Out With the Guidelines?

This column has in the past commented on the shortcomings of guidelines in a number of clinical situations, including the expanded use of coronary interventional procedures and the implantation of automatic defibrillators in heart failure.

Guidelines designed as advisory to clinicians have achieved almost biblical, or, if you will, Koranic, status in the care of patients.

As a result, many physicians young and old have developed a state of mental paralysis, unable to make clinical decisions based on pathophysiology, pharmacology, or social status of the patient. They go through a maze of decision trees that often lead to a clinical judgment without fully considering the patient's status.

The process is reminiscent of the Yogi Berra aphorism, “When you come to a fork in the road, take it.”

The recent analysis by the Duke Clinical Research Institute of the ACC/AHA Clinical Practice Guidelines process over the last 25 years provides an incisive examination of the weakness of this otherwise well meaning process (JAMA 2009;301:831-41).

Guidelines were created largely to incorporate the results of randomized clinical trials (RCTs) into clinical care. Such trials, designed to prove the efficacy of new drugs and devices, represent major investments by pharmaceutical companies, which are focused on testing their products in the most advantageous environment. As a result, many patients with comorbid illnesses are excluded to provide a clear message about the drugs' specificity in a given illness.

Those of us who straddle the clinical research world and clinical practice know very well patients included in trials bear little resemblance to those we treat at the bedside or in the clinic. Most Medicare beneficiaries have multiple comorbidities that affect both therapy and outcome. Nevertheless, the guidelines have used RCT results when available.

Unfortunately, RCTs cover only a small fraction of clinical practice. In lieu of RCT data, “expert opinion and standard of care” were used to fill the gaps in evidence, which comprise most of the clinical decision process.

This reliance on expert opinion in guideline committees also has been challenged as a potential source of bias. Industry has had such profound impact on clinical research and trials that it is almost impossible to find physicians with expertise who do not have some potential bias as a result of research support, advisory committees, speakers' bureaus, and consulting fees. Many of these experts included in guideline panels have honest scientific investment in therapeutic concepts, which also have the potential to cloud their judgment.

Unfortunately, there is a paucity of guidelines that are even supported by RCTs. According to the Duke analysis, with a class I recommendation, defined as when “there is evidence and/or general agreement that a given procedure is useful and effective” only 19% had a class A level of evidence, defined as being supported with RCT data. In 36% of the class I recommendations, “expert opinion, case studies or standard of care” —class C evidence level—was the support. The analysis further suggested that as guidelines expanded over time, the use of class II recommendations, where there is “conflicting evidence and/or a diversion of opinion” has increased.

To put this in the context of current clinical practice, the report cited data from the ACC National Registry showing that 30% of all the percutaneous coronary interventions and 39% of those performed after an acute MI fell under class II indications. This represented 115,000 and 45,000 procedures, respectively.

The development of standards for medical therapy is important for the care of our patients, and RCTs are crucial to understanding the safety and benefit of new therapies. Much of the success in reducing the mortality in heart disease has been the result of the incorporation of RCT results into bedside care.

Nevertheless, our current process needs to be improved. We need to develop a mechanism for testing therapy in the real world dominated by an increasingly aging patient population with myriad comorbid diseases.

The current attempts by the Centers for Medicare and Medicaid Services to assess the efficacy and safety of new therapies have been inadequate. The development of registries such as the ACC's ICD Registry, has not met the full potential of this process.

We need randomized trial data of real patients as a new therapy is introduced in clinical care. In the absence of that, the collection of registry data must be strengthened and the analyses made available to the medical community as soon as possible.

This column has in the past commented on the shortcomings of guidelines in a number of clinical situations, including the expanded use of coronary interventional procedures and the implantation of automatic defibrillators in heart failure.

Guidelines designed as advisory to clinicians have achieved almost biblical, or, if you will, Koranic, status in the care of patients.

As a result, many physicians young and old have developed a state of mental paralysis, unable to make clinical decisions based on pathophysiology, pharmacology, or social status of the patient. They go through a maze of decision trees that often lead to a clinical judgment without fully considering the patient's status.

The process is reminiscent of the Yogi Berra aphorism, “When you come to a fork in the road, take it.”

The recent analysis by the Duke Clinical Research Institute of the ACC/AHA Clinical Practice Guidelines process over the last 25 years provides an incisive examination of the weakness of this otherwise well meaning process (JAMA 2009;301:831-41).

Guidelines were created largely to incorporate the results of randomized clinical trials (RCTs) into clinical care. Such trials, designed to prove the efficacy of new drugs and devices, represent major investments by pharmaceutical companies, which are focused on testing their products in the most advantageous environment. As a result, many patients with comorbid illnesses are excluded to provide a clear message about the drugs' specificity in a given illness.

Those of us who straddle the clinical research world and clinical practice know very well patients included in trials bear little resemblance to those we treat at the bedside or in the clinic. Most Medicare beneficiaries have multiple comorbidities that affect both therapy and outcome. Nevertheless, the guidelines have used RCT results when available.

Unfortunately, RCTs cover only a small fraction of clinical practice. In lieu of RCT data, “expert opinion and standard of care” were used to fill the gaps in evidence, which comprise most of the clinical decision process.

