Promoting Factors for Physical Activity in Children with Asthma Explored through Concept Mapping

For children with asthma, physical activity (PA) can decrease the impact of their asthma. Thus far, effective PA promoting interventions for this group are lacking. To develop an intervention, the current study aimed to identify perspectives on physical activity of children with asthma, their parents, and healthcare providers. Children with asthma between 8 and 12 years old (n = 25), their parents (n = 17), and healthcare providers (n = 21) participated in a concept mapping study. Participants generated ideas that would help children with asthma to become more physically active. They sorted all ideas and rated their importance on influencing PA. Clusters were created with multidimensional scaling and cluster analysis. The researchers labelled the clusters as either environmental or personal factors using the Physical Activity for people with a Disability model. In total, 26 unique clusters were generated, of which 17 were labelled as environmental factors and 9 as personal factors. Important factors that promote physical activity in children with asthma according to all participating groups are asthma control, stimulating environments and relatives, and adapted facilities suiting the child’s needs. These factors, supported by the future users, enable developing an intervention that helps healthcare providers to promote PA in children with asthma.


Introduction
Physical activity (PA) is beneficial for children's health [1]. This especially holds for children who suffer from a chronic disease [2,3]. Asthma is the most frequently diagnosed chronic disease among Children with asthma who engage only a small amount of time in PA are referred to healthcare providers, such as paediatric physical therapists and pulmonary nurse practitioners. There they participate in exercise training, learn how to cope with exercise barriers, and are motivated to be more physically active. Healthcare providers indicate that they need interventions that are effective in improving and maintaining PA for children with asthma on the long term. To develop a potentially effective intervention, it is important to implement promoting factors that suit the needs of children with asthma. In the current literature, these factors are mostly studied from the perspective of parents and experts instead of children and the focus is often on barriers instead of facilitators [15,16]. Thus, better insight in promoting factors from the perspective of children with asthma is needed. Furthermore, studying the factors from the perspective of parents and healthcare providers is also important since parents know their children best and play an important role in the PA behaviour of their child, and healthcare providers play an important role in promoting PA in these children and controlling their asthma.
Better insight into the factors that promote their PA is required to develop an intervention that increases PA levels in children with asthma. Here, increasing PA levels is about increasing PA in general and not specifically about sports activities. In the process of developing an intervention that fits the needs of these children, insight into these factors is required. Therefore, three stakeholder groups are important: children with asthma themselves, who are the experts of their own behaviour; parents of children with asthma, who know how to stimulate their children; and healthcare providers of children with asthma, who have to implement the intervention in their care and have important knowledge about the medical barriers. Accordingly, the aim of our study is to explore the promoting factors of PA in children with asthma through the perspective of children with asthma, their parents, and their healthcare providers to implement these factors in the intervention.
Concept mapping (CM) is a helpful participatory method to acquire knowledge from a participant's point of view. It combines qualitative and quantitative methods by collecting data in a structured way and analysing this data quantitatively [17,18]. To implement the most important factors in an intervention, the aim of our study is to explore the promoting factors of PA in children with asthma from the perspective of children with asthma, their parents, and their healthcare providers.

Recruitment
Children and their parents were recruited through paediatric pulmonologists and paediatric physiotherapists in three different hospitals in The Netherlands specialised in paediatric asthma (Amsterdam University Medical Center, Amsterdam; Tergooi Hospital, Hilversum; and Kinderkliniek, Almere). Children who were diagnosed with asthma, aged between 8 and 12 years old, and either active or inactive could be included. Children who met the inclusion criteria received a study information letter from their physician. Their physician reported their age, activity level, and asthma status. The activity level was based on their PA participation in a general week and was labelled by their physician as low (not doing sports at all and not playing outside regularly), moderate (playing sports once a week and playing outside regularly), and high (playing sports at least twice a week and playing outside regularly) levels of PA. The asthma status, which corresponds to the level of asthma control, was defined as mild/well controlled, moderate/partially controlled, or severe/poorly controlled. All children are treated by a paediatric pulmonologist in a specialised hospital who prescribes medication according to the "Dutch Paediatric Society" guidelines for asthma in children [19]. The treatment includes taking medication before PA to prevent or diminish exercise induced symptoms.
Written informed consent was obtained from both parents or legal guardians for participation of a child. A separate individual written informed consent was acquired for participation of a parent. In total, 28 children and 23 parents gave written informed consent and, respectively, 25 and 17 actually participated.
Healthcare providers could be included if they were either physical therapist or lung nurse practitioner and if they had experience in treating children with asthma. Information letters were sent to potential participants via both a Dutch paediatric physical therapist profession group and the paediatricians of the participating hospitals. A written informed consent was obtained for participation. In total, 23 healthcare providers gave consent and started the study; two dropped out due to illness.
The AMC Medical Ethical Committee approved the research protocol with trial number METC 2017 191. This study was registered in the Dutch Trial Registry with trial number NTR6658 on 21 August 2017.

