Life-sustaining treatment decisions in pediatric intensive care: an Italian survey on ethical concerns

Objectives To investigate how life-sustaining treatment (LST) decisions are made and identify problematic ethical concerns confronted by physicians and nurses in pediatric intensive care within Italy. Methods An 88-question online survey was created, based on a previous qualitative study conducted by this team. The survey was designed to identify how LST decisions were managed; contrasting actual practices with what participants think practices should be. Replies from physicians and nurses were compared, to identify potential inter-professional ethical tensions. The study also identified participants’ principal ethical concerns. Moreover, open-ended questions elicited qualitative perspectives on participants’ views. The survey was pilot-tested and refined before initiation of the study. Results 31 physicians and 65 nurses participated in the study. Participants were recruited from pediatric intensive care units across five Italian cities; i.e., Florence, Milan, Padua, Rome, Verona. Statistically significant differences were identified for (a) virtually all questions contrasting actual practices with what participants think practices should be and (b) 14 questions contrasting physician replies with those of nurses. Physicians and nurses identified the absence of legislative standards for LST withdrawal as a highly problematic ethical concern. Physicians also identified bearing responsibility for LST decisions as a major concern. Qualitative descriptions further demonstrated that these Italian pediatric intensive care clinicians encounter significantly distressing ethical problems in their practice. Conclusions The results of this study highlight a need for the development of (a) strategies for improving team processes regarding LST decisions, so they can be better aligned with how clinicians think decisions should be made, and (b) Italian LST decision-making standards that can help ensure optimal ethical practices.

Background NB: this study was conducted before the COVID-19 pandemic. Therefore, pandemic-related ethical concerns are not reflected in this investigation It is widely recognized that critical illness in childhood commonly requires complex health care that frequently gives rise to challenging ethical concerns [1][2][3][4]. A leading ethical concern relates to the use or withdrawal of life-sustaining treatments (LSTs), such as assisted ventilation (invasive or non-invasive), chest compressions, inotropic support of circulatory function, renal replacement therapies, parenteral or enteral nutrition or hydration, extra-corporeal membrane oxygenation, and selected surgical interventions.
In many Western countries, ethical standards relating to treatment decisions for children are based on the child's 'best interests' [5,6]. Best interests is commonly defined as the treatment option that offers the greatest proportion of benefit in relation to burden. Generally, these standards recognize that any LST can be withheld or withdrawn depending on the balance of benefits and burdens for the child [1,6,7]. However, a child's best interests are frequently difficult to determine because it can be unclear which benefits and burdens should carry the greatest weight. Moreover, it sometimes unclear what decisional authority and responsibility should be borne by different stakeholders when making LST decisions with regard for critically ill children; e.g., parents, physicians, nurses and other health care providers (HCPs), as well as childpatients themselves. A statement published by the Italian Society of Neonatal and Pediatric Anesthesia and Intensive Care states that the physician in charge of the patient's care and the unit head bear the main responsibility for the final decision, although the participation of other staff and the parents should be sought [2].
This investigation followed a previous qualitative study by the authors in Italy [8], wherein focus groups with 16 physicians and 26 nurses as well as individual interviews with 9 parents were conducted. Findings uncovered the 'private worlds' of pediatric intensive care unit (PICU) physicians, nurses and parents. As they struggled through complex ethical dilemmas, they all suffered tremendously and privately. Physicians struggled with the weight of responsibility and solitude in making LST decisions. Nurses struggled with feelings of exclusion from decisions regarding the patients and families that they cared for. Physicians and nurses were distressed by legal barriers to LST withdrawal. Parents struggled with their dependence on physicians and nurses to provide care for their child, striving to understand what was happening to their child.
Aside from this 2011 study, very little empirical research has examined ethical concerns in Italian pediatric intensive care. Our previous study demonstrated that there are significant and under-examined ethical concerns in the PICU that require further investigation.
The objective of this study was to investigate how ethical concerns are managed in Italian pediatric intensive care. Specifically, we examined how LST decisions are made and sought to identify the most problematic ethical concerns confronted by Italian physicians and nurses in the PICU.

