At present, there is a certain overmedicalization of society that at times results in diagnostic and therapeutic interventions that are poorly supported by evidence that may entail risks and discomfort for patients excessive health care costs and unrealistic health care expectations. Thus, a significant proportion of health care interventions yield little benefit to the patient and may even cause harm. Berwick1 declared that clinical practices of little value may account for 25%–33% of health care expenditures in every country in the world. Similarly, Graham et al.2 established that between 20% and 25% of performed diagnostic and therapeutic procedures are not based on evidence, and therefore may have a deleterious impact on the health of the patient.

Thus, it is no surprise that strategies have been developed to curtail unsafe, ineffective or inefficient practices, such as the “do not do” recommendations of the National Institute for Health and Care Excellence in the United Kingdome or the “Choosing Wisely” initiative in the United States, aiming to promote a more effective and responsible use of health care resources3,4.

In this context, in 2013, the Ministry of Health of Spain, the GuíaSalud national guideline system within the national health system and the Sociedad Española de Medicina Interna (SEMI, Spanish Society of Internal Medicine)5 launched the Commitment for Quality in Spanish Scientific Associations project with the participation of several scientific societies in Spain with the aim of reducing unnecessary health care interventions. In the framework of this project, each scientific society proposed 5 “do not do” recommendations for their corresponding fields. The initial round of the project involved 12 scientific and medical societies, but many more have joined since, including the Asociación Española de Pediatría (Spanish Association of Pediatrics)6.

In the field of intensive care, the Sociedad Española de Medicina Intensiva Crítica y Unidades Coronarias (SEMICYUC, Spanish Society of Critical Intensive Medicine and Coronary Units) published a set of 65 “do not do” recommendations applicable to patients admitted to intensive care units7.

While this issue has been addressed separately in the intensive care and paediatrics fields, there are currently no “do not do” recommendations for patients requiring admission to a paediatric intensive care unit (PICU) in Spain.

The aim of the study was the identification and selection by a group of experts of a set of “do not do” recommendations (DNDRs) for paediatric intensive care in Spain in a consensus-based process.

The study was organised in 2 phases: the first devoted to gathering potential recommendations, and the second to developing the DNDRs. The project was coordinated by the working group on safety of the Sociedad Española de Cuidados Intensivos Pediátricos (SECIP, Spanish Society of Paediatric Intensive Care) and was conducted entirely by electronic mail between the participants.

The 2 phases of the study developed from June 2019 as follows:

• 1 -

  Collection of potential recommendations (9 meses). The coordinators of the 12 working groups of the SECIP disseminated information about the project and received and handled the DNDRs submitted by members regarding any area of paediatric intensive care (PIC). The coordinators asked for proposals to be based on the highest possible level of evidence and to be accompanied by an explanation of the rationale supporting the proposed DNDR. The goal, as detailed in the Commitment for Quality in Spanish Scientific Associations project,5 was to identify “unnecessary interventions, either because there is no evidence of their efficacy, their effectiveness is low or questionable, they are not cost-effective or they are not a priority” in the field of PIC.

  Participants entered their proposals in an Excel spreadsheet designed for the purpose. The initial collection of DNDRs also included the recommendations that the SEMICYUC had established in each of its knowledge areas, since the recommendations issued by this association for intensive care in adults could guide and enrich the work at hand.

• 2 -

  Evaluation, selection and draft of the recommendations (12 months). After potential recommendations had been collected, a group of members of the working groups of the SECIP who volunteered to collaborate evaluated them to make the final selection. The recommendations were selected by the classic Delphi method.

  In the second phase, the basic working tool continued to be the Excel spreadsheet, in which the necessary information to describe each DNDR was added to facilitate the assessment of each proposed recommendation: wording of each DNDR, rationale accounting for the care practice that it aims to change and bibliographic references on which the recommendation is based.

The evaluators applied 3 evaluation criteria: the prevalence of the practice the recommendation sought to modify, the severity of the potential effects of the practice, and the ease with which the practice could be modified. Each evaluator rated each of these criteria on a scale of 1–9 for each DNDR based on their knowledge and experience. For example, if a DNDR referred to a very infrequent practice, 1 or 2 points could be given; if the consequences of the practice were very severe, the recommendation could receive 8 or 9 points, and if it appeared that the practice would be very difficult to change or eradicate, 1 or 2 points could be assigned. The greater the sum total of these ratings, the greater the priority assigned to the DNDR.

Evaluators were encouraged to write comments supporting their assessment of each DNDR. All the comments were provided to the evaluators in the second round so that they could finetune their assessments in successive rounds.

If evaluators wished to provide supplementary bibliographic references, they could add them to the spreadsheet in a column devoted to that purpose.

Thus, in the second round of the Delphi process, each evaluator was able to review the ratings of each criterion (expressed as means and percentiles) and the comments of other evaluators in order to make any necessary adjustments to his or her own ratings.

Last of all, for each DNDR, evaluators have to state whether they considered that the recommendation should pass to the next round or be eliminated (“pass” or “not pass” variable, rated as 1 or 0, respectively) (Fig. 1).

Figure 1.

Detail of the Excel file used for the initial selection and evaluation was performed in the Delphi process.

(0.39MB).

Whether DNDRs passed from one round to the next in the Delphi process was decided based on the mean value of the pass/no pass variable and the corresponding coefficient of variation (CV), which was considered acceptable if it was less than 80%. Those that passed were added to the Excel spreadsheet for the DNDRs to be considered in the second round of the Delphi process, which also included the anonymised comments of the raters (Fig. 2).

