Improving Guideline Adherence
A Randomized Trial Evaluating Strategies to Increase β-Blocker Use in Heart Failure
Background— The dissemination of clinical practice guidelines often has not been accompanied by desired improvements in guideline adherence. This study evaluated interventions for implementing a new practice guideline advocating the use of β-blockers for heart failure patients.
Methods and Results— This was a randomized controlled trial involving heart failure patients (n=169) with an ejection fraction ≤45% and no contraindications to β-blockers. Patients’ primary providers were randomized in a stratified design to 1 of 3 interventions: (1) control: provider education; (2) provider and patient notification: computerized provider reminders and patient letters advocating β-blockers; and (3) nurse facilitator: supervised nurse to initiate and titrate β-blockers. The primary outcome, the proportion of patients who were initiated or uptitrated and maintained on β-blockers, analyzed by intention to treat, was achieved in 67% (36 of 54) of patients in the nurse facilitator group compared with 16% (10 of 64) in the provider/patient notification and 27% (14 of 51) in the control groups (P<0.001 for the comparisons between the nurse facilitator group and both other groups). The proportion of patients on target β-blocker doses at the study end (median follow-up, 12 months) was also highest in the nurse facilitator group (43%) compared with the control (10%) and provider/patient notification groups (2%) (P<0.001). There were no differences in adverse events among groups.
Conclusions— The use of a nurse facilitator was a successful approach for implementing a β-blocker guideline in heart failure patients. The use of provider education, clinical reminders, and patient education was of limited value in this setting.
Received December 17, 2002; revision received March 17, 2003; accepted March 17, 2003.
Clinical practice guidelines are designed to translate medical research and expert opinion into recommendations for everyday practice, but health care providers are often slow to incorporate these recommendations.1–3 For example, despite firm evidence for a mortality benefit of angiotensin-converting enzyme (ACE) inhibitors in chronic heart failure (CHF), they remain underutilized.4–6 Although ACE inhibitor use has risen in recent years,7 the consequences of these delays likely resulted in a substantial excess of morbidity and mortality and in increased health care costs.
β-Blockers greatly reduce mortality and morbidity in patients with CHF8–10 and are recommended in all current CHF guidelines.11–13 However, their use for this indication has not penetrated most practice settings. Indeed, because most practicing physicians were taught that β-blockers are contraindicated in CHF and because initiating these agents is complex and time consuming, it is likely that the incorporation of these recommendations will occur slowly and that innovative approaches will be required to enhance their use.14
Prior attempts at guideline implementation have demonstrated that active interventions, rather than passive dissemination of guidelines, are more successful in changing processes of care.15 Other strategies associated with better guideline adherence include timely patient-specific reminders, focused treatment objectives, and involvement of patients as partners in their medical management.16 The use of a nurse case manager for inpatients with CHF has also been shown to improve guideline adherence.17
Therefore, we conducted a randomized trial to evaluate the effectiveness of two interventions to increase β-blocker utilization in patients with CHF.
Experimental Design and Sample
We conducted a randomized trial to determine whether two intervention strategies, a nurse facilitator and a combination of patient-specific computer reminders and patient letters, could improve the utilization of β-blockers in appropriate, stable outpatients with CHF compared with an aggressive provider education program alone.
Eligible practitioners included general internists, cardiologists, and other internal medicine subspecialists, medical residents, and nurse practitioners who provided primary care for patients with CHF at the San Francisco Veterans Affairs Medical Center (SFVAMC). The investigators in this study and their patients, including those followed in the Heart Failure Clinic, were excluded.
A stratified randomization using computer-generated random numbers was used to ensure an even distribution by provider training and CHF caseloads across the 3 arms of the trial. Randomization occurred at the provider rather than patient level because of the likelihood of contamination if a provider’s patients were assigned to different arms of the trial. The trial period started on February 1, 2000, and patients and providers were followed for 1 year after they were identified as candidates for β-blockers or until the trial’s completion on April 16, 2001, whichever came earlier (median follow-up, 12 months; mean±SD, 11±2 months). The Committee on Human Research at the University of California, San Francisco approved this study, and informed written consent was obtained from each provider.
Potential subjects were identified initially from outpatient diagnostic codes for CHF between October 1, 1998, and September 30, 1999. Patients were eligible for inclusion if they were receiving primary care at the SFVAMC at the start of the intervention period (February 1, 2000), met Framingham criteria for CHF,18 and had left ventricular ejection fraction (EF) ≤45% or moderate or severe left ventricular systolic dysfunction on their most recent evaluation.
