Ethical and legal principles

The project group first studied the relevant national and organisational documents,19–24 and additional published papers. According to Beauchamp and Childress's approach,25 which was also adopted by the American National Association of Neonatal Nurses in their position statement on nurse involvement in ethical decisions,5 the following four principles are helpful in solving ethical dilemmas: beneficence,—that is, healthcare professionals should balance benefits of treatment against the risks and costs for the patient; non-maleficence—that is, healthcare professionals should avoid causing harm to the patient; although most treatment involves some harm, this should not be disproportionate to the benefits of possible treatment; distributive, procedural and legal justice and autonomy, referring to parents being the legal representatives of their child. The medical team, however, also has direct legal responsibility to the child. When, based on medical arguments, treatment is obviously futile and/or continuing treatment would harm the child disproportionally, physicians are not allowed to start, and are even obliged to stop, treatment. On the other hand, if parents ask to stop treatment, but the benefits for the child are evident to the medical team, the parents' wish can be disregarded. In these cases the parents should be informed cautiously, which requires excellent communicative skills.26 When, however, the team is in doubt about the benefits of certain treatment, the parents' opinion is essential in deciding whether treatment is in the child's best interest. Five quality-of-life criteria are considered to be helpful20–22: expected communicative skills; potential of self-care; degree of hospital dependency; degree of suffering and expected lifespan. In the newly formulated policy, MEDM takes place within the context of these ethical principles and within legal boundaries.

Intervention

• MEDM meetings are scheduled on the second and fourth Tuesdays of every month. When healthcare providers or parents have doubts about the moral justification of a child's treatment, the case is scheduled for the next MEDM. Fictitious cases may be discussed when there is no actual patient. Ad hoc meetings are called when deliberations cannot be delayed until the next scheduled meeting;

• The coordinating nurse and physician select the patient to be discussed and prepare the meeting by a checklist guaranteeing that all steps are taken (eg, everyone is informed; the chair is invited, etc).

• The dilemma the team confronts (should we do A or B, or possibly C?) is analysed, following the steps of the Utrecht model.27 This model was chosen because its five-step structure guides the discussion. The model is ‘to the point’ for our purposes; it encourages all professionals involved to contribute to the discussion. Additionally, it invites the chair to summarise and conclude on a step before moving on to the next. All this allows for more controlled discussion. The five steps are also followed in reporting.

  • Exploration: a representative of every professional group involved (physician, nurse, social worker and pastor) informs the other team members about the important aspects to be considered, providing a broad perspective of the patient’s medical and nursing problems, as well as the psychosocial, cultural and religious context of the child and the family;

  • Agreement on the ethical dilemma and investigation of possible solutions: the dilemma that was described in advance is reconsidered and the chair verifies whether the initial question best describes the imminent dilemma; if not, the participants search for a better phrasing in the light of the information received;

  • Analysis: appraisal of possible solutions by describing the effects of different choices for the child and the parents, discussion among the participants about opinions, thoughts and arguments, listening to each other's points of view and trying to understand contradictory thoughts;

  • Decision-making: pros and cons are weighed, participants are invited to agree or disagree and explain why they do so. Subsequently, a decision is made, which is preferably based on consensus, but ultimately the physician in charge of the patient is responsible for the decision, having taken into account the other professionals' points of view. If subsequently the child’s’ condition changes such that the situation is discussed again and a different conclusion is reached, this should be documented quickly and clearly. When an attending physician disagrees with the team decision for personal reasons, he or she should assign treatment to a colleague.

  • Planning actions: for example, deciding on the person(s) who will inform the parents and how, scheduling a subsequent meeting, or guaranteeing the child's comfort with medical and non-medical interventions.

