Costantini O, Huck K, Carlson MD, Boyd K, Buchter CM, Raiz P, Cooper GS. Impact of a Guideline-Based Disease Management Team on Outcomes of Hospitalized Patients With Congestive Heart Failure. Arch Intern Med. 2001;161(2):177-182. doi:10.1001/archinte.161.2.177
Congestive heart failure is the most common reason for hospitalization in the United States, and guidelines to improve the quality of care for patients with congestive heart failure have been developed. However, adherence is typically low. We hypothesized that a guideline-based care management team would result in greater quality and efficiency of care than guidelines alone.
A faculty cardiologist and nurse care manager at an academic medical center reviewed each patient's data and made guideline-based recommendations. Hospital length of stay, total costs, and use of recommended guidelines were compared between 173 patients before team implementation but with available guidelines, 283 care-managed patients, and 126 concurrent non–care-managed patients.
Care-managed patients achieved higher rates of use of angiotensin-converting enzyme inhibitor than baseline or non–care-managed patients (95%, 60%, and 75%, respectively; P<.001), as well as increased adherence to guidelines for daily weight monitoring and assessment of left ventricular function. Hospital length of stay was lower (median, 3, 4, and 5 days, respectively; P<.001) as were costs of hospitalization (median, $2934, $3209, and $4830, respectively; P<.01). These differences persisted after adjustment for severity of illness.
When compared with dissemination of guidelines alone, an active care management approach was associated with significant improvements in quality and efficiency of care for hospitalized patients with congestive heart failure.
IN THE UNITED STATES, congestive heart failure (CHF) is the leading discharge diagnosis by diagnosis related group, accounting for nearly 1 million hospitalizations and causing 200 000 deaths each year.1 Annual expenditures for CHF are estimated to be $38 billion, of which $23 billion is for inpatient care.2 Medicare expenditures for CHF exceed those for acute myocardial infarction and cancer combined.
In the past decade, a variety of therapies, including angiotensin-converting enzyme (ACE) inhibitors and certain β-adrenoceptor antagonists, have been shown to improve functional status and survival for patients with CHF.3- 6 Specific diagnostic tests (echocardiography, cardiac gated blood pool scan, etc) have also been shown to improve heart failure management. In addition, experts have reached consensus and have published guidelines regarding the appropriate management of heart failure.2,7,8 However, despite compelling medical evidence, the drugs, tests, and recommendations have not been used optimally in all patients.1,9,10
Although a care path for CHF that included comprehensive and specific recommendations for patient care on each hospital day had been implemented at our institution in 1995, the adherence was low. For example, in 1996, only 25% of patients with heart failure were enrolled and treated according to the care path. Given the previous problems with implementation, we developed simple guidelines targeted at specific patient-care decisions for the treatment of CHF caused by left ventricular systolic dysfunction. Consensus among caregivers was gained by involving nurses, primary care physicians, and the full- and part-time faculty in the design process. Finally, we proposed to implement the guidelines by means of a physician expert and a nurse care manager who reviewed each patient's case daily and made recommendations when appropriate regarding patient treatment. We hypothesized that this process would increase the use of the recommendations outlined in the national guidelines, improve the quality of care, and reduce length of stay for patients with CHF by ensuring early discharge when appropriate.
The CHF disease management program was developed at University Hospitals of Cleveland, Cleveland, Ohio, a 950-bed urban tertiary care hospital that includes both full-time and part-time faculty. Approximately 50% to 60% of patients are cared for by house staff, and the remainder by nurse practitioners or physician assistants. Although patients with CHF are typically admitted to general medical floors, the admitting physician can opt to admit the patient to a telemetry ward or intensive care bed, if clinically indicated. However, with the exception of the cardiac intensive care unit, formal involvement of a cardiologist in patient care is left to the discretion of the attending physician.
Before the design of the new program, we conducted a series of interviews with nurses, house officers, primary care physicians, and cardiologists to ascertain reasons thought to be responsible for the limited success of the previous care path. Although cardiologists had designed the care path, consensus had not been gained among primary care physicians. Many physicians and nurses were unaware of the existence of a heart failure care path, and it was thought to be too complicated and cumbersome. Physicians were not involved in the care path implementation process. All of those interviewed thought that absence of physician participation in care path implementation had limited the program's success.
