Comparison of endoscope-assisted and microscope-assisted type I tympanoplasty; a systematic review and meta-analysis

In this study, we aimed to make a comparison between endoscopic and microscopic type I tympanoplasty in relation to the duration of surgery, outcomes and complications through a meta-analysis.

Inclusion criteria for our meta-analysis were as follows: (1) journal articles published in English; (2) articles concerned with TM perforation especially due to COM; (3) studies describing myringoplasty or type I tympanoplasty, and comparing surgical outcomes of the endoscope with the microscope; (4) either temporalis fascia or perichondrium as a source of the graft tissue; (5) graft placement method via underlay technique.

Exclusion criteria for our meta-analysis were as follows: (1) articles describing other types of otitis media (e.g. acute OM or OM with effusion) or other pathologies (e.g. cholesteatoma or middle ear tumours); (2) the aid of an endoscope holder; (3) cadaver studies; (4) animal studies; (5) irrelevant publications to our study.

The outcome measures, which we considered in terms of efficacy and safety, were average operative time (intraoperative outcome); average postoperative air-bone gaps (ABGs) improvement and graft uptake rate (primary efficacy outcomes); complications rate (secondary safety outcomes).

We performed a systematic search for all available studies comparing surgical outcomes of the endoscope with the microscope in the databases of PubMed, Cochrane library Ovid, Scopus, Google scholar, and ClinicalTrials; dating from inception until 22 November 2019. We used the following keywords (in different combinations): microscopic, endoscopic, type I tympanoplasty, myringoplasty, chronic otitis media. Review articles and bibliographies of each identified study were searched for additional references that may contain further related studies.

Abstracts of articles identified using the above search strategy were reviewed; articles that appeared to fulfill the inclusion criteria were retrieved in full. We excluded duplicate records and irrelevant reports at this stage. When there was a doubt, a second reviewer assessed the article, and a consensus was reached.

Data were independently extracted onto a previously edited Excel table by two reviewers and cross-checked; any discrepancies were resolved by consensus. For the meta-analysis, we retrieved the following information: author, year of publication, study design, number of patients, and outcomes regarding efficacy and safety.

Data entry, processing and statistical analysis were carried out using Review Manager 5.3 (RevMan 2014) [10]. A meta-analysis was performed to calculate direct estimates of each treatment technique. Interventions for patients, who achieved favourable outcomes, were pooled to evaluate efficacy that was measured by standardized mean difference (SMD) with a 95% confidence interval (CI) for operative time and postoperative ABG improvement; and odds ratio (OR) with 95% CI for graft uptake rate. In addition, interventions for patients, who reached serious adverse events, were pooled to evaluate safety that was measured by OR with 95% CI for complications rate.

According to the heterogeneity of treatment effect across trials using the Chi² test results and I²-statistics; a fixed-effect model (P ≥ 0.1) or random-effects model (P < 0.1) was used. In addition, we used a random-effects model for subgroup analysis.

We assessed the publication bias across studies using the funnel plot method for each pooled analysis that included more than or equal to 10 studies [11].

Figure 1 represents the PRISMA flow diagram for the review process and study selection. We found 150 records by searching the database; of these, sixty records were left after removing the duplicates and after the exclusion of ninety records based on the title and the abstract review. We searched 60 articles for eligibility by full-text review; 15 articles cannot be accessed or obtain full-text; 10 studies were reviews and case reports; eight studies did not describe the functional outcome; the desired procedure was not used in five studies leaving 22 studies [12‐33] that met all inclusion criteria.

Table 1 shows the demographic characteristics and clinical data of all 22 studies [12‐33]. Regarding their design, these 22 studies included ten randomized controlled trials, six prospective comparative studies, five retrospective comparative studies and one retrospective cohort study. The enrolled studies were published between 2008 and 2019. The total number of interventions was 1578 interventions; with 766 interventions in the endoscopic group, and 812 in the microscopic group.

We found nineteen studies that reported operative time with a total number of interventions (n = 1414). The average operative time was (77.7 ± 24.5) min for the ET group and (91.7 ± 18.8) min for the MT group. In the pooled analysis (Fig. 2A), endoscopic tympanoplasty (ET) showed a highly significant decrease in mean operative time compared to microscopic tympanoplasty (MT) (SMD: −1.33; 95% CI −1.95 to −0.72; p < 0.0001). However, highly significant heterogeneity (I² = 96%, p < 0.00001) and publication bias were found (Fig. 2B).