Questionnaire development
An online survey questionnaire was developed by the research team. The questionnaire was designed with the survey software LimeSurvey, on a secure password-protected server. The LimeSurvey online survey tool was hosted on a McGill University (Montreal, Canada) server and maintained by the Service Centre Tools Implementation group.
Themes that were identified in our initial qualitative study were used to develop the questionnaire. The aim was to develop a questionnaire that would: (a) be succinct and not require more than 10 min to complete; (b) collect some general descriptive information about participants; (c) document participant's perceptions about actual practices regarding LST decision-making and their thoughts about how these should be made; and (d) identify ethical concerns that participants consider most problematic.
The questionnaire was designed to collect data for two comparative analyses: (a) responses between physicians and nurses, as well as (b) reported 'actual' and 'should be' practices for all participants.
Upon completion of the first version of the questionnaire, a first pilot testing of the online questionnaire was conducted with three Italian PICU nurses, to assess the clarity, time-requirement, and technical functionality of the online survey. The survey was further adapted and a second pilot testing was conducted with 2 physicians and 2 nurses working within an Italian PICU. The second pilot test examined the following questions (English translation of pilot test conducted in Italian): (a) How long did it take to complete the questionnaire?; (b) In your opinion, is this time excessive or adequate?; (c) Did you find it difficult to answer any questions? If so, which ones and for what reason? (response options: difficult to understand; difficult to relate to my reality; too complex); (d) What would you change in the questionnaire structure? (i.e., sections to be deleted and/or added and/ or modified); and (e) Do you want to add your own comments on the questionnaire?
The final version of the online questionnaire consisted of 88 items regarding LST decisions, as well as questions on participants' demographic background.

Sampling and participant recruitment
It was recognized that social ethical viewpoints and underlying moral values could vary across different cities and regions in Italy. It was also believed that ethical views on various clinical practices could vary across settings. It was important to ensure that sampling for this study would include multiple settings. Therefore, a total of five PICUs from five cities were recruited to participate; i.e., Florence, Milan, Padua, Rome, Verona (Table 1). Although PICUs from southern Italy were also invited to collaborate, none agreed to participate during the study's recruitment period despite repeated requests.
Moreover, participating PICUs were solicited in a manner that could ensure a mix of PICUs with anesthetist-intensivists as well as pediatrician-intensivists. These two different training backgrounds were believed to be potentially associated with different clinical practice approaches; although this had not been systematically documented.
The diverse mix of different cities and physician training backgrounds were used solely to ensure that the participating sample was inclusive of these PICU diversities. These factors were not examined for statistical differences, as this would require a significantly larger sample size and more complex analyses.
Two lead physicians on this research team are leaders within the Italian PICU community; each brought a different training background to the study (Biban: pediatrician-intensivist; Giannini: anesthetist-intensivist). One of the nurses on the research team (Bonaldi) is an active member of the Italian PICU nursing community. The two lead physicians (Biban; Giannini) prepared a list of PICUs in Italy that met the sampling requirements described above, striving to recruit a minimum of 50 physicians and 50 nurses (i.e.: based on the design of the survey scales, t-test analyses were planned to examine statistical significance. For a d = 0.5, where we considered a difference of 1.0 between physicians and nurses as moderately significant on a 5-point scale and a Power of 0.8; a minimum total sample size of 100 was required; 50 physicians and 50 nurses). The medical director for each identified PICU was contacted, to solicit the PICU's participation in the study. For each PICU where the medical director agreed to participate, the lead Italian nurse on the team (Bonaldi) contacted the nurse manager in that PICU to solicit the participation of the nurses in that PICU.
The goal was that the medical director and the nurse manager in each participating PICU would promote the study in their PICU and help recruit physicians and nurses to participate in the study. The medical directors and nurse managers were sent a short announcement that they could distribute among physicians and nurses. The announcement provided a brief description of the study and indicated a direct link to the online questionnaire. These announcements were circulated by email and/or by hard copy, depending on the preferences of each PICU.
Following repeated recruitment measures over the course of several months, recruitment was terminated, as the investigators were concerned that an overly prolonged data collection period could result in 'data contamination' of potential practice changes over time. A total of 31 physicians and 65 nurses were recruited.

