The findings of the present meta-analysis showed no statistical difference between single-visit and multiple-visit treatment modalities. This means that there is no difference between the treatment regimens, or, alternatively, the sample size is too small to demonstrate a true difference. Trope et al. [66] demonstrated a radiographic healing rate of 64 % in single-visit versus 74 % in multiple-visit treatment. However, the difference was not significant. They calculated the need for a sample size of 354 teeth in each treatment group for demonstrating significant differences based on power set at 80 %. The number of treated teeth in the present meta-analysis, 351 in single-visit and 315 in multiple-visit treatment, approached the calculated number by Trope. A larger sample size is desired, giving a higher power in the statistical analysis, but there are lack of randomized clinical trial in multi-centre settings in the field of endodontology [94]. In this meta-analysis 65 teeth out of 351 (18 %) in the single-visit group and 63 out of 315 (20 %) in the multiple-visit group were not healed (Table 5), given a difference of 1.5 % between the two treatment groups. According to calculations in Stata, 10844 teeth are needed in each treatment groups to achieve a significance different between the two treatment regimens at this low level.
The present meta-analysis included clinical follow-up studies, and was not limited to randomized clinical trials. Clinical follow-up studies are lower in the hierarchy of evidence. Only three [50, 55, 57] of the eight included studies in the current meta-analysis were randomized controlled trials. Of the 666 teeth included in present meta-analysis, 434 were included in the three abovementioned RCTs. The remaining five studies were follow-up studies. A systematic review and meta-analysis by Sathorn at el. [5] in 2005 aimed to answer if single-visit treatment results in a lower healing rate then multiple-visit treatment with calcium hydroxide dressing. They included only three studies [4, 18, 66], all of them ordinary follow-up studies and concluded that there was no statistically significance in healing between single-visit and multiple-visit root canal treatment, which is in agreement with the findings of the present study with an extension of included studies.
Lucena et al. reported quality assessment of randomized clinical trials limited to English language articles published in the field of endodontology between 1997 and 2012 as poor. Sample description, no adherence to Helsinki Declaration regarding ethical aspects, lack of information on randomization methods and no previous sample size estimation or inappropriate sample size estimation and inappropriate reports of statistical analysis contributed to the weakness and limitations [95]. However, the present study showed no difference comparing radiographic outcome of RTCs and non-RCTs.
Clinical trials in endodontics cannot easily be completely blinded. For example, when comparing single-visit versus multiple-visit treatment, both patients and dentists are inevitably aware of their randomized assignment. However, a blinded assessment of outcome can often be achieved, as recorded in six of the included studies in the present meta-analysis [4, 18, 50, 55, 57, 66, 73]. In six of the eight included studies, the evaluation of radiographic outcome was carried out blinded and not performed by the operators. One study [18] mentioned that the radiographs were judged by the operators involved in the treatment, however the examiners did not know whether the tooth belonged to the single-visit or the multiple-visit group. Dorasani et al. [81] did not clearly describe the radiographic evaluation procedure. On the other hand, many investigators believe that the data obtained are less reliable than those obtained from double-blind trials [96].
Although published literature is the main source of evidence for clinical decision-making, there are still concerns that many studies, in particular those with negative results, are not published. Since there is a tendency that studies with positive results would more likely be published than studies with negative results, publication bias is a concern for most meta-analyses and systematic reviews. There was no test of publication bias performed in our meta-analysis, due to the very small number of identified randomized controlled trials. The literature search in PubMed and EMBASE databases used most relevant terms related to the null hypothesis, thus, we claim that a possible selection bias was minimized. Nonetheless, language limitations and size of apical lesion are examples for limitations that can contribute to selection bias.
In all included studies, the patients were randomly assigned to each treatment groups, and root canal treatment was performed according to standardized protocols, using rubber dam and establishment of asepsis. The teeth in five of the included studies [4, 50, 66, 73, 81] were treated by one operator, in one study by two operators [18] an in another study by four operators [57]. In one of the included studies [55] the number of operators were reported to be more than one. However, it was not exactly clear how many persons that had been involved in the treatment processes. A low number of operators in each study may increase the internal validity of each of them.
Five of the included studies used the PAI-index [50, 55, 66, 73, 81], which is reasonably accurate, reproducible and able to discriminate between sub-populations. According to Ørstavik et al. [38] PAI-score 3 was seen as disease, by which changes in bone structure could be detected with some mineral loss. It is applicable to use the PAI-score for the analysis of periapical radiographs in clinical trials, epidemiological and in retrospective studies of endodontic outcome [38]. Thus, included studies using the PAI-index should be comparable. By doing separate calculations on the studies using the PAI-index for final assessment of outcome, the overall OR were close to the overall OR for all included studies. As such, the different indices are relatively close when it comes to evaluation outcome on intraoral radiographs reported on a dichotomous scale as healed or not-healed. The important aspect for the present meta-analysis is that, regardless of the index used, the same index was used for both study groups in each study.
In the present meta-analysis, only teeth with definite preoperative apical periodontitis were included to make the comparison between the treatment protocols viable. There was no evidence of a difference, in terms of radiological healing between teeth treated in a single-visit compared with teeth following a multiple-visit protocol (OR=1.10). Our finding is consistent with the results of a meta-analysis from 2005 where only three of the eight identified studies in this meta-analysis were analyzed [5].
The observation time in four of the included studies was 12 months [55, 66, 73, 81], in two of them 24 months [50, 57] and in the remaining two studies almost five years [4, 18]. However, it has been concluded that simple calculation of success rates would overestimate the chance of complete periapical healing within the first years after therapy, but underestimate it over longer observation periods [97]. It has also been demonstrated that there is a gradual increase of periapical healing over a period of four years [4, 18]. Several studies choose a 12-months follow-up evaluation as an end-point due to the resource-intensive nature of clinical studies and difficulty of controlling patient loss over time [98-100]. By using the PAI-index, evidence of periapical changes in bone density associated with healing should be apparent at 12 months and longer observation times might not be necessary [101]. The 1‐year follow‐up was first recommended by Ørstavik [102], who showed a peak in incidence of healing or emerging chronic apical periodontitis (CAP) at 1 year. Risk assessments at 2, 3, and 4 years did not indicate an added risk that filled roots developed CAP during this period. Although complete healing of preoperative CAP in some instances required 4 years to heal, initial signs of healing (although incomplete) were visible in at least 89% of all healed teeth after 1 year. The fact that the included studies had a follow-up time of 12 month or more adds to the validity of the study.
Except for the study by Paredes-Vieyra & Enriquez [50], which presented a significant risk (odds ratio of 3.5) in favour of single-visit endodontics, all other included studies showed no significant difference between the treatment modalities. The same finding was reported in a meta-analysis by Schwendicke & Gostemeyer [44]. In this study, the aim was to study several factors such as pain and flare-ups related to one-visit versus multiple- visit root canal treatment. Their main inclusion criteria was randomized controlled trials or controlled trials without signs of selection bias and studies comparing single-visit with multiple-visit root canal treatment in permanent teeth with closed apices, regardless of the preoperative condition. In two other systematic reviews, the radiographic healing in single-visit root canal treatment appeared to be slightly more effective than multiple-visit treatment, however without reaching statistical significance [5, 7].