Our systematic electronic literature search yielded 30 relevant publications that used a qualitative design in their study (Table 2) [34-63]. Articles reporting on these studies were published between 2004 and 2021 (inclusive). The studies were conducted in 16 countries, representing four continents (North America, Europe, Oceania, and Asia), indicating that considerations on withholding/withdrawing LST in paediatric patients occur on a worldwide scale. Seven studies took place in the USA [39, 40, 43, 49, 55-57]; four each in the UK [41, 45, 47, 53] and Switzerland [34, 36, 50, 52]; three each in France [34, 36, 38] and the Netherlands [44, 51, 63]; two each in Canada [38, 61], Australia [35, 59], Mexico [54, 62], Germany [46, 50], Belgium [34, 36], Luxembourg [34, 36], and Sweden [42, 48]; and one each in Italy , Romania , Japan , and Thailand . Five studies were carried out in more than one country [34, 36, 37, 50, 52].
Relevant data were collected through different qualitative methods. Most studies used semi-structured interviews to sample physicians’ perceptions. Other studies used focus group discussions, open-ended questions, and conversations.
The analysis approach was reported in 23 studies [35, 37-39, 41, 42, 44-47, 49, 50, 52-59, 61-63]. Thirteen studies used thematic analysis [35, 37-39, 41, 45, 49, 52-55, 62, 63]. Six studies used content analysis to analyse qualitative data [42, 46, 50, 57, 58, 61]. Two studies used thematic analysis as well as content analysis [47, 56].
As a whole, 499 physicians participated in the studies reported on in the included articles. Of these 499 physicians, 116 were paediatric intensivists, 77 were paediatric nephrologists, 71 were paediatric haematologists or oncologists, 21 were general paediatricians, 17 were neonatologists, 13 were paediatric neurologists, 6 were paediatric cardiologists, 6 were general practitioners, 1 was a paediatric chronic disease specialist, 1 was a paediatric rehabilitation specialist, and 1 was a paediatric revalidation specialist. The practicing specialty of the remaining 169 physicians was not stated. The study sample size ranged from 2 to 69. Fourteen of the articles reported the gender of the physicians [34, 36-38, 44, 48-51, 54-58]. With the exception of two studies [51, 54], most of them had more male physician subjects.
Seven studies included only physicians [36, 49, 50, 55-58]; the remainder of the studies included other healthcare professionals, parents, or additionally children. Of the 30 included articles, five reported the physicians’ attitudes regarding withholding/withdrawing LST [39, 46, 56, 57, 63]. The remaining articles reported physicians’ experiences about withholding/withdrawing LST.
Table 3 summarises the results on quality assessment of all included publications. Twenty-two studies were deemed high quality, and eight were evaluated as moderate quality. All included studies were carried out using appropriate methodologies. With a few exceptions, the majority of studies contained a clear statement of aims and findings; most of the studies obtained ethical approval from ethics committees and informed consent from participants. Most of the studies developed significant recruitment strategies and rigorously analysed their data. However, in 23 studies, researchers considered their own roles, potential biases, reactions to events and implications of changes in the study design inadequately.
Our QUAGOL-guided analysis identified four components of physicians’ perceptions on the decision-making process about withholding/withdrawing LST in paediatric patients: stakeholders, structure, ethical values, and influencing factors (Table 4). We used these four components to organise the presentation of our findings.
Stakeholders involved in the decision-making process
Most of the included publications reported the physicians’ perceptions on who they felt the key stakeholders are that should participate in the decision-making, what the nature of their roles are, and how these stakeholders experience their participation in the decision-making process.
Who should be involved in the decision-making process and what are their roles?
We found that almost all physicians held the viewpoint that decisions about withholding/withdrawing LST should be made jointly by physicians, parents, and patients (given that the latter are old enough to participate meaningfully). How the physicians defined ‘shared decision-making’ varied, depending on a given stakeholder’s specific role and level of involvement in the decision-making process.
