Complex Decision Making about Dialysis in Critically Ill Older Adults with AKI

The first hospital wards in the United States devoted exclusively to the care of critically ill patients opened in 1958 at the University of Southern California Medical Center in Los Angeles and Baltimore City Hospital (1). Over the last 60 years, intensive care unit (ICU) services have become widely available to a growing number and diversity of acutely ill patients, offering an ever-expanding menu of intensive treatments to support or replace lost organ function. The widespread availability of these treatments has raised new and challenging questions about “where to draw the line” when offering and discontinuing intensive interventions intended to prolong life, especially for those patients who are least likely to benefit (2).

AKI is extremely common in patients who are sick enough to be admitted to the ICU (3), and kidney dialysis is one of many treatments that might be initiated during an ICU stay. Although outcomes among critically ill older adults with AKI are known to be poor (4,5), there can be substantial heterogeneity in the risk of adverse outcomes among members of this group (6,7). There can also be marked differences in the preferences, goals, and values of individual patients when faced with potentially life-altering treatment choices (8). In this context, engaging patients and their families in a shared decision-making process is essential to supporting a patient-centered approach to care. However, available evidence suggests that, despite growing support for shared decision making, medical necessity and clinicians’ sense of patients’ best interest often drive decision making when the stakes are high (2,9).

In this issue of the Clinical Journal of the American Society of Nephrology, Bagshaw et al. (10) report the results of a multicenter, prospective, observational study of critically ill older adults (≥65 years old) with severe AKI (defined using Kidney Disease Improving Global Outcomes criteria) admitted to one of 16 Canadian tertiary care centers. Among members of this cohort, the authors described rates of receipt of dialysis, triggers for dialysis initiation, 90-day survival, renal recovery, and quality of life at 6 and 12 months after enrollment (10). Additionally, the attending physician caring for each patient was asked the following hypothetical question at the time of enrollment: “If indicated for the management of this patient’s AKI, would there be willingness to offer kidney replacement therapy?”

Among 499 older adults with severe AKI—the majority of whom also received mechanical ventilation and/or vasoactive therapy—46% were treated with dialysis during their ICU stay. The most common triggers for dialysis were oligoanuria, fluid overload, and acidemia. Recovery of kidney function was the most common reason that dialysis was not initiated (67%). Other reasons included death or clinical deterioration before starting treatment (9%), a plan to withhold or withdraw life-sustaining therapy (10%), dialysis was felt to be inconsistent with the patient’s goals of care (24%), and the clinician felt that the patient was too old (1%) or unlikely to benefit (7%).

In multivariable analysis, older age, greater severity of cardiovascular disease, and residence in a nursing home were associated with a lower likelihood of receiving dialysis, whereas higher sequential organ failure assessment score, metabolic/endocrine diagnoses, and receipt of vasopressors were associated with a higher likelihood. Overall, 65% of cohort members survived to ICU discharge (at which time 24% were receiving dialysis), 52% survived to hospital discharge (at which time 17% were receiving dialysis), and 35% were still alive 1 year after enrollment (at which time 6% were receiving dialysis). Importantly, adjusted survival was severely limited regardless of whether patients received dialysis (90-day survival was 50% for those treated with dialysis versus 51% for those not treated with dialysis), and there were no significant differences between groups in health-related quality of life at 6 and 12 months after enrollment.

For 138 cohort members (28%), the attending physician indicated at the time of enrollment that he or she would not be willing to offer dialysis if this was indicated. During follow-up, only 9% of these patients received dialysis compared with 60% of the 361 patients whose physicians had indicated that they would be willing to offer dialysis. Among patients for whom they would not have been willing, reasons for not starting dialysis were more likely to have been because this was felt to be inconsistent with the patient’s goals of care or unlikely to be beneficial. In multivariable analysis, a higher Charlson comorbidity score, documented goals of care at ICU admission, and receipt of vasopressors were negatively associated with physicians’ willingness to offer dialysis, whereas higher sequential organ failure assessment score and metabolic/endocrine diagnostic category were positively associated.

