one.

The implementation of the DySMA consists of a patient history section and an examination section (see Appendix 3). An SLP assessed the DySMA for all children, and the OrSAT [13] was used as a comparison tool. The instrument was chosen because it is most similar to the DySMA in terms of the aim, the construct measured, and the scoring. The Neuromuscular Disease Swallowing Status Scale NdSSS [22] and the p-FOIS [6] were also captured. Although a CSE was also performed in symptomatic children, it is not reported as a comparative outcome due to the lack of standardization [9].

Step 1: Rapid Review and preparation of the item pool

After evaluating the six included studies, two oral intake scales (NdSSS [22, 23] and p-FOIS [6]), and one oral and swallowing abilities instrument (OrSAT) [13] were identified. One instrument, the HINE [24], includes an item on sucking/swallowing, as well as the Floppy Infant Module [25]. A total of nine outcome parameters could be identified in the included publications and were reported in the results as part of the instruments used: oral intake/ type of nutrition [6, 13, 23], tongue fasciculation [23, 26], difficulty managing secretions [23], weak sucking [5, 23, 26, 27], weak voice [5, 23], ineffective cough [5], wet voice/breathing [5], coughing while feeding [5, 13], fatigue while feeding [5, 13].

Based on the clinical experience of the research group and two preliminary studies [11, 19], additional items were added on gastrointestinal problems like vomiting and reflux, compensation while feeding, and eating behavior. These studies were published after the rapid review search period and, therefore, could not be included in the review.

From a clinical perspective, the questionnaire needed to be short and also usable by non-specialists in dysphagia. This resulted in a rather small item pool. After the first formulation session (FS1), 36 unique items classified into the following ten categories were included in the Delphi process: 1) type of nutrition, 2) compensation, 3) secretion management, 4) choking, 5) eating behavior, 6) fatigue, 7) gastrointestinal problems, 8) strength (voice and cough), 9) non-nutritive sucking and, 10) atrophy (see Appendix 2 for detailed item list/detailed process). A nominal rating system (yes/no) was adopted for the items of the tool.

Step 2: Modified Delphi method

Online survey

Due to the acquisition via large mailing lists, it is unclear how many potential participants were initially reached. A total of 28 participants attempted to complete the questionnaire. Of these, 23 met the inclusion criteria. Finally, 19 questionnaires were filled in completely and used for analysis. Six participants had also taken part in the interviews in the first round.

Of the 19 participants, eight were SLPs with a median experience of three years (range 0–19 years) for the target group. Seven participants were PTs with a median experience of 28 years (range 6–32 years). The remaining four were NPs with a median experience in children with SMA younger than 24 months of 11 years (range 7–25 years).

Most participants worked in centers or hospitals with a neuropediatric focus (n = 13), three were employed in a clinic with a different focus, and three worked in an outpatient practice.

Consensus on relevance

Nine of the ten categories were deemed relevant and did not exceed the IQR threshold value of 1.25. The category gastrointestinal problems crossed the threshold with an IQR of 2 and were discussed in the following FS3 (see Appendix 2).

Of 31 items, a clear consensus on relevance was found for 26 items (IQR <  1.25) (see Appendix 2). Four items were finally excluded after discussion in the last FS and after a cognitive debriefing (CD) with users. Appendix 2 shows that the items bites of food, noticeably high/narrow palate, vomiting, and extremely picky eating habits were retained even though the threshold was exceeded. The application-related CD showed that parents frequently reported these parameters and that they represented a typical symptom for the target group. In addition, after discussion, the items requires many small meals, eats very slowly/takes a long time for a meal, falls asleep while eating, and suspected reflux were excluded because they overlapped with other items and could not be clearly scored in the application.

Consensus on wording

Consensus was reached on the wording for six categories; there was also minimal rephrasing for two items. Four categories were reworded because the 75% consensus threshold was not reached (see Appendix 2). Comments on the rewording in the open text field concerned the changing, omitting, or adding of individual words, which are not listed here in detail.

Framework

Regarding the framework, it was decided in the last FS that categories 2-6 are automatically scored as 0 if oral nutrition is no longer possible (type of nutrition≤1). This way, ambiguities in scoring during cognitive debriefing are eliminated (see Appendix 3 DySMA). Ambiguities in assessing children who were not fed orally also occurred in earlier prototypes and the implementation of the OrSAT, so this point was important to users.

The final pilot version of the DySMA consists of two parts (patient history and examination), ten categories (see Appendix 2), and 36 items. In the first Delphi round, there was consensus that the tool does not require direct observation of food intake (see Appendix 2). For this reason, all items related directly to food intake and the secretion management category were finally assigned to the patient history. All items that are considered non-nutritive were assigned to the examination section. These also require evaluation by the examiner for accurate assessment (e.g., whether the jaw opening is limited or if there is significant paradoxical breathing). A score between 0 and 35 points can be achieved; the higher the score, the more advanced the swallowing development (see Appendix 3).