A large number of physicians stated that they played a major role in the decision-making [34, 37, 38, 40-44, 48, 51, 52, 54, 59, 60, 62, 63]. In three publications, most of the physicians preferred the decision-making to be shared with colleagues; that is, via medical team collaborations [34, 38, 52]. In the early stages of the decision-making, physicians described their primary roles as the one diagnosing diseases, initiating and promoting open discussions with parents about their child’s condition (especially if they foresaw a potential deterioration), facilitating the development of a future care plan for the child, and allocating care resources [38, 39, 43, 44, 48, 52, 54, 59, 60, 63]. Later on in the decision-making, physicians felt medically responsible for making the final decision on withholding/withdrawing LST [34, 37, 38, 40-42, 44, 51, 52, 54, 60, 62]. While most physicians preferred sharing their final decisions with parents, and even with patients if appropriate [34, 38-40, 42, 47, 50-54, 57, 60-63], some of them viewed decisions about futile LSTs as strictly medical and preferred not sharing their decisions of withholding/withdrawing LST with parents [44, 51, 52, 54].
Physicians reported playing other roles. For example, they oriented and reassured parents during the decision-making process [34, 36, 38-40, 42-44, 48, 51, 52, 54, 55, 60, 63]. Physicians provided parents with appropriate information about their child’s critical condition to assist them in making decisions that are beneficial for the child [34, 36, 39, 40, 42-44, 48, 52, 54, 55, 60, 63]. Physicians empathised with parents, reassuring them that their decisions were correct in order to ease the parents’ feelings of guilt [34, 36, 38, 40, 44, 55, 63].
Overall, physicians agreed that parents should participate in the decision-making on withholding/withdrawing LST [34, 36, 38-40, 42, 47, 50-56, 60-63]. They felt that these decisions are very personal for parents and that they should reflect family values . Therefore, parents were given opportunities to discuss their values and preferences with members of the healthcare team [34, 36, 38-40, 42, 47, 50-56, 60-63]. Especially in cases of poor prognoses or unavailable standard treatments, physicians felt that the parents and patients should play a prominent role in the decision-making [51, 52, 61]. The main reason was that physicians knew that giving parents a key role was in the best interest of the child [47, 51, 55, 56]. As parents know their child the best, they could more accurately interpret and communicate their child’s behaviours, judge the degree of their child’s suffering, and express concerns based on their observations [47, 51, 56].
Some physicians reported that some parents authorised them to make decisions about withholding/withdrawing LST in order to avoid being ultimately responsible for these decisions [34, 37, 38, 40, 50, 52, 54, 63]. In general, physicians struggled with placing the burden of major EOL decisions on parents, because most parents do not have sufficient medical knowledge to fully understand all the implications of their decisions, and they would have to live with the consequences of their decisions [34, 40, 53, 55, 61, 63].
Only eight publications discussed what physicians did or could do to involve paediatric patients in the decision-making process, such as informing them about their critical condition and considering their opinion about future care plans [34, 47, 50, 52, 54, 57, 61, 62]. Seven of these publications targeted at adolescents [34, 47, 50, 52, 54, 57, 62]. Because of their older age (compared to a child’s), respect for autonomy was regarded as more important for adolescent patients [50, 57, 62]. However, some physicians were hesitant to involve paediatric patients in the decision-making process [36, 37, 46, 50, 52-54, 57, 62], because they were concerned about medicolegal liabilities and the adolescent’s capacity to make decisions [36, 37, 46, 50, 53, 54, 57]. For instance, in Mexico the law clearly states that parents are legally responsible for making decisions about withholding/withdrawing LST for their child, but it does not specify the age at which minors should be involved in the decision-making process .
How is involvement in the decision-making process experienced?
Our analysis revealed that, in general, most physicians experienced emotional stress derived from their involvement during the decision-making process about withholding/withdrawing LST. They also experienced difficulties and challenges along with the parents. However, some also reported positive experiences. The details of these experiences are described next.
Physicians experienced both positive and negative feelings when being involved in the decision-making process about withholding/withdrawing LST [36-39, 41, 46, 47, 50, 57, 58, 61]. Some physicians repeatedly characterised their experiences as being on a ‘mission’ to care for severely ill children [44, 58]. Other physicians reported feeling more confident, because they had acquired more clinical experience  or feeling more secure when a clear legislative framework for decision-making was in place [36, 37, 50].
