In this investigation, neither a statistically significant evidence nor a clinically relevant difference was found to underline the superiority of perioperative antibiotic prophylaxis for preventing postoperative infective events in the case of routine removal of impacted wisdom teeth without local inflammation.
The development of antimicrobials paved the way for the age of modern medicine. Their effectiveness has allowed the performance of life-saving medical key procedures during the last 8 decades, e.g., joint replacement, cesarean sections, gastrointestinal surgeries, or oncological and immune system suppressing interventions securely and successfully [24]. However, besides an increased number of medical procedures, an aging society, and poor sanitation and hygiene in low and middle-income countries, inadequate and frequent use of antimicrobials destabilize a supposedly safe system [9, 25]. Thus, this study was conducted in the sense of resistance during prophylaxis and to promote the waiver of antibiotics whenever possible. Due to ambiguity, it has to be mentioned that further clinical trials are reported to be necessary [8]. Several published studies identified effects which significantly speak for [4, 14, 23, 26] and against [15, 27,28,29,30,31,32] antibiotic prophylaxis. Indeed, the necessity is still given to perform prospective, randomized, placebo-controlled, and practical clinical trials [23] to find a clear consensus about prophylactic antibiotic administration for healthy patients [8]. It has to be highlighted that the number needed to treat to prevent one SSI is rather high and lies at 143, according to a cohort study by Lang et al. [23]. Alternatively, present guidelines still support a perioperative antibiosis [35, 44], whereas the decision suggests to stay with the surgeon and observe each patient’s risk [35]. Therefore, it should be mentioned when considering antibiotic administration that a differentiation between not inflamed, clean–contaminated, and inflamed or even dirty conditions, such as pericoronitis, fistulas, infected cysts, or purulent secretion, has to be performed not only for the prevention of systemic [45] but also for local wound healing complications [35, 44].
This present study attempted to decrease the bias as much as possible by using a randomized, double-blind, placebo-controlled study design in clean–contaminated conditions, as performed in other studies over the last decade [14, 15, 28, 30, 31]. With split-mouth design less frequently shown in the literature [15, 29], patients acted as their controls, resulting in a higher homogeneity within the study population. As mentioned previously [31], the risk of opposite medical interaction was minimized with the prohibition of antibiotic intake 3 months before the start and the same waiting time between the first and second interventions. Furthermore, the mean age of 21 years [29] resembles the featured phase of prophylactic third molar intervention [33]. Similar but slightly older age distribution can be found with 26.4, 23.0, and 28.5 years in other studies [23, 28, 31].
Further in this study, only upper and lower third molar surgery cases were included at two appointments (18, 48 vs 28, 38). Although this is in contrast to others [15, 28, 31], it resembles the routine protocol at the study center and as previously performed [29]. Impacted third molars had to be of medium difficulty, preferably class B and class 2, according to Pell and Gregory [19], to achieve a balanced degree of osteotomy and similar surgery duration among the study population. Regarding the knowledge that surgical trauma potentially leads to a higher degree of tissue injury followed by an increased “inflammatory response” [34], a limited selection of three experienced oral surgeons was allocated in this study.
