19.1 Infrastructure and Policy

The underlying structure of a health-care system can impact patient care in important ways. Organizational priorities, ICU-specific culture, and the physical layout and practical resources of ICUs can influence the frequency and quality of SDM. And although unit- and hospital-level structural measures alone do not guarantee that SDM occurs, they can provide an indirect indication of the value placed on SDM within an institution when patient-level data collection is prohibitively expensive or labor intensive.

Physical infrastructure measures may involve assessing the availability of structural resources that promote family and proxy decision maker presence in the ICU, as well as resources that facilitate patient participation in SDM. Examples of physical infrastructure recommended to facilitate SDM include a dedicated and appropriately furnished room for meetings between clinicians and family members, and individual patient rooms that include comfortable spaces and facilities for visitors (i.e., showers) so that patient proxies are more likely to be available for discussions. Other unit-level measures of the resources that promote patient involvement in SDM include availability of speaking valves for ventilated patients, other resources to facilitate speech during invasive mechanical ventilation (i.e., expertise with deflated cuff speech), and communication support devices such as communication boards.^1–3

Infrastructure related to the electronic medical record (EMR) can also provide insights into the system-level facilitation of SDM. Elements of EMR design that support SDM include dedicated, easily accessible, and editable features that (1) store legal documents such as advance directives and portable medical orders related to life-sustaining treatments (i.e., physician orders for life-sustaining treatment forms), (2) store the identity and contact information of patient health-care proxies, and (3) document the content and outcomes of discussions that take place between clinicians, patients, their proxies, and other friends and family members.

The interprofessional staffing of ICUs varies greatly and can impact both the timing and quality of SDM. Interpreters, social workers, palliative care consultants, chaplains, and clinical ethics consultants often play important roles in SDM. However, not all hospitals and ICUs have each of these professionals on staff, and their hours may be limited to weekday business hours.

Finally, policies written and implemented at the ICU, hospital, and institutional levels can explicitly address SDM, or indirectly signal how patients and families are viewed. For example, some institutions have decision-making protocols for high-risk procedures such as extracorporeal life support or ventricular assist devices. Policies and protocols that support SDM could require family meetings, palliative care consultation, ethics consultation, or signed agreements about the administration and duration of therapies. An investigator using this approach to understanding attitudes toward SDM in a health system might survey ICUs about unit-level policies regarding visiting hours, the number of family members who can be in a patient’s room, family presence during rounds, and family presence during invasive procedures or cardiopulmonary resuscitation. The presence of SDM curriculum for medical trainees and the educational resources dedicated to ensuring physicians are skilled at engaging with proxies when making decisions also signal the cultural importance of SDM within teaching institutions.

19.2 The Decision-Making Process

Before assessing what roles two parties played in the process of making a decision about healthcare, investigators must first determine what needs to be decided and whether the decision warrants SDM. Most decisions made in an ICU are appropriately not made in collaboration with patients or their families. These include medical decisions like what ventilator settings to apply, what dose of antibiotics or analgesics to give, and what diagnostic test to order. According to a 2016 joint policy statement from The American College of Critical Care Medicine and the American Thoracic Society, “clinicians should engage in a SDM process to define overall goals of care and when making major treatment decisions that may be affected by personal values, goals, and preferences.”⁴ Although there is general consensus that decisions about code status, as well as withholding and withdrawing life support fit these criteria, deciding what other interventions constitute preference-sensitive decisions is not always so straightforward. Investigators may choose to focus on a single procedure that clearly requires the consideration of a patient’s goals and limits, such as a tracheostomy for patients with prolonged mechanical ventilation and a poor prognosis. Alternatively, a modified Delphi process can help to achieve consensus among stakeholders about a panel of interventions requiring SDM.⁵

An investigator’s next challenge is identifying each potential opportunity for SDM about the decision of interest. In a prospective study, this means anticipating when the clinical team will engage the patient or their surrogate. A member of the research team may need to round or coordinate closely with the clinical team throughout the day so they are prepared to record or observe family meetings or similar interactions. In a retrospective study, investigators must rely on the EMR for evidence of a SDM process. However, caution is warranted when using the EMR to identify instances of SDM because documentation may be incomplete or inconsistent.

