Age-Appropriate Advance Care Planning in Children Diagnosed with a Life-Limiting Condition: A Systematic Review

Pediatric advance care planning (pACP) is an important strategy to support patient-centered care. It is known to be difficult, yet paramount, to involve the child in pACP while adjusting treatment to age and the corresponding stage of development. This systematic review was aimed to evaluate the age appropriateness of pACP interventions by assessing their characteristics, content, and evidence. CINAHL, Embase and MEDLINE were searched from 1 January 1998 to 31 August 2020 in order to identify peer-reviewed articles containing strategies and tools to facilitate pACP in both children (0–18 years) with life-limiting conditions and their families. An assessment of quality was performed using Cochrane tools and COREQ. The full protocol is available as PROSPERO CRD42020152243. Thirty-one articles describing 18 unique pACP tools were included. Most tools were developed for adolescents and young adults. In most cases, the interventions tried to assess the child’s and family’s preferences concerning their current and future hopes, wishes, and goals of the care. This was aimed to enhance communication about these preferences between children, their families, and health-care providers and to improve engagement in pACP. The relevance of an age-appropriate approach was mentioned in most articles, but this was mainly implicit. Seven articles implemented age-appropriate elements. Six factors influencing age appropriateness were identified. Tools to support pACP integrated age-appropriate elements to a very limited extent. They mainly focused on adolescents. The involvement of children of all ages may need a more comprehensive approach.


Introduction
Children with life-limiting conditions often receive highly complex care over a long period of time. This care may include high-risk treatments with severe side effects and palliative care services. The medical conditions and care needs of these children often interfere with their daily life, including their social activities [1]. These children live with the burden of invasive treatment procedures, hospital admissions, and (often) side-effects from therapies [2,3]. However, these children are not routinely asked about their experiences regarding living with illness [3]. In addition, the child's voice is not systematically included in decision making or when discussing treatment preferences [3]. The involvement of parents and children, in a way appropriate for both age and level of development [4], in decision making is considered obligatory in family-centered health care. However, involving children and families is challenging in practice. Uncertainty about prognoses,

Study Selection
Articles published in English in peer-reviewed journals between 1 January 1998 and 31 August 2020 were eligible for inclusion if they reported on a well-described strategy or tool for supporting pACP. pACP was defined as a strategy to identify preferences and goals for future care and treatment [24] by connecting the expertise of the child and family with the expertise of the medical team [14][15][16]. Exclusion criteria were systematic reviews, articles published before 1998, and articles reporting on prenatal advance care planning. The full texts of potentially eligible studies were independently assessed by three reviewers. Disagreements were resolved in discussion with members of the research team. If an article did not provide a comprehensive description of the tool, then more detailed information was requested from the first author by email.

Data Extraction and Quality Assessment
Data extraction was conducted by two authors using a predesigned form [25]. Data regarding the content of the tool, the person conducting the conversation about ACP, its target population, and the items and outcomes related to age appropriateness were extracted. Two authors independently evaluated the studies' methodological rigor by using the appropriate tool. Disagreements were resolved through discussion. We used the Cochrane Collaboration's risk of bias tool for randomized controlled trials [26]. This Children 2022, 9,830 4 of 27 enabled us to evaluate the following: random sequence generation, allocation concealment, the blinding of participants, the blinding of outcome assessments, incomplete outcome data, and selective reporting. One or zero points were allocated when there was a low or high bias risk, respectively. An unclear risk of bias was noted with a question mark, resulting in zero points. A total score of six was achievable. Observational studies were evaluated with an adapted version of the Cochrane bias tool. This enabled us to appraise the selection of study population, the comparability of study groups, the standardization of intervention protocols, the standardization of outcome measurements, any missing data, any confounders, and any selective outcome reporting [26]. Points were assigned as mentioned above. A total score ranging from zero to seven was counted. Qualitative studies were evaluated using the COmprehensive consolidated criteria for REporting Qualitative research (COREQ), assessing 32 criteria concerning three domains: the research team and reflexivity, the study design, and the analysis and findings [27]. Scores of one, 0.5, and zero points were assigned when the score was, respectively, properly described in the manuscript, incomplete, and not described. Assessments of both the risk of bias and the quality of reporting were conducted for mixed-method study designs. A few articles were not critically appraised due to their narrative, non-empirical study design. This review was exploratory in nature, so inclusion was not affected by the quality of selected papers [28].

