CHAPTER 15 ACCESSING AND READING DENTAL PUBLIC HEALTH RESEARCH: EVIDENCE-BASED DENTAL PRACTICE
A revolution has taken place in health care. No longer is society willing to tolerate the health care practitioner, no matter how skilled, who relies solely on experience, personal judgment, and the training received in school to make decisions regarding patient care. It is expected that the practitioner’s selection of treatment modalities, materials, and responses to patient inquiries regarding current practices and controversies in dentistry and medicine will be consistent with the best knowledge and research available in the field. As stated in the Institute of Medicine’s 1995 report, Dental Education at the Crossroads, oral health care delivery must be an integral aspect of comprehensive health care, including primary care. “Dental education must be scientifically based and undertaken in an environment in which the creation and acquisition of new scientific and clinical knowledge are valued and actively pursued.”1 For the health care provider, learning must be a lifelong endeavor and cannot end with the attainment of a dental degree, a residency program, or specialty training.
The profession and the public expect the modern dentist to be informed and able to respond to patients’ questions regarding current controversies in oral health care, public health practices, and personal oral health maintenance recommendations. Bader and colleagues2 maintain that the incorporation of dental research findings by clinicians into their patient care is central to the maximization of treatment benefits and minimization of treatment harm.
Health policy developers in the public health sector and third-party payers in the private sector expect health care providers to be able to demonstrate the efficacy and appropriateness of their treatment decisions. Because of rising health care costs, a growing medically and dentally vulnerable aging population, and an increasing demand for health care, there is a rapidly increasing demand that practitioners demonstrate the “cost-effective” aspects of their treatment decisions.3 The dental profession has received a great deal of criticism in the public press because of findings that enormous variations exist in treatment recommendations and health care practice. These variations have been attributed to (1) poor science underlying the clinical decisions, (2) poor quality of clinical care decisions, and (3) variations in clinical skills.4 To counter these criticisms and to respond to the challenge of modern health care, the dentist must combine evidence-based information with practical clinical experience when engaging in the process of diagnosis, treatment planning, and treatment.
How is the dentist expected to meet this demand and continue to provide dental services in both quality and quantity adequate to meet patient demand and economic reality? The flood of new articles, books, and reports is overwhelming. Not only must information relevant to practice decisions be identified and obtained, it must also be digested, evaluated, interpreted, and prioritized for its clinical validity and practicality. In addition, the dentist, as part of the public health delivery system, must also be aware of current trends in the public health sector. Infection control, environmental intervention, public health education, and changes in strategies and regulations regarding third-party and government funding for dental services all affect daily practice.
This chapter provides some concrete suggestions for accomplishing the difficult challenge of being informed and aware of current standards of care. Basic issues and methods in dental research are discussed first. The chapter then examines current criteria for evidence in the literature and reviews the types of studies that are accepted as evidence and provides criteria for weighting their importance. The chapter concludes with a discussion of strategies for searching and identifying studies that are helpful in answering the questions important to practice decisions and relevant to patients’ concerns.
Research involves a systematic investigation designed to answer questions about the events we observe. The goal is to develop accurate explanations and predictions for future occurrences of these events. It is a process of discovering and documenting “regularities” in the universe. This process is intended to provide valid, consistent, and reliable predictions, explanations, and descriptions that can be used to understand, plan, and make decisions. Typically, a research study does not prove that a phenomenon or causal relationship exists but rather demonstrates that the events or population characteristics documented are not the result of an artifact created by a study but a consistent finding that can provide a basis for sound conclusions.
In most studies, a set of conditions (independent variables) is specified and then a second set of factors (dependent variables) is measured. The question posed is, “How does the dependent variable change as the values or conditions of the independent variable are manipulated or selected?” Much of research design is focused on eliminating alternative, competing explanations for observed or hypothesized relationships. The term for this effort is capturing the variance. Essentially, we are trying to identify the consistent percentage of the total variation in the dependent variable (effect size) that is due to the changes in the independent or causal variable. Other factors (intervening variables, including experimenter error and chance) will increase or decrease the amount of change observed. The researcher is attempting to isolate and quantify that portion of the change which is reliably (not created as an artifact or chance event that occurred in the study) affected by the variations in the independent variable. Because of this, standards of evidence emphasize the importance of replication of study findings rather than depending on the findings of a single study.