Nevertheless, many more physicians expressed negative feelings, including struggling with carrying the weight of responsibility for making decisions to withhold/withdraw LST [36, 37, 41] and fearing therapeutic obstinacy when they knew a given treatment would not benefit the patient [37, 41, 44, 46, 57]. Physicians also reported being fearful of making errors [37-39, 46, 57], especially when they were unfamiliar with the child as a patient . They were also frightened when they were unfamiliar with the child’s best interests but were obliged to make decisions with little time available. Some physicians reported feeling isolated and lonely when making these complex and delicate decisions [37, 41].
Physicians faced three major challenges during the decision-making process. The main challenge was transitioning between curative treatment and palliative care [35, 42, 55]. Physicians were unsure when and how to initiate this transition because of uncertain prognoses and blurred boundaries between curative and palliative care [35, 42, 55].
Another major challenge linked to this transition process was tension felt in the physician-parent relationship. De Graves & Aranda reported that communicating with parents is considered to be the biggest challenge in paediatric oncology . Our present analyses corroborated those findings. Physicians were concerned about when and how to discuss EOL decisions with parents and how to provide them with support [51, 53, 55]. As parents may still hope their child can be cured, they may not be ready to accept palliative care . In the study of Boss et al. , for instance, physicians felt that families and the physician-parent relationship might become strained or damaged by discussing withholding/withdrawing LST with parents if their child has a less serious or mild condition.
A final challenge for physicians was determining when and how children should participate in treatment decisions [36-38, 46, 50, 52, 54, 57, 62]. In the study of Fay et al. , physicians reported that patients were afraid to ask questions about their condition, because their parents did not tell them anything about their condition. Hence, parents revealed only what they wanted to reveal to their child rather than revealing what their child actually wanted to know . In this context, parents assessed their child’s capacity to make decisions . Understandably, physicians were sceptical of their child’s autonomy, as they doubted whether a minor could really understand and process the necessary information needed to make a well-informed decision . The great variability in different minor patients’ capacity for making informed decisions only increased the complexity of deciding whether to include paediatric patients in the decision-making process .
Some physicians also described what parents experienced during the decision-making process about withholding/withdrawing LST. Physicians said parents experienced positive, negative, or mixed feelings when involved in the decision-making process. In de Vos et al. , physicians reported that most parents understood and were not surprised about the inevitable need to withhold/withdraw LST. However, in other studies, physicians stated that parents faced difficult situations. For instance, being confronted with therapeutic futility could make parents psychologically vulnerable and could hinder them from participating in the decision-making [39, 54]. Physicians perceived that for many families the decision-making process was a ‘roller-coaster ride’ of emotional ups and downs, interspersed with despair and grief as the possibility of death became a distinct reality .
Structure of the decision-making process about withholding/withdrawing LST
Our analyses showed that from the physicians’ perspective, the decision-making process could be divided into three stages: (1) early preparation via advance care planning (ACP); (2) information giving and receiving (this is the main decision-making process where stakeholders exchange and discuss information); and (3) arriving at the final decision about withholding/withdrawing LST. These will be described in more detail next.
Early preparation for the decision-making via ACP
Physicians identified major aims of as well as facilitators of and barriers to the ACP process
The first aim of ACP was to discuss values, preferences, and goals of care with parents and the child through an exchange of relevant information [57, 61]. In this way, physicians were well-informed about what treatments the parents and the child would find appropriate at the end of life . The second aim of ACP was to improve the quality of care and to make decisions that were in the best interests of the child; that is, by avoiding unnecessary suffering resulting from emergency or intensive care interventions [46, 47].
Physicians mentioned four key facilitators in the early preparation stage of decision-making via ACP: (1) shared decision-making between healthcare professionals (HCPs) and parents ; (2) supportive setting (i.e., comfortable location, and sufficient time and opportunities for all team and family members to meet) ; (3) early, ongoing conversations starting at the time of diagnosis and when patients were still well enough to discuss their condition, values, preferences, and goals [57, 61]; (4) involvement of HCPs with special training and sufficient expertise in ACP discussions [57, 61].
Physicians perceived three barriers that could hinder the early preparation stage of decision-making via ACP. The first key barrier related to uncertainty issues regarding the early recognition of the life-threatening condition of the child. Physicians were regularly confronted with a lack of diagnostic precision and with insufficient time to evaluate the child’s illness trajectory as precisely as they desired . Consequently, ACP conversations were started only when patients were already seriously ill [46, 47, 57].
