Customize your JAMA Network experience by selecting one or more topics from the list below.
Wong SPY, Vig EK, Taylor JS, et al. Timing of Initiation of Maintenance DialysisA Qualitative Analysis of the Electronic Medical Records of a National Cohort of Patients From the Department of Veterans Affairs. JAMA Intern Med. 2016;176(2):228–235. doi:10.1001/jamainternmed.2015.7412
Copyright 2016 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.
There is often considerable uncertainty about the optimal time to initiate maintenance dialysis in individual patients and little medical evidence to guide this decision.
To gain a better understanding of the factors influencing the timing of initiation of dialysis in clinical practice.
Design, Setting, and Participants
A qualitative analysis was conducted using the electronic medical records from the Department of Veterans Affairs (VA) of a national random sample of 1691 patients for whom the decision to initiate maintenance dialysis occurred in the VA between January 1, 2000, and December 31, 2009. Data analysis took place from June 1 to November 30, 2014.
Main Outcomes and Measures
Central themes related to the timing of initiation of dialysis as documented in patients’ electronic medical records.
Of the 1691 patients, 1264 (74.7%) initiated dialysis as inpatients and 1228 (72.6%) initiated dialysis with a hemodialysis catheter. Cohort members met with a nephrologist during an outpatient clinic visit a median of 3 times (interquartile range, 0-6) in the year prior to initiation of dialysis. The mean (SD) estimated glomerular filtration rate at the time of initiation for cohort members was 10.4 (5.7) mL/min/1.73 m2. The timing of initiation of dialysis reflected the complex interplay of at least 3 interrelated and dynamic processes. The first was physician practices, which ranged from practices intended to prepare patients for dialysis to those intended to forestall the need for dialysis by managing the signs and symptoms of uremia with medical interventions. The second process was sources of momentum. Initiation of dialysis was often precipitated by clinical events involving acute illness or medical procedures. In these settings, the imperative to treat often seemed to override patient choice. The third process was patient-physician dynamics. Interactions between patients and physicians were sometimes adversarial, and physician recommendations to initiate dialysis sometimes seemed to conflict with patient priorities.
Conclusions and Relevance
The initiation of maintenance dialysis reflects the care practices of individual physicians, sources of momentum for initiation of dialysis, interactions between patients and physicians, and the complex interplay of these dynamic processes over time. Our findings suggest opportunities to improve communication between patients and physicians and to better align these processes with patients’ values, goals, and preferences.
In recent decades, maintenance dialysis has been initiated progressively earlier in the course of chronic kidney disease (CKD).1-5 The reasons for this trend are not well understood. Although there are recognized indications for dialysis, such as life-threatening electrolyte abnormalities and signs and symptoms of uremia, in practice, there is often considerable uncertainty about the optimal time to initiate dialysis in individual patients and little evidence to guide this decision.6
Observational studies examining the association between the timing of initiation of dialysis and subsequent outcomes have contradictory findings. Most studies have reported higher mortality when dialysis is started earlier in the course of CKD or at a higher estimated glomerular filtration rate.7-14 This finding may be partially explained by the older age14 and higher prevalence of complex comorbid conditions and signs of malnutrition8,9,11,14 among patients who initiate dialysis at a higher estimated glomerular filtration rate, but higher mortality also has been variously attributed to myocardial stunning, accelerated loss of native kidney function, and a systemic inflammatory state associated with dialysis.11 Other studies, however, have reported no difference15 or improved survival16-18 when dialysis is started earlier in the course of CKD. The putative benefits of dialysis, such as better control of volume status, increased clearance of uremic toxins, and avoidance of severe uremic complications, also have been cited as a rationale for early initiation.19,20 The Initiating Dialysis Early and Late (IDEAL) trial, in which patients were randomized to start dialysis at either a higher (10-15 mL/min/1.73 m2) or lower (5-7 mL/min/1.73 m2) targeted, estimated glomerular filtration rate, found no difference in survival or other clinical outcomes between the 2 study arms,21 but patients randomized to start early spent more of their remaining lifetime receiving dialysis and incurred higher total health care costs.22
Prior work has focused primarily on associations between estimated glomerular filtration rate at initiation of dialysis and measured patient characteristics and outcomes.7-18,21,22 These studies do not address with sufficient granularity how decisions about when to start dialysis are made in clinical practice. An earlier study used the medical records of a national cohort of patients receiving care within the Department of Veterans Affairs (VA) to examine patients’ clinical presentation around the time maintenance dialysis was initiated.5 Among members of this cohort, a trend toward earlier initiation of dialysis from 2000 to 2009 was not explained by changes in clinical indications, such as signs and symptoms consistent with advanced CKD or presentation with acute illness around the time of initiation of dialysis during the same 10-year period. To better understand the driving forces behind decisions about timing of dialysis, we performed an in-depth qualitative analysis of the medical records of the members of this same cohort.