Procedures
The CM method consists of six steps, which are shown in Figure 2. Data were collected between November 2017 and March 2018. Descriptive participant characteristics, means and standard deviations were calculated with R version 3.4.1.

Concept Mapping Sessions with Children and Parents
The CM sessions for the children and parents were organised on two different dates. Sixteen children, divided into two groups of seven and nine children, and twelve parents participated on the first day. Nine children, divided into two groups of four and five children, and five parents participated on the second day. This resulted in a total of four groups of children and two groups of parents with whom CM sessions were held. All groups were facilitated by one or two trained researchers. Each researcher had received a total of eight hours of training, consisting of reading the concept mapping manual and attending an oral presentation provided by an expert in concept mapping. Thereafter, a pilot test of the concept mapping session was performed with a group of five healthy children. The final concept mapping sessions took approximately five hours and were split into two parts with a fun and interactive workshop in between.
To get familiar with the CM method and the subject of asthma and PA, all groups started with a warming up question: "Which activities belong to the definition of physical activity?". Thereafter, the seeding statement was presented in two ways: "What would help a child with asthma to become more physically active" and "A child with asthma would be more physically active if . . . ". Each participant individually wrote down as many ideas they could think of in response to the seeding statement. Subsequently, the participants shared their ideas one by one in a group brainstorm, during which participants could come up with new ideas as well. The ideas were checked on uniqueness and clarity by the group, resulting in a list of unique ideas.
The unique ideas were subsequently printed on separate cards for each participant. Each participant individually grouped all ideas into piles based on similarity, and named all piles. A minimum of two piles was required with each pile containing at least two ideas. An "other" or "miscellaneous" pile was prohibited. Subsequently, each participant rated the importance of all ideas in being helpful to become more physically active, using a five-point Likert scale ranging from very unimportant (1) to very important (5).

Concept Mapping Sessions with Healthcare Providers
CM sessions with healthcare providers were performed online with a self-developed web application. The process of the CM session was comparable to the session of the children and parents, except for the group brainstorm since the professionals individually used the online web application. After the researchers deleted duplicates, each participant received the list of all unique ideas from all healthcare providers together. Each participant was asked to check the list on the clarity of each idea. They were stimulated to add new ideas when they were inspired by other ideas on the list. Based on their comments and additions, the researchers generated the final list of unique ideas that was presented in the web application. Next, participants were asked to perform the sorting, naming and rating tasks using the web application.

Generating the Concept Maps
For each participant of a group, a matrix was built based on the piles they composed. In this, a cell pi, jq contained 1 in case the participant stored idea i and j in the same pile, and 0 if the ideas were placed in different piles. Then, the matrices of all participants in the same group (i.e., all matrices of the participants that sorted the same list of unique ideas) were summed. Subsequently, the result was divided by the number of participants. This resulted in a k gˆkg distance matrix per group, with k g being the number of ideas within a group, in which all pairwise distances between all ideas were represented. Since it is impossible to visualise a k-dimensional space, the distance matrix had to be transformed into a two-dimensional space. This was done using the multidimensional scaling (MDS) algorithm [20]. The stress factor indicates how well the original pairwise distances in the k-dimensional space are preserved in the two-dimensional space. The MDS algorithm optimises this stress factor resulting in a two-dimensional space in which the original pairwise distances are represented to the best extent possible. Subsequently, the relative distances in the two-dimensional space were visualised. In this representation, ideas that participants often placed in the same pile appear closer together than ideas that were less often grouped together. Clusters of ideas were made with hierarchical cluster analysis. Specifically, Ward's method, which minimises the within-cluster variance, was used. The default number of generated clusters was 6. The average importance values for each cluster was calculated based on the importance rating of the underlying ideas.
RCMap was applied to create the concept maps from the sorted ideas of the participants [21]. This open source software is implemented for R, of which we used version 3.4.1.