Data Synthesis and Analysis
The researchers listed the characteristics and content of the pACP tools and their reported empirical outcomes. A narrative synthesis was provided to summarize the results [29]. Any age-appropriate elements and related theoretical groundings were identified by using a qualitative approach. Age-appropriate elements were defined as components of the tool that were adapted to a specific age and corresponding stage of cognitive development. It was reported whether elements were adapted on the basis of age groups in general or, specifically, on the development capacities that matched a specific stage of development. Fragments of articles related to age appropriateness were extracted. The open coding of these fragments resulted in a list of codes related to age appropriateness. Overarching concepts that describe factors influencing age appropriateness in the context of pACP were identified [30]. The protocol of this review is registered in the public registry PROSPERO, with registration number CRD42020152243.

Risk of Bias and Quality of Reporting
Tables A1-A3 of Appendix A show an overview of the scores per article, with re to the risk of bias and an assessment of the quality of reporting. The total scores per s are presented below (Tables 2-6: Article Characteristics). The six randomized contr trials had a median total score of 4 out of 6 (range: 3-5). All six articles could not their participants and therefore did not meet this criterion. For observational studie 6) and the quantitative parts of mixed-method studies (n = 5), the median scores w (range: 2-5) and 4 (range: 2-5), respectively. The qualitative studies (n = 4) had a me total score of 10 out of 32 (range: [4][5][6][7][8][9][10][11][12][13][14][15][16][17][18]. For mixed-method studies (n = 5), the median score of the qualitative part was 8.5 (range: 4.5-12).

Box 1.
Example of implicit description of the concept of age appropriateness.
"Most adolescents aged 14 years and older do not differ from adults in their capacity to make informed treatment decisions, and their understanding of death is no less mature than that of adults" [42] (p. 2).
Few articles referred to the concept in an explicit way by describing any implications of using the concept [11,32,33,56,59]. An explicit description of the age-appropriate concept is presented in Box 2.

Box 2.
Example of explicit description of the concept of age appropriateness.
"Developmentally, the AYA period is characterized by emerging abstract thinking and an evolving sense of vulnerability. Given this complex developmental stage, AYA patients may benefit from the use of specialized tools to facilitate abstract consideration of factors involved in decision making" [59] (p. 2).
Although most articles referred to age appropriateness as a concept in some way, this was generally not translated into specific elements of the described tools nor specified for different levels of development. Twelve articles provided general recommendations to implement age appropriateness in pACP tools [11,32,37,48-51,54-56,58,59]. Fourteen studies claimed their tools to be age-appropriate [11,32,33,36,[48][49][50][51][52][53]56,57,59,60], yet only seven articles implemented elements adjusted to the age of their population. Most of these articles adapted the language to the child's age [11,48,50,51,56,57], used age-appropriate images [49,50], or added a glossary [48,56]. One article referred to their pACP guide as containing family-centered language [57]. These elements mainly focused on adolescents and young adults [48,49,51] or did not specify a particular age of their target population [56,57]. None of the articles explained why these adaptations meet the development needs or capacities of studied children nor explored the development needs of children in general.
Another way to contribute to the age-appropriate concept was by evaluating a tool for its age appropriateness. Most articles did not report any empirical study data regarding the age appropriateness of the tool used by participants. Twenty-eight studies described or evaluated the effectiveness and the child and family preferences of their tools, but none of them specifically evaluated their age appropriateness [11,[31][32][33][34][35][36][37][38][39][41][42][43][44][45][46][47][50][51][52][53][54][55][56][57][58][59][60]. Only three studies examined the age appropriateness of their tool by asking adolescents and young adults if the tool was considered appropriate for themselves and other participants of their age [40,48,49]. The development stage or capacities to participate in the pACP of the children were not described or researched. None of the articles examined age appropriateness in young children. Age-appropriate outcomes were reported by providers in one article [40] and by AYAs in three articles [40,48,49]. These studies showed that AYAs considered pACP tools to be age-appropriate [40,49] and could be introduced before the age of 18 [40]. AYAs experienced the tools as helpful [49]. Only one article examined the perspective of the providers, revealing that pACP tools were considered less appropriate for AYAs and therefore contradicting AYAs' opinion on age appropriateness. About half of the providers reported thinking that pACP conversations should occur after the age of 18 [40].

