The “evidence pyramid” illustrates how different types of studies basically are assessed for their potential ability to provide different levels of scientific evidence for a specific clinical question
188.8.131.52 Case Reports
A simple way of clinical research is the description of clinical cases, which may show unique or unusual features of the condition or outcome of therapy. Such case analyses are quite common in endodontic journals and at congresses. It is the only mean by which specific or even unexpected clinical events can be described and are therefore important as further examinations may be initiated. The limitation of case presentations is obvious. Information from a single case cannot be transferred to other patients because of the wide variation and many factors not controlled or checked for.
184.108.40.206 Case Series
Series of cases provide better information. Larger groups of patients with a particular disease or condition subjected to treatment are studied. The involvement of chance can be checked by statistical analysis. Yet, the efficiency of the clinical procedure cannot be ascertained to be better or equal to any other method if no control group is available. Sometimes, inclusion of data from prior studies or other authors’ results is used for comparison purposes. However, this procedure will not bring particular strong evidence to the report, as the conditions, under which the studies were conducted, may not have been very similar.
220.127.116.11 Case-Control Studies
Case-control studies also belong to the arsenal of methods of observational studies available for clinical research. A case-control study assesses persons with a condition (or another outcome variable) of interest and a suitable control group of persons without the condition (comparison group, reference group). The potential relationship of a suspected risk factor or an attribute to the condition is examined by comparing the affected and non-affected subjects with regard to how frequently the factor or attribute is present (or, if quantitative, the levels of the attribute) in each of the groups. For an example Kim et al.  studied the influence of a presence of an isthmus area when performing endodontic surgery in maxillary and mandibular first molars. Of the 106 teeth, 72 teeth had an isthmus, and 34 did not. The analysis revealed that the cumulative 4 years survival rate after surgery was 61.5 and 87.4%, respectively, when an isthmus was present and absent.
18.104.22.168 Prospective Cohort Studies
Retrospective studies of any nature will not do well as they suffer the risk of having limited or no control of a number of aspects relevant to outcome. To provide good evidence, clinical research requires being prospective. A type of clinical research design that has gained increased attention in recent years is the prospective cohort observational study . Cohort refers to a group of patients. This study design implies that a large sample of patients can be treated and then be assembled for follow-up examinations. An important factor of such studies is that a fairly large number of clinics have to be included in order for the report to gain generalizability. Yet the study design has the advantage to allow inclusion of general dentists and therefore will give availability to aspects on endodontics at which we have very little understanding.
Given that a comprehensive registry of patients under treatment currently is under way in many countries, valuable basic information on the efficiency of procedures can be gained . However, since validated data on diagnoses, treatment protocols and other essential details often are lacking, this type of study cannot give answers to more detailed clinical questions.
22.214.171.124 Randomized Controlled Trial
Randomized controlled trials (RCTs) are genuine clinical experiments. Two or more groups of subjects receive different measures, are followed forward in time and are compared using an outcome assessment. The distinguishing feature of an RCT is that patients are allocated to test and control procedures in a strictly randomized manner. This kind of study usually observes the effect of a single variable only. All other variables (background variables, confounders) are controlled by the chance effect in both the test and the control group. To be appropriate RCT requires further a precalculated minimal number of patients to be included in order to ensure that a statistical significant difference between the test and the control procedure can be ascertained. RCTs are indeed powerful tools as many of the biasses that affect nonrandomized trials can be eliminated.
In Fig. 9.2 the set-up of three different RCTs is sketched. From a scientific perspective, these would be highly desirable to implement in order to provide significantly better knowledge of the prognosis for apical periodontitis in root-filled teeth.
Basic designs of three RCTs that should be highly preferred to be carried out in order to increase the level of scientific evidence for crucial questions regarding apical periodontitis in root-filled teeth. (a) Comparing the outcome of periapical tissues following surgical versus non-surgical retreatment. (b) Comparing “functional retention” following surgical retreatment versus extraction and replacement with an implant. (c) Comparing important outcomes following retreatment versus “no intervention”
Yet, the long time they require to conduct and the high costs make RCTs difficult to perform. On a careful consideration, we must realize that RCTs may not attain evidence-based research very easily for important clinical questions for endodontic retreatment. In fact, RCTs are ideal for testing the effects of drugs because it can use placebo and be controlled double blinded. However, for assessment of important aspects of surgical interventions, such as endodontic retreatment procedures, several predicaments occur. For example, it would probably be difficult to enrol a sufficient number of root-filled teeth with apical periodontitis similar enough to be randomized to extraction and implant or non-surgical retreatment. But even if possible, it is likely that the values and expectations of the patients, the dentists and the evaluators could influence the assessment of the outcome since neither patients, dentists nor evaluators can be blinded to the allocated treatment (Fig. 9.2a, b). Another concern is the long follow-up time necessary to a meaningful outcome comparison assessment. If, for example, surgical or non-surgical intervention is compared to extraction implant placement in a randomized manner (Fig. 9.2b), the crucial and interesting comparison is at hand only after several years [6, 7].