A disease management team was assembled to achieve the primary goals of adhering to nationally published guidelines for the care of patients with heart failure and of identifying opportunities to improve quality and efficiency of care for hospitalized patients. This team initially consisted of a nurse care manager, a faculty cardiologist, and a physician representative from the 3 major part-time faculty groups with admitting privileges. The team reviewed published literature and national care guidelines to develop evidence-based recommendations for the inpatient treatment of patients with CHF. In addition, surveys were mailed to all admitting physicians to gain consensus regarding several care issues that might influence practice patterns, including monitoring intake and output vs daily weight measurements, and the use of monitored beds. Based on the literature review and establishment of physician consensus, we then identified 3 principal interventions for improvement of quality and/or efficiency of care. These included the use of ACE inhibitor medications, documentation of left ventricular function by echocardiography, and the consistent use of admission and daily weights. Specific guidelines are listed below.
A. Treatment with ACE inhibitors
Recommended for all patients with left ventricular systolic dysfunction
Contraindications include history of intolerance or angioedema, serum potassium level greater than 5.5 mmol/L, symptomatic hypotension, systolic blood pressure less than 90 mm Hg, and serum creatinine level greater than 221 µmol/L (2.5 mg/dL)
For patients with documented intolerance to ACE inhibitors or for patients symptomatic during treatment with maximally tolerated doses of ACE inhibitors and diuretics, add hydralazine hydrochloride and isosorbide dinitrate or angiotensin II receptor blockade
B. Use of echocardiogram
When there was no previous assessment of left ventricular function
When there is marked change in clinical status suggesting deterioration of left ventricular function (not applicable if dysfunction is known to be severe)
When there is new or markedly changed cardiac murmur
C. Implementation of daily weights
Obtain admission weight on all newly admitted patients
Perform daily weights in early morning before daily rounds
After the guidelines were disseminated, the nurse care manager screened newly admitted patients to identify cases of CHF appropriate for care management. A care management sheet was developed to aid the physicians caring for the patient in summarizing patient data and assessing progress. This 1-page sheet included the history, previous heart failure evaluation, medications, and clinical presentation. It also tracked the daily response to medical treatment with an ongoing summary of pertinent factors, such as vital signs, daily weight, dose of diuretics given, important laboratory values, and medication changes. The nurse care manager and faculty cardiologist reviewed the clinical response and treatment plan, offering daily recommendations, based on the previously developed care guidelines. Because individual patients may respond differently to interventions such as ACE inhibitors, there was variability in recommendations from patient to patient. The recommendations were written on the care management sheet, which was placed in the patient's chart. If the treatment plan by the clinicians varied significantly from the guidelines, the nurse care manager contacted the house officer, nurse practitioner, or physician assistant directly to discuss the recommendations. On occasion, the faculty cardiologist also contacted the attending physician to review the current guidelines and recommendations.
The nurse care manager also interviewed the patients to provide patient education, to assess discharge needs, and to evaluate the patient's ability to comply with a prescribed plan. Education on self-management was provided, with emphasis on diet, medications, fluid and weight monitoring, recognition of symptoms, and activity and exercise guidelines. Patients were also screened as potential candidates for clinical trials, cardiac rehabilitation, and cardiac transplant evaluation, and were referred as needed to ancillary services such as physical therapy, social services, and dietary and substance abuse counseling.
The care-managed group consisted of all patients with CHF discharged during the second through fourth quarters of 1997 in whom guideline-based recommendations were made, regardless of whether they were adhered to. Two comparison groups were also included. First, the baseline period included patients with a principal diagnosis of CHF who were discharged during the first quarter of 1997, which was immediately prior to implementation of the disease management program. Note that, during this period, a set of comprehensive guidelines had been in place for almost 2 years, but the other program features (ie, daily nurse and physician expert involvement) had not. Second, all patients discharged during the second to fourth quarters with a principal diagnosis of CHF but who were not examined and followed up by the team composed the non–care-managed group. These patients were missed by the screening process primarily because the nurse care manager was unavailable, the admitting diagnosis did not seem likely to be related to heart failure, or the patient was in "observation" status and therefore did not appear on the admitting list. In this group, written care guidelines were available for the clinician to use independently, but there was no active intervention undertaken by the CHF care management team. Medical record review was also used to verify the diagnosis of CHF for all patients in the baseline period and for non–care-managed patients.
The nurse care manager collected disease-specific clinical data on all care-managed patients, including use and dosage of ACE inhibitors, compliance with daily weight measurement, and use and appropriateness of echocardiograms. Similar data were collected through medical record review for all patients during the baseline period and for non–care-managed patients for the second through fourth quarters of 1997. The hospital's financial database was also used to measure resource utilization in both care-managed and non–care-managed patients during the study period. Data elements included length of hospital stay, total hospital costs, and intensive care unit costs, measured with the Cost Management Information System software (Eclipsys Corporation, Delray Beach, Fla).
