Attitudes of pediatric intensive care unit physicians towards the use of cognitive aids: a qualitative study

We employed a descriptive, qualitative research design with thematic content analysis of semi-structured interviews of practicing PICU physicians.

The first author (MW) recruited 15 participants from his professional network of PICU physicians working across Canada. Physicians were invited to participate via direct email or through their division director. This email included an information and consent form, with verbal consent obtained at the time of the interview. Our purposeful sampling approach sought participant diversity according to age, gender, primary language, experience, and practice locations. At least one physician from each of the six contacted centers agreed to participate. Interviews were anonymized and the audio recordings and transcriptions were stored on a password protected computer. This study was approved by the research ethics board of the Centre Hospitalier Universitaire de Québec.

MW conducted individual semi-structured phone interviews in either French or English, depending on the preference of the participant. The interview grid included nine main questions (see Appendix 1), accompanied by follow-up or clarification questions. As described by Francis et al. [21], the interview was based on the TPB as a theoretical framework with questions addressing attitudes towards the use of cognitive aids, social norms around cognitive aid use, and their perception of control over their capacity to adopt cognitive aid use themselves or influence their use by others. At the end of the interview, and separate from the TBP framework, we asked participants what features they would desire from cognitive aids designed to assist with their practice. Two investigators (MW and LC) developed the grid using clinical experience in a PICU environment (MW) and expertise applying the TPB for qualitative medical research (LC) [22, 23]. To ensure that the interview was comprehensible and well organized, the first two interviews served as pilot interviews. For this process, participants were asked to provide feedback on the clarity and structure of the questions. Interviews were also analyzed for interviewer interaction with participants. Following these pilot sessions, minor revisions were made to the grid including changes to question order, and a reduction in perceived repetition of interview themes. Since pilot participants expressed good understanding of the research goal and questions, and MW and LC judged that their responses did not substantially differ from later participants, their responses were included in the final analysis. All interviews were audio recorded and transcribed verbatim by trained research assistants.

Interview transcriptions were coded independently by a research assistant and two researchers (MW and LC). An inductive analysis was applied, with new codes created each time a participant was noted to have given a novel response to an interview questions. If a concept had already been mentioned by another participant, that statement was assigned a pre-existing code. The interviews were analyzed by the coders individually, and then in meetings where the three coders compared their results. One to three interviews were coded at each meeting. In cases of disagreement, all three returned to the transcripts and lists of codes in order to generate consensus. During analysis MW provided content expertise and explanation of clinical scenarios, where LC provided expertise in qualitative analytic methods. This iterative process led to the creation of new codes, improvement of code definitions, and code merging. Any merging of codes was also done by consensus among MW and LC when judged to contain sufficient overlap. Data saturation was monitored by evaluating the frequency of new codes with subsequent interviews. Once saturation was achieved, the frequency that each code was mentioned was noted. LC and MW then grouped codes into themes based on code title and the messages conveyed by participant quotations. Considering the qualitative nature and small sample size, no demographic-based comparisons were planned or performed, though basic participant demographic data was collected.

As one of the major determinants of intention in the TPB framework, attitudes towards an intervention are key prior to acceptance of a new behavior. Several of the questions in the questionnaire were targeted at eliciting participant attitudes toward the use of cognitive aids in their PICU practice.

All participants reported using some form of cognitive aid in their personal lives, and 10/14 respondents reported using some form of cognitive aid in their professional lives. The types of professional aids mentioned by participants are summarized in Table 3. Cognitive aids were frequently cited as useful in situations that are rare, error prone, high risk or stressful. For example, 13/14 participants perceived cognitive aids such as PALS cards as useful in resuscitation settings. One participant mentioned that “when people are stressed and they’re not thinking things clearly, there are a lot of things that I know right now that I can tell you and then, if I’m stressed out, I’m going to start forgetting little details.”

Aids were frequently mentioned as useful for clinicians with little experience in a PICU setting (7/14). Other situations where aids were judged useful included (one respondent each): repetitive clinical settings, PICUs where clinical practice varies significantly between practitioners, clinical settings where the evidence base is well developed, and environments with a high risk of distraction.

The majority (10/14) of participants mentioned the benefits of standardization of practice within a team, with a subsequent decrease in communication confusion between physician and allied health team members. One participant stated, “I think that for uniformity of clinical practice, I find that desirable…Especially in a unit where doctors change service at night and on weekends. If these doctors have completely different practices…that can lead to confusion on the part of the nursing staff.” Thus, cognitive aids were seen by some as tools that encourage uniform, accepted practices throughout a unit.