Interpreting the Concept Maps
Four researchers (A.B., K.B., A.T., and A.Br.) independently explored the clusters on interpretation and analysed whether more or fewer clusters would represent the participant's ideas better. The final clusters were named by the researchers based on the pile names of all participants.
To place all clusters in a theoretical framework, two researchers (A.B., K.B.) categorised all clusters of all concept maps according to the categories of the PAD model. A cluster was labelled as either a personal factor (attitude, intention, self-efficacy, health condition, or personal facilitators and barriers) or an environmental factor (social influence, or environmental facilitators and barriers). To visualise the overlap and differences between all stakeholder groups, we represented the unique clusters in a Venn diagram.

Participant Characteristics
The 25 participating children were on average 9.6˘1.0 years old and 60% were female. Of these children, 14 had mild asthma symptoms, five had moderate asthma symptoms, and six were diagnosed with severe asthma. There was only one child that had low levels of PA, seventeen children had moderate levels of PA, and seven children had high levels of PA.
The 17 participating parents were on average 43.4˘6.3 years old and 61% were female. Ten parents had a child with mild asthma, four parents had a child with moderate asthma, and three parents had a child with severe asthma. Thirteen parents had a child with moderate levels of PA, and four parents had a child with high levels of PA. No parents of children with low levels of PA participated.
The 21 healthcare providers that completed the study were on average 40.8˘11.3 years old and 85% were female. They had on average 15.9˘10.9 years of work experience in their profession and 7.9˘6.8 years of work experience with children with asthma.

Children's Clusters
In total, the four groups of children generated 51, 49, 25, and 36 ideas that were subsequently categorised into, respectively, ten, seven, four, and five clusters. Figure 3 shows an example of a concept map of one group of children. Table 1 shows the corresponding ideas to this concept map. The concept maps and lists of ideas of the other three groups of children can be found in Appendix A. Table 2 presents all clusters of the four groups of children together with their average importance rate and standard deviation (SD), and the label according to the PAD model.
The most important clusters for the four groups of children were getting a positive feeling of PA, having good asthma control, play outside more often, and get rewards for doing exercises or PA. Children in all groups mentioned that playing outside more often would stimulate their PA. Moreover, children in three out of four groups mentioned that making physical activities joyful or cool would stimulate them. Clusters regarding getting positive feelings of PA, being motivated by others, having access to sports facilities, and adding competitive elements to PA were mentioned by two out of four groups of children. Three out of four groups rated a cluster that was categorised as personal as most important, while one group rated an environmental cluster as most important.  Table 1. Table 1. All ideas, together with their importance rates and standard deviation, of the first group of children categorised per cluster. The numbers of the ideas correspond to the numbers in the concept map shown in Figure 3.

Parents' Clusters
Two groups of parents generated 37 and 36 ideas, which were grouped into, respectively, eight and five clusters. The concept maps and lists of ideas of both groups of parents can be found in Appendix A. Table 3 shows all clusters with their average importance rating and SD, and the label according to the PAD model. The most important clusters for the two groups of parents were about external factors that positively affect asthma control, and the child having sufficient knowledge about asthma, PA, and medication. Both these clusters were labelled as personal factors. There was only one overlapping cluster between both groups of parents, which regarded the child being in an environment that stimulates PA.

Healthcare Providers' Clusters
The group of healthcare providers generated 116 ideas in total, which were grouped into thirteen clusters. The concept map and list of ideas of the healthcare providers can be found in Appendix A. Table 3 shows all final cluster names. Of all thirteen clusters, five were categorised as personal factors, and nine regarded environmental factors. The cluster named "child's relatives have sufficient knowledge about asthma, PA, and medication" was rated as most important. Figure 4 shows the Venn diagram of all unique clusters generated by children, parents and healthcare providers. In total, nine clusters regarded personal factors, whereas seventeen clusters regarded environmental factors. Although most clusters were labelled as environmental factors, Tables 2 and 3 show that five out of seven groups rated a cluster representing a personal factor as most important. The children identified nine unique clusters, i.e., that neither of the other stakeholder groups came up with. The parents and healthcare providers identified, respectively, six and five unique clusters. The cluster "setting (realistic) goals" overlapped for the children and healthcare providers, while the cluster that overlapped for the parents and healthcare providers was "knowledge of child about asthma, PA, and medication". No overlapping cluster was found between the children and the parents. The four clusters "having good asthma control", "physical activity and exercises that fit the child's needs and capabilities", "being stimulated and motivated by family/others", and 'having stimulating situations and (school) environments" overlapped for all three stakeholder groups.