Perioperative antibiotic prophylaxis is the administration of antibiotics before, during, or after a diagnostic, therapeutic, or surgical operation to prevent infectious adverse effects [35]. Therefore, antibiotics are not thought to replace good surgical techniques; rather, they should accompany interventional procedures [35]. Within the oral cavity, mainly mixed infections occur. The ideal antibiotic should be non-toxic, easy to apply, and broadly effective against gram-positive, gram-negative, and anaerobic bacteria, such as amoxicillin [15], which is the most widespread in Europe and still resembles the key antibiotic in medicine [10]. In case of incompatibility, cephalosporins and macrolides/lincosamides are frequently prescribed [36]. Thus, correlating with others [15, 28, 30, 31] and according to the common empirically based clinical approach [10], the choice in this study fell on preoperatively initiated amoxicillin, orally administered for 4 days, as described by Payer et al. [21]. Although concerning the active ingredient, Sayd et al. [37] did not find a significant difference between amoxicillin–clavulanic acid and azithromycin in 108 patients, similar to Adde et al. [27] at 71 participants in the comparison of amoxicillin with clindamycin. Equally, as Reiland et al. [38] did regarding the application in a retrospective cohort study of 1895 samples, analyzing the difference per oral and intravenous administration. Iglesias et al. [39] suggested the noninclusion of clavulanate due to a significantly higher rate of gastrointestinal complications (5.5%) (diarrhea from Clostridium difficile infection) and resistance promotion [10]. The pre- or postoperative start of the medication regimen is also discussed in the literature [41]. Although Lopez-Cedrun et al. [14] could not find a statistically relevant difference in a sample of 123 patients, including a parallel-group study, we relied on the sufficient plasma levels suggested by Allegranzi et al. due to a 60-min preoperative antibiotic administration [41]. Thus, former recommendations [40, 41] and the recently published studies in this field [15, 28, 29, 31] were followed. Generally, a maximum duration of 24 h in surgery and a single-shot antibiotic use with a possible additional intraoperative dose are recommended, which in particular, focused on the prevention of resistance development [8]. This contrasts with the prolonged protocol used in this study, which however resembles a general procedure in oral surgery [15, 28]. However, this issue should not be further contentious. This study provided another important jigsaw piece to prove the nonsuperiority of antibiotic prophylaxis, opposite to the placebo medication in clean–contaminated sites in routine wisdom tooth surgeries as previously noted [21].
In this study, the primary outcome variable was the presence of SSIs. From our perspective, SSI means the occurrence of local inflammation at the extraction site with cloudy secretion and a tendency of propagation or the presence of an abscess with purulent secretion. In both cases, a patient needs dental care. Local inflammations were treated with disinfecting local irrigation, while abscesses underwent an incision with gauze drainage and further antibiosis. Dry sockets were not included in our clinical investigation because this clinical picture shows neither the signs of extraoral swelling, fever, and trismus nor a purulent secretion and may not result in a life-threatening situation. Thus, they cannot be clinically investigated when flaps are closed completely. The definition of SSIs varies in different reports, including swelling, pain, increased body temperature, or c-reactive protein levels [15, 28, 42]. Therefore, besides different study designs, it is challenging to directly compare the results of this investigation with others. Overall, we identified an infection rate of 11% within our study population, which is higher than the reported 0.03% by Milani et al. [31], 1% by Xue et al. [15], 4% by Lopez-Celdrun et al. [14], and 5.7% by Lang et al. [23]. In this context, noting especially the studies concerned with the lowest values [15, 31], potential sources of bias should be discussed. On one hand, the sample size calculation was not described in detail and may be too low with 20–30 in each group, and on the other hand, postoperative assessment methods were not transparently described [15, 31]. Concerning the effect of antibiotics versus placebo in our study, no significant difference occurred between the two treatment groups. In detail, eight patients were recognized in the placebo group, which is in accordance with the works of Bezerra et al. (11.76%) [29], Lopez-Celdrun et al. (12.5%) [14], Artegoita et al. (12.9%) [17], Monaco et al. (14.82%) [4], and Lacasa et al. (16%) [16]. The three inflammations of our EG were also rather confirmed with the studies of Monaco et al. (3.12%) [4], Lang et al. (5%) [23], Lacasa et al. (5.3%) [16], and Milani et al. (6.2%) [31]. In the present trial, the infection rates seem to be multifactorial. First, only third molars without any sign of inflammation were removed to observe the specification of clean–contaminated situations on the day of surgery. Signs of inflammation change the conditions toward contaminated or even dirty conditions, which modifies the antibiotic regime [35, 44]. Second, the patients’ preparation with aseptic mouthwashes and measures against the progression of swelling, such as preoperatively administered corticosteroids and gentle intraoperative tissue management, may impact infection prophylaxis instead of perioperative antibiotic administration.
