How Do We Develop Trustworthy Clinical Practice Guidelines? Straight Talk with Sanjay Kaul, MD

Over the past three decades, the ACC, together with the American Heart Association (AHA), has taken a leadership role in developing clinical practice guidelines (CPGs) to aid the health care provider in adopting best-practice approach to patient care. Since 1984, more than 70 CPGs encompassing over 8,000 recommendations have been developed for 23 CV topics. In 2009, a report examining the scientific evidence underlying these guidelines concluded that recommendations were largely developed from lower levels of evidence or expert opinion.1 In a scathing indictment of the guideline process, the accompanying editorial exhorted the clinicians and policy makers "to reject calls for adherence to guidelines unless there is evidence of appropriate changes in the guideline process."2

On March 23, 2011, the Institute of Medicine (IOM), under the commission of the Medicare Improvements for Patients and Providers Act of 2008, released standards for developing rigorous and trustworthy CPGs to optimize patient care.3 The poor compliance with IOM standards is consistent with a recent report demonstrating barely one-third of the guidelines issued by subspecialty societies satisfy more than 50% of the IOM standards.4 Although substantial progress has been made in the past few years, it is clear much work remains to be done to improve the quality and the reliability of the guideline process.

To this end, several key issues merit consideration. First, factors that modify the quality of evidence should be explicitly, rigorously, and consistently considered in a standardized manner to refine the evaluation process. Important limitations such as missing data, loss to follow-up and the attendant violation of intention-to-treat principle, early stopping for benefit, inconsistency and indirectness of results, and use of "soft" endpoints serve to weaken the evidence and should therefore result in downgrading of recommendations. However, a large treatment effect, a dose-response gradient, and the use of "hard" endpoints strengthen the evidence and should prompt upgrading of recommendations accordingly.

Second, the current guideline process relies heavily on conventional statistical significance which tends to overstate the strength of association. Complementary approaches are available, such as the Bayesian inference that utilizes the Bayes factor, a more desirable metric than P value to quantify the strength of evidence. Practical guidelines for interpreting Bayes factors should be formally incorporated in evaluating the strength of evidence.

Third, the clarity of the balance between clinically important, not just statistically significant, benefits and harms should influence the strength of guideline recommendations. Although the strength of recommendation criterion in the guideline grid is defined by the benefit-risk trade-off, a transparent and explicit benefit-risk assessment is not routinely considered. A formal evaluation of clinical importance, given the overall benefit-risk, and ideally benefit-risk-cost, profile of each therapeutic intervention should be brought to bear on the development of guidelines.

Fourth, a systematic review team comprised of methodological experts (statisticians, decision and information scientists, and health economists) should independently appraise the evidence de novo and interact with the clinical and scientific experts, as well as with patients, patient advocates, and consumer representatives, in developing clinically-relevant and patient-centered guidelines. Evidence-based guidelines optimize patient care when they inform clinical judgment and incorporate patient values and preferences. To minimize potential intellectual bias, investigators of guideline-relevant clinical trials (currently included, but not permitted to vote) should only serve as advisors as part of the external review process.

Fifth, an external agency such as the Agency for Healthcare Research and Quality (AHRQ) should be charged with the authority to certify guidelines and only those that meet the IOM standards be allowed publication in journals.

Finally, guidelines should be efficient, user-friendly and parsimonious (avoid voluminous report), timely (keep pace with advances), disseminated at the point of clinical contact (embedded in electronic medical records), and used as financial incentives (evidence-based reimbursement).

In conclusion, guidelines that are driven by scientifically documented and rigorously appraised high-quality evidence are more likely to be accepted by the stakeholders, thereby reducing the variability in care and improving the quality and reducing the cost of care. It is our collective responsibility to ensure that the ACCF/AHA guidelines subscribe to the standards set forth by the IOM. Until then, the trustworthiness of guidelines will be limited.

1. Tricoci P, et al. JAMA. 2009;301:831-41.
2. Shaneyfelt TM, et al. JAMA. 2009;301:868-9.
3. Institute of Medicine. "Clinical Practice Guidelines We Can Trust."
4. Kung J, et al. Arch Intern Med. 2012 October 22. [Epub ahead of print]

Sanjay Kaul, MD, is director of the Vascular Physiology and Thrombosis Research Laboratory at the Burns and Allen Research Institute at Cedars-Sinai Medical Center in Los Angeles. Dr. Kaul also directs the Cardiology Fellowship Training Program and the Cardiology Consult Service.

Keywords: Patient Care, Health Personnel, Follow-Up Studies, Health Services Research, Institute of Medicine (U.S.), Electronic Health Records, Reproducibility of Results, Intention, Expert Testimony, United States Agency for Healthcare Research and Quality, Bayes Theorem, Publishing, Medicare, Risk Assessment

< Back to Listings