Author + information
- Received June 1, 2010
- Revision received July 15, 2010
- Accepted July 27, 2010
- Published online December 14, 2010.
- ↵⁎Reprint requests and correspondence:
Dr. Jalal K. Ghali, DMC Cardiovascular Institute, 3990 John R. Street, Suite 9370, Wayne State University, Detroit, Michigan 48201
Guidelines have rather quickly assumed a central role in health care delivery in the U.S. They have become the foundation on which performance measures are built and, therefore, a major player in assessing the quality of care provided by individuals and institutions, the ramifications of which involve reputation, reimbursement, and litigation. We are concerned, however, that in our enthusiasm for collectively endorsing these guidelines, we are marginalizing the importance of physician judgment and inadvertently risking the conversion of guidelines into “cookbooks.” We believe that this viewpoint, while unequivocally acknowledging the fundamental importance of guidelines, simultaneously provides a critically important perspective on the potential for misuse of both guidelines and performance measures. Further, we hope that publication of this viewpoint will help temper enthusiasm for overzealous conversion of guidelines into performance measures, thereby restoring the vital role of physician judgment and insight into patient management.
In the past 3 decades, countless randomized trials, meta-analyses, and large observational studies have led to major advances in the care of patients with a diverse array of medical disorders, including heart failure (HF). These studies have spawned the era of evidence-based medicine, and the development and dissemination of authoritative guidelines designed to facilitate patient management under a wide range of clinical situations. Optimally, guideline recommendations are derived from the results of major clinical trials, but it is important to recognize that for a large proportion of important clinical questions for which definitive trials are not available, recommendations are based primarily on expert opinion, which frequently requires modification as new information becomes available. Nonetheless, guidelines are an authoritative resource for practicing physicians, and they serve a vital role in minimizing variations in care by encouraging the implementation of evidence-based interventions while discouraging potentially harmful ones. Although they are meant to supplement rather than replace clinical judgment, they have been widely perceived and accepted, perhaps somewhat simplistically and rigidly, as defining appropriate and inappropriate medical care.
Implementation of guidelines into routine clinical practice has often been suboptimal at best, and as a result of which, many patients, including those with HF, frequently receive substandard care (1). This observation has led to the development of performance measures for evaluating and reporting quality of care at the level of the hospital and individual practitioner (2,3), and to the design and testing of methodologies for improving quality of care through closer adherence to established guidelines (4–7). Although performance measures incorporate clinical reasoning in determining the appropriateness of an intervention, the value of performance measures as a means for improving clinical outcomes remains unproven (8,9). More recently, in addition to public reporting (10,11), pay-for-performance initiatives have been implemented (with more on the way) that reward institutions and physicians for providing “better” care, as judged by adherence to specific performance metrics (11–13), and many institutions and programs have incorporated their performance results into promotional materials. Indeed, an “industry” has evolved to certify and promote programs based on these measures. Thus, the development of guidelines has provided a benchmark upon which performance measures have been built, and has clearly signaled a new era in health care delivery.
Few would argue with the conceptual underpinnings of evidence-based medicine, practice guidelines, performance measures, or even a restructuring of the reimbursement system to better align payment with the quality of care provided (rather than with volume, the principal driver of reimbursement under the current system). However, it must be recognized that there are important limitations to the existing evidence base, practice guidelines derived from that evidence base, performance measures synthesized from less-than-perfect guidelines, and reimbursement strategies that rely too heavily on adherence to performance measures. Indeed, overzealous application of guidelines, fueled in part by vigorous promotion by professional organizations and in some cases by industry (which often has its own financial incentives for promoting guidelines that encourage use of their products), as well as by provider incentives to achieve higher-quality ratings and, perhaps, higher reimbursement, may potentially have effects that are directly opposite to those intended. Thus, by diminishing the importance of physician judgment in managing the individual patient, thereby fostering “cookbook medicine,” specific interventions may be applied in patients who are unlikely to benefit or who are potentially at risk for harm (14).