A majority (10/14) of participants also mentioned that aids can decrease the amount of forgotten details, and 6/14 thought this could lead directly to improvement in patient safety. Some (5/14) cited complex bedside rounds with multiple, routine details that are easily forgotten, but that could be collated into a daily checklist. These potential advantages were to thought to be particularly useful for junior physicians or learners by 7/14 respondents. One senior participant summarized the attitude that while the patient safety benefits might be present for all physicians, less experienced users might benefit the most by stating: “It’s not something that I use commonly. Though, I expect people junior to me to use them, which I kind of find entertaining now that I say it out loud.” Creation of these aids was also seen by one participant as a convenient way to summarize complex literature and make it available for routine decision making.

The most frequently perceived disadvantage to PICU cognitive aid use was limitations to physician reflection or judgment (7/14). There was a sense that aids could become a distractor from the patient’s actual clinical condition (3/14), or impede reaction time (2/14). Participants mentioned that aids might create a dependence on the tool, inhibit learning or decrease effective memorization. One participant said that that “the perverse effect of all these aids is that you prohibit your brain to reflect because everything becomes a little automatic.” Two participants mentioned that these limitations might pose a risk to patient safety if the aid limited judgment to the point of applying an inappropriate protocol. Concern over loss of clinical judgment was also discussed in terms of the design of future cognitive aids. Some participants objected to the requirement of following a protocol (6/14). One participant said, “I would say I’m for it, with the limit that you always need to leave room for clinical judgment because these things are always going to be susceptible to technical errors.”

Participants expressed a desire to examine evidence supporting cognitive aid use in multiple ways. Several stated that if there was a clear scientific consensus that use of cognitive aids improved patient outcomes, they should be adopted, and possibly even imposed by regulators or hospital administers (8/14). Even those supporting that position, however, cautioned that such evidence should include outcomes on all aspects of cognitive aid use, such as the respondent who stated that she would like to see “what is the impact on the entire PICU system, including nursing and physician time.” Similarly, some participants stated they would be opposed to cognitive aid use without data supporting their efficacy (3/14)

Most (11/14) respondents reported frequent cognitive aid use by themselves and their colleagues. The most common formats included standardized orders, or checklists for techniques such as extracorporeal membrane oxygenation (a high risk therapy of mechanical cardiopulmonary bypass employed for extremely ill patients). Most expressed no opinion towards this use by their colleagues, though 2/14 had favorable opinions such as the participant who said, “I don’t think less of anybody who looks at a cognitive aid, in fact, I think more of them. I think that it’s just admitting. ‘Yeah I’m human, I can’t remember, I’ve got to look it up’.”

Most (8/14) participants had at least one colleague who had expressed positive opinions towards the use of cognitive aids by encouraging their use or pointing out their benefits. Several described having cognitive aid “champions” in their units, who often encouraged cognitive aid usage as a way to improve patient safety (4/14).

Several, however, had heard negative opinions towards cognitive aids efficacy expressed by their colleagues, specifically skepticism towards cognitive aids’ effectiveness (5/14). One participant even reported, “I think that sometimes it can even be a subject of a little bit of teasing, because there are some for who using a cognitive aid really isn’t their style.”

Participants almost uniformly reported that the presence of a family at the bedside would not influence cognitive aid use (12/14). Half of those who would not be influenced by the family stated that it would increase transparency towards the family, including a participant who said, “I think it sends a strong message of attention to details and focus on patient safety to the family.” Still, a few participants expressed concerns that families might perceive them as incompetent if they were obligated, in the words of one participant, to “pull out my little checklist” in the treatment of a critically ill child (3/14).

Nearly all physicians said they would support an aid that would encourage non-physicians to remind physicians of discovered risks to patient safety (13/14). This finding was in line with other comments, such as a participant who mentioned that the presence of cognitive aids seems to diminish the effect of the professional hierarchy, allowing allied health personnel to more freely question physician treatment plans. This possible reduction of hierarchy extended to resuscitation teams, where non-physician team members using an aid might feel more comfortable to point out deviations in treatment algorithms (3/14). Nearly all who saw this as advantageous agreed with the participant who stated, “it doesn’t matter who recognizes a problem, if there’s a problem, it needs to be addressed.”