Bagshaw et al. (10) are to be commended for their work to shed light on dialysis decision making in critically ill older adults, which is an extremely complex and dynamic process. Strengths of this study include a prospective design that relied on triangulation of information from the medical record, interviews with cohort members or their surrogate decision makers, and questions posed to the clinicians caring for them. This approach allowed for the collection of detailed information that is highly relevant to the care of older adults and rarely available in studies of this sort (e.g., residence, functional status, home support, frailty, and quality of life). The authors also collected valuable information about triggers for starting dialysis, reasons for not starting dialysis, and patients’ documented goals of care. Furthermore, study follow-up extended beyond the ICU stay to capture a range of outcomes of potential importance to patients and their families, including survival and need for ongoing kidney replacement therapy at the time of ICU discharge, hospital discharge, and 6 and 12 months after study enrollment as well as patients’ place of residence and health-related quality of life at 6 and 12 months.

The study offers several novel insights about the provision and outcomes of kidney replacement therapy among critically ill older adults. Findings reaffirm previous reports of extremely high mortality rates in this population (4,5), but also extend this observation by showing that these poor outcomes persist regardless of whether patients receive dialysis, a finding that perhaps signals the many other factors that can shape outcomes in this population. It is also striking that most patients treated with dialysis were also receiving mechanical ventilation and vasopressor support, suggesting that decisions about dialysis in this population rarely unfold in isolation and highlighting the importance of a collaborative approach among providers. Study findings also indicate that the risk of death among members of this cohort far outweighed their risk for dialysis dependence such that only 6% of the approximately one third of cohort members who survived for a year were on dialysis at that time. With the caveat that outcome frequency may vary across centers, health systems, and countries, this kind of prognostic information may be extremely helpful in supporting discussions about goals of care and shared decision making with critically ill older adults with AKI and their families.

The description of physician willingness to offer dialysis if this were to be indicated for treatment of AKI is an especially novel and interesting feature of this study. Prior studies have shown that, even after accounting for differences in comorbidity, older adults are less likely to receive intensive therapies in acute care settings, raising concern about clinician bias in shaping treatment decisions (6). Contrary to what might have been expected on the basis of prior reports (6), advanced age was an uncommon reason that dialysis was not started among members of this cohort. Nevertheless, the results of this study do raise concern that decisions about dialysis might have been driven more by provider- and system-level factors than by the goals and values of individual patients. Although not examined in multivariable analysis, how clinicians responded to the hypothetical question about whether there was willingness to offer dialysis seemed to be one of the strongest (if not the strongest) predictor(s) of whether patients went on to receive dialysis. Because clinicians were not asked to explain why they were unwilling to offer dialysis in each instance, it is not clear what considerations shaped their responses to this question. Certainly, reasons cited for not starting dialysis seem to suggest that concerns about whether dialysis would be beneficial or consistent with patients’ goals of care were a prominent concern. However, these patients were actually less likely to have documented goals of care, and most of those for whom goals had been documented did want to receive cardiopulmonary resuscitation and ICU-level care. These results call for more work to understand what factors shape physician willingness to offer dialysis to critically ill older adults and the extent to which this reflects a shared decision-making process that honors the values, goals, and preferences of individual patients and their family members versus a more paternalistic approach, in which decisions are driven by clinician practice style and perceptions of patients’ best interest (8).

In conclusion, the study by Bagshaw et al. (10) highlights both the very high mortality rates and substantial complexity of clinical decision making among critically ill older adults with AKI. These findings argue for humility and compassion in caring for these patients and their families and offer information that might be helpful in contextualizing discussions about goals of care and treatment preferences. They also highlight the need for more work to understand what factors shape clinician willingness to offer dialysis and identify opportunities to improve shared decision making around dialysis initiation for older adults in critical care settings.

Disclosures

C.R.B. has nothing to disclose. A.M.O. reports personal fees from Fresenius Medical Care, personal fees from UpToDate, personal fees from Dialysis Clinics Inc., personal fees from Hammersmith Hospital, and personal fees from University of Pennsylvania, outside the submitted work.