The development of clinical practice guidelines in oncology presents unique problems and challenges due to the heterogeneity of disease presentations and the complexity of therapeutic decisions. Guidelines should incorporate
The development of clinical practice guidelines in oncology presents unique problems and challenges due to the heterogeneity of disease presentations and the complexity of therapeutic decisions. Guidelines should incorporate not only standard outcomes measures such as survival but, if feasible, quality of life measures, patient preferences, and cost-effectiveness analyses. Although clinical guidelines are being promulgated as a means of assisting physicians in making therapeutic decisions, another purpose of guidelines implementation reflects the goals of managed care, ie, to promote practice conformity and reduce the cost of care. Methods to monitor physicians' adherence to guidelines are expected to increase as a means of ensuring compliance. This article reviews the process of developing two common types of guidelines used in oncology: path guidelines (using stage II breast cancer as a paradigm) and boundary guidelines (such as the ASCO growth factor guidelines).
A s the US health-care system increasingly relies on managed care mechanisms to direct patient management, new methods are needed to assist providers, payers, and policymakers in implementing this transformation. A method that currently figures prominently in the planning of many health organizations is the derivation of clinical practice guidelines .
Given the heterogeneity of disease presentations and the complexity of therapeutic decisions, the development of guidelines in oncology presents unique problems and challenges. As in all aspects of health-care planning, concepts and objectives must be precisely defined at the outset, so that the planning process has clear direction and a solid underpinning.
The classic definition of clinical practice guidelines is "systematically developed statements to assist practitioners and patients in decisions about appropriate health care for a specific clinical circumstance". This definition is useful in that it focuses guidelines developers on the concept of guidelines as an assistance to clinical decision making. As guidelines are developed, reference to this aim will assist oncologists in deciding whether the recommendations are useful. If a clinician can use a guideline in arriving at a medical decision, then it has been framed correctly. If, because of vagueness or lack of specific direction, it does not provide any substantive guidance, the guideline is not providing meaningful assistance in patient management.
In reality, the recent emphasis on guidelines stems from a more concrete goal than assistance in clinical decision making. The reason for the tremendous energies being directed toward guidelines development is the belief that they are a vehicle for assuring conformity of practice [3,4]. A more pragmatic definition of guidelines might be that they are a set of directives reflecting the most appropriate management of a clinical situation, whose goal is to promote conformity in the delivery of medical care. While not absolute, this definition makes guidelines more prescriptive and more realistically reflects the underlying reasons for the resurgent interest in this area.
The assumption that guidelines can actually promote and result in conformity must still be viewed as unproved . Although, theoretically, there is good reason to believe that they will achieve this goal, the appropriate evaluation and assessment studies have not been performed, and at this stage the entire process must be considered experimental. The reasons for striving for conformity in practice through the use of guidelines stem from the major aims of managed care and may be listed as follows:
1. The use of clinical practice guidelines will eliminate inappropriate interventions . This concept is based primarily on health services research studies that have demonstrated the high prevalence of inappropriate procedures using evaluations by experts.
2. The cost of care can be more accurately gauged. One of the prime reasons that many groups are currently developing guidelines is to develop standard packages of care for which costs can be analyzed by financial and administrative experts, so that they can be used in managed care contracts. The methodology for these costing analyses is still not well defined, and for a guideline to obtain the specificity and comprehensiveness that will permit the accurate estimate of standard costs will require extensive economic modeling.
3. The cost of care may be driven down . One of the presumptions of conformity to guidelines prescriptions is that it will prevent overutilization of medical services. While this may certainly be true, it is also true that guidelines may uncover areas of underutilization of medical care and lead to increased medical costs . If the more widespread use of guidelines leads to universal compliance with currently accepted guidelines for adjuvant therapy of early-stage breast cancer, the cost of medical care, at least in the short term, may increase. Although economic modeling would suggest that this is a "good buy"[ 9], the overall direction of health-care spending may be up, not down, as a result of conformity to a guideline.