We conducted a post hoc qualitative analysis of the electronic medical records of a national random sample of 1691 patients receiving care within the VA who initiated maintenance dialysis from January 1, 2000, to December 31, 2009. We identified members of the cohort using a combination of administrative data and information from the medical record collected within the year prior to initiation of dialysis, as described previously.5 Patients were included in this cohort only if the decision to initiate dialysis occurred within the VA and if there was sufficient documentation in their medical record to understand the clinical context in which dialysis was initiated. The VA Puget Sound Healthcare System Institutional Review Board approved this study and waived the requirement to obtain informed consent from patients.
The VA maintains a comprehensive, nationwide, electronic medical record system for all patients who receive care at any of its facilities; the records include progress notes for all inpatient and outpatient clinical encounters. One of us (A.M.O., a nephrologist with >15 years of clinical experience) reviewed the medical record of each patient around the time of initiation of dialysis and abstracted passages from progress notes containing information relevant to the decision to initiate dialysis, as previously described.5 To assess the completeness of medical record abstraction for ascertaining information relevant to decisions about dialysis, a second investigator (S.P.Y.W., a senior nephrology fellow) independently abstracted passages from the medical records of a randomly selected sample of 50 patients. Side-by-side comparison indicated that passages independently abstracted by both investigators were largely similar (identical passages were retrieved for 42 of the 50 patients).
Using inductive content analysis (an unstructured method of inquiry that facilitates discovery of previously unidentified factors pertaining to a phenomenon),23 the 2 investigators then independently reviewed abstracted passages from all patients, coding for factors that seemed relevant to decisions about the timing of initiation of dialysis in individual patients. Together, the investigators reviewed all codes and their corresponding passages for discrepancies in interpretation of the passages, and assignment of codes was deliberated until consensus was reached. The investigators then iteratively examined codes for patterns and trends and combined related codes into larger thematic categories. A third investigator (E.K.V., a geriatrician, palliative care physician, and ethicist) independently reviewed thematic categories for consistency and whether they were well grounded in their constituent codes and passages.24 All 3 investigators then collectively refined the final thematic schema. We used Atlas.ti qualitative analysis software, version 7 (GmbH), to facilitate organization of codes and passages. Data analysis took place from June 1 to November 30, 2014.
Most of the 1691 patients initiated dialysis as inpatients (1264 [74.7%]) and with a hemodialysis catheter (1228 [72.6%]) (Table 1).5 In qualitative analysis, we identified 3 dominant, overlapping themes relevant to understanding the timing of initiation of dialysis: physician practices, sources of momentum, and patient-physician dynamics. To illustrate each of these themes, Tables 2, 3, and 4 provide, respectively, exemplary quotes abstracted from the medical records of individual patients.
Timing of initiation of dialysis reflected differing physician practices, ranging from those directed at preparing patients for dialysis to those focused on forestalling the need for dialysis by using more conservative medical therapies to reverse or halt the loss of renal function and manage the signs and symptoms of advanced CKD (Table 2).
The decision to pursue dialysis usually involved a series of interdependent steps to prepare patients for dialysis. Examples include steps to help patients select a dialysis modality (hemodialysis or peritoneal dialysis) and secure permanent dialysis access (eg, arteriovenous fistula or graft or peritoneal catheter).
