In this article, we review the tools needed to enable clinicians to evaluate evidence and to assimilate changes in clinical practice as new information results in revisions in generally accepted knowledge.

	EVIDENCE-BASED TREATMENT DECISIONS

TOP ABSTRACT EVIDENCE-BASED TREATMENT... TOOL NO. 1: DIFFERENCES... TOOL NO. 2: CRITICALLY... TOOL NO. 3: UNDERSTANDING... TOOL NO. 4: SELECTING... APPLYING THE TOOLS CONCLUSIONS REFERENCES

 
The importance of the quality of evidence in making therapeutic decisions is illustrated by the revision in the once generally accepted knowledge regarding the cardiovascular effects of vasoconstrictors in local anesthetic solutions. For decades, practicing dentists believed the axiom that the amount of epinephrine in local anesthetic solution is less than the amount that would be released from a patient’s adrenal glands owing to pain if the vasoconstrictor were excluded.^1–3 Hence, the amount of epinephrine in the local anesthetic—up to 200 micrograms—was considered appropriate for healthy adult patients⁴ and concentrations of 1:50,000 epinephrine were recommended for patients with cardiovascular disease.⁵

More recent recommendations for using epinephrine in clinical dental practice are not in full agreement,⁶ with recommendations ranging from contraindications to epinephrine in a patient with controlled hypertension taking medications with known epinephrine interactions⁷ to caution against using epinephrine in patients with uncontrolled hypertension.⁸ As with any drug, the dose administered is an important consideration.⁹ A series of clinical studies demonstrated that measurable increases in circulating epinephrine levels accompany administration of epinephrine-containing local anesthetics that clearly can be distinguished from the levels attained when anesthetic solution without epinephrine is administered.^10–12 The greater the dose administered, the greater the change in circulating levels and the greater the cardiovascular effects.^9,13

Since the 1990s, the decision to use an epinephrine-containing local anesthetic in a particular patient has been based on a dentist’s knowledge of the physiological effects of epinephrine balanced against the possible cardiovascular risk for the patient.^6,14 Injection of 100 to 150 mg of epinephrine, common during oral and periodontal surgery, produces venous plasma concentrations equivalent to that present during heavy exercise.¹⁵ Absorption of the small amounts of epinephrine contained in one to two dental cartridges evokes modest, but reproducible, increases in stroke volume and cardiac output, with comparable decreases in peripheral vascular resistance.^16,17 Large doses of epinephrine-containing local anesthetic, either inadvertently administered intravascularly or due to a high total administered dose, can produce dramatic increases in blood pressure and changes in heart rate (tachycardia and dysrhythmias). These changes are more likely to occur in patients with pre-existing cardiovascular disease and in patients taking medications likely to augment adrenergic actions (for example, tricyclic antidepressants and nonspecific ß-adrenergic blockers.¹⁸

Assertions made in the 1960s based on animal data and textbook assumptions that were extrapolated to humans without any quantification of actual circulating epinephrine levels have been supplanted by scientific evidence.^9–12 Development of reliable catecholamine measurements in human blood¹⁹ combined with controlled clinical trials provided direct evidence for the revision of this widely held tenet—with a directly opposite conclusion. This example illustrates the fact that the therapeutic decision-making process is highly dependent on the quality of evidence (Box 1²⁰) that is considered in making a judgment and application of that evidence to patient care.

	TOOL NO. 1: DIFFERENCES BETWEEN CLINICAL PRACTICE AND CLINICAL RESEARCH

TOP ABSTRACT EVIDENCE-BASED TREATMENT... TOOL NO. 1: DIFFERENCES... TOOL NO. 2: CRITICALLY... TOOL NO. 3: UNDERSTANDING... TOOL NO. 4: SELECTING... APPLYING THE TOOLS CONCLUSIONS REFERENCES

View larger version (83K): [in this window] [in a new window]   Figure 1. Similarities and distinctions between clinical practice and clinical research.

The clinician assumes that the therapy is successful on the basis of a subjective evaluation of the outcomes (that is, the treatment worked for a number of patients). By contrast, the clinical investigator makes assessments without knowledge of the treatments that patients receive in order to control his or her bias, as well as that of the patient, regarding the expected outcome. Clinical success of a new treatment in practice is judged qualitatively based on recollection of patients who were treated with an older therapy. Therapeutic efficacy in a clinical trial is quantified in comparison with baseline measures or in groups of patients receiving different treatments in parallel, and the findings are analyzed statistically to take into account chance occurrences and regression of symptoms over time that can mimic clinical success.

