Computerized decision support for anemia management in hemodialysis patients

Abstract

In 2005, Dialysis Clinic Inc. (DCI) implemented a computerized decision support program for anemia treatment in end-stage renal disease. The program has evolved over time and remains critical to patient care despite the changing scientific, regulatory, and reimbursement environment. Our goal is to optimize anemia management and decrease nursing time devoted to anemia management. We have separate protocols for administration of erythropoietin and iron. Patients on both the erythropoietin and iron protocols achieved hemoglobin distributions more favorable than those on neither protocol and used lower doses of erythropoietin and iron. Nursing time devoted to anemia management decreased substantially. Widespread adoption of the computerized decision support program has allowed DCI to meet our anemia goals while increasing time for our nurses to do important clinical work.

Introduction

In 2005 Dialysis Clinic Inc. (DCI), a national not-for-profit dialysis provider, implemented its first computerized decision support protocols for anemia treatment in end-stage renal disease. Eight years later, these protocols remain critical to patient care despite the major changes in the scientific, regulatory and reimbursement environment. The primary goal of our computerized decision support program is to optimize anemia management while providing more time for our healthcare professionals to attend to other important patient needs.

The protocols address issues common to all dialysis facilities. Prior to the implementation of our decision support program there was considerable variation in anemia management. Some physicians reviewed and responded to patients’ monthly laboratory data immediately. Other physicians responded more slowly and required prompting. The resulting delays allowed potential opportunities for dose adjustment to slip by for several weeks. Some facilities deputized anemia managers to prompt physicians or to adjust doses according to written treatment protocols. Even the most assiduous physicians and nurses had difficulty integrating many months of hemoglobin, serum ferritin, and transferrin saturation values with corresponding erythropoietin and iron doses.

In developing our computer support program we took as a point of departure an existing written erythropoietin protocol, based on the package insert. This protocol was nominally used by a cluster of facilities with results that were considered excellent at the time, with the majority of hemoglobin values equal to or greater than 11 g/dL.

Upon review, our corporate nurses recognized that this written protocol was not consistently followed across the organization by the facilities that subscribed to it. Therefore, it was unclear whether the protocol we had selected was itself excellent, or whether it was in use in facilities with excellent practitioners. We briefly considered performing audits to measure adherence to the protocol. However, we realized that the results were likely to be inconclusive, because of sample size and the many ways in which orders might depart from the written protocol. Therefore, our initial computerized erythropoietin protocol was designed to make order changes in response to hemoglobin results more rapid and consistent. Currently, our decision support consists of two protocols, one for erythropoietin and a second for iron administration. We piloted the erythropoietin protocol in a few facilities whose medical directors were members of the DCI Information Technology Steering Committee. This was a quality improvement effort, not a research project, and our results are not generalizable to facilities that lack DCI’s specific information management infrastructure.

It soon became evident that protocol-driven anemia management achieved hemoglobin distributions at least as narrow as did manual treatment. Not surprisingly, an increasing number of facilities adopted the erythropoietin protocol. In early 2006, the Centers for Medicare & Medicaid Services liberalized its payment policy with respect to erythropoietin doses given after hemoglobin values exceeded 13 g/dL. We changed the protocol to allow more liberal administration of erythropoietin. We thought that this new policy would decrease hemoglobin cycling that resulted from frequent interruptions in erythropoietin administration. However, several months later we became concerned that the revised protocol might be too aggressive. Therefore, at our October 2006 annual meeting, the medical director steering committee decided to return to the original, more conservative, protocol. This chain of events reflects a major strength of computerized protocols, the ability to rapidly assess the effects of changes and make the necessary adjustments.

Over eight years we have changed the erythropoietin protocol 15 times, making more than half the changes in the first two years. Since December 2006, we have adjusted the protocol annually, except in 2011, when we made an additional modification in response to the U.S Food and Drug Administration erythropoietin label change. We have introduced finer gradations in dose changes in response to the slope of hemoglobin concentrations over time, and have made allowance for variation in hemoglobin testing frequency, allowing default testing either every two or four weeks. The protocol mandates weekly hemoglobin determinations when erythropoietin administration is interrupted. Physicians choose the ceiling hemoglobin above which no erythropoietin is given, the ceiling weekly erythropoietin dose, and the frequency and route of administration.

In a survey conducted after the implementation of the erythropoietin protocol our nurses reported that the time they devoted to anemia management had decreased 50%. In 2010 we introduced an iron protocol, which further reduced nursing time spent on anemia management. The iron protocol incorporates the type of iron preparation, target values for transferrin saturation, and ceiling values for serum ferritin concentration above which no iron is given. The protocol specifies a range of values for these parameters and the nephrologist chooses which to follow. The protocol appropriately delays blood sampling for measurement of transferrin saturation and serum ferritin following the most recent iron administration.

Patients are assigned to the protocols by physician orders, each of which states the protocol explicitly. The computer executes their intentions promptly and consistently. We have taken several steps to ensure that individual physicians retain their ability to individualize anemia management. First, DCI considers assignment of a patient to a protocol a medical decision, which should be made by the individual nephrologist in collaboration with the patient. Accordingly, physicians retain the freedom to decide for whom the protocol is appropriate. Each protocol incorporates several treatment options, allowing the physician to individualize values of protocol parameters at the patient level. The physician may over-ride any individual order and may remove the patient from protocol treatment at any time. Second, at our semiannual medical directors meeting we share outcomes associated with the treatment protocols, provide updates on existing protocols and seek input on existing protocols and suggestions for improvement. Third, protocol adoption is encouraged, but it is not required. In the early years of protocol deployment, if a corporate review identified a facility as not treating anemia effectively under local practices, we strongly encouraged use of the computer protocols. Now, virtually all facilities choose to use the protocols. Fourth, we assess anemia management monthly using corporate-wide and facility-specific control charts.

More than 80% of our hemodialysis patients are now treated with both the erythropoietin and the iron protocols. Computerized protocols have virtually eliminated medication wastage. Below we compare results observed in patients on both the erythropoietin and iron protocols (n=7399) with those patients who were on neither protocol (n=434) for all 12 months of 2012. The mean dose of erythropoietin among patients on both protocols was only 83% of the dose received by patients on neither protocol. Similarly, iron dose among patients on both protocols was only 77% of that administered to patients on neither protocol.

The distributions of hemoglobin values among patients assigned to both protocols and neither protocol are depicted in the figure. Patients on both protocols were more likely to have hemoglobin values between 10 and 12 g/dL (74%) compared to patients assigned to neither protocol (70%). Patients assigned to both protocols were almost two-thirds as likely to have hemoglobin concentrations less than 10 g/dL, and half as likely to have hemoglobin concentrations less than 9 g/dL compared to patients assigned to neither protocol. The percent of patients with hemoglobin values above 13 g/dL was similar among patients on both protocols versus neither protocol.

Conclusions

In summary, in collaboration with affiliated physicians, and building on a solid information system, DCI has developed and implemented an effective decision support program for anemia management. Patients who were on both the erythropoietin and iron protocols achieved hemoglobin distributions that were more favorable than those on neither protocol and did so using lower doses of both erythropoietin and iron. Widespread adoption of the computerized decision support program has allowed DCI to meet our anemia goals while providing more time for our professional staff to do important clinical work. -by Philip Zager, MD; Klemens B. Meyer, MD; Vlad ladik, MS; SUsan Paine, MPH; Doug Johnson, MD