Aligning the timing of preparation with patients’ trajectories of illness was a common challenge (Table 2). Although some patients were able to smoothly transition to dialysis after completing key preparatory steps (quote 1a), most were faced with the decision to initiate dialysis earlier than anticipated (quote 1b). Where patients were in the course of preparation sometimes influenced the timing of initiation. In some instances, initiation of dialysis was deferred until permanent access was ready (quote 1c). In other patients, the presence of functional access for dialysis seemed to have an almost disinhibitory effect, tipping the balance in favor of initiation of dialysis in situations in which there did not seem to be pressing clinical indications (quote 1d).
Often, dialysis was initiated after physicians had attempted a series of medical interventions intended to reverse or halt the loss of kidney function and/or treat the signs and symptoms of advanced CKD. Dietary changes and titration of diuretics, ion-binding resins, and alkali therapy were frequently prescribed with the intent to manage signs and symptoms until dialysis could be arranged or to postpone the need for dialysis.
A plan for when dialysis would be initiated was often made based on how patients responded to these interventions (quote 1e) (Table 2). The time frame for evaluating the effect of medical interventions seemed to vary depending on clinical context. In the outpatient setting, patients’ response was usually reassessed during clinic visits over the course of weeks to months. In the inpatient setting, the time frame was often compressed to hours or days before a determination was made regarding whether to start dialysis.
Physicians seemed to vary in their willingness to offer alternatives to dialysis. Some physicians tried to accommodate patients’ preference to continue with medical therapies rather than to start dialysis (quote 1f) (Table 2). One physician spoke of trade-offs between using medical therapies vs starting dialysis (quote 1g). Other physicians seemed to view the signs and symptoms of advanced CKD more as “indications” signaling the “need” for dialysis rather than as treatment targets in their own right (quote 1h).
Although known triggers for initiation of dialysis, such as electrolyte abnormalities and signs and symptoms consistent with uremia, were documented in the medical records of most cohort members around the time dialysis was started,5 we also identified other sources of momentum for initiation of dialysis (Table 3).
Acute illness was the most prominent source of momentum for initiation of dialysis. For patients with life-threatening illness, physicians often described an urgent “need” for dialysis, and the imperative to treat appeared to supplant patients’ choice in the matter (quote 2a) (Table 3).
Even in the absence of urgent clinical indications, simply being in the hospital seemed to increase the likelihood of initiation of dialysis. Patients with advanced CKD who were admitted to the hospital were often labeled by inpatient health care teams as “predialysis” or “approaching dialysis,” and the question of when to initiate dialysis was routinely considered as part of the assessment and care plan. Hospitalization for an illness unrelated to kidney disease was sometimes explicitly viewed by physicians as an “opportunity” to coordinate the initiation of dialysis for patients who experienced obstacles to accessing routine medical care. For instance, for one patient who was hospitalized for treatment of psychiatric illness and who had an unstable housing situation, physicians recommended that dialysis access be placed during the hospitalization because “placement will be easier from an inpatient status.” This patient ultimately started dialysis during the same hospitalization. We also found examples in which physicians seemed to seize the “opportunity” to initiate dialysis during a hospital admission in patients who had been reluctant to start dialysis (quote 2b) (Table 3).
Another source of momentum was physicians’ perceived need to optimize patients’ clinical status for an upcoming high-risk procedure, such as surgery or angiography, with the assumption that outcomes would be more favorable if dialysis were started in advance of the procedure (quote 2c) (Table 3). In some patients, dialysis was “required” by physicians before the relevant procedure would be performed (quote 2d).
Certain preparatory interventions sometimes had the unintended effect of precipitating initiation of dialysis. For example, several patients experienced worsening renal function or symptoms after surgery to place a dialysis access, prompting initiation of dialysis (quote 2e) (Table 3).
Timing of the initiation of dialysis also seemed to be shaped by the complex interactions between patients and physicians. There often appeared to be a push-pull relationship between patients and physicians, in which patients were described in progress notes as being “resistant” or “reluctant” to prepare for or start dialysis (Table 4). Physicians seemed most concerned about safety and survival, seeing any delay in initiation of dialysis as potentially hazardous, while patients seemed to have a range of priorities and concerns.