The second barrier relates to difficulties HCPs experience in reaching consensus about appropriate care goals [47, 57]. Different clinical judgements about clinical risk management and the effectiveness of medical interventions and different ethical attitudes among HCPs represented a barrier in their desire to reach a consensus in the healthcare team .
The third barrier involved four factors that physicians perceived to hinder HCPs and parents from reaching consensus about care goals [47, 57]: (1) physicians’ not fully knowing the care goals of parents and patients ; (2) parents’ varying and vacillating perspectives towards EOL decisions ; (3) parents’ unrealistic expectations about PICU admissions ; and (4) parents’ cultural, religious, and social background .
Decision-making process with information delivery and receipt
When physicians diagnosed that a child in their care was approaching the EOL, they initiated EOL discussions alone with the parents or in conjunction with the paramedical team. Shared understanding of the medical situation of the child and the way relevant information is exchanged and discussed were identified as elements that contributed to the quality of the decision-making.
Understanding the medical situation of the child
We identified four essential steps that physicians considered to be important to promote a shared understanding of the child’s medical situation. First, during meetings with families, physicians initially discussed what the parents understood about the child’s illness and the progression of their condition. They did this to assess whether parents could accept the eventuality of their child’s prognosis .
Second, physicians provided a medical update of the child’s current condition and prognosis based on their clinical observations of the child’s discomfort, ability to communicate, and responses to advanced medical treatment [39, 43, 44, 48, 60, 63]. Asking and answering questions about the efficacy and futility of possible interventions enabled physicians to enhance their understanding of the clinical and caring situation of the child (e.g., experience of pain, quality of life) and that of the family’s (e.g., psychosocial situation, ethical concerns) . This exchanging of information also better prepared parents for possible future decisions that they might face about withholding/withdrawing LST [34, 36, 39, 44, 54, 60, 63].
Third, physicians considered it to be important to promote a shared understanding of the child’s medical situation related to disease progression. Physicians perceived acute deterioration as the most common reason to initiate the discussion about withholding/withdrawing LST. Physicians created opportunities for parents to discuss their child’s deterioration, as these discussions might facilitate future care planning [51, 59].
Fourth, before final decisions could be made, physicians thought it was important first to provide parents with an update on possible treatment options (e.g., maximum therapeutic treatments, withholding/withdrawing LST) and to lay out the consequences of each treatment option for the child (e.g., negative effects of paediatric intensive care admissions) [43, 48, 60, 63]. In the study of de Vos et al. , most physicians asked parents at the end of a family meeting whether they had additional questions.
Manner of exchanging and discussing relevant information
Besides the content of the information being exchanged, physicians emphasised that it was important to be aware of how to exchange and discuss relevant information on withholding/withdrawing LST during the different stages of the decision-making process. Physicians emphasised that information needed to be conveyed in a simple and concise way to help parents understand the medical perspective [34, 53, 60, 61]. Therefore, they suggested using clear and unambiguous language without medical jargon and to use visual aids, such as medical imaging scans [34, 60, 61].
Physicians also stated that it was important not to withhold or water down information about the child’s condition; rather, it was important to paint an accurate, complete picture of the child’s condition and to do so repeatedly and in a consistent way [49, 54, 60, 61]. Finally, physicians reported that communication with parents should be adapted to the evolving situation of the child and the parents’ state of mind [38, 53, 54, 56]. For instance, if the parents disagreed with the physician’s views on the best interests of the child, physicians said they should try to reframe the care plan in terms they thought would be more acceptable to parents . However, those authors also said that physicians should also be more direct about withholding/withdrawing LST . Moreover, when a patient’s condition was severe, they said that the delivery of information should be unvarnished and blunt .
Making final decisions
When it came to the stage of making a final decision about whether to withhold/withdraw LST at the end of life of a paediatric patient, physicians considered two major options.
When medical treatments became clearly futile at EOL, the majority of physicians supported a final decision concerning withholding/withdrawing LST to limit the child’s suffering [37, 44, 63] and to improve the child’s living conditions [36, 37, 54]. To realise this improvement, physicians said that care goals should be reset . They acknowledged that the final decision to withhold/withdraw LST can be extremely difficult for parents. However, parents who initially argued for extraordinary treatments to be tried, eventually agreed that there were no longer reasonable options to consider when their child failed to respond [37, 44, 60]. Some parents preferred to withhold/withdraw LST and take their child back home so EOL care and death could happen there [54, 60] (sometimes expressed as keeping their child’s soul at home ).