4. Quality of care can be defined in terms of adherence to guidelines. In response to the mandates of managed care, the issuing of "report cards" and the credentialing of physicians and health-care providers are becoming increasingly important. The derivation of medical review criteria and performance measures will lead to comparisons of physicians' practices with standards for adherence to guidelines and assignment of levels of quality of care to specific providers . The determination of appropriate indicators of guidelines adherence and the implementation of efficient computerized systems to monitor these indicators will become areas of intense investigation in the next few years. On balance, it is probable that conformity, insofar as it eliminates inappropriate practices, will benefit patients, but the assumption must be made that it will not be a straight-jacket that precludes necessary modifications to accommodate individual patient needs.
The guideline process involves not only development but also validation, dissemination, and assessment of impact. The current approaches to guidelines development have been well described by Woolf . Informal consensus methods consist of a group of experts deriving a set of guidelines based on their expertise, with no explicit definition of the data used or the process for reaching consensus. The formal consensus process involves bringing together recognized experts for a formal presentation of available data. The experts are then left to their own devices on how to digest the data and use it in arriving at their recommendations. The NIH consensus process is a prominent example of the use of this methodology.
The third approach, evidence-based guidelines, is the method currently considered to be the soundest for deriving objective guidelines. The process utilizes a formal review of the literature, with evaluation of the strength of the evidence based on explicit criteria. This categorization and presentation of the data allow clinicians using the guidelines to assess the strength of the evidence underlying the recommendations and, by extension, the strength of the specific guideline.
The fourth method, as promulgated by David Eddy and coworkers, has been termed the explicit method. . This method attempts to assess the magnitude and probability of outcomes, and use the results of this analysis in choosing between alternative practices. Unfortunately, for the bulk of oncology clinical situations, the quantitative data that would permit guidelines developers to use this process are not available.
There is a plethora of terminology related to systematically derived tools to assist in managed care. It is useful to distinguish between two major classes: clinical practice guidelines, which detail the appropriate tests and treatments, and critical pathways, which lay out the plan of how to perform these tests and treatments. Critical pathways are basically centered on the timing of medical procedures  and are probably best derived at the local level. Clinical practice guidelines may be divided into two types: path guidelines and boundary guidelines, which are discussed below.
During the guidelines development process, it is imperative to define the outcome measures that will be used in deriving the recommendations prior to embarking on the specific guidelines effort . The most relevant outcomes in oncology are survival and improved quality of life. Although the methodology for assessing quality of life is still in a developmental stage, in many instances, data exist that allow quality of life outcome measures to be factored into guidelines development. The two other outcomes of importance in oncology--cost effectiveness and patient preference--are not currently used to any significant degree in guidelines development. The methodology for incorporating patient preference into guidelines in a systematic way must still be developed . Economic analyses, for the most part, are cost-minimization studies. Since data from cost-effectiveness or cost-utility studies are not available for most oncologic practices or procedures, it may prove difficult to incorporate economic considerations into oncology guidelines .
An example of the importance of determining the relevant outcomes prior to deriving guidelines was apparent in the deliberations of the American Society of Clinical Oncology (ASCO) growth factor guidelines panel . The panel's recommendations depended on whether the outcomes to be considered were secondary outcomes (such as decreased hospital days, decreased days on antibiotics, or a decrease in infections) or the primary outcome of increased survival. In many boundary guidelines processes, the first several hours of discussion must focus on the appropriate outcomes to use in framing the guidelines. If consensus about outcomes is not achieved, the panel is at risk of deriving a set of recommendations that are internally inconsistent, if not frankly contradictory.
The Ontario Cancer Group has recently published a review of the guidelines process that can serve as a useful primer for those embarking on guidelines development . The first step in the process involves the derivation of evidence-based recommendations from a thorough literature review performed by the guidelines staff. (The Ontario group recommends that dedicated staff positions be available for the guidelines process.) Physicians then develop the guidelines by adapting the evidence-based recommendations, incorporating their perceptions of the strength of the recommendations and prevailing views about appropriate practice. The interposition of local practice considerations allows guidelines to reflect local strengths and deficiencies.