Dental health research, including oral public health research, can be divided into three types: descriptive studies, experimental studies, and meta-analytic studies. Box 15-1 provides examples of these three types of research.
BOX 15-1 Types of Public Health Research
Descriptive studies are designed to document or characterize observations resulting from data collection (e.g., surveys) and systematic observation of a target population or population sample. For example, what is the decayed, missing, and filled surfaces level in the population of children, between ages 10 and 15 years, who have not been exposed to a consistent fluoridated water supply? Documentation includes the following: epidemiologic research, in which rates (incidence and prevalence) of disease in a “target” population are documented; surveys, in which information is elicited regarding population characteristics; and demographic studies, in which the population is characterized and described in terms of the distributions of such characteristics as age, sex, economic status, education, unmet need, and oral health status.
A second type of descriptive study reports relationships between behaviors or characteristics and disease rates. Risk assessment, in which the rates of disease among individuals exposed to a particular pathogenic cause are compared with the rates of those not exposed (e.g., rates of cancer among smokers as compared with rates of nonsmokers), and correlational studies, in which rates of disease are correlated with characteristics such as economic status, education, and health-related behaviors, are examples of this type of study.
Experimental studies are designed to test whether a hypothesized relationship between the independent variable(s) (conditions) and a dependent or outcome variable is causally related. For example, does changing the acidity level of the saliva affect the demineralization process that is occurring in the mouth? Experimental research involves controlled studies, either in the “field” or in the laboratory, which are designed to “test” hypothesized causal relations. For example, two matched groups are compared in an experimental design in which one group is treated with an experimental drug and the second is given a placebo. The groups are then compared as to their recovery time from a particular disease. A “field experimental study” might involve a “natural experiment” in which two communities, one of which has a fluoridated water supply and the other of which has a water supply with a very low fluoride concentration, are compared over a number of years in terms of their caries incidence. A more controlled field experiment might involve a direct intervention into one community (e.g., a screening for oral cancer) and an indirect intervention into another community (e.g., an education program about oral cancer) with the goal of testing for the cost-effectiveness of a particular set of health promotion strategies. A controlled clinical study might be one in which one group uses a particular mouthwash and the other group uses another mouthwash and the two groups are compared for gingival health and calculus accumulation. The research study might occur in vivo or in vitro. The highest level of experimental research includes randomized small clinical trials in which the experiment is repeated in a variety of settings instead of in one setting with a large number of subjects.5
Meta-analysis involves a “secondary research method” in which a series of primary research studies is systematically explored for consistent findings across research studies. It involves the use of statistical techniques to compare and integrate the findings of a large series of independent projects.6 Although meta-analysis was formally documented in 1976 and the statistical techniques employed have appeared in the literature over the last 60 years, the use of meta-analysis as a primary resource represents a change created by the evidence-based research movement. Meta-analysis is used to collect, analyze, and compare the results of multiple primary studies to develop an overall conclusion or summary of their findings. It not only summarizes and statistically tests the findings, but also identifies phenomena that are consistent and stable. By considering and weighting for the sample size, effect size, and research design in each study, the stability and statistical significance of the conclusions can be evaluated.7
Each of the research approaches considered here has value in evidence-based dentistry if conducted appropriately. Criteria for evaluating a research study are discussed later in this chapter. However, these studies can be classified in terms of their “level of evidence,” or credibility. Box 15-2 provides a generalized ranking of published articles in terms of rigor and vulnerability to artifact, bias, and random effects. The lowest level of evidence comes from case descriptions, uncontrolled observational studies, and random surveys. In these studies, the researcher has not systematically controlled the conditions under which the observations occurred so that the results are often produced by the chance events surrounding the study. The second level of evidence comes from studies in which the researcher has taken care to control for internal biasing influences. Included in these studies are surveys based on stratified, random sampling; field studies in which the evaluators are calibrated; and longitudinal studies in which appropriate controls are used to ensure validity. In studies with a greater level of evidence, the researcher has increased the level of replication, controls for sampling error, and controls for contaminating variables (by instituting control groups, matched control groups, and random assignment to condition). The highest level of evidence is presented when the researcher combines a number of studies so that the effects and artifacts of any one study are counteracted by other studies or statistical techniques so that what remains is a stable estimate of the examined phenomena.