Exclusions included current treatment with β-blockers at predefined target doses (carvedilol ≥50 mg/d, metoprolol tartrate ≥100 mg/d, or atenolol ≥100 mg/d) or at lower doses if there was documented intolerance at higher doses. Patients were also excluded for contraindications to β-blockers, including asthma, severe obstructive lung disease (associated with chronic oral corticosteroid treatment, home oxygen use, or forced expiratory volume in first second [FEV1] <1 L), diabetes with hypoglycemia, greater than first-degree atrioventricular block, symptomatic sick sinus syndrome, heart rate ≤60 beats per minute, systolic blood pressure ≤90 mm Hg, prior documented intolerance to a β-blocker, refractory CHF with persistent volume overload, or advanced comorbid conditions.
A total of 1247 subjects were identified by an outpatient CHF diagnosis, of which 390 were excluded for lack of primary care at the VA (n=265, including 49 patients who had died) or having a primary investigator as a provider (n=134). After chart review using prespecified criteria, 209 of the remaining 857 patients were eligible for the study after exclusions for the reasons indicated in Figure 1. As required by the University of California, San Francisco Committee on Human Research, potential participants were notified and provided a printed postcard to indicate their refusal to participate in this project; 40 patients declined.
Before the provider randomization and interventions, all providers received substantial education on the use of β-blockers in heart failure patients, including Grand Rounds presentations, noon conferences, and individual clinic conferences. In addition, guidelines for β-blocker initiation and uptitration were distributed by letter and e-mail and posted on the hospital intranet.
Providers were randomly assigned to 1 of the 3 groups. The control group received no additional intervention beyond the educational program. In the nurse facilitator group, a study nurse practitioner (NP) supervised by 2 cardiologists (M.A. and B.M.) was assigned to providers. After obtaining approval from the primary provider, the NP assumed responsibility for initiating, titrating, and stabilizing appropriate CHF patients on β-blockers. After reaching the target or the maximum tolerated β-blocker dose, the patient was returned to his or her primary care provider for all additional care.
Provider and Patient Notification
Providers in this arm of the study were given a list of their CHF patients who were potential candidates for β-blocker therapy. Computer alerts appeared when the provider accessed these patients’ electronic record for the first 2 visits after randomization. Patients in this arm of the study were also sent letters informing them that they might be candidates for β-blocker therapy and that they should discuss this treatment with their primary provider.
The SFVAMC β-blocker guideline suggested the use of carvedilol because of the availability of tablet strengths appropriate for initiation in CHF patients. Metoprolol tartrate was frequently used at the discretion of the provider.
The primary prespecified outcomes, which were assessed 1 year after the patients were entered or at closeout of the study (whichever came earlier), were the proportion of patients who were initiated or uptitrated and maintained on β-blockers and the proportion of patients reaching target doses of β-blockers using intention to treat analysis. Target daily doses were defined as carvedilol 50 mg, metoprolol tartrate 100 mg, or atenolol 100 mg (atenolol was used only in patients who were taking subtarget doses of this agent before the initiation of the program). Information was collected on hospitalizations, emergency room visits, and deaths during the trial period as an indicator of safety but not efficacy.
An independent research assistant assessed the use of β-blocker therapy by reviewing pharmacy records and computerized progress notes. In addition, the mean time to initiation and uptitration of β-blockers was determined for each provider group. Hospitalizations, emergency room visits, and deaths were identified by chart abstraction and by patient contact at 3-month intervals. In addition, deaths were ascertained using the Department of Veterans Affairs death registry (the Beneficiary Identification and Records Locator Subsystem Death File).
Because we assumed that many nonideal candidates for β-blocker therapy would be included, we estimated that the best intervention would result in a 70% rate of β-blocker initiation or uptitration to target dose. The intended sample size of 80 patients per group was selected to provide approximately 85% power to detect an absolute 25% or greater difference between the best and next best intervention with a 0.05 type I error. Because we were able to enroll a total of only 169 subjects after exclusions during the time window of this study, the actual power was less than intended.
χ2 tests were used to compare categorical variables; ANOVA was used to compare continuous variables across groups. Between-group differences in the primary outcome measures were compared using a logistic regression model that accounted for clustering by the provider and included covariates that differed among the groups (at P<0.20). The proportions of patients who experienced hospitalizations, emergency visits, or death after β-blocker initiation or uptitration were analyzed using the 2-tailed Fisher’s exact test because of the low event rates. Differences in the time to reach target β-blocker doses among the treatment groups were compared by life table analysis (Kaplan-Meier curves) and the log-rank test. All analyses were done using Stata software (version 6.0, Stata).