In the next step of the process, the guidelines are reviewed by external reviewers (community physicians) to highlight any local biases that may invalidate the recommendations. Following this review, the guidelines are analyzed as a policy document. It is at this late stage that economic assessment of the guidelines takes place. A recommendation, even though appropriate, that required resources beyond the capacity of the provider would need to be modified. This step may require sophisticated cost-effective analyses. The final guideline is promulgated with plans for systematic review.
The interactive nature of the process, coupled with the technical support to accomplish the task, has led to a method by which the physicians feel ownership in the guidelines. Although the process is time-consuming (generally about 3 months for a single set of guidelines), the Ontario group reports that most participants feel the time was well spent.
The type of guidelines most frequently developed by oncology groups at this time is the path guideline (Figure 1). The path guideline is a management plan that enables providers to make sequential decisions about testing or therapy for a given clinical situation.
Guidelines usually are stage specific for each tumor site. A typical path guideline would include workup and staging procedures, primary treatment, adjuvant treatment, follow-up and surveillance schedules, salvage therapy for recurrent disease, and supportive therapy. The major issue confronting path guideline developers is the degree of complexity that should be incorporated into the guidelines .
An examination of a hypothetical path guideline for stage II breast cancer will serve as a useful template for highlighting the problems and issues that confront developers of these guidelines.
Initial Workup--There would undoubtedly be a general consensus that the initial workup for stage II breast cancer should include a physical examination, CBC, platelets, chest x-ray, and mammogram. The controversial elements requiring discussion would be whether to include a bone scan and abdominal computed tomography (CT). This controversy serves as an excellent introduction to the guidelines process, since evidence is available to allow the developers to arrive at evidence-based recommendations. Thus, analysis might center on data from a study showing that bone scans are positive in less than 5% of women with stage II breast cancer ; this would lead to a guidelines recommendation that bone scans should be excluded in the initial evaluation.
Initial Therapy--The classic path guideline would list two major alternatives for the initial treatment of stage II breast cancer (Figure 2A). The first alternative, breast-conserving surgery, consists of one of the variants of partial mastectomy, axillary dissection, and radiation therapy. The other alternative would be a modified radical mastectomy, perhaps with radiation therapy for four or more positive nodes. This path reflects the NIH Consensus Panel pronouncement that the two therapeutic alternatives are equivalent, and patient preference should guide the decision-making process .
The difficulty in utilizing this path guideline arises when clinical factors may alter the equivalency of the two options. The recommendation does not offer guidance in subsets of stage II breast cancer in which the alternatives are not weighted equally. Do the guidelines apply to a woman with a large tumor in a small breast? What are the recommendations for an 85-year-old woman with Alzheimer's disease? A patient with scleroderma? Each of these groups might benefit from an expansion of the guidelines.
Although it is recognized that guidelines cannot cover every clinical situation, there is probably a benefit in trying to include enough branching in the path to be meaningful to the practicing decision maker. One approach would be to expand the initial therapy guidelines to include the schema shown in Figure 2B. "Old" patients, defined by comorbid or chronologic criteria, might have much more conservative treatment options. For those who are able to undergo major surgery, two groups might be outlined: (1) a group in which there are limited alternatives, such as patients with large tumors and small breasts, or those with collagen diseases, and (2) a group of patients who are truly candidates for either procedure (lumpectomy or mastectomy).
Alternative Adjuvant Therapy--The classic path guideline for stage II breast cancer would probably define adjuvant therapy for tumors larger than 1 cm. For patients who are over age 50 and have an estrogen-receptor (ER) positive tumor, tamoxifen (Nolvadex) would be recommended. For those under age 50, or over age 50 and ER negative, chemotherapy with a broad range of regimens, such as FAC, AC, CMF, and CMFVP, would be detailed. It is at this stage that economic analysis may become important. The cost of providing six courses of FAC with the Adriamycin delivered by continuous infusion is several thousand dollars more than four courses of Adriamycin-cyclophosphamide utilizing bolus Adriamycin. At the policymaking stage, the guidelines might be modified to accommodate the economic implications of the options.