Considering the central role of guidelines, it is critical to examine the evidence base for the HF guidelines and define their limitations. A recent analysis of the distribution of levels of evidence for recommendations in American College of Cardiology/American Heart Association (ACC/AHA) clinical practice guidelines (15) found that for the HF guidelines, Level of Evidence: A (multiple clinical trials and/or meta-analyses) accounted for 26.4% of recommendations, Level of Evidence: B (a single randomized trial or nonrandomized studies) for 19.4%, and Level of Evidence: C (expert opinion and observational studies) for 54.3%. Similarly, the 2009 focused update of the ACC/AHA guidelines for the diagnosis and management of patients with HF (16) contains a total of 157 discrete recommendations, of which only 35 (22%) are supported by Level of Evidence: A. Another 31 recommendations (20%) are based on Level of Evidence: B, whereas the majority, 91 (58%), are based primarily on expert consensus opinion.
Recommendations with the highest level of evidence are based on randomized clinical trials (RCTs) and are therefore subject to all the limitations of RCTs, which are generally performed in environments and patient populations that differ greatly from those of usual clinical practice. This inherent selection bias of RCTs results in under-representation of the elderly, women, and minorities; conversely, patients who are likely to be adherent to the prescribed treatment and follow-up are over-represented (17). Comparing patients in RCTs to those in the community reveals remarkable differences, including younger age (mean age 64 years vs. 73 years), under-representation of women (23% vs. 50%), and markedly limited ethnic/racial diversity (18). Furthermore, trial patients tend to be healthier, with fewer concomitant comorbidities, such as renal insufficiency, chronic lung disease, obesity, depression, and neurocognitive disorders. These factors may significantly impact prognosis and the risk–benefit ratio of specific therapies, in many cases more profoundly than the primary illness. Thus, it is never safe to assume that treatments of proven efficacy in younger, healthier, predominantly male, and white patients will provide equivalent benefit in older patients, women, or ethnically diverse populations. For example, in a meta-analysis of the angiotensin-converting enzyme inhibitor (ACEI) trials, Flather et al. (19) reported that among 1,066 patients over 75 years of age, ACEIs had no discernible effect on mortality. More recently, in a meta-analysis of 5 trials, Ghanbari et al. (20) reported that implantable cardioverter-defibrillators (ICDs) provided no mortality benefit in women. Despite these studies, current guidelines rarely distinguish use of these therapies on the basis of age, sex, or concomitant conditions.
Concerns about the importance of race/ethnicity and the potential for harm are exemplified by the use of aldosterone antagonists. Current ACC/AHA guidelines recommend the use of an aldosterone antagonist in “selected patients with moderately severe to severe symptoms of HF and reduced left ventricular ejection fraction who can be carefully monitored for preserved renal function and normal potassium concentration” (Level of Evidence: B) (16). This recommendation is based on 2 randomized trials, both of which enrolled carefully selected patients, with few subjects over 75 years of age, a minority of women, and very few African-Americans (21,22). Yet, the guidelines do not exclude these subgroups from the recommendation to initiate aldosterone antagonist therapy or emphasize the need for caution in subgroups for which experience is limited. This is noteworthy, since a recently published analysis from the A-HeFT (African-American Heart Failure Trial) found that aldosterone antagonists were not associated with a survival benefit in African-American patients who were not receiving hydralazine and nitrates (23).
A further concern is that patients treated with an aldosterone antagonist in the context of a well-designed trial received rigorous follow-up to monitor for adverse effects, and it cannot be expected that such meticulous follow-up would be matched in clinical practice. This is a particularly important issue, since post-discharge follow-up of HF patients is often inadequate (24). Indeed, the lack of adequate caution in patient selection and assurance of rigorous follow-up is the likely explanation for the striking rise in hyperkalemia and excess early mortality following publication of the RALES (Randomized Aldactone Evaluation Study) in 1999, as reported in a large observational study conducted in Canada in a “real-world” HF population (mean age 78 years, 50% women) (25). Moreover, spironolactone was not associated with reductions in either all-cause mortality or HF admissions in this analysis.
These concerns notwithstanding, a recent analysis from the AHA-sponsored Get With The Guidelines program reported marked underutilization of aldosterone antagonists in patients hospitalized with acute decompensated HF, and the authors concluded that “stronger uptake of aldosterone antagonist therapy…may be warranted” (26). Notably, the value of spironolactone in patients hospitalized with decompensated HF, who often have fluctuating renal function and electrolyte levels, has not been tested, and the potential for harm induced by widespread initiation of spironolactone in these patients may be substantial. Furthermore, patients hospitalized with HF represent a heterogeneous group, including some patients who experience substantial clinical improvement following hospital discharge, as a result of which, they may no longer fulfill guideline criteria for aldosterone antagonist therapy and other interventions.