Figure 2.

Detail of the Excel file used for the second round of the Delphi process.

(0.56MB).

The first phase of the study yielded a total of 182 DNDRs submitted by 30 members of SECIP working groups. Only one working group (Cardiology) did not actively participate by submitting proposed recommendations. The distribution of proposed recommendations by area of knowledge was heterogeneous (Table 1).

The evaluation with the Delphi method was conducted on a voluntary basis by 14 members of the SECIP in different regions of Spain, most of who had already participated in the first phase. The work experience of these evaluators varied widely (ranging from 2 to 24 years in PIC), and all were staff physicians except for 2 section chiefs. Their evaluations reduced the initial set of recommendations to 85 DNDRs (Table 1), and after the second round, the definitive selection included a total of 32 DNDRs, 6 of which were SEMICYUC recommendations (Table 2).

Table 3 presents the final set of DNDRs of the SECIP selected after the second Delphi round, excluding the 6 recommendations of the SEMICYUC. For each of them, the table includes the mean pass/no pass value, CV and the rational and bibliographic references that support it.

The health care quality dimensions addressed most frequently by the final recommendations were clinical effectiveness and patient safety.

In this study, we developed a set of “do not do” recommendations for the field of paediatric intensive care in Spain through the collaboration of a large number of paediatric intensive care specialists throughout the country. To do this, we used the Delphi method, commonly applied to quality control studies, to establish the general opinion of a group of experts following a consensus process, thereby completing a project of the SECIP, which defined a set of quality indicators a few years ago with a similar methodology38. The study also provided an overall of PIC in Spain, which is very useful given the different levels of care that PICUs offer in Spain, among which only a minority manage certain conditions (transplantation, cardiac surgery, etc.). In addition, although originally the Delphi method was very costly on account of the successive gatherings of experts it required, in recent years it has been used by groups of experts that are geographically distant11, which is particularly practical in certain times, such as the present period, in which the SARS-CoV-2 pandemic precluded any scientific gathering.

We ought to highlight the diversity of the professional profiles of the experts who participated in the study. Rather than considering it a drawback, we think that it reflects the reality of our scientific society (the SECIP), a particularly young society the scope of which sometimes overlaps with other fields of paediatric care.

On the other hand, we believe that this study offers contributions genuinely focused on paediatric care as regards what “not to do”. Not only did we not simply adopt the DNDRs of the SEMICYUC7, as evinced by the fact that only 6 of them made it to the last phase of the study, but the final set includes DNDRs that are truly innovative or pioneered in PIC. Two examples are the recommendations not to conduct medical and nursing rounds separately or not to impose strict or restricted hours for family visits. These are practices that are being implemented in intensive care as part of the effort to deliver more humane care in this specialty.

Something worth highlighting is that the dimensions predominantly addressed in the selection of DNDRs were patient safety and clinical effectiveness. This probably reflects the nature of intensive care, which seeks to solve problems without creating additional ones. This can be seen in measures taken to avoid iatrogenicity due to the use of hypotonic saline solutions, to avoid malnutrition during the PICU stay, to prevent the irrational use of antibiotics or to avoid delaying initiation of renal replacement therapy when indicated, to provide a few examples of the DNDRs with the highest agreement and which refer to situations that may still be too frequent in Spanish PICUs.

Another aspect worth mentioning is that some of the recommendations seek not only to reduce risk, but also to increase the efficiency of our work, for instance, avoiding routine laboratory testing or head computed tomography (CT) scans in patients with traumatic head injury showing no signs of deterioration.

Equally interesting is the “paediatric nature” of this set of DNDRs, which focus not only on the patient but also on the close environment, for instance, the recommendation not to routinely make parents leave when invasive procedures are going to be performed or the recommendation to avoid conveying information in an improvised or non-standardised manner. In fact, we found some overlap in the recommendations made from different areas of expertise, such as the recommendation to not hinder the presence of family members during procedure issued from the areas of resuscitation, ethics and safety. We chose to keep all such proposals because we believed that they reflected efforts from different subspecialties of intensive care toward the same goal, which is to facilitate the presence of parents and caregivers in our units, perhaps, as noted above, a practice not as extended as would be ideal in paediatric intensive care.

We also observed recurrent proposals from different sources of DNDRs meant to improve communication and information in the PICU (joint rounds, presence of family members, standardised handoff practices…), which could contribute to the prevention of adverse events in Spanish PICUs, given how frequently communication problems are involved in the development of adverse events in inpatient units39.

Our study is not free of limitations, some of them intrinsic to its methodology, such as the heterogeneity in the professional experience of the participants or the variable involvement of members of different working groups over time. In addition, the expert opinions were limited by the current evidence supporting the proposed recommendations, which also varied across the set of recommendations obtained in the study. Thus, this work reflects a specific time and particular circumstances. Lastly, the method itself, based on the asynchronous exchange of information by electronic mail as opposed to in-person gatherings, obstructed the process in some instances and resulted in delays relative to the deadlines originally established, which posed a challenge to the coordinators of the study.

We believe that this study underscores both undesirable practices that ought to be corrected in Spanish PICUs as well as practices that have yet to be generalised and whose implementation could help improved the safety and quality of paediatric intensive care in Spain. This set of recommendations could be a care quality improvement tool, helping reduce variability in paediatric intensive care in Spain, prevent the iatrogenic effects of unnecessary interventions and promote adequate and responsible use of health care resources as well as the search for care practices that are efficient and satisfactory for our patients and their families.

This research did not receive any external funding.

The authors have no conflicts of interest to declare.