The nonsuperiority of the antibiotic regime also seems to be underlined by the presented results concerning trismus and swelling. Although in a study of 293 participants [30], perioperative antibiosis showed a statistically significantly lower amount of swelling compared to placebo, it was equally found that without antibiotic use, the swelling also significantly decreased until postoperative day 7 [42]. Furthermore, Xue et al. [15] reported no significant difference between amoxicillin and placebo administration 2 and 10 days (p = 0.110; p = 1.000) after the third molar surgery, as similarly done by Lacasa et al. [16] from the third postoperative day. Furthermore, there is no significant difference between EG and CG in analog face measurement and an additional digital comparison of face scans, with mean volumes of 14.79 ml versus 13.87 ml on postoperative day 1 and 5.47 ml and 4.59 ml on day 7. Regarding those parallel measurements, which to the best of our knowledge have never been performed in this form before, we believe we can provide an alternative digital method that is worth continuing in future investigations and sufficient evidence that speaks against the prophylactic use of antibiotics. The latter is the same underlined by trismus development, not reflecting any significant difference at any postoperative time point (d 1: p = 0.399; d 7: p = 0.570), as reported in an earlier RCT [31] at postoperative day 4 and 7.
Patient-centered outcomes are an essential source of information in medicine. In a recent review [5], pain after the third molar removal was highlighted as having the most influence on patients’ quality of life. Besides clinical parameters (pus secretion, swelling, and trismus), patients’ subjective impressions concerning bleeding, swelling, and pain were investigated. Neither swelling nor pain revealed any significant result between the EG and CG, which correlates with the course of the needed pain medication. Milani et al. [31] reported a similar finding with no significant difference in pain between the groups with antibiotics and placebo medication on postoperative days 4 and 7. An additional argument can be presented with this knowledge, and straightforward advice is possible when patients ask for antibiotic prophylaxis, believing in rapid recovery with less pain. From our point of view, these results also justify our opinion of antibiotic reluctance concerning alveolitis, as no difference occurred in the patient-related outcome measures, which were evaluated for postoperative pain. The only exception presented is the “bleeding” parameter on the day of surgery. There is evidence that antibiotics administered along with oral anticoagulants lead to an increased incidence of postoperative bleeding [43]. However, the effect in our study cannot be explained because only healthy subjects were included here. Thus, this parameter seems to need further investigation.
One minor limitation of this study is the overall dropout of 9 patients. Anaphylactic reactions and gastrointestinal complaints resembled the minority opposite, a lack of motivation to return. An 8.5% of patients were lost in the first study phase, whereas a further 6.8% missed the second intervention. Nevertheless, the calculated limit of necessary samples and surgeries was fulfilled, from which it is concluded to be able to present reliable results. Furthermore, compared with two other randomized clinical trials [28, 29], our case number (50 patients with 100 surgeries) exceeds that of Bezerra et al. [29]. They only included 34 patients in their split-mouth study, similar to Artegoita et al. [28]. At the expense of the total case number, but to the benefit of the unbiased split-mouth analysis, the decision was made to only include patients (number, n = 50) who had both interventions completed, resulting in 100 interventions (Fig. 1). Another limitation of this study might be the relatively short follow-up period of 7 postoperative days. The thought was to follow our routine clinical protocol [22] and to combine final check-ups and suture removal in one appointment to increase the patient’s compliance to return to follow-up. However, this comparatively short period might have influenced the low rate of SSIs. An infection may occur, in the case of delayed wound closure beyond the 7-day follow-up, which was not investigated in this study. Some studies described the usefulness of prolonged investigation periods of 10 days up to 8 weeks [14, 15, 28], whereas others have applied the same as in this study [30, 31, 37]. However, in our opinion, a delayed infection due to wound dehiscence may be induced by shifted food residues, which do not have a causal relationship to the effect of a perioperative antibiosis. For a third limitation, 26% of digital face scans were not usable. This happened due to technical problems (e.g., caused by motion artifacts or artifacts due to long beards and irritating hairstyles). As a result, cases where no superimposition was appropriate were ruled out. Nevertheless, the case number seems to be balanced with a lack of 13 in the EG and 13 in the CG. Although it could be shown that the application of this tool is more than equal to the analog measurement, the technical aspects, including an increased effort of time, equipment, and costs, should be considered.