Statistical analysis
Given that (a) the required sample sizes for t-test analyses were not attained and (b) all data were rated on a 5-point scale (ranging from 1 = strongly disagree to 5 = strongly agree) and were not normally distributed, data were analyzed with nonparametric robust statistical methods by a specialized statistician who ran a series of between-person and within-person comparisons [9]. The statistician was naive to the specific hypothesis for each comparison. All comparisons conducted were based on the masked-coded variables provided by the investigators. All robust tests were conducted using the 'WRS' or the 'WRS2' packages in R version 3.6. For each series of comparisons, a sample R code for the first comparison is provided.
Two statistical comparisons were conducted. The first involved a series of analyses comparing physician with nurse responses for all 88 questions in the survey. For this set of independent mean comparisons, Yuen's modified t-test [10] for independent trimmed means with 5000 bootstrap was used [9][10][11]. In order to adjust for multiple comparisons, the Benjamin-Hochberg procedure was used [12,13]. The original p-values are reported in the Appendix.
A second series of analyses were conducted to examine 27 pairs of questions in the survey for nurses and physicians separately. For this set of dependent mean comparisons, a procedure using 20% trimmed mean with a 5000 percentile bootstrap was used [14].

Ethical considerations
The online questionnaire indicated that completion of the survey would represent the participants' consent for their replies to be used for the study. No additional consent procedure was required.
The online questionnaire was designed in a manner that ensured the personal identity of each participant was not identifiable. Moreover, the online survey data was stored on a secure password-protected universitybased server at McGill University, in Montreal (Canada), where one of the researchers was located.
The study received research ethics approval from the Ethics Committee for Clinical Experimentation for the Province of Verona and Rovigo, Azienda Ospedaliera Universitaria Integrata Verona (i.e., approval code number: 795CESC). Table 2 outlines descriptive data regarding respondents who participated in the study, which included 31 physicians and 65 nurses practising in 5 different PICUs in different regions of Italy. Table 3 indicates survey questions where there was a statistically significant difference in responses between nurses and physicians. These differences were noted in 14 items, out of the total 88 items on the survey questionnaire. These differences related to:

Results
(a) whether (i) discussions or responsibility regarding LST decisions involved other physicians within the team (these differences were reported for actual practices as well as for how practices should be) or (ii) if noninitiation of LST is permitted (nurses rated all these items lower than physicians); (b) (i) LST decision-making criteria are actually based on a patient's non-survival (referring to LST withdrawal) or prolonged suffering (referring to LST withdrawal and non-initiation) (nurses rated these items higher than physicians); (ii) LST decision-making criteria should be based on a patient's non-response to treatment (referring to LST withdrawal and non-initiation) or a patient's non-survival (referring to LST withdrawal) (nurses rated all these items lower than physicians); (c) concerns about harms caused to patients as well as families because of LST decisions (nurses rated these items higher than physicians).
The second series of analyses examined 27 pairs of questions in the survey for nurses and physicians separately. In this analysis, survey items relating to participants' ratings of actual practices regarding LST decisions were compared with their ratings of how they thought LST decisions should be made. These comparisons were analyzed separately for nurses and physicians. For all of these 'actual/should' comparisons, comparing 27 pairs of survey questions, statistically significant differences were found for all comparisons among nurses and almost all among physicians, with the exception of three comparisons among the latter highlighted in Table 4. These three comparisons without statistical differences among physicians relate to questions regarding (a) sharing responsibility with parents for noninitiation of LST; (b) non-initiation of LST or (c) LST withdrawal, if parents requested the latter two.
Questions 24 (items NO24-A to NO24G) and 26 (items NO26-A to NO24-J) asked participants to rate 17 ethical challenges in terms of whether they were problematic. Both nurses and physicians rated all items as      All comparisons were statistically significant among nurses and among physicians, with the following exceptions There were no statistically significant differences, among physicians only, for the following paired questions problematic, rating all items above 3 and some at 4 or above, on a 5-point scale. Physicians and nurses identified the absence of legislative standards for LST withdrawal as a highly problematic ethical concern. Physicians also identified bearing responsibility for LST decisions as a major concern. Question 28 included 3 items asking participants to rate their familiarity with and utilization of the recommendations on the initiation, continuation and withdrawal of LSTs developed in recent years by SIAARTI (The Italian Society of Anesthesiology, Analgesia, Resuscitation and Intensive Care) and SARNePI (The Italian society for neonatal and pediatric anesthesia and resuscitation) (2). Nurses and physicians rated all 3 items below 3 on a 5-point scale, demonstrating low levels familiarity and utilization. Table 5 outlines verbatim exemplars of qualitative data collected in open-ended survey questions, providing additional individual perspectives on some of the survey questions. These data provide more personal accounts of difficult ethical struggles experienced by participants.