• A standardised electronic form is introduced (online appendix 1), incorporating the same five phases of the Utrecht model. These form the ‘leading thread’ for the preparation, deliberation and report of ethical case deliberation. In addition, the Nijmegen method28 shaped the forms' first, explorative phase, because this method elaborates on the roles of the participating professionals in more detail and was especially developed for ethical deliberations about children. It includes: medical diagnosis, diagnostics and results, prognosis, treatment effects, nursing problems, effects of nursing interventions and psycho-social effects of the disease for child and family (online appendix 1, phase 1: exploration). Completing this form provides a shared understanding of participants' roles and unequivocal presentation of the patient case and the ethical problem, explicitly from the four professional perspectives: medical, nursing, psychosocial and religious. The introductory section and the first phase of the form are completed before the meeting. Adaptations may be made if the meeting yields more or different information. The responsible physician afterwards completes the form's phases 2 to 5, prints and signs the form, which is then saved into the patient's electronic medical file;

• An ethicist chairs the meeting and facilitates ethical deliberation. Being from another department, this ethicist may be perceived as a more impartial chair than a physician who is directly responsible for clinical patient care and is involved in the team. Acting impartially, the ethicist could help team members to fully explore the patient case, following the steps of the proposed method.

• Participants who feel not (yet) confident with the procedure receive practical help from members of a working group of five nurses and two physicians with a special interest in MEDM.

Structured multidisciplinary MEDM differs from clinical ethics committee meetings: in MEDM, all participants but the chair, who is primarily responsible for the process, are directly involved ‘caretakers’. Clinical ethics committees are consulted in complicated or exceptional situations that require external expertise; such committees usually include one or more ethicist(s), lawyer(s), physician(s), nurse(s), social worker(s), pastor(s), manager(s) and sometimes lay person(s).29 ,30

Implementation and evaluation

The intervention was implemented as follows:

• In introductory training sessions by the end of 2008, professionals of the NICU received information about legal and ethical aspects of MEDM in the Netherlands from a lawyer and an ethicist. Furthermore, one of the project group members introduced the procedure and a smaller group of attendees discussed a fictitious patient case, while the others observed.31 ,32

For lack of a suitable instrument, we constructed a 15-item questionnaire to assess opinions on effectiveness of the intervention. A four-point Likert scale was provided for the response, ranging from 1 (fully disagree) to 4 (fully agree); higher scores were indicating more positive judgements. Exploratory factor analysis revealed an underlying four-factor structure.

Sample

At the start of the introductory training sessions, all 105 participants were asked to complete the 15-item questionnaire anonymously; 92/13 were female/male, mean age was 38.7 (SD 9.1; n=103), mean job tenure at this hospital was 10.5 years (SD 7.6; n=93). Project group members were excluded. The new procedure came into effect in February 2009. Eight months and 16 MEDM sessions later, the same questionnaire was distributed to all 113 professionals employed in the NICU at that time; again anonymously; 100/13 were female/male, mean age was 38.9 (SD 8.9; n=109), mean job tenure at this hospital was 11.0 years (SD 7.3; n=100). Project group and working group members were excluded.

Statistical analysis

Differences in professional representation of the respondents before and after implementation of the new procedure were tested with a Fisher's exact test; the significance level was set at p=0.05 (two-tailed). To compare before and after questionnaire scores, means (SDs) and standardised mean differences (SMDs) were calculated for the factors and the separate items. Analogous to Cohen's D, SMD=0.20 was considered a small effect, SMD=0.50 a medium effect and SMD=0.80 a large effect.33 Since all distributions were (close to) normal, t-tests were performed to test the differences for significance. Because factor analysis suggested that the empirical structure is four dimensional, the significance level of the t-tests was divided by four to correct for multiple testing and set at p=0.0125 (two-tailed). In comparing scores on the four factors for the two largest groups of participants, nurses and physicians, before implementation and after 8 months, the intraindividual changes across time could not be assessed because the study was conducted fully anonymously. Therefore, the method of two-way analysis of variance for independent observations was executed to evaluate the differences between professionals (physicians, nurses) and additionally we estimated the changes across time and differences between the professionals across time by assuming that the correlations would be 0.50, while also taking into account that 71% of the professionals were assessed twice. Consequently, we used the t-test for related observations to test the changes across time. Subsequently, as the standard errors could be estimated for these changes, t-tests for independent observations were performed to evaluate differences between the professionals on the changes; the latter representing the possible interaction effect. Significance levels were set at p=0.05 (two-tailed).