Surveillance--A typical schedule for surveillance would include physical examination every 3 months for 2 years, and then every 6 months for 3 years; a chest x-ray every 12 months for 5 years; and mammography every 12 months. An ASCO survey revealed that many physicians are complying with these recommendations for asymptomatic postmenopausal stage I and II breast cancer patients . However, according to this survey, a substantial number of oncologists are performing additional studies: 28% recommended bone scans, 38% CEAs, 21% CA15-3s, and 10% CT scans.
One of the major issues in guidelines development is validation, defined as the demonstration that a guideline does, in fact, lead to the desired outcome. Surveillance and follow-up of breast cancer patients represents one of the few areas in oncology in which a guideline may be said to have been validated. In a multicenter randomized controlled trial, the Givio group in Italy compared the effects of clinical follow-up regimens and intensive follow-up schedules on survival and health-related quality of life in breast cancer patients . The study population consisted of 1,320 women with operable breast cancer who were less than 70 years of age and treated on the same treatment protocol. The clinical regimen consisted only of physicals and mammograms, whereas the intensive regimen added liver function tests every 3 months, chest x-ray every 6 months, bone scan every 12 months, and liver ultrasound every 12 months.
Five-year survival was 80% for the intensive group and 78% for the clinical group. In addition, there were no differences between the two groups in quality of life, based on measurements of patients' perception of quality of life, body image, emotional well-being, and satisfaction with care. Thus, utilizing the two major oncology outcomes, survival and quality of life, this guideline can be considered to have been validated.
Management of Metastatic Disease--Figure 3A depicts a typical path guideline for recurrent disease following stage II breast cancer. This schema delineates the choice between the use of hormone therapy and cytotoxic chemotherapy as salvage therapy. Unfortunately, for the clinician confronted with a patient, this simple branching does not begin to encompass the spectrum of choices involved in the care of this complex disease.
Figure 3B delineates the 11 most likely sites of metastases in recurrent breast cancer. Subsets of these sites, based on specific metastatic clinical presentations, lead to 23 different clinical scenarios for the management of metastatic breast cancer (Figure 3C), and separate guidelines could be written for each of these 23 paths. Many will encompass just one or two choices, but the standardization of care for well-recognized entities, such as leptomeningeal carcinoma, pleural effusions, or lesions of the femur, would be of major usefulness in clinical decision making. Critical care pathways can then be written to provide detailed plans for accomplishing the treatment recommendations, which, in turn, can allow cost accounting.
In summary, although path guidelines would appear to offer a reasonable way to ensure conformity of care for patients with stage II breast cancer, it is apparent that these path guidelines may have to include fairly extensive detailing in order to cover the full range of clinical scenarios the oncologist may face. Further work will be necessary to demonstrate the usefulness of this approach.
The second major type of guidelines to be considered is the boundary guideline. Essentially, boundary guidelines attempt to define the limits of appropriate practice. They are most often applied to specific modalities or procedures, and their aim is to determine when and how the modality should be used. An example is the ASCO growth factor guidelines . A typical boundary guideline would be the one for primary prophylaxis, which recommends that "primary administration of CSFs be reserved for patients expected to experience levels of neutropenic sepsis with an expected incidence greater than 40%."
Parameters--In deriving boundary guidelines, which often focus on the use of modalities with controversial or incomplete evidence for their usage, several considerations should be entertained. The first is that the guidelines should consider all appropriate alternatives. If possible, the guidelines should set the parameters for the preferential use of each alternative. Thus, the ASCO growth factor guidelines pointed out that an alternative to the use of growth factors is to reduce the dose of cytotoxic therapy, thereby lowering toxic effects and making growth factors unnecessary. Although the ASCO panel recognized dose reduction as an alternative approach, they were not explicit about the conditions under which the use of the alternative was preferred. This may be compared with the Ontario, Canada, growth factor guidelines, which specifically limit their use to curative therapy and recommend dose reduction in palliative or adjuvant therapy (M. Levine, Ontario Cancer Foundation, personal communication).
