Uncertainty at the Limits of Viability: A Qualitative Study of Antenatal Consultations

This qualitative study is informed by applied thematic analysis (ATA),¹⁵  an inductive approach that is used to analyze textual data to provide description, understanding, and pragmatic discussion that goes toward solving practical problems.¹⁵  Theoretical models derived by the ATA process are systematically grounded in the process of data analysis and in the data themselves.¹⁵  To examine how uncertainty in the gray zone of viability is managed in shared decision-making, the data sources for this study were the anonymized transcriptions of 25 real-time audio recordings of 25 gray zone antenatal consultations recorded naturalistically at the bedside. The study was approved by the Mayo Clinic’s Institutional Review Board (institutional review board 15-003365).

The antenatal consultations took place between either a neonatologist or neonatology fellow (referred to as neonatologist) and pregnant women. Recorded antenatal consultations were purposively sampled by using the following inclusion criteria. Neonatologists were eligible for inclusion if they practiced in the study hospital and were responsible for undertaking antenatal consultations. The coprimary investigator (C.A.C.) participated in the study given the critical role in patient care as 1 of 8 attending neonatologists at the study site. The senior author provided an additional layer of expertise as a clinician trained in palliative care and advanced communication. Pregnant women were eligible for inclusion if they were English speaking and admitted to the study hospital for complications of pregnancy that resulted in a diagnosis of anticipated delivery in the gray zone of viability. The gray zone was defined as 22 0/7 to 24 6/7 weeks’ completed gestation or when additional factors prompted shared decision-making regarding the provision or withholding of resuscitation. As such, our study included 1 pregnant woman at 25 1/7 weeks’ gestation, with a pregnancy complicated by severe intrauterine growth restriction.

Over 18 months, 28 pregnant women were approached by the study team in person after permission from the patient’s care team was obtained, which considered their decisional capacity and appropriateness of the timing of the consent process so as to respect the stressful setting for the patients and as not to interfere with necessary care. Individuals obtaining consent provided study information to the pregnant woman and, if present, her support person, in both verbal form (reading of a consent script) and written form (providing patients with a study brochure and consent form for reference).

Twenty-five of the 28 pregnant women approached provided written consent to participate. Additional eligible pregnant women were not approached for consent because of precipitous labor and delivery or unavailability of the study team at the time of presentation. Audio recorders were switched on at the start of the consultation and stopped at the end. Participants were informed that they could request the audio recording be turned off at any time; no recording was interrupted in our study.

Participant demographic data, including sex, race, age, marital status, pregnancy gestation, and birth outcome, were collected via chart review by the study team.

Transcriptions of the audio recordings of the antenatal consultations were anonymized while delineating the role of the individual speaking, assigned a randomly generated reference number, and transcribed by expert qualitative transcriptionists. Analysis did not separate codes derived from responses of the neonatologist, pregnant woman, or support person.

ATA advocates that analysis benefits from the expertise of multidisciplinary perspectives.¹⁵  The research team for this study consisted of two neonatologists (C.A.C. and B.D.K.), two medical sociologists with expertise in neonatal medicine (K.C.) and conversation analysis (K.S.S.), and two pediatric palliative care specialists (C.A.C. and M.J.T.). The research team used ATA to interpret the transcribed antenatal consultations in a systematic way.¹⁵  First, the team (blinded to the consulting neonatologist), read 5 transcripts to discuss, determine, and agree on the major themes arising from the text that were of interest to understanding the pragmatic concern in clinical practice of shared decision-making in high uncertainty. Second, taking these major themes as a basis, the resultant code book was developed by the coprimary investigator (K.C.) who then returned to the remaining 20 transcripts to complete preliminary coding. Third, preliminary coding of each transcript was refined by group consensus through a process of monthly group meetings over a period of 15 months. These discussions were used to achieve consensus over transcript coding and to ensure analysis answered relevant and practical questions relating to neonatology clinical practices and antenatal consultations in the gray zone of viability. A total of 4 themes were derived through our analysis, of which we report on 1, namely, uncertainty.

The study group included 25 pregnant women with a median age of 28 years (SD 5.3), 72% of whom were married. Of these women, the self-identified race distribution consisted of white (19 of 25), Black (2 of 25), Hispanic (2 of 25), and Indian (2 of 25) women. The majority (22 of 25) of antenatal consultations included the woman’s support person. Support persons included expectant fathers, significant others, expectant grandparents, and close friends present either in person or over the phone at the time of the consultation. Indication for prenatal consultations varied (Table 1). The most common indication was preterm premature rupture of membranes. Median gestational age at the time of consultation was 23 0/7 weeks (range 21 2/7–25 1/7 weeks, SD 5.88 days). A total of 8 neonatologists and 2 neonatology fellows conducted consultations. Of these, 20% were female and 100% were white. Equal weight was placed on each neonatologist’s consultations, including those conducted by the coprimary investigator (C.A.C.), with the number of consultations completed by each physician varying between 1 and 3 of 25 consultations (Table 1).