Step 3: Pilot-testing of the DySMA

The characteristics of the three pilot groups are shown in Table 1. In groups I and II, there were more girls than boys overall. Further detailed information, such as the CHOP-INTEND [16] and long-term trajectories, has been published elsewhere [11, 19] or will be published after the completion of the DySMA study proper.

Figure 2 shows the DySMA total scores of the three samples in the top row, with the OrSAT measurements below. It should be noted that, strictly speaking, no progress is shown for healthy controls, but only separate measurements of individual infants and toddlers of different ages are shown.

Fig. 2

Graphic representation of the development of swallowing measured with the DySMA (a-c) and the OrSAT (d-f); a; d) in a healthy control sample; each point represents one case, b; e) in an SMA sample, the children were clinically pre-symptomatic before the start of therapy, c; f) in an SMA sample, the children were symptomatic before the start of treatment, the lines represent the course of a case.

The DySMA score of healthy controls ranges from 23 to 35 (35 is the maximum score), and the OrSAT value ranges from 5 to 12 (12 is the maximum score); for both tools, the maximum value is reached from 22 months. For all children, the oral intake scales remained at the maximum value.

In pre-symptomatic children, the DySMA ranges from 17 to 32 and the OrSAT from 4 to 11. None of these children reached the maximum value. However, it should be noted that the maximum age in this group is 17 months, not 22 months like in the healthy control group. The NdSSS ranged from 7 to 8, and the p-FOIS from 4 to 6.

The symptomatic sample had a DySMA range from 1 to 27 and an OrSAT range of 1 to 10. The NdSSS ranged from 1 to 8, and the p-FOIS from 1 to 6. As in the pre-symptomatic sample, the maximum age was 17 months.

It is evident that the development of swallowing is more pathological in the symptomatically treated group than in the other groups. The trajectories in the healthy controls and the pre-symptomatically treated children could be similarly represented with the DySMA and the OrSAT. However, the OrSAT trajectory was somewhat steeper due to the distribution of scores. The symptomatic group started with a lower OrSAT score and stagnated or improved over time, whereas the DySMA tended to worsen (Fig. 2).

Limitations and considerations for future studies

As far as the Delphi method is concerned, the small number of participants was a clear weakness. According to the COSMIN recommendations [17], a sample size of less than 30 persons is insufficient. In German-speaking countries, however, there has been little but increasing expertise in the field of swallowing in infants and toddlers. At the same time, there is little experience with children with SMA under 24 months of age. In addition, newborn screening in Germany only began in autumn 2021, so the number of cases is still low. For this reason, 19 individuals can be considered adequate. The same applies to the sample size of the pilot sample.

Regarding the classification of the groups, it should be mentioned that the pre-symptomatic sample was clinically classified and not based on device measurements, such as compound motor action potential (CMAP).

It is important to note here that ideally, children with SMA should receive a detailed swallowing examination by an SLP, which cannot be replaced by performing the DySMA. Due to a general shortage of SLPs in Germany in this area and the underrepresentation of SLPs in SMA centers and research, an interprofessional approach will be necessary. In our online survey sample, we saw that PT (Mdn 28y) and NP (Mdn 11y) currently have significantly longer experience with the target group than SLP (Mdn 3y); there is still some catching up to do here. Regarding the pilot testing of the DySMA, it must be taken into account that SLPs used the tool in our study. Although two PTs and two NPs tested the applicability of the tool and contributed to the final version in a cognitive debriefing, in the future, the reliability between PT, NP, and SLP must also be checked in order to make a final statement about the interprofessional applicability.

Concerning the DySMA framework, it can undoubtedly be critically discussed that it has to do without direct observation of food intake and that it is a combined tool of patient history and examination. In the formulation sessions, it was discussed that parents of children with SMA already have to fill in several Parent-Reported Outcome Measures (PROMS) at the examination appointments (at least, this is the case in our center). In the experience of the examiners, it was easier to ask questions and give examples in an oral patient history. However, it also became apparent that some items require specific assessment by the examiners (examination section, e.g., reduced maximum jaw opening) as they cannot be judged effortlessly by parents.

DySMA tool’s next steps are to check the reliability and validation using larger samples (DySMA study) and norming using healthy controls (DySMA norm study). In addition, the English translation needs to be reviewed by an international expert consensus and made available internationally. Due to the condition’s rarity and the so far poorly-researched target group under 24 months of age, collaboration with expert groups worldwide is valuable here.