This reliance on expert opinion in guideline committees also has been challenged as a potential source of bias. Industry has had such profound impact on clinical research and trials that it is almost impossible to find physicians with expertise who do not have some potential bias as a result of research support, advisory committees, speakers' bureaus, and consulting fees. Many of these experts included in guideline panels have honest scientific investment in therapeutic concepts, which also have the potential to cloud their judgment.

Unfortunately, there is a paucity of guidelines that are even supported by RCTs. According to the Duke analysis, with a class I recommendation, defined as when “there is evidence and/or general agreement that a given procedure is useful and effective” only 19% had a class A level of evidence, defined as being supported with RCT data. In 36% of the class I recommendations, “expert opinion, case studies or standard of care” —class C evidence level—was the support. The analysis further suggested that as guidelines expanded over time, the use of class II recommendations, where there is “conflicting evidence and/or a diversion of opinion” has increased.

To put this in the context of current clinical practice, the report cited data from the ACC National Registry showing that 30% of all the percutaneous coronary interventions and 39% of those performed after an acute MI fell under class II indications. This represented 115,000 and 45,000 procedures, respectively.

The development of standards for medical therapy is important for the care of our patients, and RCTs are crucial to understanding the safety and benefit of new therapies. Much of the success in reducing the mortality in heart disease has been the result of the incorporation of RCT results into bedside care.

Nevertheless, our current process needs to be improved. We need to develop a mechanism for testing therapy in the real world dominated by an increasingly aging patient population with myriad comorbid diseases.

The current attempts by the Centers for Medicare and Medicaid Services to assess the efficacy and safety of new therapies have been inadequate. The development of registries such as the ACC's ICD Registry, has not met the full potential of this process.

We need randomized trial data of real patients as a new therapy is introduced in clinical care. In the absence of that, the collection of registry data must be strengthened and the analyses made available to the medical community as soon as possible.

This column has in the past commented on the shortcomings of guidelines in a number of clinical situations, including the expanded use of coronary interventional procedures and the implantation of automatic defibrillators in heart failure.

Guidelines designed as advisory to clinicians have achieved almost biblical, or, if you will, Koranic, status in the care of patients.

As a result, many physicians young and old have developed a state of mental paralysis, unable to make clinical decisions based on pathophysiology, pharmacology, or social status of the patient. They go through a maze of decision trees that often lead to a clinical judgment without fully considering the patient's status.

The process is reminiscent of the Yogi Berra aphorism, “When you come to a fork in the road, take it.”

The recent analysis by the Duke Clinical Research Institute of the ACC/AHA Clinical Practice Guidelines process over the last 25 years provides an incisive examination of the weakness of this otherwise well meaning process (JAMA 2009;301:831-41).

Guidelines were created largely to incorporate the results of randomized clinical trials (RCTs) into clinical care. Such trials, designed to prove the efficacy of new drugs and devices, represent major investments by pharmaceutical companies, which are focused on testing their products in the most advantageous environment. As a result, many patients with comorbid illnesses are excluded to provide a clear message about the drugs' specificity in a given illness.

Those of us who straddle the clinical research world and clinical practice know very well patients included in trials bear little resemblance to those we treat at the bedside or in the clinic. Most Medicare beneficiaries have multiple comorbidities that affect both therapy and outcome. Nevertheless, the guidelines have used RCT results when available.

Unfortunately, RCTs cover only a small fraction of clinical practice. In lieu of RCT data, “expert opinion and standard of care” were used to fill the gaps in evidence, which comprise most of the clinical decision process.

This reliance on expert opinion in guideline committees also has been challenged as a potential source of bias. Industry has had such profound impact on clinical research and trials that it is almost impossible to find physicians with expertise who do not have some potential bias as a result of research support, advisory committees, speakers' bureaus, and consulting fees. Many of these experts included in guideline panels have honest scientific investment in therapeutic concepts, which also have the potential to cloud their judgment.

Unfortunately, there is a paucity of guidelines that are even supported by RCTs. According to the Duke analysis, with a class I recommendation, defined as when “there is evidence and/or general agreement that a given procedure is useful and effective” only 19% had a class A level of evidence, defined as being supported with RCT data. In 36% of the class I recommendations, “expert opinion, case studies or standard of care” —class C evidence level—was the support. The analysis further suggested that as guidelines expanded over time, the use of class II recommendations, where there is “conflicting evidence and/or a diversion of opinion” has increased.

To put this in the context of current clinical practice, the report cited data from the ACC National Registry showing that 30% of all the percutaneous coronary interventions and 39% of those performed after an acute MI fell under class II indications. This represented 115,000 and 45,000 procedures, respectively.

The development of standards for medical therapy is important for the care of our patients, and RCTs are crucial to understanding the safety and benefit of new therapies. Much of the success in reducing the mortality in heart disease has been the result of the incorporation of RCT results into bedside care.

Nevertheless, our current process needs to be improved. We need to develop a mechanism for testing therapy in the real world dominated by an increasingly aging patient population with myriad comorbid diseases.

The current attempts by the Centers for Medicare and Medicaid Services to assess the efficacy and safety of new therapies have been inadequate. The development of registries such as the ACC's ICD Registry, has not met the full potential of this process.

We need randomized trial data of real patients as a new therapy is introduced in clinical care. In the absence of that, the collection of registry data must be strengthened and the analyses made available to the medical community as soon as possible.