Among 91 potential providers, 2 refused participation and 15 had no patients with CHF. The remaining 74 providers (74% internists, 12% cardiologists, and 14% nurse practitioners) were randomly assigned to 1 of the 3 arms of the trial: (1) control (25 providers, 51 patients); (2) nurse facilitator (25 providers, 54 patients); and (3) provider and patient notification (24 providers, 64 patients). The distribution of provider specialty and training level was similar among the groups (Table 1).
Approximately one third of patients were taking low-dose β-blockers at the start of the study. All patients had an EF ≤45%, and 36% of patients had an EF ≤30% or an EF qualitatively described as severely abnormal by echocardiography. Baseline characteristics of patients were similar in the 3 intervention arms, with the exception that more patients in the control group were receiving chronic hemodialysis because of the inclusion of a nephrologist in this group (Table 2).
In the nurse facilitator group, two thirds of subjects were either initiated or uptitrated on β-blockers, including more than 60% of subjects who were β-blocker naive at entry; in contrast, fewer than one third of patients were initiated or uptitrated in the other 2 study arms (Table 3). More subjects in the nurse facilitator group reached target doses as well (Table 3).
Among the patients who reached target doses of β-blockers, the mean length of time from initiation to target dose was also lowest in the nurse facilitator arm (5.9 months) compared with the provider and patient notification group (9.3 months) and the control group (8.5 months) (P<0.001) (Figure 2). This relatively long interval for uptitration reflected the preference for scheduling titration visits to coincide with other medical appointments.
The proportions of patients requiring an emergency room visit or hospitalization during follow-up were high in all 3 groups, but there were no differences in rates of adverse events based on randomization assignment (Table 4). There were few deaths, with the highest number occurring in the control group, which also had the highest number of patients undergoing chronic hemodialysis, of whom 2 died.
This clinical trial compared 3 approaches to implementing β-blockers in patients with systolic heart failure and demonstrated that a nurse facilitator could safely and conveniently improve utilization of β-blockers compared with a provider and patient notification strategy or provider education. The intensive program of provider education alone resulted in relatively low rates of new β-blocker use (29%), and most of the newly initiated patients did not achieve recommended target doses. Furthermore, β-blocker utilization was not enhanced by a multifaceted approach of giving practitioners a list of potential candidates for β-blocker therapy, providing patient-specific computer reminders timed with patient visits, and encouraging the identified patients to discuss this therapy with their primary provider. These findings are consistent with a growing body of literature indicating that improving adherence to complex guidelines requires active interventions rather than passive dissemination of information.
Methods to improve guideline implementation have focused largely on changing individual provider behavior rather than on the organizational support and environment of providers. However, a consensus panel by experienced implementation experts concluded that implementation efforts focused primarily on the clinician were unlikely to be successful.19 These experts rated presence of organized systems in the clinic as the most important factor for effective change. Use of dedicated advance practice nurses to identify patients who have not met guideline goals and to work with the provider to implement practice guidelines is an example of a system-oriented approach.
The most common barriers to guideline implementation include lack of awareness, inertia of previous practice, lack of agreement, and external barriers, including lack of time or staff support.15 The objective of initiating β-blockers faced several of these obstacles. Although strongly evidence-based, the use of β-blockers in heart failure is novel and counterintuitive for many providers. The implementation of β-blocker therapy is also viewed as difficult and time consuming, because it requires frequent visits for uptitration and careful monitoring of symptoms, physical findings, and side effects.
We believe the nurse facilitator intervention was successful because it overcame these specific barriers and provided a helpful service for the provider, making it possible to initiate β-blockers without additional monitoring and time expenditure on the part of the provider.
Although the nurse facilitator group had superior results to the other intervention groups, only 67% of the assigned patients were initiated or uptitrated, and only 43% reached target β-blocker doses by the end of the study. Within the Nurse Facilitator group, one fourth of patients identified as potential candidates for β-blockers were subsequently documented to have contraindications to β-blockers not noted by chart review (n=12, primarily bradycardia, pulmonary disease with wheezing, or history of nonadherence to clinic visits) or refused β-blocker initiation (n=3). Accounting for these, it is noteworthy that virtually all of the ideal candidates could be initiated or uptitrated with the nurse facilitator intervention. Comparable numbers are not available for the other provider groups, because the providers did not routinely document why β-blockers were not initiated or uptitrated.