Application of Modifiers--A second major area of focus in developing boundary guidelines should be the application of modifiers to the guidelines. Modifiers may be defined as special circumstances that fall outside the boundaries of the guideline and allow for deviation. It is well recognized that individual patient situations will legitimately allow for a deviation from the use of a guideline. In many instances, there are well-defined clinical scenarios that fall into this category, and the guideline may want to recognize these. Whenever possible, the guideline should try to delineate these exceptions as explicitly as possible.
For example, the ASCO growth factor guidelines state that clinicians may choose to deviate from the primary prophylaxis guideline and use growth factors when the incidence of neutropenic sepsis is expected to be less than 20%, if there are preexisting risk factors, such as existing neutropenia, extensive prior chemotherapy, prior pelvic radiation, or decrease in immune function. As mentioned above, if feasible, the factors determining the exception should be defined as explicitly as possible. Thus, extensive prior chemotherapy might be categorized as the use of stem-cell ablating agents, such as mitomycin (Mutamycin) or nitrosoureas, for at least 4 months, thereby more precisely defining this modifier.
Language Specificity--The third area that requires meticulous attention in deriving boundary guidelines is the specificity of the language used. The Center for Health Policy Research at George Washington University has published an excellent treatise concerning the issues of language in guidelines . In guidelines recommendations, the use of the words "shall," "should," and "may" must be carefully applied. "Shall" denotes an imperative, and mandates following a specific recommendation. Deviation from this recommendation would be deemed inappropriate care. "Should" denotes a preference but allows the guidelines user leeway by evoking appropriate modifiers. "May" denotes options for which sufficient information is not available to make specific recommendations. The use of "may" in a guideline provides very little assistance to the physician confronted with a clinical decision.
The second issue in guideline language is that of vagueness versus ambiguity. Particular recommendations in a clinical practice guideline may often be vague because the data do not permit further specificity. This is acceptable, and the guideline should provide information about why it is vague on that issue and which type of data is needed to clarify the pronouncement. Ambiguity is not the same as vagueness, and refers to language that allows different interpretations of the same sentence by two readers. Ambiguity must be studiously avoided in guideline derivation, and the use of skilled editors will assist immeasurably in preventing semantic mishaps.
Once guidelines are developed, they must be implemented. Physicians' concerns usually center around issues of rigidity, top-down authority, and loss of clinical autonomy . Given these negative factors, promulgation of a guideline does not automatically lead to its acceptance and to changes in physician behavior. Education alone may not be sufficient to induce physicians to conform to the guidelines . Endorsement from respected leaders has been shown to assist in the implementation of guidelines. Feedback to the physicians, in the form of reminders for individual patients, may be a useful tool in promoting compliance .
As the managed care environment increases, it is to be expected that more and more administrative mechanisms, eg, pharmacy computer systems, will be put in place to monitor whether specific choices conform to accepted guidelines . As guidelines become more ingrained in the managed care system, the use of rewards and punishments for physicians, based on compliance with guidelines, may become the most powerful incentive to their use.
An encouraging note is found in a recent evaluation of physicians' attitudes toward guidelines . The survey showed that clinicians believe practice guidelines would be good educational tools and convenient sources for information. A high proportion of physicians (70%) thought that guidelines were intended to improve quality, and 65% thought they were likely to do so. In the area of cost savings, although 61% of the physicians felt that guidelines were being derived to decrease costs, only 22% felt they would actually accomplish this task.
These considerations lead to the conclusion that the development of clinical practice guidelines in oncology remains a largely uncharted area. The ultimate structure that these guidelines will take and the means for implementing them remain uncertain, but certainly, on a theoretical basis, they would appear to offer a feasible way to approach the mandates of a managed health-care environment.
1. Coleman RL: Promoting quality through managed care. Am J Med Qual 7(4):100-105, 1992.
2. Field MJ, Lohr KN (eds): Clinical practice guidelines for a new agency. Institute of Medicine, Committee on Clinical Practice Guidelines. Washington, DC, National Academy Press, 1990.
3. Tingley FW: The use of guidelines to reduce costs and improve quality: A perspective from the insurers. Jt Comm J Qual Improv 19(8):330-334, 1993.