Principal Results
This CM study explored children's, parents', and healthcare providers' perspectives on factors promoting PA in children with asthma. According to all stakeholder groups, factors that promote PA in children with asthma are: having good asthma control, having physical activities and exercises that fit the children's needs and capabilities, being stimulated and motivated by family members/others, and having stimulating situations and (school) environments.
The results of this study will be used to develop an intervention, for instance a smartphone application or a game, that helps healthcare providers to stimulate PA in children with asthma. Both the categories of the PAD model and the differences and similarities visualised in the Venn diagram will help to translate the promoting factors into requirements for the tool to be developed. Especially factors regarding intention, attitude, and self-efficacy are important for behavioural change. Moreover, we have special interest in factors that were mentioned by all stakeholder groups, since independently generating the same factors emphasises the value of these factors.
Fun, playing together, and being stimulated and motivated by family or others have been found to be important factors in stimulating PA levels of children with asthma. The factors fun and playing together were mentioned by the children, whereas the factor about stimulation and motivation from family or others were mentioned by all three stakeholders. The result regarding the factor fun was comparable to the results of Noonan et al. (2016) who showed that 10-11-year-old primary school children indicated the factors fun, enjoyment, competence, and PA provision as important for their (out-of-school) PA participation [22]. The factors regarding playing together and motivation and stimulation from family or others were not found for children with asthma specifically, but they were found in children with a disability [23].
Interestingly, both the healthcare providers and parents felt that children should know more about asthma and the effects of asthma on PA and medication to become more physically active, whereas this was not mentioned by the children themselves. When we noticed this discrepancy on the study days, we asked the children after completion of the CM session whether they felt that learning more about asthma, PA and medication would help them to become more physically active. The overall reaction was that they did not think so. In fact, many children indicated that they already knew everything about it. Their parents explained that the children indeed feel that they know a lot about it, but parents felt it was not enough, or that children were not able to convert their knowledge to corresponding behaviour. This was confirmed by results from a previous interview study among 15 children (7-17 years old) with asthma and their parents which showed that children do have knowledge about asthma-triggers, but need to be more aware of asthma symptoms and increase their knowledge about medication use [24]. In addition to the discrepancy between the children and the adults, a discrepancy between the parents and the healthcare providers was found. Although they both felt that increasing the children's knowledge about asthma would stimulate PA, they did not feel the same about the knowledge of the children's relatives regarding asthma. The healthcare providers felt that the children's relatives, including the parents, should know more about asthma and the effect of asthma on PA and medication to stimulate the children's PA, whereas the parents only felt that their children lack this knowledge.
Another interesting result was that no overlapping clusters were found between the children and their parents. There can be various reasons for this result. Parents may for instance have other insights in the disease and they may reflect on situations differently. On the other hand, we did find overlapping clusters between the children and the healthcare providers, so the age difference between the children and parents does not seem to be a reason for finding no overlapping clusters. The lack of overlapping factors may indicate that parents do not exactly know what will promote PA for their own child. Not knowing promoting factors of their own child might be a PA barrier on its own. Buffart et al. (2009) also used the PAD model to assess stimulating factors of PA. This particular study assessed barriers and facilitators in young adults with a childhood-onset physical disability. They showed that PA related environmental facilitators in young adults with physical disabilities can be diverse while personal facilitators are limited to personal benefits and perceived rewards [25]. It therefore is not remarkable that our study resulted in more clusters with environmental factors than clusters with personal factors. Our results might suggest that environmental factors are more important for the stakeholders in promoting PA. On the contrary, five out of seven groups rating a personal factor cluster as most important might indicate that personal factors are more prominent in their ability to improve PA behaviour in children with asthma. Although the participants felt that solutions regarding their personal behaviour or skills might impact most on PA, they could think of more environmental solutions.
To the best of our knowledge, only one study assessed the influencing factors of PA in children with asthma [15]. Kornblit et al. (2008) examined the perceptions of barriers to PA and improvement opportunities in children with asthma. In contrast to our study, it was only about parental perceptions and the children's perceptions were not studied. Although they focussed on the barriers, whereas we focussed on the facilitators, the results of both studies were comparable regarding the parents. Both the parents in the study of Kornblit et al. and the parents in our current study mentioned that children would be more physically active if schools would have more appropriate physical activities and a stimulating environment. Moreover, knowledge of asthma at schools was reported as important in both studies. Parents felt that their children would be more physically active if schools would know more about asthma symptoms in relation to PA, and when teachers would stimulate their children to take their asthma medications if necessary.