Finally, RCTs for the most central clinical questions have ethical challenges. For example, perhaps the utmost important and vital RCT would be to randomly allocate individuals with asymptomatic apical periodontitis in root-filled teeth to retreatment or monitoring the condition without intervention (Fig. 9.2c). The many difficulties involved in investigating the most relevant clinical questions with an RCT design have resulted in a limited number of publications over the years. And most of them are investigating relatively trivial issues  but still without evidence-based answers even to these.
126.96.36.199 Systematic Reviews and Meta-Analyses
There are several approaches to summarize the scientific basis for clinical practice. In recent years, when thanks to developments of computer and IT technology, large amounts of data and literature can be both searched and retrieved within a very short period of time, so-called systematic reviews have become increasingly common. By definition the review must be conducted in a systematic way and contain at least four components:
Formulation of a clear question (or several clear questions)
Searching and identifying relevant research
Collecting and critically analysing included reports
Summarizing results, making conclusions and giving recommendations as to how to proceed in the clinical setting
Systematic reviews are a special type of review article, which can be considered to provide the highest level of evidence when several similar RCTs on the same clinical question are utilized. Meta-analysis is a specialized type of systematic review, where data are pooled for a quantitative rather than a qualitative analysis. This type of study can provide the highest level of evidence, if the report is limited to proper RCTs. However, a meta-analysis cannot give a higher-quality evidence than that which exist in the studies included in the analysis.
9.2.2 Assessing the Quality of Available Research
In assessing the scientific quality of a clinical research report, a number of factors are essential. These aspects sum up into an account of internal validity (the degree to which the results of a study are correct for the sample of patients being studied) and the extent of external validity (generalizability) (the degree to which the results of an observation hold true in other settings).
188.8.131.52 The PICO Concept
A good starting poin t to use for evaluating the quality of a RCT is the PICO concept (Fig. 9.3). It stands for population, intervention, control procedure and outcome measure. The PICO model can also be adopted for other types of studies both for planning and for evaluating individual studies, for example, when pursuing a systematic review. At each “letter” there are however many pitfalls that have to be avoided if the study is to produce results of high internal and external validity. PICO helps the researcher or evaluator to systematically evaluate all the phases of a study.
The PICO concept applied for a RCT on important outcomes following retreatment versus “no intervention” of root-filled teeth with asymptomatic apical periodontitis
184.108.40.206 Biasses in Clinical Research
“Bias” is the term for a process at any stage of inference tending to produce results and conclusions that deviate from the true condition systematically. The quality of studies is subject to the risk of being limited by numerous biasses. The problem affects all kinds of reports including the top articles in the evidence pyramid (prospective cohort studies and RCTs) (Fig. 9.1). Biasses are in four wide-ranging categories, viz. sampling bias, selection bias, measurement bias and confounding bias .
Sampling bias arises when the sample of patients is systematically different from those suitable for the research question or the clinical use of the information. For example, studies on the outcome of endodontic retreatment have exclusively been conducted in dental schools or specialist centres. An important question is if these teeth are representative of “root filled teeth with apical periodontitis” in general? Perhaps the teeth treated are a sample of “suitable teeth” for referral and treatment . When reporting a clinical study, it is always important to accurately describe the inclusion and exclusion criteria for the subjects included in the study.
Selection bias arises when comparisons are made on groups that differ in ways, other than the factors under study. Groups of patients often differ in many ways by age, sex, general health and severity of disease. If we compare the outcome of two groups that differ on a specific issue of interest (e.g. surgical versus non-surgical retreatment) but are dissimilar in any other way and this difference itself is related to the outcome of interest, the comparison between the groups will be biassed. Thus, little can be concluded from the results. In our example of surgical versus non-surgical retreatment, if “easy cases with easy access” (perhaps premolars and incisors) are more frequent in the non-surgical group, the outcome may be systematically better or poorer. Randomization is the best way to overcome these difficulties. The randomization procedures must then be performed without manipulation and be clearly described in the methods of the study.
Measurement bias arises, when the means or methods of measurement are different among the groups of patients. This is the reason why historical comparisons (data from other reports) often are invalid. Another problem may be the lack of common criteria for evaluating the outcome. For example, when comparing results of non-surgical and surgical endodontic procedures, there is no mutually recognized way to interpret “healing from “no healing” in radiographs. The problem with intra- and interobserver variation must also be handled in an appropriate way by using blinded and independent evaluators.
Confounding bias arises when two factors are associated with each other, and the effect of one is confused with or distorted by the effect of another, not measured or controlled, factor. For example, if survival of a group of teeth, which had a surgical retreatment, is compared with a group where non-surgical retreatment was conducted. Perhaps the result showed a significantly higher survival after 10 years in the non-surgical retreatment group. Yet in further analyses of the data, it was revealed that in the non-surgical group, a new crown was placed more frequently postoperatively than in the surgical group. Consequently, it may be that the placement of the new crown rather than the choice of treatment explained the observed difference in outcome.