Discussion
Data generated by this investigation have corroborated international research results regarding the many significant ethical challenges confronted by PICU HCPs, as well as our own previous qualitative research within Italy. The magnitude of these ethical challenges among participants in this study was revealed through a particular feature of our study design. Comparatively analyzing participants' reports of actual LST decision-making practices contrasted with their views on how these decisions should be madedirectly comparing actual with shouldhelped bring to light the many facets of current practices that participants considered ethically inadequate. Indeed, ethical tensions were identified across all the realms of LST decision-making practices that were examined. Results drawn from these multiple Italian sites as well as inter-professional participants (i.e., nurses and physicians) suggest that PICU teams are commonly confronting significant ethical difficulties that are inadequately addressed. Our results also demonstrated that some ethical challenges are experienced differently according to professional perspectives -i.e., nursing or medicinecorroborating our earlier qualitative research which revealed the many differences in roles, responsibilities, and ethical difficulties encountered within these two professions [8]. Quantitative results were further illuminated by qualitative data.
These results highlight needed substantive and procedural advances regarding ethical aspects of PICU practice (e.g., policies, practice standards). Substantively, following from results reported in our previous study, PICU HCPs are troubled by the lack of clear legal or ethical standards regarding the permissibility of withdrawing LST for children. This is especially noteworthy when compared to some other countries where there are no legal or ethical distinctions between non-initiation and withdrawal of LST, basing such decisions on a case-by-case basis in terms of the best interests of the child in question [5,6]. Moreover, substantive standards could also clarify the formal role and responsibility that parents should have regarding LST decisions in the PICU. Some LST decision-making standards already existed at the time of the study, such as the SIAA RTI guidelines [2]. Such standards were developed to serve as professional practice supports, without explicit grounding in Italian legal norms. Yet, participants demonstrated a low level of awareness of these standards. Recognized national professional societies could lead the further development of such standards and the promotion of their eventual legal recognition by legislators. Indeed, although the recent Italian LAW n. 219, 22 December 2017 -operational since January 16, 2018 -established new rules on informed consent and advance care planning, these have not explicitly defined LST decision standards for pediatrics.
In terms of needed procedural advancements, some differences were noted between nurses' and physicians' accounts of how LST decisions were made, while some participants reported that psychological and clinical ethics supports are needed to assist HCPs and parents to navigate these complex decisions that involve numerous stakeholders. Procedural advancements could include the development of practice standards and institutional policies that promote the involvement of consultants with psychological and clinical ethics expertise to support PICU teams and families to help ensure open discussions, collaborative and respectful communication, constructive reconciliation of disagreements, individual and group opportunities to address experiences with moral distress, and treatment decision-making aligned with relevant national and international legal, ethical, and professional standards [15]. This can include the development of policies on the use of consultations with a clinical ethics committee or consultant for cases that are actually or likely to be ethically troublesome.
Educational activities should be organized within hospital centers and within regional and national conferences to help PICU HCPs learn about substantive and procedural strategies for addressing ethical concerns within their practice.
Future research should extend the investigation reported here in additional PICUs throughout Italy. Such research should solicit participation from PICUs in southern Italy, where ethical views and practices regarding LSTs may differ from those in northern and central Italy. We acknowledge the non-participation of southern Italian PICUs as a limitation of the study. Moreover, future research should investigate LST decisions over time, examining the course of actual clinical practices in relation to what clinicians think should be done.
Another limitation of this study was that the sample sizes required for t-test analyses could not be attained • I believe that the decision to limit or suspend life support should ALWAYS be made collectively by parents, nurses and doctors, and that doctors should give accurate information to parents to enable them to make informed decisions (RN) • I think it is not up to us to judge what is dignified or not, we are no one to decide that a person 'must' die, we are no one even to say that they 'must' live (RN) • We should always have the intellectual honesty to communicate the real situation and be able to share with the whole team and parents (even the patient if we are dealing with a teenager) and evaluate case by case the best treatment and solutions (RN) • Giving false hopes or harassing defenseless people is cowardly and disrespectful (RN) • Seek to share decisions with parents (MD) No. 25: Problematic aspects of LST decisions. In your experience; what are the most problematic aspects? [Please specify] • In general, the main problem lies not in the parents who best of all understand the suffering of the