Lastly, the perception that a treatment works "in my hands" often is regarded by practitioners with certainty, while a clinical research finding requires confirmation in at least one other well-controlled trial before it can be considered as suggesting or demonstrating a therapeutic effect. Where the two processes converge is in considering how well the study group represents the general population, which is termed "generalizability."

Why such a dichotomy for similar processes? Box 2 summarizes many of the therapeutic factors that have been shown to contribute to clinical success in addition to a genuine therapeutic effect. Any and all of these factors may result in a reduction in symptoms independent of a therapeutic effect, thereby obscuring the ability to attribute the outcome to the intervention. The evaluation of therapeutic efficacy also is likely to result in a perception of clinical success if the factors identified in Box 2 are not controlled for. Failure to appreciate these differences between clinical practice and clinical research may lead to ineffective or harmful treatments being presented to patients as safe and effective, thus exposing them to the risk of iatrogenic injury and putting the clinician at risk of professional liability for injuries resulting from use of a nonvalidated treatment. This latter concern is illustrated by the more than 670 lawsuits directed at one clinician for placing temporomandibular joint (TMJ) implants in the 1970s and 1980s.²¹

	TOOL NO. 2: CRITICALLY EVALUATING THE EVIDENCE

TOP ABSTRACT EVIDENCE-BASED TREATMENT... TOOL NO. 1: DIFFERENCES... TOOL NO. 2: CRITICALLY... TOOL NO. 3: UNDERSTANDING... TOOL NO. 4: SELECTING... APPLYING THE TOOLS CONCLUSIONS REFERENCES

View larger version (58K): [in this window] [in a new window]   Figure 2. Progression of therapeutic interventions (adapted with permission of Elsevier from Dionne22).

– The treatment has documented safety and efficacy.

– The treatment has documented safety but equivocal efficacy.

– The treatment is efficacious, but safety has not been documented.

– The treatment is neither safe nor efficacious.

Nonsteroidal anti-inflammatory drugs. If several trials indicate that the treatment is effective and has minimum toxicity, the clinical community considers it to be a validated therapeutic practice. An example of this outcome is the use of nonsteroidal anti-inflammatory drugs (NSAIDs) for the control of acute pain. If the treatment is found not to be effective or toxicity becomes evident, the drug is withdrawn from the market, which occurred with several orally administered NSAIDs (that is, zomepirac, suprofen and the selective cyclo-oxygenase-2 [COX-2] inhibitor rofecoxib), or labeling restrictions are imposed, as was done for orally administered ketorolac and the COX-2 inhibitor class. Persistence of a therapeutic approach that has been documented to be ineffective or unsafe is an irrational clinical practice.

The need to integrate new evidence into clinical practice is illustrated by rofecoxib, a selective COX-2 inhibitor. At standard therapeutic dosages, conventional NSAIDs such as ibuprofen exert their analgesic and anti-inflammatory effects through COX-2 inhibition, yet at the same time they also block COX-1. This latter action, in turn, leads to diminished production of the prostaglandins that regulate homeostatic functions. As a consequence, NSAIDs produce substantial platelet effects, along with the potential for renal and gastrointestinal toxicity, which may interfere with clinical use.

The discovery and characterization of the COX-2 form of cyclo-oxygenase led to the hypothesis that selective inhibition of COX-2 would provide the potent anti-inflammatory and analgesic effects of an NSAID without influencing COX-1 and its important physiological functions. On the basis of this hypothesis, and motivated by the significant morbidity and mortality associated with NSAIDs,^23–25 researchers made substantial efforts to develop specific inhibitors of COX-2. Although a number of studies suggested that the original paradigm regarding the roles of COX-1 and COX-2 was overly simplistic, COX-2 inhibitors are highly selective for COX-2 suppression at the dosages administered clinically, with minimal effects on COX-1 activity.

Therapeutic process revised. Rofecoxib was approved for treatment of osteoarthritis, acute pain in adults and primary dysmenorrhea,^26,27 but it was withdrawn from the market in 2004 owing to an increased incidence of cardiovascular events when administered on a long-term basis. Thus, a therapeutic process that was well-founded on scientific evidence and evaluated carefully in thousands of patients before being introduced to clinical practice was revised as more evidence from tens of thousands of patients exposed to the drug for longer periods revealed previously undetected evidence of significant cardiovascular morbidity and mortality.

Moreover, re-examination of the cardiovascular effects of other selective COX-2 inhibitors such as celecoxib and valdecoxib resulted in the U.S. Food and Drug Administration’s adding warnings to their package labeling. This scenario illustrates the need to continually re-evaluate generally accepted clinical practices in light of emerging evidence and to base therapeutic decisions on the best available evidence, not to rely solely on expert opinion or the longevity of a clinical practice.