Factors Influencing Age Appropriateness
We identified four factors related to age appropriateness that might influence the pACP approach: willingness to participate, ability to participate, social identity, and legal responsibilities. How these factors function at certain development stages was not clearly described. Table 9 shows an overview of these factors per article. Articles were marked with an 'x' when contributing to this factor. Sixteen articles stated that children, especially AYAs, show a willingness to participate in pACP conversations [11,[31][32][33][34][35][36]39,40,42,43,48,51,53,56,59]. Articles explored the child's willingness to participate by asking this to the children themselves and their parents. Willingness reflects the motivation of the child to be involved in a pACP conversation and clarifies to what extent this may be so. Many articles cited previous research on this subject, which showed that adolescents and young adults have a desire to participate in pACP. Few declared the same desire among young children and teenagers [11,34,43,56].
Five articles described a developing social identity in adolescence as a factor related to age appropriateness [32,48,51,55,56]. During adolescence, children develop an awareness of themselves and others, which influences children's preferences and goals in pACP.
Some articles described the law requesting an advance directive, or living will, starting from a certain age [31,33,34,38,39,48,49,53,56]. These legal documents or conversations were sometimes described as part of the pACP conversations. Laws determine what is considered a legal age in participating in own health-care decisions. In some articles, younger age groups (18 years old or younger) were excluded from such topics or conversations [31,33,34,38,39,48,53,56].

Discussion
To our knowledge, this is the first systematic review examining age-appropriate characteristics and outcomes in pACP interventions for children with life-limiting conditions. Thirty-one articles reporting on 18 unique pACP tools were identified. Although pACP is aimed to emphasize the preferences and goals of children and their parents, the voices of children are explored by the interventions to a very limited extent. Two thirds of the studied articles referred to the age-appropriate concept; however, none of the studies comprehensively examined the development stage of their target population. Few interventions contained elements adjusted to the development of the child, or evaluated the intervention on age appropriateness [40,48,49]. The factors contributing to age appropriateness identified from the studies we investigated were: willingness to participate, ability to participate, developing a social identity, and legal responsibilities.