To determine if potential outcome differences could be attributed to case mix rather than the interventions, 2 severity measures were used. First, the hospital's financial database included the All Patient Refined Diagnosis Related Group (APR-DRG) classification,11 which was based on International Classification of Diseases, Ninth Revision, Clinical Modification12 discharge diagnosis codes and was developed to predict resource consumption for a specific diagnosis related group. The APR-DRG assigned patients to a 4-point scale of severity of illness (1, mild; 2, moderate; 3, severe; and 4, catastrophic) based on the presence or absence of relevant codes. Second, a previously developed and validated disease-specific multivariable model was used to predict hospital length of stay. This measure was obtained from the Cleveland Health Quality Choice (CHQC) coalition13,14 and included data elements from the history, physical examination, and admission laboratory studies. On the basis of the model coefficients, a predicted length of stay was determined for each patient and used to assess severity of illness in subsequent analyses.
Comparisons of the use of specific interventions (ie, ACE inhibitor) and outcomes in care-managed and non–care-managed patients during concurrent periods, as well as care-managed patients and patients from the baseline period, were made with the χ2 test and Wilcoxon rank sum test for categorical and continuous variables, respectively. Because of the skewed distribution, average length of stay and costs were expressed as median and interquartile range (Q1-Q3). Within a given stratum of APR-DRG severity score, length of stay and costs were also compared between care-managed and non–care-managed patients by means of the Wilcoxon rank sum test. Finally, 2 multivariable linear regression models included care management and severity of illness as determined by the CHQC database as independent variables and either length of stay or total costs as dependent variables. The coefficient associated with care management indicated its independent association with outcome.
We identified 173 patients with CHF who were discharged during the baseline period (first quarter of 1997) and for whom a written care path was available. After implementation of the disease management team (second to fourth quarters of 1997), 283 patients constituted the care-managed group, whereas 126 patients constituted the non–care-managed group. The characteristics of the 3 patient groups are shown in Table 1. There were no differences between the 3 groups in age and sex distribution, but patients who were non–care-managed were less likely to have Medicare and more likely to have Medicaid as their primary insurance. Non–care-managed patients were also more likely than the care-managed group to have normal left ventricular function as measured by echocardiography, and had a somewhat higher severity of illness according to the APR-DRG classification. In contrast, severity of illness according to the CHQC model was somewhat greater during the baseline period than for care-managed patients, but similar for non–care-managed patients.
When compared with baseline and non–care-managed patients, the care-managed group showed significant improvements in the targeted clinical measures of quality of care (Table 2), including the use of ACE inhibitors at hospital discharge, recording of hospital admission and daily weights, and assessment of left ventricular function. Hospital length of stay of the patients in the care-managed group was also significantly lower than for the baseline period or the non–care-managed group (Table 2). The reduction in length of stay was not attenuated by an increase in the 30-day readmission rate or in-hospital mortality rate. The cost per case was lower for care-managed patients than for non–care-managed patients or patients who were treated during the baseline period (Table 2). Finally, care-managed patients had a significantly lower cost per intensive care unit admission than the other 2 groups.
When patients were stratified according to severity of illness as determined by the APR-DRG score, care management was generally associated with shorter hospital stay and lower total costs (Table 3). With the exception of the lowest severity stratum, which accounted for only 20 patients, both length of stay and hospital cost per case were significantly lower (P<.05) in care-managed patients. As the severity level increased, there was a progressively greater improvement in length of stay and hospital cost associated with care management. Within each stratum, all 3 groups were similar with regard to severity of illness according to the CHQC index.
Multivariable models were also used to determine the independent association of care management with both length of stay and cost of care. In a linear regression model adjusting for admission severity of illness according to the CHQC model, care management was associated with a 2.2-day reduction in hospital length of stay (P<.001). In a second model, care management was associated with a $2204 reduction in severity-adjusted total hospital costs (P<.001).
This study provides evidence that an active care management approach is superior to clinical practice guidelines alone in improving quality and efficiency of care in hospitalized patients with congestive heart failure. Implementation of this program, which focused on a finite number of evidence-based interventions, was associated with measurable improvements in use of recommended therapies and diagnostic studies, as well as reductions in resource consumption. Moreover, these differences were observed for both concurrent non–care-managed patients and for patients admitted before program implementation and persisted after adjustment for patient severity of illness.
In an attempt to improve physicians' decision making and quality of care, clinical practice guidelines have been designed to define the best practice for specific clinical situations. The American Medical Association lists more than 1000 medical practice guidelines for health care delivery.15 The recent impetus for guidelines arises from the recognition of large practice variation among physicians, the perception of overuse of expensive resources in health care, and the accelerating growth of managed care and other drivers of cost containment and quality assurance.15- 17 Despite the fact that clinical guidelines have been developed throughout the country for many common disease processes, physicians are reluctant to embrace them. Most studies report a 50% or less compliance with established national guidelines.9,18 Efficacy of guidelines depends in part on the type of practitioners who implement them (specialist vs primary care),1 the type of hospital in which guidelines are used, the complexity of the guidelines ("user-friendliness"), and the availability of the guidelines to practitioners.
