Background
Continuous renal replacement therapy (CRRT) is often employed in hemodynamically unstable patients in the intensive care unit (ICU) with severe acute kidney injury (AKI) and end stage renal disease (ESRD). [
1,
2] In-hospital mortality in this patient population is high, consistently reported as over 50 %. [
3‐
7] However, patients with severe AKI that survive their hospital stay have an excellent chance of recovering renal function; over 85 % of these patients do not require indefinite renal replacement therapy [
3,
7,
8].
An accurate understanding of prognosis allows patients and their health care proxies (HCPs) to make informed decisions about goals of care. This understanding is critical to maintain good patient-provider relationships, and may be associated with less psychosocial distress for patients and HCPs. [
9‐
11] The majority of studies assessing patient and provider expectations of illness are in oncologic disease. These data suggests that both groups overestimate likelihood of survival, with patients being more optimistic than their physicians overall. [
12‐
16] Similarly, a recent survey study among nephrologists and patients with ESRD showed that patients dramatically overestimate their probability of long-term survival and were more optimistic about their candidacy for renal transplantation compared to their physicians [
17].
Maintaining effective communication and making shared treatment decisions pose many challenges in the ICU setting. [
18‐
23] The initiation of CRRT presents an opportunity for patients/HCPs and ICU providers to discuss the logistics and implications of starting a new procedure and often leads to questions about expectations of illness. Despite many attempts to identify potential risk factors for poor outcomes in this patient population, there is no widely accepted predictive model, and thus the potential for communication mismatch exists. [
3,
24,
25] There are currently no studies in the literature that assess the expectations of survival and renal recovery of patients/HCPs, physicians, and critical care nurses around patients receiving CRRT. We sought to assess this potential mismatch of the prognostic assessment between these groups using a survey tool that assesses all three groups.
Methods
Setting and Participants
Patients who were treated with CRRT in the medical or cardiac ICUs at Massachusetts General Hospital (MGH) between February and August 2013 were eligible for this study. Patients were excluded if the primary medical team did not feel that a survey study was appropriate given the clinical context of the patient’s care (e.g., if patient’s care was limited to comfort measures). Surveys were performed within the first 48 hrs of a patient starting CRRT. Surveys were administered either verbally or on paper depending on participant preference. If patients were unable to complete the survey (due to their medical situation, e.g. intubation or sedation), their HCPs were interviewed instead. No member of the research team was directly involved in patient care for any of the participants in this study. All surveys were performed in English and all respondents self-reported fluency in English.
MGH is a tertiary care hospital with 1008 total beds and 126 ICU beds available for CRRT, 34 of which are in the medical or cardiac ICU. CRRT is provided as continuous veno-venous hemofiltration using machines from NxStage Medical (Lawrence, MA). All nurses who perform CRRT have received specialized training and have at least 18 months of critical care nursing experience at MGH. Decisions to initiate CRRT are made by nephrologists in consultation with intensivists. CRRT is performed in lieu of intermittent hemodialysis as per local standard of care. Consent and education about CRRT was performed as per standard of care by the consulting nephrology team prior to the initiation of CRRT. No formal script or additional educational material around CRRT was provided to those participating in this study by the study team.
Survey Methodology
A member of the research team identified potential index patient cases who were initiated on CRRT within 48 hrs by reviewing the ICU and nephrology consult censuses. Each survey triad included (1) a patient on CRRT (or their HCP), (2) an ICU physician from the primary care team, and (3) an ICU nurse currently employing CRRT for that patient. Patients/HCPs were given a six question survey designed by the research team that included one open ended question, two “true or false” questions, and three multiple choice questions (Table
1). Surveys assessed understanding of CRRT (questions 1–3), expectations of survival and need for indefinite dialysis (questions 4–5), and how well the care team explained CRRT (question 6). In order to maximize comparability within survey triads, multiple-choice questions were presented as four choices of percentage quartiles. Following each patient/HCP survey, a member of the primary medical ICU team and a primary nurse were independently asked three analogous questions about general expectations of survival, likelihood of requiring indefinite dialysis, and assessment of patient/HCP understanding of CRRT. All participants were asked to give generalized answers about patients on CRRT rather than the index case in order to minimize bias. Providers who cared for more than one patient enrolled in this study were only surveyed once. Sufficient survey-naïve providers were available for all index cases. Physicians surveyed were critical care attendings, critical care or cardiology fellows, or internal medicine trainees (post graduate year one through four) and were selected based on survey-naïve status and clinical availability. Medical records were reviewed for relevant demographic and medical information about the patient’s admission. Primary admitting diagnosis for each patient was determined from the medical record discharge summary.