By “evidence-based” dentistry we mean the “integration of best research evidence with clinical expertise and patient values.”8 In the appropriate application of the evidence-based approach, the practitioner will combine the best evidence available, practical experience, current standards of care, and the limitations and possibilities of the specific situation to make a judicious decision. In Box 15-3, we suggest a hierarchy of research evidence authority based on the vulnerability of various types of studies to bias (validity) and chance artifacts. Later in this chapter, we will discuss strategies for reading and evaluating a research study for its rigor and design; the reader, however, must also consider the relevance, practicality, and heuristics of the study when making a final decision regarding the appropriateness of findings for use.
BOX 15-3 Evidence Hierarchy
Adapted from Sutherland S: Evidence-based dentistry: part IV. Research design and levels of evidence, J Can Dent Assoc 67:375, 2001 and the Oxford Centre for Evidence-based Medicine Levels of Evidence, May 2001. cebm.jr2.ox.ac.uk/docs/levels.html.
The strongest evidence is replication of the study findings. Independent replications provide clear evidence that the finding is not just a random event. The researcher demonstrates that the finding is not just an artifact of a single study but a consistent event. Systematic reviews of research findings (meta-analysis) represents this improved level of evidence when the guidelines for study selection and statistical compilation are followed. Systematic replication of findings using controlled studies provides evidence for the stability and validity of the reported phenomena. By combining randomized, selected studies (or the universe of available studies) from a pool of studies meeting explicit, predetermined, experimental design criteria, the researcher can counteract and eliminate bias that occurs in each individual study reviewed.
Evidence is stronger to the degree to which each research study controls internal and external contaminating influences so as to be able to identify the specific causal factors and population characteristics in question. For example, a survey may return highly biased results if only those individuals who feel strongly about a particular issue respond. If one were testing a new medication, it would be inappropriate to select for testing only patients who had a high probability of getting well while excluding those who were sicker or less responsive. These external and internal factors may be controlled statistically or by the use of experimental controls (use of systematic observation, control groups, randomized sample selection, and reliability checks on the data collected).
Ranking and selecting studies for review imply the application of rules of evidence. To the degree that the experimenter controls for bias, the study gains in validity and reliability. The experimenter can control for sample bias by using a randomized sample selection with specific stated rules for selection or rejection of the subject. The experimenter controls for experimenter bias by “blinding the experimenter” to the experimental (independent variable) conditions and by “calibrating” the observer to eliminate response bias. The experimenter controls for preexisting conditions and influences such as placebo effects, subject bias, and “maturation” by use of experimental controls. The more systematic and carefully controlled the study, the more likely the outcomes will be the result of the study process rather than of an artifact or a chance finding.
Randomized, controlled trials represent the next highest level of evidence. This level of evidence is supported by single (nonreplicated) experimental studies in which the experimental and control conditions are clearly specified and in which assignment to the experimental and control conditions is random. Epidemiologic surveys in which the population is sampled systematically (random, stratified sampling) and the observers are calibrated serve as the next level of evidence. Nonrandomized studies with controls such as case-controlled studies and field studies form the next level of evidence. Studies using historical controls but using randomized sampling or selection serve as the next level. Cohort studies in which disease risk assignments are made using correlational analysis are next in terms of evidence. Case reports and related anecdotal or descriptive evidence are next. Finally, the reports of expert committees and the opinion of experts form the lowest level of evidence.