child but in the orientation contrary to the withdrawal of care that denotes the culture of doctors, in particular of the senior physician responsible for the PICU, who never wants to involve the Clinical Ethics Committee in any way and leaves the whole burden of decisions and interviews with parents to the doctor on duty, generally young physicians on night duty. After a death, none of the doctors ever want to talk about the case again. Moreover, even some young doctors, just to avoid problems, are willing to sustain 'accanimento terapeutico'. The nursing staff, on the other hand, is always more sensitive and available for meetings to discuss such cases (MD) • Caused by not feeling protected (RN) • The opinions of members of the treating team cannot always be aligned. In these circumstances the opinion of the ethics committee is useful in orienting and choosing a common line, even if not always fully shared by everyone. Sometimes a strong parental opinion can force the team to maintain or continue care that is futile or does not ensure a minimum quality of life for the child (MD) • Unfortunately, in our reality the withdrawal of some vital supports is not always accepted by everyone and therefore sometimes a limitation of treatments is decided (rather than withdrawal) (MD) • Different theories and ideologies of the various doctors on the team (RN) • Often we are afraid of the consequences and prejudices of people, the law often does not even protect professionals. The choice of 'accanimento terapeutico' is therefore understandable at times but only for personal protection. With the 'living will' something could change for adults, but for pediatrics I am not optimistic (RN) • 'accanimento terapeutico' is used as defensive medicine (RN) No 27: Other: Please specify (if forced to cause 'accanimento terapeutico' for other reasons) • The massive waste of economic resources is really a HUGE problem in my opinion. It's a question I ask myself every day! (MD) • The absence of a CLEAR legislative framework also gives way to a thousand interpretations and above all does not indicate a common approach. The lack of a true ethics consultation (the American model for example) is a serious problem. The [name of hospital is anonymized] Ethics Committee is composed of random people with no experience in resuscitation, and the only intensivist involved is not in the least taken into consideration by the top intensivists who are definitively pro-'accanimento terapeutico'. The problem is serious and it is the principal cause of burnout among medical and nursing staff (MD) No: 29: Other Comments: • Every single case deserves a collegial discussion. In emergency situations, we often find ourselves in the position of having to start life support, even invasive interventions. It is not always easy then when the case becomes oriented toward a poor prognosis and the withdrawal of LST should be undertaken (MD) • I am a simple nurse and in the face of life events, where we have to decide, I find myself in difficulty regarding the certainty dictated by people superior to me. I believe that in suffering there is no man capable of deciding whether he is right or not, whether he is a head physician or a nurse. Faced with a life touched by a profound problem, where rationality leads us to decide, I listen and let myself be carried away by Faith that helps me to live linked to principles that are important to me (RN) • Greater support on a psychological and emotional level for staff and parents in the post-mortem and better decision-making would be useful (RN) • Many circumstances are interfered with by 'team' orders and by the fear of those responsible for running into medico-legal situations that could expose them to criticism and denunciations (MD) • What is missing, in addition to the advice of an ethicist which fortunately would be requested only a few times a year, is NEEDED PSYCHOLOGICAL SUPPORT for the critical care team which would serve to consolidate and amalgamate complex decisions by analyzing the positions of individual members and possibly solving impasses with individuals who have a conflicting view given their subjective experience with end-of-life problems (MD) • How to establish the concept of "a dignified life" in a manner that is valid for the whole team (RN) NB: Excerpts of all qualitative data are presented, to demonstrate a range of views disclosed by nurses and physicians NB: All survey questions and replies have been translated to English from original Italian survey • 'Accanimento terapeutico': This is an Italian expression referring to persistent needless excessively burdensome interventions, for which there is no directly equivalent term in English.
within a reasonable timeframe. On the other hand, sound data analyses could still be conducted with nonparametric robust statistical methods [9].

Conclusion
The results of this study highlight a need for the development of (a) strategies for improving team processes regarding LST decisions, so they can be better aligned with how clinicians think decisions should be made, and (b) Italian LST decision-making standards that can help ensure optimal ethical practices.

Detailed Statistical Analyses
NB: All robust tests were conducted using the`WRS' or the`WRS2' packages in R version 3.6. For each series of comparisons, a sample R code for the rst comparison is provided.

Primary Research Question (RQ1)
This series of analyses were conducted to compare doctors' vs. nurses' responses for all 88 questions in the survey. For this set of independent mean comparisons, Yuen's (1974)

Second Research Question (RQ2)
This series of analyses were conducted to examine 27 pairs of questions in the survey for nurses and doctors separately. For this set of dependent mean comparisons, procedure using 20% trimmed mean with 5000 percentile bootstrap was used (Wilcox, 2017