Another recommendation that raises concern in the translation of guidelines into performance measures is the use of ICDs. At present, there is no evidence from clinical trials that ICDs reduce mortality in patients 80 years of age or older (27,28), nor as noted previously, is there convincing evidence of benefit in women (20). Yet, current guideline recommendations do not distinguish indications for ICD insertion based on age or sex (16). Thus, a 98-year-old woman with Class III HF symptoms and a left ventricular ejection fraction of 28% should be considered for an ICD (Level of Evidence: A!) in the absence of contraindications. Moreover, the physician who fails to offer an ICD to such a patient risks being labeled as practicing “bad medicine,” with potentially serious ramifications for his or her practice and reimbursement.
A further concern is the subjectivity inherent in guideline recommendations based on consensus opinion of experts. A case in point is the addition of an angiotensin-receptor blocker in patients with reduced ejection fraction who remain symptomatic despite treatment with ACEIs and beta-blockers. Four separate guideline committees, upon reviewing and analyzing the same data, have provided different recommendations and assigned different levels of evidence, ranging from Class I, Level of Evidence: A to Class II, Level of Evidence: B (Table 1) (16,29–31).
In summary, evidence-based medicine, practice guidelines, quality assurance programs, and performance measures unequivocally represent major advances towards ensuring that the largest number of patients receives the highest quality of care. However, they also create the potential for a “slippery slope” by inadvertently marginalizing the importance of physician judgment in providing truly patient-centered care. The current enthusiasm for translating practice guidelines into routine clinical care needs to be tempered by a better understanding of the limitations of the guidelines and by an appreciation for the variable quality of care and patient adherence that impact early outcomes as patients transition from hospital to outpatient care.
Moreover, we believe that the push to convert guidelines into an ever-growing number of performance measures with implications for reimbursement should be undertaken with a greater degree of circumspection. As we move from practice guidelines to performance measures, it will be important to “mind the gap” between what we truly know, what we think we know, and what we would like to know, and between the rigor of patient selection and follow-up in RCTs and the much more variable management that occurs in clinical practice, lest we engage in more harm than good. In this context, it must be recognized that guidelines are merely guidelines, and that optimal patient care still requires a healthy dose of physician judgment.
Dr. Ghali has received research grants from Astellas Pharma US, CardioKine, Biogen Idec, and Amgen; is on the advisory panels of CardioKine, Biogen Idec, and Otsuka; on the Speakers' Bureau of Otsuka; is a consultant for Amgen; receives compensation from CardioMEMS; and is on the Data and Safety Monitoring Board of CardioMEMS. Dr. Massie has received a research grant from Novartis; consultant fees from Novartis, Duke Clinical Research Institute, Cytokinetics, and Nile Therapeutics; compensation fees from Takeda, Corthera, and Biotronik; and speaking honorarium from Boston Scientific. Dr. Mann is a consultant for Medtronic. Dr. Rich has received a research grant from Astellas Pharma US. This paper is being simultaneously copublished in the Journal of Cardiac Failure.
- Abbreviations and Acronyms
- American College of Cardiology/American Heart Association
- angiotensin-converting enzyme inhibitor
- heart failure
- implantable cardioverter-defibrillator
- randomized clinical trial
- Received June 1, 2010.
- Revision received July 15, 2010.
- Accepted July 27, 2010.
- American College of Cardiology Foundation
- Stafford R.S.,
- Radley D.C.
- ↵Specifications Manual for National Hospital Quality Measures Version 2.4: Measure Information Forms—Heart Failure (HF) (11/30/07). Joint Commission on Accreditation of Healthcare Organizations (JCAHO) http://www.jcaho.org/pms/core+measures/aligned_manual.htm. Accessed May 7, 2010.
- Bonow R.O.,
- Bennett S.,
- Casey D.E. Jr..,
- et al.
- Ansari M.,
- Shlipak M.G.,
- Heidenreich P.A.,
- et al.
- Mazimba S.,
- Grant N.,
- Makola D.,
- et al.
- Woolf S.H.,
- Grol R.,
- Hutchinson A.,
- Eccles M.,
- Grimshaw J.
- Jessup M.,
- Abraham W.T.,
- Casey D.E.,
- et al.
- Havranek E.P.,
- Peterson P.N.
- Arnold J.M.,
- Howlett J.G.,
- Ducharme A.,
- et al.
- Dickstein K.,
- Cohen-Solal A.,
- Filippatos G.,
- et al.