Table 1
Survey – Subjects were asked one open-ended question (question 1, patient/HCP only) and five multiple-choice questions (questions 1–6)
1 | What is your understanding of this machine’s purpose? | Open Ended |
Patient/HCP only |
2 | CRRT is a procedure done on patients who have kidney failure where a machine replaces the job of the kidneys. | True or False |
Correct Answer: True |
Patient/HCP only |
3 | CRRT helps the kidneys heal faster | True or False |
Correct Answer: False |
Patient/HCP only |
4 | How many patients who are treated with CRRT in the intensive care unit will need to be on dialysis forever after leaving the hospital? | Multiple Choice |
Choices: <25 %, 25-49 %, 50-75 %, ≥ 75 % |
5 | How many patients who are treated with CRRT in the intensive care unit will survive and leave the hospital? | Multiple Choice |
Choices: <25 %, 25-49 %, 50-75 %, ≥ 75 % |
6 | For Patients/HCP: How well did the doctors do in explaining why I (or my family member) needs CRRT? | Multiple Choice |
| For MD and Nurse: How well do you feel the patient or HCP understands why they need CRRT? | Choices: Completely, mostly, a little, not at all |
Statistical Analysis
A prior study at MGH assessing all patients requiring CRRT from 2008 to 2011 reported the probability of survival to hospital discharge between 25 % and 50 % and the probability of requiring indefinite dialysis once discharged at less than 25 %. [
3] These results are supported by multiple prior studies. [
4,
5,
7,
8,
26‐
34] These responses were considered consistent with the literature when adjudicating responses.
Immersion crystallization and codebook analysis were performed to identify important and recurring themes for open-ended answers in question 1. [
35,
36] Descriptive analysis was performed for questions 2, 3, and 6. Univariate Fisher’s exact analysis was performed for questions 4 and 5 to compare responses between patient/HCP, physician, and nurse subgroups. Univariate rank sum and Fisher’s exact analyses were used for continuous and categorical variables, respectively, to analyze for correlation between patient/provider demographic or medical background information and correct answer choices. STATA version 12.1 (StataCorp LP, College Station, TX) was used for all statistical analysis.
A pre-determined interim analysis was performed after 20 survey triads were completed. After an additional 12 triads were completed without significant change to preliminary results study recruitment was deemed complete.
Ethics Statement
Written informed consent was obtained from all study participants. All responses and patient/provider information were de-identified except to members of the research team. The Partners Human Research Committee for human subjects approved the study. All clinical investigation was conducted according to the principles expressed in the Declaration of Helsinki.
Discussion
Our study describes the single-center experience of mismatch around patient/HCP and provider perceptions of CRRT using a non-validated survey tool and descriptive analysis techniques. Of note, most patients in this cohort were too ill to participate in a survey, and therefore 88 % of responders from the patient/HCP group were HCPs. While HCPs often must make decisions for patients in the ICU, it should be noted that HCPs represented the majority of the patient/HCP subgroup, and may not necessarily have the same views or understanding as the patients themselves.
One of the most prominent findings in this study was that all three subgroups – patients/HCPs, ICU physicians, and nurses – provided answers that were inconsistent with published literature around survival and likelihood of requiring long-term dialysis. Forty-four percent of ICU physicians and nurses identified in-hospital mortality rates consistently with the literature, as did just 6 % of patients/HCPs. All three subgroups were similarly discrepant in assessing the likelihood of requiring long-term dialysis compared to prior studies. While both nurses and physicians were significantly more consistent with published probabilities of survival than patients/HCPs, these responses highlight a need for further education for patients and providers alike. Our results are consistent with prior literature examining attending and resident physicians’ accuracy of prognosis. Survey studies around patients with acute congestive heart failure, late stage cancer, and critical illness requiring ICU admission described similarly low rates of prognostic accuracy. [
13,
37‐
39] At our study site, consulting nephrologists, rather than ICU physicians or nurses, obtain informed consent for CRRT. This study did not assess the degree of communication between consulting nephrologists and the ICU care team and patients/HCPs. The consent process is a natural time for patients/HCPs to ask questions about survival and potential long-term dialysis needs; this interaction may serve as another potential target for investigation of perceptions of CRRT.