Although some patients were described as “ambivalent” toward or “accepting” of dialysis when physicians indicated it was time to start, many more patients viewed dialysis as a treatment of “last resort” and asked to postpone dialysis when physicians recommended initiation (quote 3a) (Table 4). Family members sometimes seemed more eager for patients to start dialysis than the patients themselves (quote 3b).
Physicians documented a range of pragmatic concerns cited by patients as reasons for postponing initiation of dialysis, which included competing social or family obligations, financial problems, or transportation difficulties (quote 3c) (Table 4). Although some physicians made note of significant barriers to dialysis for individual patients, these concerns did not always seem to affect their recommendation (quote 3d). At times, physicians pressed patients to start dialysis despite patients’ clear reluctance (quote 3e).
For many patients, timing of initiation of dialysis was tied to a broader decision about whether to initiate dialysis at all. It was not uncommon for patients to have “refused” or “declined” offers to initiate dialysis on multiple occasions before eventually starting treatment. Patients’ willingness to start dialysis seemed to be shaped by their changing experience of illness over time (quote 3f) (Table 4). Some patients seemed to need to become very sick before agreeing to start dialysis (quote 3g). Physicians described using various arguments to “persuade” or “convince” patients to start dialysis (quote 3h). One patient refused treatment altogether when his physician would not give him the time to consider his choices (quote 3i).
Most physicians seemed to view dialysis as a beneficial treatment and were skeptical when patients said they did not want to start dialysis (quote 3j) (Table 4). Some physicians perceived their patients to be “in denial” about or have “poor insight” into the severity of their kidney failure and the “inevitable need” for dialysis (quote 3k). Missing appointments with nephrologists and vascular surgeons for placement of dialysis access was often cast by physicians as “noncompliant” behavior or a means of “avoiding” dialysis. Other patients seemed to defer decisions about initiation of dialysis to their physicians (quote 3l). The question of whether patients were “competent” to make decisions about dialysis seemed to arise most often in situations in which patients rejected recommendations to start dialysis (quote 3m). In situations in which patients presented with cognitive impairment, it was more common to find documentation of concerns about whether patients had uremic encephalopathy as a compelling indication for dialysis than it was to find documented concerns about their capacity to participate in decisions about dialysis (quote 3n).
Patients’ treatment goals and values were rarely documented in the medical record. When they were documented, many patients emphasized quality of life and autonomy. In contrast with patients, physicians tended to have a safety-conscious approach toward initiation of dialysis and focused on the perceived hazards of delaying dialysis. Although patients were typically asked to provide informed consent before dialysis was initiated, the decision was frequently cast by physicians as one of “dialysis vs death” or grave disability (Table 4) (quote 3o). One physician documented that his patient felt that “he ‘has no choice’ if his other alternative is ‘dying.’”
The medical record serves as important documentation of health care interactions, accountability, and communication; it provides a unique vantage point from which to examine how treatment decisions are made about the timing of initiation of dialysis. Our analysis suggests that the initiation of maintenance dialysis reflects the complex interplay of at least 3 interrelated and dynamic processes: the care practices of individual physicians, sources of momentum for dialysis initiation, and interactions between patients and physicians.
Dialysis is associated with distinct benefits and harms for individual patients; clinical practice guidelines recommend that decisions about whether and when to initiate dialysis be shared between patients and physicians.6,25,26 In a shared decision-making model, treatment decisions integrate patients’ values, goals, and preferences with physicians’ expertise on prognosis and the risks and benefits of available treatments. Our findings spotlight the substantial challenges to achieving this ideal in real-world clinical settings.