The use of nurses to assist in guideline implementation has been studied in several other conditions, including diabetes,20 depression,21 and hyperlipidemia.22 In diabetes, the use of a structured care approach with a diabetes nurse educator improved glycemic control, lipid levels, smoking cessation, and treatment satisfaction.20 In CHF patients, nurse managers can improve utilization of guideline recommendations for ACE inhibitor therapy, daily weight monitoring, and left ventricular function assessment in heart failure.17,23 We used a nurse practitioner in our study, because physical examinations play an important role in patient selection and monitoring during β-blocker initiation.
Although a few patients in the provider education and the provider notification groups were initiated on β-blockers during the follow-up period, it is uncertain whether this reflected the study interventions or the gradual evolution of clinical practice. The inability of computer reminders and patient notification to improve β-blocker utilization was disappointing. Among the large numbers of clinical trials evaluating the utility of computer reminders for improving prevention efforts in medicine, most have found positive results.24–26 However, two meta-analyses have demonstrated that computer reminders were of particular use when the prevention activity was simple to use (eg, influenza vaccine) but were less effective for more time-consuming activities, such as PAP smears. In this context, the lack of effect of computer reminders on the use of β-blocker therapy in heart failure patients is less surprising.27,28 In addition, our findings reinforce those of prior investigators that any intervention effort aimed at improving utilization of a particular guideline must be tailored to the unique barriers of the clinical activity and the practice setting. This study also demonstrates the importance of pilot testing clinical reminders before widely implementing them with the implicit assumption that they will be effective.
The other component of the provider and patient notification group in our study, letters to patients suggesting that they discuss the use of β-blockers with their provider, apparently was also unsuccessful. This approach was meant to empower patients to be proactive in their care, although it had no apparent effect in our study. One prior study in Britain looking at postal prompts in coronary artery disease patients had similarly disappointing results.29
Our study was conducted at a single academically affiliated Veterans Affairs medical center, and the results may in part reflect this setting and the types of providers and patients associated with it. Additional studies are needed to determine the generalizability of our findings, although the use of educational interventions, computer-generated reminders, and advanced practice nurses are increasingly commonplace in other practice environments. The most important barriers to implementing β-blocker therapy, the identification of appropriate candidates for the intervention and the ability and willingness to devote the requisite time and effort, are universal problems in medical practice.
A substantial number of patients were found to be ineligible for this study. Many of the reasons are obvious, for example, patients were no longer followed or deceased, patients were already receiving target doses of β-blockers, or patients had EF >45%. Others were excluded for methodological reasons, including management by study investigators or in the heart failure program or lack of accessibility because of geography. However, 104 patients were excluded because they met prespecified chart review criteria for not being eligible for β-blockers. It is possible that some of these patients may have been candidates and have been successfully treated.
Despite the substantial effort and resources spent to demonstrate treatment efficacy in clinical trials and to develop evidence-based clinical practice guidelines, their implementation into medical practice remains limited. Without effective implementation, these guidelines will have little effect on patient outcomes. We found that an intervention based on a nurse facilitator significantly enhanced the rate of initiation and uptitration of β-blockers. The creation of systems to help support desired clinician behavior, such as the use of a nurse facilitator, seems to be an effective strategy to improve guideline adherence.
This work was supported by a Quality Enhancement Research Initiative (QUERI) grant from the Office of VA Health Services Research and Development. Drs Ansari, Shlipak, and Heidenreich are supported by Career Development Awards from the VA Health Services Research and Development Service. The authors thank Rhonda Benotti, BA, Anna Beyer, MD, and Mamatha Nagabushunum, MD, for their assistance with data collection and chart review and Jean Bullard, RN, MS, for her assistance with identification of the study cohort.
Dr Massie has served as a consultant to AstraZeneca and GlaxoSmithKline, which manufacture 2 of the β-blockers used in this study.
Konstam M, Dracup K, Baker D, et al. Heart Failure: Evaluation and Care of Patients With Left Ventricular Systolic Dysfunction. Rockville, Md: Agency for Health Care Policy and Research Publication; 1994.
Hunt S, Baker DW, Chin MH, et al. ACC/AHA guidelines for the evaluation and management of chronic heart failure in the adult: executive summary: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol. 2001; 38: 2101–2113.
Task Force for the diagnosis and treatment of chronic heart failure, European Society of Cardiology. Guidelines for the diagnosis and treatment of chronic heart failure. Eur Heart J. 2001; 22: 1527–1560.
Cannon DS, Allen SN. A comparison of the effects of computer and manual reminders on compliance with a mental health clinical practice guideline. J Am Med Inform Assoc. 2000; 7: 196–203.
Feder G, Griffiths C, Eldridge S, et al. Effect of postal prompts to patients and general practitioners on the quality of primary care after a coronary event (POST): randomised controlled trial. BMJ. 1999; 318: 1522–1526.