Defining Age Appropriateness in pACP
In this review, we have defined age appropriateness as the level of cognitive development of a child corresponding to a certain age. Cognitive development can differ between individuals of the same age and can fluctuate in children with life-limiting conditions [61][62][63]. Age-appropriate pACP tools would therefore benefit from adjustments to the development stage of a child with a life-limiting illness.
In this review, we identified different factors (willingness to participate, ability to participate, legal responsibilities and social identity) that might characterize or influence the development stage. Piaget described different stages in cognitive development in children [17]. The literature is not, however, clear about whether these stages could be used for children with life-limiting illnesses. However, it does provide general information on the development capacities and the comprehension of topics related to ACP. As ACP is intended to be used to discuss future care preferences, children might benefit from an understanding of the medical concepts involved and also from a greater role in medical decision making. A review of medical decision making in children and adolescents showed that four cognitive capacities are needed: communicating a choice, understanding, reasoning, and appreciation [64]. In this way, the development stage of the child, with corresponding cognitive capacities, determines their ability in medical decision making. Expressing a choice, the first criterion, can either be accomplished via language or non-verbal communication [65,66]. Starting from the age of five years, children have a proper understanding of language, and this is a first step towards medical decision making. Non-verbal communication helps in assessing a child's preferences but is excluded as legal consent [67]. The second criterion, understanding information on one's own medical treatment, requires different neurological capacities in decision making [65,66]. Previous studies have shown that children from the ages of seven to ten years can orient and maintain attention [68][69][70], those from six to 12 years old can memorize [71,72], and those from the age of ten years old can recall received information [73][74][75]. Aside from understanding information on treatment, the comprehension of understanding of concepts such as illness, life, and death depends on the cognitive understanding of death as a biological act [76][77][78][79][80]. This can be fully understood starting from the ages of five to seven years old [78][79][80][81]. The understanding of sub-concepts such as irreversibility, universality, personal mortality, inevitability, causality, and unpredictability might even begin at the age of three [77][78][79][80][81]. The articles we researched provided some basis to this theory, implying that children, starting from a young age, should not be excluded from these topics in pACP. Children from the ages of six to eight years can logically reason [82,83] about decision-making consequences, including risks and benefits, which is the third criterion [65,66,84]. This capacity further develops in adolescence, therefore meaning that they can understand more complex issues [82]. Few of the articles we researched mentioned that children can understand the consequences of decisions, indicating that children are able to reason regarding logically their own pACP decisions and should therefore be included in weighing different treatment options. The last criterion, appreciation, indicates that children from the ages of three to four years start recognizing their own norms and values, as well as the effect of these on their own life [67,85,86]. This implies that preferences and hopes in pACP could be explored in a more simple manner and early in childhood. Most studied articles focused on pACP interventions for older children and may have underestimated the value for younger children. Adolescence is considered an interesting development phase in decision making in which children develop a social identity and awareness of themselves and their peers [87]. They highly value the acceptance of peers, which influences decision making [88]. Adolescents make more decisions offering swift rewards in the presence of the other peers [89]. Altogether, Grootens-Wiegers et al. stated that children from the age of 12 are expected to be competent in decision making [64]. Legally, children from the age of 12 years are allowed to make joint decisions on medical issues with their parents. From the age of 16, they can make decisions on their own [90]. In the USA starting from the age of 18 they are allowed to give informed consent for participating in clinical trials [91]. pACP can play a valuable role in preparing children for decision making. However, ACP was not developed for contemporaneous medical decision making; rather, it was developed for preparing certain decisions in the future. In this way, children can participate and have a feeling of control in their own disease process. This can only be achieved when the child's level of development is assessed as part of pACP or prior to the initiation of pACP.
We identified cognitive impairment as a factor influencing the concept of development. Approximately half of the children with life-limiting illnesses also suffer a degree of cognitive impairment [92]. Multiple articles excluded children with cognitive impairment, indicating that it could complicate participation in pACP. Cognitive impairment is defined as a deficiency in cognitive function consisting of multiple capacities: memory, general intelligence, learning new things, language, orientation, perception, attention, and concentration and/or judgment [93]. Cognitive impairment is a broadly used term in which one or more cognitive functions are affected in general. Engaging these children in pACP would therefore benefit from adaptations to the development of their cognitive functions.
Most pACP articles on children with life-limiting conditions reported that pACP interventions need to be culturally appropriate. Preferences in discussing pACP topics differ between cultures [94]. However, the literature is not clear what is considered ageappropriate pACP participation in different cultures. Cognitive development, and therefore level of the child's participation in conversations and decision making, is influenced by cultural differences that affect parenting roles, government guidelines and education [95]. To involve children in pACP in an adequate and age-appropriate manner, their ethnical background should be considered.
Even though defining different stages in the development of children with life-limiting illnesses would be helpful, evidence indicates that the development, cognitive or otherwise, of a child is an individual, fluctuating, culturally-dependent and differentiating process with regard to different topics [61][62][63]. The development of a child can progress or regress individually during the life or disease process. When a child deteriorates, cognitive development can regress and changes in level of development and rate of assent could appear [90]. On the other hand, clinical experience has shown that children, especially adolescents, with life-limiting illnesses often seem to have a better understanding of death and dying compared to healthy children of their age [61]. This can be due to their greater experience with death than other children of their age [62]. However, the literature is inconsistent. Experiencing death through media [62] and what parents teach their children about biology and natural processes [63] stimulates children with regard to the concept of death and dying [62,63]. Diverse ACP topics might have different levels of development in one individual. A comprehension of life and death is only one aspect of the topics discussed in pACP. Other topics, such as preferences and hopes in general, could be easier for children to talk about and could be comprehended on another level of development. The comprehension of diverse topics in pACP differs in the development in children and should therefore determine their level of participation in that topic. The comprehension of one's own body, for example, develops between the ages of four to six years [96], while the understanding of death develops later, between the ages of five and seven years [62,[78][79][80][81].