Richards et al.  described three strategies used by physicians to help parents deal with these difficult decisions. First, they limited the range of options that they thought were futile, because they did not want to give the parents false hope. Second, when parents felt that they could not make decisions about withholding/withdrawing LST because of religious reasons (even if they agreed with the physician’s decision), physicians helped them make it. Third, physicians delayed making a final decision, because some parents needed to witness their child’s dying to understand that death was inevitable.
Our analysis revealed that a minority of physicians supported the decision to continue LST, even in acute situations or when they were in doubt of the child’s condition. Opting to continue LST was the standard protocol to ensure that patients had been afforded all possible opportunities to be cured [37, 38, 52, 56]. A small number of physicians argued that it was important to continue LST, because they felt they possessed no right to decide whether any person should live or die [36, 44]. According to physicians in the study of Jongaramraung et al. , some parents wanted extraordinary therapeutic treatment to be tried, because they were not ready to accept the likely fact that their child was dying.
Ethical values that are balanced in the decision-making process
Physicians reported that they felt a need to balance their ethical values, including advocating for the best interests of the child and those of the parents. Within this aim, they articulated their values and commitments as professionals to make ethically sound decisions.
Best interests of the child
Physicians emphasised that advocating for the best interest of the child was the main driver of their decision-making when considering LST at their paediatric patients’ EOL [47, 53-55]. Specifically, our analysis revealed that physicians recalled their experiences of handling similar cases in the past and compared them against published cases. This kind of self-evaluation helped them reach a decision by balancing the expected benefits and burdens of LST [36, 50, 56]. They perceived specific factors related to the patients’ clinical condition as being essential during the decision-making process: (1) laboratory results and radiological investigations of the child ; (2) medical history, diagnosis, and acute deterioration of the child [50, 51]; (3) the child’s response to already-attempted treatments and the possible reversibility of current symptoms [36, 50, 60]; and (4) the risks of complications from new treatments being considered . Physicians reported trying to avoid prolonging the child’s suffering; and by carefully observing the child’s comfort, they aimed to improve their current and future quality of life [36, 44, 54, 56, 58]. They did this by considering their wishes and desire to continue to live [34, 50]. Physicians adopted an advocacy role when presenting to parents clinical perspectives related to the best interests of the child .
Physicians also mentioned that it was not always easy to advocate for the best interests of the child [37, 42, 58]. Several obstacles in this regard that physicians faced were difficulties in imagining the pain of a child’s suffering from particular diseases (e.g., neuromuscular disease) ; hesitations about what was best for the patients in a particular situation or in a need to comply with legal regulations [37, 46, 50, 53].
Best interests of parents
Physicians also stressed the importance of advocating for the best interests of the parents [47, 50, 53, 55, 56]. Therefore, parents’ attitudes towards withholding/withdrawing LST, as well as their well-being, were considered in the decision-making process [34, 36, 38-40, 42, 50-54, 59-63]. Physicians reported that parents needed some time to accept their child’s condition before making a final decision . They needed to gain a sense of control over the decisions made to reach a sense of peace about their decision . Therefore, some physicians tried to instil hope in the parents, because hope that their child would survive often exists—despite disease progression—even up to moments just before death . They considered how the consequences of their decisions would affect parents, trying to avoid placing more burden of responsibility on the parents, or leading them towards a direction that would bring future regrets [34, 37, 38, 40, 50, 55, 63].
Factors influencing decision-making
Physicians described some factors related to patients, parents, and physicians that facilitate or impede the decision-making process about withholding/withdrawing LST at their paediatric patients’ EOL. We refer to these as facilitators and barriers.
Physicians stated that a good collaborative relationship with the medical team and parents was essential in helping them make decisions about withholding/withdrawing LST at EOL.
For parents, physicians mentioned three facilitators. One facilitator was routinising LST discussions and sharing their decisions with parents. Physicians perceived this as being very helpful for maintaining a long-lasting trusting relationship with them [39, 43, 51]. Although disagreements often arose during discussions with parents, physicians emphasised not all disagreements were concerning. Some disagreements stimulated physicians to think about more suitable alternatives for the child’s specific situation . A second facilitator was providing practical and psychosocial support to parents by encouraging them to spend more time with their child and to take advantage of psychosocial resources, including other families and potentially helpful persons (e.g., a priest) . A third facilitator for the decision-making process was parents’ experience with and understanding of (embodied knowledge) their child’s previous treatments [51, 54, 57].
