Evidence-based dental practice: searching the literature and writing a report

Writing the Discussion (Leake et al., 2001):
A discussion should be divided into subsections that each analyzes some aspect of the studies. Some common ways to organize the subsections are: by research design, by methods used, by population studied, by results, or (less common) chronologically.   This excerpt from Dr. Leake’s paper is organized according to the five main methods used to prevent root caries in the seven studies. Notice that the same information is given in each sub-section:

(1) how many studies
(2) their design and the strength of that design
(3) their findings
(4) their limitations/strengths

7.3          Prevention of Root Caries
Seven studies were retained which provide evidence that root caries is preventable. These methods include:

• Water fluoridation
• Fluoride dentifrice
• Topical APF gel
• Fluoridated mouthrinses
• Saliva stimulation

7.3.1      Water Fluoridation
(1)Fluorides in water have been demonstrated as effective in three studies (Burt et al., 1986; Hunt et al., 1989; Stamm et al., 1990). (2)All of the studies are descriptive which would ordinarily mean that the evidence is of level III. However, since they are dealing with residence history, the studies might be considered as longitudinal which would raise their quality measure to Level II.  (3)All find a difference between residents in fluoridated and non-fluoridated communities, (4)but in Iowa (Hunt et al., 1989) the difference is only apparent for those who have lived 41 or more years in the fluoridated community.

7.3.2      Fluoride Dentifrice
(1)Only one study (Jensen & Kohout, 1988) has demonstrated the effect of dentifrice on root caries among older adults in Iowa, USA. (2)The study is a randomized clinical trial and therefore of strong design (4)but of only one year duration thus limiting its validity.  (3)Over the span of the study a 67 percent lower incidence of root caries was reported in the dentifrice group, with a 12-month caries incidence of 0.14 for the dentifrice group and 0.43 for the control group.

Another Sample of a Discussion, with Conclusion (Peressini, 2003):
A Discussion section should interpret the results in order to answer the question.

(1) It should offer an overall evaluation of the strength/weakness of the entire body of studies and say why.

(2) It should clearly list the major strengths and/or weaknesses of the studies in the Results section. In the excerpt below from Dr. Peressini’s paper, notice that the items in the list are separated with semi-colons (;).  This is a very common strategy. Notice too that she begins with a transition (‘In particular’) that moves the reader from the general evaluation to the specific details.

(3) Make recommendations for future research, especially in cases where the studies have not been very strong, or have not been successful in answering the question

(4) Come to a conclusion (usually introduced with “In conclusion”) that gives the answer to the question: Does the evidence in the literature allow us to make a recommendation for best practice and if so, what is that recommendation? In the case of Dr. Peressini’s review, she felt the evidence was too weak to make a recommendation for practice; instead, she makes recommendations for future research.

Discussion
(1)According to this literature review, the evidence for pacifier use as a risk factor for ECC is generally weak because of inconsistencies in study methods. (2)In particular, the definition of  ECC varied among the studies or was not stated; the criteria used to identify carious lesions varied among researchers or were not stated; other factors that might have been causally related to ECC were not consistently examined; the factor of interest in this review, pacifier use, was inconsistently described among studies, in that the studies did not clearly report duration and frequency of use, past or present use, or use of sweetening; and the age of the children varied among studies. Accordingly, it is difficult to compare the often conflicting and ambiguous results of these studies.
.... [section omitted]
(3)Future research investigating pacifier use as a potential risk factor in ECC should employ stronger study designs. Some of the pitfalls and weaknesses of previous studies could be avoided by implementing the following recommendations: [a list of recommendations follows]
 (4)In conclusion, the evidence does not suggest a strong or consistent association between pacifier use and ECC. More studies (with either a cohort or a case-control design) that control for other relevant variables and that clearly investigate the role of the pacifier alone must be conducted before a clear recommendation on this subject can be made.