Six-Month Prognostic Tool May Improve End-of-Life Care for Hemodialysis Patients
By Pauline Anderson
Medscape Medical News
December 4, 2009 — Researchers have developed and validated a new tool for predicting survival of patients receiving hemodialysis (HD).
The new prognostic model, which is easy to use and includes 5 variables, will break down one of the main barriers preventing nephrologists from communicating bad news to patients — lack of an accurate prognostic tool, said study author Michael J. Germain, MD, from the Department of Medicine, Section of Nephrology, Baystate Medical Center, Springfield, Massachusetts.
"One of the reasons doctors don't like giving a prognosis to patents is that they're uncomfortable because they don't feel they have accurate information about prognosis; it's just kind of guessing," Dr. Germain said.
The goal of this research was to conduct a longitudinal prospective cohort study and to develop and validate an integrated prognostic model with sufficient accuracy to predict 6-month survival for patients receiving HD.
To build the new prognostic model, Dr. Germain and colleagues used actuarial factors (eg, age, albumin, comorbid disorders), clinician predictions (the "surprise" question [SQ]: "Would I be surprised if this patients died?"), and mortality analyses.
The researchers developed a derivation cohort of 512 patients receiving HD at 5 dialysis clinics in western Massachusetts. From chart reviews and interviews, the researchers gathered information on actuarial predictors and asked nephrologists whether they would be surprised if their patient died within the next 6 months.
They then collected similar information for a validation cohort of 514 patients at 8 New England HD clinics (the 5 original plus 3 additional clinics).
At the completion of the study, 54.9% of the "no" group of patients (those for whom nephrologists had answered 'no' to the SQ) had died compared with 17% of the "yes" group patients, with an unadjusted hazard ratio (HR) of 4.88.
SQ was 1 of 5 variables that were independently associated with early mortality in the derivation cohort. The SQ (not surprised vs surprised) had an HR of 2.71. The other variables were older age (HR for a 10-year increase, 1.36), dementia (HR, 2.24), peripheral vascular disease (HR, 1.88), and albumin (HR expressed for a 1-U increase, 0.27).
When using a model that integrates all 5 variables, the researchers determined that the area under the curve for predictions of 6-month survival in the derivation group was 0.87 (95% confidence interval, 0.82 - 0.92). The figures for 12- and 18-month survival were 0.82 and 0.79, respectively.
When testing this prognostic model in the validation cohort, the researchers found that the comparison prognostic figure for 6-month mortality in this group was 0.80 (95% confidence interval, 0.73 - 0.88).
"The integrated prognostic model lends itself to risk stratification of patients, it is more specific and sensitive than any of its components (e.g., the SQ), and it seems to be a considerable improvement over other existing instruments at predicting survival in the dialysis population," the authors write.
Furthermore, this prognostic model is sufficiently accurate for both research and clinical applications, they said.
To use the tool, physicians would simply open a spreadsheet on their computer or log onto a Web site and type in the numbers and answers in the appropriate 5 fields, and they will immediately get the prognosis. "It will literally take just 5 seconds to do," said Dr. Germain. "Or they could have it on a handheld device — we could have it as an iPhone application."
Although nephrologists do not want to extinguish a patient's hopes by delivering a poor prognosis, research shows that more than 90% of dialysis patients want to know their prognosis, and they want to hear the news from their own physician, said Dr. Germain. "Patients and families are relieved, and they don't get depressed or have a negative response to hearing their prognosis. Even after a patient has died, his or her family feels they are better adapted to the grief when told the prognosis ahead of time."
The researchers are planning a larger study that will compare a group of patients receiving usual care with another group receiving their prognosis derived using the new tool from both a nephrologist and a social worker. The main outcome of this multicenter study will be to see whether this prognostic tool affects hospital and palliative care, said Dr. Germain.
Adding a social worker, who has communication training, to the team helps in delicately relaying bad news. It also eliminates another barrier for nephrologists — lack of time.
Because both cohorts were drawn from a single geographical area (New England), "it remains to be seen whether the resultant model may be as accurate in other regions or in clinics with differing populations," the researchers said.
They also noted that there were not enough physicians in the study to allow for meaningful analysis of potential associations between nephrologist training, experience, and accuracy of SQ predictions.
The prognostic model may eventually be strengthened with the inclusion of other variables — for example, C-reactive protein, hemoglobin, and calcium levels and body mass index — said the authors.
The authors have disclosed no relevant financial relationships.
Clin J Am Soc Nephrol. Published online December 3, 2009.