Strengths and Limitations
A strength of this study is the combination of qualitative and quantitative methods using CM. The CM method turned out to be a solid method to answer our research question. Moreover, the CM sessions were not only valuable to answer our research question, but also valuable for the children and parents. Many of them felt that they benefited from talking with fellow patients or fellow parents of a child with asthma.
Another strength is that we developed a tool to perform CM online, which enabled a number of healthcare providers to participate, while they would not have been able to participate in case of face-to-face sessions. The developed tool to perform CM online has promising possibilities for investigating other complex research questions or other participants. The generalisability of the results further strengthened our study. Most of our results are not typically asthma related, but show factors related to the patient's environment and to coping strategies involving the patient itself, parents, teachers, and sports instructors. It is known that these factors apply for many other children with a chronic disease [26].
A limitation of the online CM procedure with healthcare providers was that a group discussion was not possible. In the construct of CM, a group brainstorm is advised to enable participants to, in this case, expend their thoughts about PA stimulation when receiving ideas from others [27]. We tried to minimise the effect of not having a group discussion by giving the participants the opportunity to add new ideas after reading the ideas of their fellow participants.
To minimise participant's burden, the CM sessions were performed on one day. Therefore, the ideas of the different groups could not be combined. As a consequence, participants sorted and rated the ideas of their own group, resulting in one concept map for each group.
Another limitation of our study is the small number of participants per group and the imbalance between level of activity and asthma severity of children who participated. Although we tried to create a heterogeneous group of child participants, a relatively large part of the group had moderate levels of PA, while there was only one child with low levels of PA. Moreover, children with mild asthma symptoms were overrepresented compared to children with moderate or severe asthma. Similarly, parents of children with mild asthma and moderate levels of PA were overrepresented. The distribution of severity in asthma symptoms represents the real distribution among children with asthma, but the distribution of levels of activity did not. We gave information letters to many children with low levels of activity, but almost none of them were interested in participating. This might be due to the fact that children who are already physically active are more intrinsically motivated and therefore like to participate in a study that is about PA. Furthermore, our results might have been influenced by the skewed distribution of severity in asthma symptoms. It could be that participants with more severe asthma mention other promoting factors than children with mild asthma. On the other hand, all of our patients are treated by a paediatric pulmonologist in a specialised hospital instead of a general practitioner due to their severity in asthma symptoms. The homogeneous level of healthcare also might have influenced the promoting factors that were mentioned by the participants.

Conclusions
According to children with asthma, their parents, and healthcare providers, important factors supported by multiple stakeholder groups were having good asthma control, having access to PA facilities and tailored exercise, being stimulated and motivated by relatives, having PA stimulating situations and (school) environments, setting realistic goals, and having sufficient knowledge about asthma in relation to PA. These findings may inform future interventions and are a good start for setting the requirements of an intervention that helps healthcare providers in promoting children with asthma to become more physically active.

Appendix A. Final Concept Maps and Ideas of All Groups
Appendix A.1. Children Group 1 Figure A1. Concept map of the first group of children. The numbers of the ideas correspond to the ideas in Table A1. Table A1. All ideas, together with their importance rates and standard deviation, of the first group of children categorised per cluster. The numbers of the ideas correspond to the numbers in the concept map shown in Figure A1.  Table A2. All ideas, together with their importance rates and standard deviation, of the third group of children categorised per cluster. The numbers of the ideas correspond to the numbers in the concept map shown in Figure A2.  Table A3. All ideas, together with their importance rates and standard deviation, of the fourth group of children categorised per cluster. The numbers of the ideas correspond to the numbers in the concept map shown in Figure A3.  Table A4. All ideas, together with their importance rates and standard deviation, of the first group of parents categorised per cluster. The numbers of the ideas correspond to the numbers in the concept map shown in Figure A4.  Table A5. All ideas, together with their importance rates and standard deviation, of the second group of parents categorised per cluster. The numbers of the ideas correspond to the numbers in the concept map shown in Figure A5. Child Table A6. All ideas, together with their importance rates and standard deviation, of the group of healthcare providers categorised per cluster. The numbers of the ideas correspond to the numbers in the concept map shown in Figure A6.