Several participants noted that if a cognitive aid initiative was developed by an individual in their group, and adopted into the culture of their unit, it was more likely to be successful (6/14). One such champion stated, “I think of all the enthusiasm, maybe that’s a little bit catchy but I’m really enthusiastic about it so eventually, I think maybe they’ll just jump on the train.” Others stated that progressive change with gradual implementation was important for aids to be integrated into the culture of a unit (5/14), with one stating that carefully integrated, well designed tools have become indispensable for several members of her colleagues. Two participants placed particular emphasis on the culture of a unit around cognitive aids including one who stated, “the cognitive aid in itself is just a piece of paper, but if it doesn’t go with the whole culture of the patient safety in the process of using it, I imagine nothing will change.”

When asked specifically about their perceived control over influence on their colleagues, nearly all felt that they had or could influence their colleagues around cognitive aid use (12/14), and only one specifically stated he or she could not. Methods used to influence others included demonstrating evidence that cognitive aids improve clinical performance or outcomes (4/14), acting as a role model (3/14), and generally promoting patient safety (2/14).

Participants were generally less receptive to methods that involved imposition of tools by hospital administrators or regulatory bodies. For the interview, imposition was defined as requirements from a body outside of the physician group with the authority to require cognitive aid usage. Expressed concerns included that these tools would limit practitioner autonomy (3/14), be implemented without physician consultation (3/14), or not respect the realities of clinical workflow (2/14). One respondent discussing frustration with imposed tools that did not fit his needs in a quaternary medical center stated, “none of us practice in the median. I don’t want to be forced to practice in the median.”

Many participants, however, were open to the idea that administrators and regulators should have the authority to impose tools when compelling evidence demonstrates that cognitive aids improve adherence to proven best clinical practice (8/14). In the case of a hospital or enterprise level cognitive aid, two stated that imposition by administrators was an effective method to standardize practice, and two others pointed out that this method was one of few that would force physician adherence. One participant noted that without evident consequences for non-adherence, uptake by those resistant to aids would be slow. Further, implementation without evidence supporting their use was a significant barrier for participants (3/14). One participant mentioned worry over possible medico-legal risk if a physician deviated from cognitive aid recommendations, especially if that aid been imposed by a hospital or regulatory authority.

The nature of our study design did not include observations, so actual use of cognitive aids could have been over or underestimated. Though the sample size was small, it represented a diverse group of Canadian PICU practitioners with a wide range of experience and types of units in which they practice. Results were monitored for saturation with each the final interviews yielding few or no new themes. The interview grid was a novel tool and underwent limited pilot testing. Pilot participants, however, were explicitly probed regarding comprehension of questions and face validity of the project, and reported no significant concerns. No control was made for the fact that the initial interviews had a slightly different order of questions, though their responses were judged to be substantially similar to those of other participants. The interviews were all conducted by MW in either French or English depending on the choice of the participant. While practicing full time in a francophone environment, his first language is English and it is possible that some subtleties might not have been fully explored during the interview. To compensate, all interviews, were analyzed by at least one native French speaker. Finally, while cognitive aid use in a PICU often effects all members of a multi-disciplinary team, the scope of this study was limited to the physician members of that team.

Despite these limitations, this study represents the first known exploration of attitudes towards cognitive aid use in a PICU environment. Considering the high stakes, highly interruptive environment of a PICU, this setting is an ideal candidate for aids that could diminish its cognitive burden. This benefit, however, is likely to limited to if cognitive aids are not designed that respect the factors that will lead to physician uptake. Well designed studies are needed to demonstrate the advantages of any newly proposed cognitive aid, both according to medical and implementation factors. Our findings also suggest that implementation would likely benefit from an iterative process, where feedback from the end users is incorporated in a way that allows for modification within the local culture and workflow.

This study was approved by the research ethics board of the Centre Hospitalier Universitaire de Québec. All participants gave verbal consent prior to participation. No identifying information related to participants was published, and the original recordings of the interviews will be destroyed 1 year after publication.

All participants gave consent to publish anonymized quotes as part of the final manuscript. Participants did not review the final manuscript, but were made aware of publication.

Considering that the transcripts and analysis of the research interviews contain data that could possibly identify participants, and that publication of the original data set was not included either in the ethics submission nor in the participant consent, this data has not been included in a public database. If further information regarding the data set is desired, please contact the corresponding author, who will discuss approval from his ethics board and participants for data sharing.