ETHICAL APPROVAL

The study was prepared following the principles of Resolution No. 466, from December 12, 2012, and complementary to the Brazilian National Health Council. The confidentiality of the participants' records was ensured. For the database organization, an identification number was adopted to de-identify the participants, guaranteeing their anonymity. The research was approved on February 25th, 2010, by the Research Ethics Committee (No. 006168/2009-70) of the Presidente Dutra University Hospital, from Maranhão Federal University (HUUFMA). All participants signed an Informed Consent Form, with participation in the study being voluntary, anonymous, and confidential.

STUDY DESIGN AND SETTING

This study is part of one broader research with a qualitative-quantitative approach, in which some qualitative results were previously presented²⁰. In this article, we present a cross-sectional investigation, using quantitative data from a survey conducted with nurses and physicians of the four

ICUs from HUUFMA, at São Luís city, Maranhão State, one of the nine states belonging to the Brazilian Legal Amazon. This is the biggest and most important teaching hospital in the state linked to the Brazilian Public Health System (Sistema Único de Saúde - SUS), that contains general diseases (12 beds), cardiology (eight beds), neonatal (20 beds), and pediatric (nine beds) ICUs. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement³⁶.

PARTICIPANTS

We intentionally sampled HUUFMA among the hospitals of the city. The research was addressed for all nurses and physicians in the Hospital (non-random sampling). Of the 118 professionals (62 nurses and 56 physicians) working in the ICU of HUUFMA, 13 (10 nurses and three physicians) were not included due to their refusal to answer the questionnaire, seven (three nurses and four physicians) for incomplete questionnaires, and one physician, because he was responsible for the research, reaching 97 (93.2%) of valid questionnaires.

Nurses and physicians were chosen because they represent the health categories that have a broader engagement with the patients and families through providing therapeutic care, performing procedures, providing information, and communicating bad news.

DATA SOURCE, DATA COLLECTION, AND VARIABLES

A self-administered structured questionnaire was used as data collection instrument. The questionnaire was given to each employee at the beginning of his/her shift in the ICU and collected at the end. It consisted of 44 questions divided into sections (Table 1). The steps used to determine the questionnaire reliability and validity were: face and content validity, and test-retest reliability. For this validation, a team of experts was consulted.

BIAS

Data were entered in the Microsoft Excel^® program by two independent typists, and these were later compared to control possible typing errors to reduce bias. Besides, data collection was performed by a single previously trained interviewer.

DATA ANALYSIS

The statistical analysis was performed using STATA^® v14.0 (Stata Corp., College Station, TX, USA). Discrete or continuous variables were summarized as the mean ± standard deviation (SD) and categorical or ordinal variables as absolute (n) and percentage (%) values. The chi-square and Kruskal-Wallis tests were used to evaluate differences in data distribution within the categories of response variables (outcomes).

Associations among the explanatory variables and outcomes were estimated using the odds ratio (OR) and the respective unadjusted and adjusted 95% confidence intervals (95%CI). Five different outcomes were considered: meanings of death (model 1); meanings of the death in terminally ill patients (model 2); attitude towards giving news of death (model 3); resuscitation of a POTP (model 4); and discussion about do-not-resuscitate orders (model 5). Models 1 and 2 represent the professionals' 'perceptions/feelings' towards death, and models 3 to 5 represent their professional 'attitudes' when confronted with dying. The remaining (explanatory) variables were grouped into blocks for input into the regression analyses in a hierarchical approach specified in figure 1. Hierarchical modeling allows the differential treatment of variables by levels of determination on the outcome, therefore considering mediation and confounding adjustments³⁷. The variables selected for the theoretical model and their distribution in the different levels of determination followed recommendations from the literature^38,39,40, in which five models (according to different outcomes) were adjusted. In models 1 and 3, estimates were calculated using multinomial regression analysis and, for the other models, it was used logistic regression analysis.

ICU: Intensive Care Unit.

Figure 1 - Theoretical model of the variables selected for the study 

Initially, univariate analyses were performed. In multivariate analyses, the variables from the distal level (block 1) were initially included in the model, keeping only those with a p-value < 0.10. Intermediate-level variables (block 2) were then included, ensuring those from the previous level and any variables from this level with p < 0.10. Proximal level variables (blocks 3-4) were then included, repeating the p-value criteria to keep variables. Thus, the associations were adjusted to variables from the same and previous levels, and the p < 0.10 was used to select variables for the adapted models, but p < 0.05 was considered to reject the null hypothesis.