	TOOL NO. 3: UNDERSTANDING WHAT CONTRIBUTES TO PERCEPTIONS OF CLINICAL SUCCESS

TOP ABSTRACT EVIDENCE-BASED TREATMENT... TOOL NO. 1: DIFFERENCES... TOOL NO. 2: CRITICALLY... TOOL NO. 3: UNDERSTANDING... TOOL NO. 4: SELECTING... APPLYING THE TOOLS CONCLUSIONS REFERENCES

 
Components of the therapeutic process and the evaluation of therapeutic efficacy may result in the appearance of an improved clinical outcome in addition to a genuine therapeutic effect (Box 2). For example, the placebo response accounts for 35 to 50 percent of the reduction in symptoms seen for therapies aimed at subjective symptoms such as pain.^28,29 Fluctuations in the disease process and spontaneous remission usually result in a reduction of symptoms, as patients seek treatment when they are at their worst and their conditions generally improve over time as a result of the natural history of the disease process. Upper respiratory infections, for instance, usually are cleared in most patients within five to seven days as the underlying viral infection resolves by endogenous immune factors. In addition, successful doctor-patient interactions generally have a positive effect on subjective symptoms, which is independent of a genuine therapeutic effect.

Components of the evaluation process for therapeutic efficacy that need to be controlled in clinical trials include the biases of the patient, the treating clinician and the study sponsor, all of whom anticipate a positive outcome for the therapeutic intervention (Box 2). Chance occurrences may result in effects in a small number of patients that are not borne out by evaluation of a larger number of subjects. Evaluation of a treatment in an uncontrolled trial, by comparison with historical data, or by comparison with a concurrent group of patients not randomly allocated to the treatments is more likely to lead to enthusiastic recommendations of a new treatment compared with the results of properly controlled trials.²⁹ Comparison of a group randomized to the new treatment with a parallel group randomized to a standard treatment or a placebo remains the gold standard by which new treatments are best evaluated.^29,30

Although it is not feasible to randomly allocate placebo medications or no treatment to patients in clinical practice, knowing the principles of how to apply the scientific method to the evaluation of therapeutic modalities enhances the likelihood that clinicians will be able to recognize a valid answer when inquiring about the effectiveness of a new treatment for use in clinical practice (that is, generalizability). Knowledge of the components of the therapeutic and evaluation processes permits astute clinicians to differentiate validated treatments from uncontrolled clinical observations, expert opinion not based on well-controlled studies or exaggerated promotional claims.

	TOOL NO. 4: SELECTING THE MOST APPROPRIATE THERAPY FOR CLINICAL PRACTICE

TOP ABSTRACT EVIDENCE-BASED TREATMENT... TOOL NO. 1: DIFFERENCES... TOOL NO. 2: CRITICALLY... TOOL NO. 3: UNDERSTANDING... TOOL NO. 4: SELECTING... APPLYING THE TOOLS CONCLUSIONS REFERENCES

 
If it is true that many current therapeutic practices have not been scientifically validated, the need to act in the face of incomplete information becomes a frequent dilemma for the clinician. However, in the absence of high-quality evidence regarding the risks and benefits of specific treatments, or when such evidence is scarce, the clinician must extrapolate beyond what is known to be true to what is likely to be true for the patient being treated.

Sackett³⁰ suggested two circumstances in which absence of treatment actually may be preferable to selecting a treatment based on inadequate evidence. The first is when false-positive diagnostic errors label the patient as having a disease when, in actuality, he or she does not. This can include physical findings that are slight deviations from normal and radiographic overinterpretation. The second circumstance is when the treatment is worse than the disease (for example, use of a TMJ implant for a patient with crepitus) or when the illness is trivial, self-limited or within the recuperative powers of the patient. If we assume that these circumstances do not apply and the patient’s condition warrants treatment, what is the best way to select a specific intervention that achieves the goal of treating the patient?

Selecting therapy via induction from biological facts or uncontrolled observations from clinical experience may result in some patients’ conditions improving owing to chance or nonspecific regression of symptoms, leading to an apparent association between the procedure and the perceived success. If the treatment is not evaluated further in a controlled trial, it may be perpetuated by general acceptance for several generations.