This study included only patients who were initiated on CRRT, as opposed to intermittent hemodialysis. While studies comparing outcomes between these two populations have shown similar rates of survival and long-term dialysis dependence, [
40] the results of this study are best generalized to the CRRT population. It should also be noted that the 97 % of patient/HCPs identified as having a high school education or greater. Those with lower education levels might have even worse prognostic estimates, though this conjecture requires further study to evaluate in the CRRT population.
Compared to their physicians, our results show that patients/HCPs overestimate the chance of survival, which is consistent with prior literature comparing patient and provider views in the non-ICU setting. [
12‐
17] Our study is novel in that it also examines ICU nurses’ perceptions of patient outcomes. Despite the fact that nurses have more direct contact with patients/HCPs than physicians, [
41,
42] we found no differences in the responses of physicians and nurses. Nurses’ understanding of illness may have a strong impact on patient/HCP perceptions and expectations, and thus highlights an additional opportunity for interventions to improve understanding of prognosis and communication in the ICU. This concept requires further directed study to examine the impact of nursing beliefs of illness on care of patients with CRRT.
Perhaps the richest patient/HCP information came in open-ended responses to question 1. Overall, responses spanned several thematic areas important to the function of CRRT. With few exceptions, patients/HCPs were generally accurate in describing some of the reasons CRRT was employed. Answers were typically short, and very few patients talked about CRRT as a life-supporting measure. Answers typically focused on the mechanical function of the device – commonly described as a filter, a “slow” treatment, or as an extracorporeal device with blood being removed then returned to the body. It is not known if this pattern of responses was because the idea of CRRT as life support (analogous to a ventilator, for example) was not emphasized during the consent or patient education process, or because this was difficult of patient/HCPs to acknowledge.
There is an interesting paradox that exists in the literature surrounding communication of advanced illness. Even after candid conversations about advanced care planning, patient/HCP understanding of prognosis and surrogate decision-making preferences has been shown to be low. [
43,
44] However, recent data suggests that many patients prefer to hear from their doctor if their overall prognosis is poor, suggesting that providers should feel both empowered and obligated to share as much prognostic information as appropriate with patients and HCPs. [
45,
46] These discussions can be challenging in the ICU setting, however, due to lack of consistent prediction models, frequently evolving treatment courses, variations in patient/HCP belief systems, and the emotional stresses inherent to the ICU. [
3,
18‐
25] At this study site, there is no standardized patient education process around CRRT, which may be a barrier to improving discordance around CRRT. Augmenting direct clinician counseling with multimedia educational tools, such as decision support videos, has been shown to be successful in improving patient/HCP understanding of illness and assisting in making decisions consistent with patients’ value systems. [
47] These tools have been used in other patient populations with poor overall prognosis, including counseling about cardiopulmonary resuscitation in the ICU. [
48‐
53] Our study highlights a significant knowledge gap and communication mismatch between patients/HCPs and ICU providers around CRRT, suggesting that there may be a role for such a tool around this intervention.
This study should be interpreted with the context of its limitations. It was performed at a single site using a small sample, thus potentially limiting generalizability. Since this study was not powered to pick up individual predictors of accuracy, no patient or HCP demographic factors correlated with answer choices. However, this study was designed in large part to be a descriptive and qualitative investigation of perceptions of illness across parallel patient and provider groups. We feel that our holistic message is more important than the power of the individual data points. ICU physicians and nurses were instructed to answer based on perceptions of overall survival and renal recovery rates (rather than for the index patient case) in order to maximize objectivity and comparability between groups. However, provider expectations for each index patient may have had an influence on answer choices due to temporal proximity to the survey, thus capturing some degree of bias. In line with this, the survival rate in this sample was 59 %, which is higher than the rates of less than 50 % that are routinely published in the literature. This is likely explained by a selection bias, as sicker patients who had a higher mortality were more likely to die (or transitioned to comfort care measures) prior to being interviewed, thus artificially improving survival in our sample. Still, this may have affected responses in all three subgroups. Because no validated survey tool previously existed that could assess the questions of this study, our research group had to create our own short survey. Prior studies assessing patient understanding of illness in hemodialysis and in the ICU have done so with similar, non-validated survey tools, [
17,
50] so we feel that are results remain interpretable. Prior to the beginning of the study, the survey tool was reviewed by two members of each subgroup in order to ensure understandability. These subjects were not included in the final analysis.
Competing interest
The authors declare that they have no competing interest.
Author contributions
ASA was the primary author of the manuscript and contributed to analysis. GH contributed to patient recruitment and analysis. RC and KC contributed to patient recruitment and analysis. EB and IB contributed to manuscript writing and study design. All authors read and approved the final manuscript.