The deterioration in the child’s physical appearance was one of the most frequently identified child-related facilitators, according to physicians. This factor enabled parents to witness their child’s deterioration and facilitated their comprehension of their child’s medical condition [53, 54]. This acceptance by the parents validated the physicians’ clinical decision to withhold/withdraw LST, with the greater aim of promoting the child’s best interests [47, 53-55].
Physicians also reported barriers that hindered their decision-making about withholding/withdrawing LST at EOL in their paediatric patients. First, lack of training in palliative care and appropriate EOL communication was the most frequently identified barrier for physicians’ decision-making [38, 42, 43, 46, 47, 52, 54, 55, 62]. Physicians reported misunderstanding when to start palliative care. For example, some thought palliative care starts only when the child is dying, while others misunderstood the legal and ethical specifics of do-not-resuscitate orders [43, 62]. Thus, they had difficulties in supporting their close HCPs or those of other units, because they themselves were ambivalent about palliative care [42, 62]. Further complicating the implementation of palliative care in children were the limited use of guidelines for making ethically appropriate decisions  and inadequately developed care models for children with life-threatening diseases .
A second barrier, as described by physicians in the Zaal-Schuller et al. study , was conflicts between physicians and parents, which significantly hindered decision-making. For instance, sometimes parents forbade a LST, even though physicians still anticipated that the child’s condition still had a realistic chance of improving. In other cases, parents wished every possible LST to be attempted, even though physicians considered all treatments to be futile.
Conflicts between physicians and parents can have different roots. The most frequently identified source for disagreements was that many parents lacked sufficient medical background knowledge to clearly evaluate the physician’s proposed course of action/inaction [35, 49, 51, 54]. Awkward, confusing, and misleading discussions with physicians ; clinical uncertainties regarding prognosis; and unforeseeable complications contributed to these misunderstandings [35, 51]. Another factor that complicated decision-making was differences in ethnic, religious, and/or linguistic backgrounds of physicians and parents [42, 51].
Physicians in the study of Forbat et al.  described three levels of conflicts between physicians and parents: mild, moderate, and severe conflicts. Mild conflicts contributed to the physician’s poor management of their relationship with families. Previous conflicts about treatment decisions caused physicians and parents to be more conscious of the potential risk of escalating disagreements. Moderate conflicts deteriorated the trust between physicians and parents. These kinds of conflicts arose because parents questioned and disagreed with the treatments proposed by physicians, leading to frequently revisited arguments. The arguments would become unshakeable when physicians and parents continued to defend their position forcefully and were unwilling even to listen or compromise. Severe conflicts disintegrated working relationships, because both physicians and parents shifted their focus from the child’s best interest to the conflict itself. Ad hominem (e.g., physical attack) and professional (e.g., reporting to the press or professional regulators) threats were carried out.
Physicians mentioned three options they used when dealing with parental dissent about withholding/withdrawing LST. First, they tried to comply with the parents’ request. Most physicians chose this option when parents decided to refuse the indicated LST [34, 38, 39, 42, 50-54, 59-63]. They accepted parents’ rights to reject physicians’ clinical recommendations and agreed with their superior ability to assess the child’s presumed wishes . Physicians were motivated to comply with parents’ requests to use non-indicated LST in order to maintain trust and to build a close relationship with parents; they wanted to discourage ‘doctor shopping’ because of the known harm that would cause to the child and to try to avoid legal problems [50, 53].
The second option for dealing with parental dissent about withholding/withdrawing LST was to re-focus on the child’s best interests and to seek a consensus decision with the parents. For example, this option was manifested by delaying to make final decisions, continuing negotiations with parents, and sharing decision-making [50, 53]. Physicians felt that resolving conflicts about a child’s best interest in the absence of court involvement really represented personal and institutional success .
The final option at the disposal of physicians was to override the parental decision, because physicians have the ethical right to refuse to act against their own conscience, their medical expertise, and their duty to deliver competent care and to avoid harm to the child .