For example, the association between malocclusion and orofacial pain has led to the assumption of causality and the therapeutic strategy of correcting the malocclusion to lessen the severity of pain by breaking the vicious cycle of muscle spasm. However, Stohler and colleagues³¹ and Obrez and Stohler³² demonstrated that producing pain experimentally in the muscles of mastication of previously pain-free volunteers results in the development of limited opening and transient occlusal discrepancies, suggesting that the association is not causal. Rather, muscle pain activates protective reflexes that limit range of motion, thereby leading to occlusal discrepancies. McNamara and colleagues³³ conducted a comprehensive review of the literature regarding the possible relationships between occlusal factors, orthodontic treatment and TM disorders; they concluded that the association is minor and recommended that permanent alterations of the occlusion should be discouraged.

However, the induction method does represent a logical approach to suggesting treatment ideas for testing in randomized clinical trials, and this may result in validation of treatments as efficacious and safe. The fatal flaw in the induction method is that it often is incapable of detecting erroneous conclusions about efficacy (for example, blood letting), even when the observations on which it operates are totally accurate (that is, some patients got better).³⁰

	APPLYING THE TOOLS

TOP ABSTRACT EVIDENCE-BASED TREATMENT... TOOL NO. 1: DIFFERENCES... TOOL NO. 2: CRITICALLY... TOOL NO. 3: UNDERSTANDING... TOOL NO. 4: SELECTING... APPLYING THE TOOLS CONCLUSIONS REFERENCES

 
Until a sufficient body of evidence becomes available, is integrated into the dental school curriculum and continuing professional education and is disseminated widely, clinicians must continue to evaluate the quality of evidence that contributes to therapeutic decision making and apply it to patient care. In an era of evidence-based clinical practice, it no longer is adequate to rely on recall of facts assimilated during professional training, to abdicate critical appraisal of the evidence supporting a therapeutic practice to a prominent clinician or an educator, or to assume that the length of time a practice has been in use implies that it has been validated. Clinicians should ask themselves the following questions when evaluating evidence used to make a therapeutic decision:

– What is the level of evidence available to make the decision (Box 1)?

– Is the therapeutic effect clearly distinguishable from placebo responses, a positive doctor-patient interaction and remission of symptoms due to fluctuations in the disease process and the natural history of the disease or symptom (Box 2)?

– What are the likely biases of the authors or sponsors of the study? Does this bias appear to have been controlled by blinded, quantitative evaluation of the outcomes (Box 2)?

– What evaluation criteria were used to measure clinical success? Is there evidence that the outcome measure is sensitive, reliable and validated?

– What is the nature of the comparison group? Was it historical (least reliable), a comparison with baseline (more reliable) or a parallel group treated identically (except for the standard therapy) to the group receiving the investigational treatment (best comparison)?

– Are the results generalizable (for example, was the group studied similar to the dentist’s patient population?)?

– Were statistical analyses conducted to rule out chance occurrences or normal variations in symptoms?

– Have the results been replicated in at least one other well-controlled clinical trial in the published literature?

– Do the stated conclusions appear logical or make sense based on the results presented?

– Was the intervention tested for safety?

– Were the study findings published in a peer-reviewed journal?

A growing number of information sources are provided by groups that evaluate the literature in the context of these questions. Box 3 provides a list of resources to aid clinicians in evaluating the evidence to apply to decision making regarding treatment.

	CONCLUSIONS

TOP ABSTRACT EVIDENCE-BASED TREATMENT... TOOL NO. 1: DIFFERENCES... TOOL NO. 2: CRITICALLY... TOOL NO. 3: UNDERSTANDING... TOOL NO. 4: SELECTING... APPLYING THE TOOLS CONCLUSIONS REFERENCES

Expertise in diagnosis and treatment selection is based largely on the decision-making process, which is a combination of the knowledge base, reasoning skills and an accumulation of experiences with patients.³⁵ The ability to make a sound clinical decision is based in large part on the quality of evidence that supports an accepted clinical practice and the practitioner’s ability to evaluate this evidence with regard to the benefits, risks and appropriateness of the treatment options. Other factors that influence clinical decision making (for example, experience with similar therapeutic situations and the practitioner’s level of clinical training) are important, but they rest heavily on the cumulative knowledge base that is integrated by the clinician in arriving at a decision.

Clinical practice requires practitioners to make frequent therapeutic decisions that are based on widely used but nonvalidated therapies, new clinical practices or changes in the evidence supporting a generally accepted therapy. While it is not feasible for a single dental practitioner to conduct large-scale, well-controlled studies to validate new drugs or devices, the ability to understand and interpret the various levels of evidence that permit an informed decision to be made are within the scope of professional practice. Growing awareness of the need for clinical practice to be evidence based likely will result in greater and higher-quality evidence that permits a reexamination of current therapies and validation of new therapies before they are introduced into clinical practice.