Recommendations for Future Research
Development is an individual, fluctuating, culturally dependent, and differentiating process between topics and cognitive functions. The stage of development, rather than age, therefore gives direction in how to appropriately engage children with regard to their development. Research is needed to identify the specific characteristics of each development stage for children with life-limiting conditions. Elements that depend upon developments can then be determined and incorporated to create age-appropriate pACP interventions. Most pACP tools lack a comprehensive inclusion or description of ageappropriate pACP elements. Current literature on pACP does not provide sufficient insight in characteristics of age-appropriate elements. This complicates the appraisal of the usefulness of current tools. Two interventions were evaluated on age appropriateness by their target population of AYAs and their providers [40,48,49]. They were considered acceptable and useful for this age group. Age-appropriate elements, such as language and images, were incorporated, but an explanation of why these elements met the child's developmental needs and capacities was lacking. Examining the intervention by the target population provides a first indication of the level of age appropriateness of a tool. However, more insight is needed regarding which development characteristics apply to certain age groups that are relevant in pACP. In this way, a framework of age-appropriate pACP elements can be designed and incorporated into existing and newly developed interventions. pACP tools can be ranked for their level of age appropriateness, which might be an indicator of high-quality family-centered tools. We were not able to apply any qualification to the level of age appropriateness of the pACP tools studied in this review without such a framework.
Creating and examining different stages in development for children with life-limiting illnesses would benefit from other fields of expertise. Experts in developmental psychology, even children themselves and their families, could determine what information or elements are considered age-appropriate [40,48,49]. The development stage should be frequently assessed because it fluctuates, which may require a separate tool. Creating an intervention that determines the level of development of a child regarding pACP topics could provide an indication of how clinicians can involve children in pACP conversations, e.g., which topics providers have to raise or questions to ask. The same fluctuating cognitive phenomena have been observed in dementia patients [97,98] in which cognitive capacities for ACP participation were assessed [97,99]. The literature shows that even patients with severe dementia can still share their preferences or wishes on a certain level [99]. Therefore, we should always explore children's cognitive development and involve children in their own disease processes on a level that corresponds with their cognitive capacities.
Future research should investigate what is considered age-appropriate participation in different cultures, which has not yet been described in the literature. In creating an age-appropriate intervention, cultural norms and values should be incorporated to involve children in a way that is appropriate for their developmental stage.

Strengths and Limitations
This systematic review contributes to the body of knowledge of the young and evolving field of pACP. The research team assessed the content and characteristics of different pACP tools regarding age appropriateness. A team offering broad expertise evaluated the different aspects of age appropriateness. This review comprises the first steps towards the incorporation of age appropriateness in pACP. The results show that age appropriateness is considered important; however, the comprehensive elaboration of this concept is still in its infancy. These findings limit the opportunities for clinical implications for current practice while emphasizing the need for ongoing research to be able to develop a comprehensive age-appropriate approach in pACP. In addition, the variety of different study designs complicates any comparison of the role of the separate intervention elements. pACP may be an upcoming field of expertise, but it is still relatively new in advance care planning and there might be more influencing factors than we have discovered. Other fields of expertise might contribute to these factors. Most reviewed articles were published and conducted in the USA. This might limit the applicability of the findings to other countries. The USA articles were mainly dominated by one research group, which might have biased the results.

Conclusions
In this review, we have summarized the age appropriateness of existing pACP tools in children with life-limiting conditions. The relevance of an age-appropriate approach was mentioned in most articles, though mainly in an implicit way. None of the articles comprehensively examined the development stage of their target population. Four factors influencing age appropriateness were identified: willingness to participate, ability to participate, developing social identity, and legal responsibilities. Three articles evaluated their tools regarding age appropriateness. The tools integrated age-appropriate elements to a very limited extent, mainly focusing on adolescents and young adults. The involvement of children of all ages in pACP needs a more comprehensive approach. Acknowledgments: We thank our information specialist, who helped us in developing the search string and conducting the search. We would like to thank Tony Sheldon and David Brunetta for reading and correcting the grammar and syntax of our review.

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.

Appendix A
Tables A1-A3 show a score overview per article, conducting the risk of bias and quality of reporting assessment. Total score (6) 5 3 3 4 3 4 + = criterion with low risk of bias, a score of one point was assigned; -= criterion with high risk of bias, a score of zero points was assigned; ? = criterion with unclear risk of bias, no score was assigned.  + = criterion with low risk of bias, a score of one point was assigned; -= criterion with high risk of bias, a score of zero points was assigned; ? = criterion with unclear risk of bias, no score was assigned; MM = mixed methods; NA: Not Applicable.    8.5 4.5 11 + = criterion was properly described, one point was assigned; +/-= criterion was incompletely described, 0.5 points were assigned; -= criterion was not described, zero points were assigned.