Clinicians’ narrative accounts of decision-making can usually be found within EMR progress notes. Progress notes may include clinician rationales, longitudinal accounts of events, and narrative comments that lend depth to understanding a patient’s hospital stay (Table 19.1). Some EMRs also have unique note types for documenting events relevant to SDM, such as family meetings. These textual data can be analyzed using qualitative methods⁶ or natural language processing.⁷ However, EMR documentation is unlikely to describe all relevant communication, and typically only provides a clinician’s perspective.

Fields within an EMR, distinct from the narrative progress notes, can indicate whether the patient has decision-making capacity, the patient’s legal healthcare proxy (or the steps taken to declare a patient unbefriended), contact information for that legal proxy, and copies of legal documents pertaining to decision-making about the patient’s care such as advance directives and physician orders for life-sustaining treatment forms.

There are dozens of instruments designed to evaluate the SDM process from the perspective of patients or proxies, providers, and observers.⁸ However, these instruments vary in quality and most have not been validated in the ICU setting. Herein, we highlight select instruments that have been most widely used or are likely to be responsive and feasible in the ICU setting and are immune to floor and ceiling effects. Investigators who are interested in whether patients or proxies possess the requisite knowledge to make an informed decision may choose to supplement these process measures with standardized questionnaires about the specific risks and benefits of the decision of interest.

Measurement instruments which ask ICU patient surrogates to report on their experience via structured questionnaires are called proxy-reported experience measures (PREMS). When using PREMS to evaluate the SDM process, a few caveats should be kept in mind. The first is that proxies prefer varying levels of control and involvement in preference-sensitive decisions. The same proxy may eschew involvement in low-stakes decisions but prefer total control of value-laden decisions about a patient’s death.⁹ Second, when asked if they shared in the decision-making process, people tend to interpret the question as asking if they are satisfied with their provider or their care,¹⁰ which is often associated with a provider’s age, accent, gender, and race. Therefore, PREMS should be interpreted cautiously. Finally, enrolling a representative sample of ICU proxies can be challenging. Patients near death and those from wealthy neighborhoods are most likely to have a family member available at the bedside.¹¹ Proxies whose loved ones receive palliative care have also historically been more willing to participate in research,¹² which may generate a cohort with disproportionately high exposure to SDM.

Despite these obstacles, PREMS are routinely used in the ICU. The six-item general communication skills score within the Quality of Communication Questionnaire¹³ has been used extensively and asks whether doctors spoke understandably, made eye contact, listened, answered questions, paid attention, and seemed to care about the patient. The Decisional Conflict Scale captures how emotionally distressed a proxy feels about a specific decision with scores ranging from 0 to 100.^{14, 15} The Patient Perception of Patient-Centeredness (PPPC) is a 14-item scale originally developed for use in the outpatient setting.¹⁶ However, modified versions of the PPPC have been used to evaluate communication interventions in the ICU setting,¹⁷ and PPPC scores are associated with satisfaction with the information provided by clinicians.¹⁸ Finally, the three-item collaboRATE scale¹⁹ is designed for evaluating the level of SDM in encounters that included multiple decisions. It includes a version for patient proxies and is available under a Creative Commons license.

Observer measures of SDM use trained assessors, who are often researchers, to assess audio- or video-recorded encounters. Recording ICU encounters and training assessors is labor intensive, time consuming, and requires informed consent from all participants. However, it also allows for a detailed evaluation of clinician behavior, including core aspects of SDM, such as highlighting the existence of a choice, eliciting patient preferences, and explaining options. One approach developed for evaluating SDM in the outpatient setting²⁰ and successfully adapted for research in the ICU setting²¹ involves a third party assessing whether physicians performed 10 behaviors central to SDM. These actions include describing treatment alternatives, discussing uncertainty, and assessing understanding. The OPTION(5) measure,²² which is applicable to a wide variety of clinical circumstances, is similar and was designed to minimize assessor burden. Although not specifically designed to assess SDM, the Patient-Centeredness of Care Scale^{16, 23} also involves experts coding how doctors respond to patient or family statements during an observed or recorded encounter.

Combining data from multiple sources may minimize the limitations inherent in each source. For example, the Dyadic OPTION instrument collects data from a patient or proxy–clinician dyad and compares how each party perceived elements of SDM during an encounter.²⁴ Comparing the patient preferences reported by ICU proxies to notes and orders in the medical record can also provide an indirect assessment of proxy involvement in decisions about care.²⁵ Comparing PREMS with the medical record, or observer methods performed by clinical researchers, can create a more complete picture of how all parties in the room perceived a decision-making event.