These findings are in accord with studies suggesting that decisions about dialysis seem to be driven much more by physician-level factors, such as regional practice style, than by individual patient characteristics.27,28 Qualitative interviews with patients and ethnographic work conducted in renal clinics have also indicated that physicians tend to frame dialysis as an inevitable treatment rather than as an explicit treatment choice with associated benefits and harms.29-32 Documentation in the medical record tended not to focus on patient concerns and priorities. Instead, physicians emphasized achieving biomedical targets (ie, level of kidney function, signs and symptoms of uremia), coordinating procedures, keeping patients safe, and prolonging life. In situations in which patients did not start dialysis when recommended or started only when they became very sick, physicians tended to view this course of action as maladaptive behavior rather than as an implicit treatment choice. Taken together, our findings call for stronger efforts to improve patient-physician communication with a view to clarifying patients’ health care values, goals, and preferences and better aligning treatment recommendations with patients’ priorities. Our findings also suggest that there is work to be done to better understand how to best support patients’ readiness to face serious illness and engage in treatment decisions about initiation of dialysis.
We used patients’ medical records to analyze decisions about dialysis. Prior qualitative studies examining decisions about dialysis have been based on information gathered from one-time interviews with patients and physicians33-35 or ethnographic work conducted in renal clinics.29,30 Interviews do not provide direct information on how these decisions occur in a clinical context or on patient-physician interactions. Direct observation may alter the behaviors of patients and physicians36 and does not always capture decisions about treatment at the time they occur. Nevertheless, our study has limitations. First, owing to the complexity of the themes identified, our findings do not lend themselves to dichotomous categorization that would be needed to quantify the frequency of each theme among cohort members. Documentation was also not uniform across all patients and physicians; therefore, we could not be certain that a particular theme was not present in individual cases. Second, we selectively presented only dominant themes; our results are not exhaustive of all themes related to the timing of initiation of dialysis. Themes also reflect interactions with the health care system and what physicians chose to document in the medical record and thus provide only indirect and limited insight on the perspectives and experience of patients. Third, our analysis examines factors relevant to understanding the timing of initiation of dialysis among patients who initiated dialysis and was not intended to address decisions about whether to initiate dialysis in patients with advanced CKD. Fourth, our study findings within the VA may not be generalizable to other health care settings or populations (especially women). Nevertheless, the VA is the largest integrated health care system in the United States, and available evidence suggests that dialysis initiation practices in the VA parallel, and are more conservative than, those found in the wider US population.2 Last, there is some evidence that the trend toward earlier initiation of dialysis may be leveling off in recent years.1,2 Thus, it is possible that our results may not reflect current practice.
Our findings offer insight into the complex processes that shape the timing of maintenance dialysis in real-world clinical settings and suggest that there may be opportunities to make these processes more patient centered.
Accepted for Publication: November 8, 2015.
Corresponding Author: Susan P. Y. Wong, MD, Department of Medicine, University of Washington, 1959 NE Pacific St, PO Box 356521, Seattle, WA 98195 (firstname.lastname@example.org).
Published Online: January 25, 2016. doi:10.1001/jamainternmed.2015.7412.
Author Contributions: Drs Wong and O’Hare had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Wong, Liu, Williams, Hebert, O’Hare.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Wong.
Critical revision of the manuscript for important intellectual content: All authors.
Obtained funding: Burrows, Liu, Williams, Hebert, O’Hare.
Administrative, technical, or material support: Burrows, Williams, O’Hare.
Study supervision: Williams, O’Hare.
Conflict of Interest Disclosures: Dr O’Hare reported receiving royalties from UpToDate. No other disclosures were reported.
Funding/Support: This study was supported by grant VA IIR 09-094 from the United States Department of Veterans Affairs and by Interagency Agreement IAA 15FED1505101-0001 between the VA Puget Sound Healthcare System and the Centers for Disease Control and Prevention. Dr Wong is supported by the Clinical Scientist in Nephrology Fellowship from the American Kidney Fund. Drs O’Hare, Hebert, and Liu reported having received research funding from the Department of Veterans Affairs Health Services Research and Development Service. Dr O’Hare also reported having received research funding from the National Institutes of Health and an honorarium from the American Society of Nephrology.
Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Disclaimer: The findings and conclusions in this report are those of the authors and do not necessarily represent the official position or policy of the Centers for Disease Control and Prevention, Department of Veterans Affairs, or US government.