Uncertainty was predominant in the thematic analysis of the data. Structurally, the theme of uncertainty was pervasive throughout the continuum of the consultations. Uncertainty was especially present in the pivotal moments of decision-making between the neonatologist and expectant parent(s) when determining if resuscitative measures should be provided for or withheld from the neonate at the gray zone of viability. Uncertainty was impactful and commonly identified by the physicians as complicating information delivery and decision-making. It was not uncommon to find talk of uncertainty in proximity to decision-making. At times, we observed expectant parents struggle through decision-making, even feeling stuck in the process as a result of uncertainty.

To further delineate this phenomenon, we categorized dimensions of uncertainty (Table 2). The table is organized by hierarchy of potential impact on decision-making, as determined by the study team. The uncertainties present in these consultations fit into 4 overarching, researcher-devised categories, including uncertainty about the timing of delivery, NICU course, individual characteristics (of the physician, expectant parent[s], and fetus or neonate), and consequences of the decision for the expectant family. Subcategorization, explanation of meaning, and illustrative examples from our data set for each type of uncertainty are shown in Table 2.

Although the sheer presence of uncertainty played a pivotal role in the encounters, the physicians’ responses to uncertainty were equally impactful. Provider responses fit into 5 broad, researcher-developed categories. From these, we subsequently devised a taxonomy of strategies used by neonatologists to actively manage uncertainty (Table 3). Each category of uncertainty was managed with different responses and/or strategies both throughout any single encounter and between encounters.

1. Provide more information: The most common provider response in the face of uncertainty was to provide additional information. It was not uncommon to observe information delivery as large sections of provider monologue, with limited interaction or response from expectant parent(s) (Fig 1).

 Neonatologists provided a variety of information (ie, risk of mortality and morbidities) and relayed them by a variety of means (ie, statistics and experiential storytelling). During each encounter, neonatologists discussed when and how more certainty may be gained. Passage of time was commonly mentioned as a modality for gaining certainty regarding outcomes. Providers also stated key checkpoints during the neonate’s life that create certainty. These included response to resuscitation, early head ultrasound findings evaluating intraventricular hemorrhage, and extubation. Providers carefully described uncertainty that still remained after these checkpoints during the clinical course.

• 2. Acknowledging the limits of medicine: Another response to uncertainty was to admit the limits of medicine and the inability to reliably predict outcomes. Providers frequently identified that uncertainty was often dependent on the boundaries of our current medical advances to converge on one anticipated outcome.

• 3. Acknowledgment and acceptance of uncertainty: Providers responded to uncertainty by means of naming uncertainty and accepting it. Often, this response was followed by the reaffirmation of the need to make a decision despite the uncertainty. The data revealed an underlying desire of expectant parents and neonatologists alike to eradicate uncertainty entirely, paired with the realization that this desire could not be actualized.

• 4. Holding hope: Providers sought to remain hopeful with expectant families during these difficult encounters. Holding hope was a tool often used by providers after outlining the spectrum of possible outcomes. The presence of uncertainty allowed for the ability to hope because negative outcomes had yet to become reality. One poignant example (Table 3) was when an expectant family expressed feeling unwanted certainty their son would die and was searching for an opportunity to reopen a window of uncertainty.

• 5. Relationship building: In the face of a difficult and time-pressured encounter, providers often turned toward building a relationship with the expectant family. They made promises to be a truthful partner who would be present along the journey to revisit conversations, reassess decisions, offer support, provide reassurance when able, and give information when needed. Neonatologists sought to empower expectant parents to make an informed decision by asking their understanding regarding the information provided. At times, sensitive to the undesirability of the subject matter, they asked permission to provide more information or explain further. Neonatologists also sought to reengage expectant parents within the consultation via various methods (ie, broad, open-ended questions; small talk; and humor). These refocused the dialogue onto the expectant parents, provided relief from the gravity of the conversation, and humanized the physician.

In this qualitative study of 25 pregnant women anticipating delivery in the gray zone of viability, uncertainty was ubiquitous. Our analysis of the types of uncertainty named in those conversations builds on the real conversations between expectant families and care teams and avoids the abstraction and artificiality of previously published high-fidelity simulation experiences¹⁶  and is free from the recall bias of previously published postconsultative interviews.¹⁷ 

Prognostic uncertainty is common in medicine and can serve different functions.^18–20  Uncertainty can be decision ratifying or that which enhances the collaborative role of physicians and expectant families in the future. Uncertainty can also be decision eroding and may serve to halt or overwhelm the ability to make decisions.¹⁴ 

The functions of uncertainty emphasize the provider’s ability to actively manage it in shared decision-making. Providers may deny uncertainty in decision-making to encourage action when they are concerned that its overabundance may paralyze decision-making.¹⁴  Acknowledgment of uncertainty by a physician serves to equalize the physician and patient within their relationship and as stakeholders.¹⁴  In this way, acknowledging uncertainty may serve to strengthen the therapeutic nature of the physician-patient relationship and allow for individualization of the encounter. Uncertainty may allow for hope when prognosis is grim. Uncertainty could be reassuring within the antenatal consultation, a function which should be maintained. It is equally important to mention that acknowledgment of uncertainty does not undermine the physician’s healing effect; conversely, it may increase it as the physician demonstrates dedication to honesty and engagement in the encounter.¹⁴ 