PERCEPTIONS/FEELINGS FACING THE DEATH (MODELS 1 AND 2)

The death of an ICU patient meant failure or fault to 26.8% of health workers, while the death of a terminally ill patient represented pain for the professional in 13.7% of the situations. The loss of an ICU patient is also considered a failure in other studies, linked to a technicist practice^18,41,42. Despite medical advances improving patient care at the end of life, strategies such as hospitalization and palliative care have only achieved a partial success. Health professionals and patients have difficulty in accepting death. Medical training focused on the use of hard technologies, with little appreciation of the physician-patient relationship, contributes to the maintenance of life at any cost². The feeling of unpreparedness, when faced with the patient's death, has been a constant challenge²⁰, requiring the reinterpretation of death and the role of medicine. Sometimes the death of a patient is considered "necessary", and is understood as an anticipated and expected event which generates tranquility and peace; therefore, a "good death"^6,7,21. The loss of patients generates feelings such as frustration, helplessness, sadness, compassion, failure, pain, and others. It is common to use defensive strategies by health professionals, such as rationalization and distancing^18,41,43.

In the specific context of the pediatric ICU, the meaning of death has different nuances^7,20,21. For health professionals, these meanings are likely determined by previous experiences, personal beliefs, and the degree of emotional involvement with the person who died. However, it is known just a little about such determinants. Studies indicate that spirituality and religion are factors that potentially affect those concepts in different societies^44,45,46. In the present study, although analyses of associations stratified by ICU specialty were not carried out, it was possible to identify that the perceptions/feelings of professionals regarding the death of a patient in a pediatric ICU did not differ from those of a general ICU, contrary to the expectative. A possible explanation is that the intensity of feelings was not measured, which was perhaps stronger in the face of the children death.

The following factors were associated with a higher chance of understanding death as a "failure or guilt" of professionals and not as "beginning/freedom" for the patient: i) having the first personal experience of death in adulthood, compared to childhood; ii) being a physician, compared to being a nurse; and iii) working in the cardiology ICU, compared to general ICU. Those who had worked longer in ICUs were more likely to understand death as an "ending" and not as a "beginning/freedom". These results are unprecedented in the literature and point to the need for participation in death rituals at an early age and more thematic discussion during medical training. Over half of the respondents revealed that their first personal contact with death occurred during childhood and adolescence; thus, the construction of beliefs related to loss occurred long before "professionalization". Early exposure appears to allow a certain familiarity with the "rites of passage" in different cultures.

In this study, only 49.5% of health workers discussed the death theme during graduation and 42.7% during postgraduation, so it can be noticed that discussing death and dying appears crucial in developing professional attitudes^{6,13,14,15,18,43,45}. The development of skills in solving problems, communication with the patient and family, and emotional control are essential for providing such care⁴⁷. Despite this, teaching communication and relationship techniques appears insufficient. Professionals must accept their vulnerability and be able to express themselves, as well as share their feelings without fear⁴⁸.

The study results show that nurses were more likely to understand the death of POTP as a "painful" experience for themselves and not as "rest or relief" for the patient. Terminal illness brings great existential suffering²². Nurses may interpret the death of POTP as "pain" for themselves more often than physicians, because of their greater involvement in the patients' daily care.

The humanistic role model suggests that empathy is a decisive element in the care of a patient. Being empathetic allows one to develop the skills to treat patients who are dying^18,22,49. When recognizing the meaning of death, the professional starts perceiving it, not just with horror and a sense of tragedy, but with responsibility for caring³⁹.

PROFESSIONAL ATTITUDES FACED WITH THE DEATH OF AN ICU PATIENT (MODELS 3, 4, AND 5)

In this study, the professional attitudes of nurses and physicians when faced with the death of an ICU patient were mainly associated with professional training and religiosity. Professionals with longer training who had worked in a higher number of ICUs were more likely to take responsibility for communicating the news of death to the family, rather than sharing that responsibility with the multidisciplinary team.

Studies on the communication of bad news are scarce. In Basel, Switzerland, the interviews with 17 nurses and 14 physicians identified a desire to talk about death with patients at the end of life. However, obstacles were identified, such as a lack of time and privacy (45.2%), confrontation with their mortality (38.7%), resistance or denial (38.7%), and the poor cognitive status of the patient (22.6%). Moreover, the excessively long working hours of ICU workers may represent an obstacle to better doctor-patient interactions⁵⁰. The adoption of bad news protocols has improved communication between professionals, patients, and families⁴⁷.