4. Leape LL: Practice guidelines and standards: An overview. Qual Rev Bull 16:42-49, 1990.
5. Grimshaw JM, Russell IT: The effect of clinical guidelines on medical practice: A systematic review of rigorous evaluations. Lancet 342:1317-1322, 1993.
6. Kane RL, Lurie N: Appropriate effectiveness: A tale of carts and horses. Qual Rev Bull 18:322-326, 1992.
7. Carpenter CE, Nash DB, Johnson NE: Evaluating the cost containment potential of clinical guidelines. Qual Rev Bull 19:119-123, 1993.
8. McKee M, Clarke A: Guidelines, enthusiasm, uncertainty, and the limits to purchasing. Br Med J 310:101-104, 1995.
9. Smith TJ, Hillner BE: The efficacy and cost-effectiveness of adjuvant therapy of early breast cancer in premenopausal women. J Clin Oncol 11:771-776, 1993.
10. AHCPR: Using clinical practice guidelines to evaluate quality of care, vol 1, pp 1-41. AHCPR Pub No 93-0045, 1993.
11. Woolf SH: Practice guidelines: A new reality in medicine: II. Methods of developing guidelines. Arch Intern Med 152:946-952, 1992.
12. Eddy DM: A Manual for Assessing Health Practices and Designing Practice Policies: The Explicit Approach. Philadelphia, American College of Physicians, 1992.
13. Zander K: Nursing care management: Strategic management of cost & quality outcomes. J Neuro Admin 18(5):23-30, 1988.
14. Evidence-based Medicine Working Group: Evidence-based care: 2. Setting guidelines: How should we manage this problem? Can Med Assoc J 150:1417-1423, 1994.
15. Hlatky MA: Patient preferences and clinical guidelines. JAMA 273:1219-1220, 1995.
16. Langley PC: The role of pharmacoeconomic guidelines for formulary approval: The Australian experience. Clin Ther 15:1154-1176, 1993.
17. American Society of Clinical Oncology: American Society of Clinical Oncology recommendations for the use of hematopoietic colony-stimulating factors: Evidence-based, clinical practice guidelines. J Clin Oncol 12:2471-2508, 1994.
18. Browman G, Levine MN, Monide EA, et al: The practice guidelines development cycle: A conceptual tool for practice guidelines development and implementation. J Clin Oncol 13:502-512, 1995.
19. Parmley WW: Clinical practice guidelines: Does the cookbook have enough recipes? JAMA 272:1374-1375, 1994.
20. Khansur T, Haick A, Patel B: Evaluation of bone scan as a screening work-up in primary and local-regional recurrence of breast cancer. Am J Clin Oncol 10:167-170, 1987.
21. National Institutes of Health Consensus Conference: Treatment of early stage breast cancer. JAMA 265:391-395, 1991.
22. Simon MS, Hoff M, Hussein M, et al: An evaluation of clinical follow-up in women with early stage breast cancer among physician members of the American Society of Clinical Oncology. Breast Cancer Res Treat 27:211-219, 1993.
23. The Givio Investigators: Impact of follow-up testing on survival and health-related quality of life in breast cancer patients: A multicenter randomized controlled trial. JAMA 271:1587-1592, 1994.
24. Spennak SM, Budetti PP, Zweig F: Use of language in clinical practice guidelines. AHCPR Contract No 282-87-6049, 1992.
25. Tunis SR, Haward RSA, Wilson MC, et al: Internists' attitudes about clinical practice guidelines. Ann Intern Med 120:956-963, 1994.
26. Lomas J, Anderson GM, Domnick-Pierre K, et al: Do practice guidelines guide practice? The effect of a consensus statement on the practice of physicians. N Engl J Med 321:1306-1311, 1989.
27. Weingarten SR, Reidinger MS, Conner L, et al: Practice guidelines and reminders to reduce duration of hospital stay for patients with chest pain. Ann Intern Med 120:257-263, 1994.
28. Johnston ME, Langton KB, Haynes B, et al: Effects of computer-based clinical decision support systems on clinician performance and patient outcome: A critical appraisal of research. Ann Intern Med 120:135-142, 1994.