More recent efforts have focused on an active multidisciplinary approach to the care of patients with common conditions such as congestive heart failure.19 In one study, patients followed up in a cardiomyopathy clinic had a significant decrease in hospital admissions and emergency department visits and an improvement in functional status when compared with those who were given usual care.19 Similar results were obtained in patients enrolled in an outpatient interactive home monitoring program20 and for patients with advanced heart failure followed up through a comprehensive active management program after discharge.2 All of these efforts have shown that a disease-specific, active management team is more likely to adhere to nationally recommended clinical guidelines than are individual physicians who may not be comfortable, for example, with initiating ACE inhibitors in patients with borderline hypotension and/or tenuous renal function.21 However, unlike these programs, in the present study, the nurse care manager and faculty cardiologist were not directly responsible for patient care.
While an active management approach has been successful in an outpatient setting, few studies have examined the effectiveness of such an approach in the hospitalized patient. Weingarten et al10 found that the use of a CHF practice guideline to reduce the length of stay in a coronary care unit was not embraced by physicians, despite verbal and written recommendations to promote the early transfer of "low-risk" patients to unmonitored beds. However, the investigators did not follow up the patients throughout the hospital course. In fact, physicians may have compensated for a statistically insignificant reduction in monitored length of stay by significantly increasing the length of stay in unmonitored beds.10 More recently, despite the well-established role of echocardiography in the diagnosis of CHF and of ACE inhibitors in its treatment, only 18% of patients were discharged with the appropriate ACE inhibitor dose prescribed,22 and only 63% of patients with newly diagnosed CHF underwent an echocardiogram.23
In this study, the nurse care manager and faculty cardiologist helped to ensure that the physicians directly involved in the care of the patient followed national guidelines. Rather than assuming care of the patient, the nurse wrote daily notes in the chart with simple suggestions, in the spirit of the national recommendations, and maintained continued interaction with caregivers. For example, from a quality standpoint, we recommended the use of echocardiography if indicated, initiating or increasing the dose of ACE inhibitors in patients with borderline hypotension and/or renal function, and using daily weights as a gauge for diuretic dosing. It is important to note that all patients, not only those who were care managed, had a clinical care path for CHF available that had been developed at the hospital approximately 2 years before the initiation of our program. However, during the baseline period, only 60% of patients received ACE inhibitors or had an echocardiogram to assess left ventricular function. It is also notable that ACE inhibitor use and daily weight recording increased in the non–care-managed patients compared with the baseline period, which may represent a "trickle-down" effect of the program. Thus, through the constant education of physicians and nursing staff who also cared for care-managed patients, the non–care-managed group, missed by our screening, probably also benefited.
We recognize several limitations of the study. First, this study was observational and not a randomized trial. Although severity adjustment was performed by 2 indexes, the non–care-managed group was composed of patients who were not captured by the screening process. Therefore, there may be potential differences in patient severity of illness that were unmeasured by our risk adjustment methods and could account for some of the observed results. Also, it is worth noting that a significant minority of patients in all 3 groups had normal left ventricular function. The majority of these patients had, as a final diagnosis, diastolic dysfunction and may have been treated as if they had left ventricular dysfunction until the results of the echocardiogram were available. Because of the nature of their disease, these patients were less likely to benefit from our active management program, which was designed primarily for systolic heart failure. Although the non–care-managed and baseline groups consisted of a greater proportion of patients with normal left ventricular function than the care-managed group (Table 2), this difference should not affect the results of our study. Given the better prognosis of patients with diastolic dysfunction, it could favor the non–care-managed group. Finally, given temporal trends in the treatment of CHF, the guidelines are periodically updated and currently differ somewhat from those during the period of this study. For example, more recent guidelines have specified a target dose for ACE inhibitors and have included treatment algorithms with β-blockers.24 The heart transplantation program that was developed after this study has also subsequently affected guideline content and resource utilization.
We conclude that, when compared with dissemination of guidelines alone, an active care management approach was associated with significant changes in practice for hospitalized patients with CHF. These differences suggest that the constant education and written recommendations were responsible for the difference in quality and efficiency of care. Future studies should address the benefit of this approach in improving long-term outcomes of this patient population, as well as its impact on the care of patients with other cardiovascular and medical diagnoses.
Accepted for publication July 20, 2000.
Corresponding author: Gregory S. Cooper, MD, Division of Gastroenterology, University Hospitals of Cleveland, 11100 Euclid Ave, Cleveland, OH 44106 (e-mail: email@example.com).