Our study can be used to challenge previous directives to deliver systematic medical information in this consult setting.¹⁰  Although these formative qualitative data cannot directly reveal the phenomenon, thorough analysis of our data has prompted us to wonder if well-meaning physicians armed with a checklist of discussion points¹⁰  may inadvertently increase the uncertainty of expectant parents using this technique. A common mistake physicians make is to assume that information is all that is needed to guide decision-making.⁷  Our data revealed blocks of text dominated by the physician, with little to no interaction from the expectant parent(s). During these blocks, large amounts of medical knowledge were relayed and, at times, included the use of medical jargon and terminology. We speculate that the lack of expectant parental involvement in these sections may be reflective of disengagement and perhaps a sense of being overwhelmed with unanticipated information. Although alternative explanations for limited participant interjection may include active listening or discomfort in interrupting the neonatologist, the etiology of this observation cannot be definitively concluded on the basis of these data.

Although uncertainty can be diminished, it can never be fully eliminated. When discussing observation of clinical encounters, Renée Fox, an American sociologist, wrote, “…uncertainty and death were the only certainties.”¹³  This is profoundly true in the antenatal decision-making for the unborn neonate at the gray zone of viability; uncertainty that even with intensive life-sustaining medical treatment the infant will survive and to what extent opposite certain death if resuscitation is withheld. It is, therefore, paramount that physicians shift from goals to eliminate uncertainty toward navigating it alongside expectant parents in these encounters.

Lantos⁷  suggests physicians use the antenatal consultation to map family preferences and values to provide information deemed necessary for shared decision-making. This will allow the physician to fulfill the call of the American Academy of Pediatrics toward individualized decision-making in these encounters³  and create a unified team that includes expectant parents whose unique needs are supported through this challenging time. Uncertainty is not solely an uncomfortable problem to be fixed but a necessary part of the process to prepare for and acknowledge in the care of these most vulnerable patients. Furthermore, uncertainty may be used as a tool to build trust-worthy relationships, navigate hope, and allow neonatologists to make recommendations consistent with the goals and values of expectant families.

Naturally occurring data, such as audio-recorded clinical conversations between patients and physicians, enable researchers to analyze the object of interest to them, that is, what actually occurs during clinical encounters in their natural settings, rather than interpreting practice through post hoc reports, such as interviews or abstracted representations of practice, such as through statistics or surveys.^15,21  Audio recordings provide rich data for the study of talk and interaction.²¹  Inductive analysis provides a descriptive and exploratory interpretative knowledge of conversation in the clinical setting,¹⁵  with resultant emergent themes derived inductively from the object of our analysis¹⁵ : an audio-recorded and transcribed conversation between neonatologist and the pregnant woman.

The main limitation to the study is generalizability. As this study was conducted at a single Midwestern regional referral center, these results may not be generalizable to other locations both within the United States and worldwide. Although the study included pregnant women from different racial and ethnic backgrounds, these data may not be representative of the ideas of the sampled racial and ethnic groups as a whole. All consults were conducted in English, and thus results may not apply to encounters conducted in other languages or via interpretive services. By its nature, a qualitative study design lacks generalizability secondary to its data being derived from a specific and limited sample of the population. Although analysis of audio-recorded encounters allows for examination of real conversations, the interpretation of these analyses is limited by the lack of later follow-up with participants to clarify meaning. Providers and participants were aware of and consented to audio recording, which may have introduced performance bias because both parties were told that a study team would be analyzing their consultation despite processes to deidentify participants and anonymize the transcripts. Although each consultation was weighted equally in analysis, including those conducted by the coprimary investigator, and consultations were analyzed in a blinded fashion, the inclusion of the respective senior author had the potential to introduce sampling bias. A final limitation is that themes were not specifically derived on the basis of the role of the individual within the encounter but considered on the basis of the consultation as a whole. Despite these limitations, this study adds a unique perspective to the medical literature secondary to the use of real-life medical encounters.

Uncertainty is pervasive within the antenatal consultation for periviable neonates. Uncertainty likely plays a significant role in decision-making toward postnatal resuscitative efforts. Although uncertainty can be diminished, it can never be fully eliminated. It is possible that well-meaning neonatologists’ intent on decreasing the uncertainty of expectant parents actually may increase it via the relay of a large amount of complex medical information, although further study is required to definitively confirm such a conclusion. Using this antenatal consultation setting to build trust in the neonatologist–expectant parental relationship may help align provider and expectant parental uncertainty and aide individualized and family-centered decision-making regarding the provision or withholding of life-sustaining medical treatment of periviable neonates.

We thank Dr Jon Tilburt for lending his expertise in the field of medical uncertainty that was pivotal to the interpretation of this data set.

ATA

applied thematic analysis