In this investigation, physicians discussed death less than nurses, and those whose first professional experience of loss was with a child were more likely to argue for the resuscitation of POTP. Nursing students were better prepared to face the deaths compared to medicine students²². However, nursing students felt less ready to talk about the death of young children compared to medical students. These findings demonstrate the importance of a broader discussion in Brazilian society about the ethical issues involved in POTP, given the limited protocols and ethical references^4,5,25.

For health students, the first contact with death is through a lifeless body in the anatomy room, i.e., in a distant, dehumanized way⁴⁸. The training curriculum should include philosophical principles and professional attitudes faced with death^13,45,51. The first discussions on the subject usually occur only during working life, after graduation, and postgraduation⁵¹, which can contribute to an attitude of indifference and detachment⁴⁸. There are essential differences between physicians' and nurses' training, because the first focuses on technical improvement, and the second emphasizes on a comprehensive and holistic approach^4,13,48. Topics such as the resuscitation of POTP and practices like euthanasia, dysthanasia, and orthothanasia are already widely discussed in some countries of the world, but not enough in Brazil^2,7,12,20,27. The findings of the present study suggest the need for improvements in curricula for dealing with patients at the end of life in Brazil, similar to some other authors^13,45,51.

Suspending intensive treatment is a difficult task in ICUs¹¹. In the pediatric ICU, 41% of deaths were linked to decisions around the withdrawal/non-adoption of life support measures and, according to this, physicians and nurses do not feel confident about the decision to remove such care^{11,15,20,45,48}. In a study conducted in Macedonia on 217 ICU physicians, most have little or no formal training (67%) in their academic life, highlighting the need to review undergraduate and graduate programs and the ongoing training of ICU professionals working with patients at the end of life^51,52. ICUs should have protocols for dealing with such situations based on multidisciplinary decisions, not just the physicians' judgment. They must consider ethical aspects in this deliberation, along with the patient's and family's wishes^5,53.

For this study, professionals who reported not giving central importance to religion were more likely to agree on the idea of discussing the resuscitation possibility with the patient's family. That is, they allowed the family to share this decision. A study conducted to analyze the influence of religion on medical decisions for terminally ill patients included 304 physicians, 386 nurses, 248 patients, and 330 family members of patients in 142 ICUs from six European countries⁴⁴, divided into Protestant, Catholic, Jewish, or without religion. Religious people tended to require more prolongation of life and less euthanasia, and professionals suggested fewer invasive procedures than patients and their families. Seventy-five percent of physicians and nurses said they respected the patient's opinion. Protestants preferred to follow the patient's wishes, and Jews were more likely to authorize ineffective treatment⁴⁴. Religion seems to be a decisive factor for health professionals in dealing with the death of ICU patients⁵⁴. It is necessary to understand that religious and spiritual practices are integral and cultural elements of fundamental importance to the holistic health of their patients if healthcare professionals are to support patients and their families in end-of-life care⁴⁵.

STUDY LIMITATIONS AND STRENGTHS

This study focuses on a relevant theme to the humanization of the end-of-life for ICU patients and the understanding of its determinants. Although we had searched a single hospital, this is a state reference facility and a teaching hospital, with large numbers of beds, and critical care professionals from different specialties, who generally work in some other ICU, contributing to increasing the representativeness of the sample. This diversity allows a more accurate diagnosis about the factors that affect the thoughts and professional attitudes of health workers when facing patients' death in ICU. Even so, due to the fact of being developed in only one center, we are not able to generalize our results, besides there is another limit related to the questionnaire validity and reliability. We just did the initial steps of the face and content validation with a team of experts, but we did not perform validation using more sophisticated statistical analysis, a strategy that was already reported by some other authors in health science^53,55. Additionally, other limitations were: the impossibility of identifying all the feelings related to the death of ICU patients, as the professionals did not specifically answer what these other feelings were; and the relatively small sample size, which did not allow analyzes stratified by ICU specialty.

We used five different outcomes to assess the meaning of death, and the professional attitudes towards an ICU patient's death. These items are not able to widely determine the meaning of death and, especially, professional attitudes. However, we focused on some emblematic and impactful decisions in maintaining life and accepting death. The low rate of losses due to the non-return of their questionnaires (6.8%), the use of self-administered questionnaires, and the use of multivariate analysis with a hierarchical approach contributed to reducing the possibility of bias. Furthermore, in this study, we obtained a great adhesion of physicians, suggesting the relevance of the theme.