Fifteen studies exhibited ABG improvement with a total number of interventions (n = 1135). Based on the pooled analysis (Fig. 3A), ET showed a non-significant difference in mean ABGs improvement compared to MT (SMD: 0.03; 95% CI −0.33 to 0.39; p = 0.87). However, highly significant heterogeneity (I² = 89%, p < 0.00001) and publication bias were found (Fig. 3B).

All 22 studies included data about graft uptake rate with a total number of interventions (n = 1559). Accrued results (Fig. 4A) showed that ET was as effective as MT (89.8% vs. 90.2%; OR: 0.95; 95% CI 0.68–1.34; p = 0.79). In addition, no heterogeneity (I² = 0%, p = 0.99) or publication bias was found (Fig. 4B).

Various postoperative complications were retrieved from enrolled studies. We analyzed the incidence of postoperative complications, with a particular focus on the infection (wet ear), wound gap, visible scar, nausea and vomiting. Incidence of postoperative complications was reported in 10 studies.

Data regarding the infection (wet ear) were available from six studies with a total of 486 interventions. Three studies reported data on the wound gap with a total of 210 interventions. Based on the pooled analysis, ET showed a non-significant difference compared to MT in terms of the rates of infection (6.3% vs 7.3%; OR: 0.94; 95% CI 0.46–1.94; p = 0.87) and wound gap (0.9% vs 6.3%; OR: 0.25; 95% CI 0.05–1.30; p = 0.10). Also, no heterogeneity was found.

Data regarding visible scar, we found three studies with a total of 220 interventions. Two studies reported data on postoperative nausea and vomiting, with a total of 160 interventions for each. In the pooled analysis, ET showed a highly significant decrease compared to MT in terms of the rates of visible scar (0% vs 72.7%; OR: 0.01; 95% CI 0.00–0.03; p < 0.00001), nausea (36.3% vs 67.5%; OR: 0.27; 95% CI 0.14–0.53; p = 0.0001), and vomiting (12.5% vs 43.8%; OR: 0.18; 95% CI 0.08–0.41; p < 0.0001). In addition, no heterogeneity was found.

The overall effect of pooling the previous five subgroups showed that the rate of complications is significantly lower in ET than in MT (8.8% vs 32%; OR: 0.24; 95% CI 0.12–0.49; p < 0.0001) (Fig. 5).

With regard to the graft uptake, no significant difference was found between ET and MT. Similar results were also reported in previous meta-analytic studies of Tseng et al. [34], Lee et al. [35] and Pap et al. [36]. In this meta-analysis, we selected included studies that used similar operative techniques such as grafting material (temporalis fascia or perichondrium) and the graft placement method by underlay technique to obtain more accurate results about graft uptake rate.

Audiological results resembled graft uptake outcomes. Not unexpectedly, ABGs improvement demonstrated no significant difference between ET and MT, despite discrepancies in hearing evaluations. Remarkable TM closure rates between ET and MT may explain comparable audiological outcomes. However, potential publication bias with highly significant heterogeneity may have negatively impacted the integrity of this analysis. Tseng et al. [34], Lee et al. [35] and Pap et al. [36] reported similar analytic results.

In agreement with Lee et al. [35] and Pap et al. [36], another significant advantage regarding ET is that the operative time for ET was significantly shorter than for MT. The surgeon's experience and the learning curve generally have an impact on the operative time. However, MT consumes more time due to frequent manipulation of the patient’s head or repeated microscope adjustment for a better view, using the post-auricular approach, or performing canaloplasty and curettage. According to Hsu et al. the relatively short time required for surgery and under anaesthesia results in significantly fewer medical resources expended on ET and decreased complications from prolonged exposure to anaesthesia [37]. In our meta-analysis, the analysis for operative time data suffers from significant heterogeneity and publication bias.

Characteristically, ET is advantageous concerning safety, minimal invasiveness and the rate of complications. Because of a wide field of view with magnification, ear surgeons have obtained minimally invasive endoscope-assisted tympanoplasty accompanied by minimal complications. In our meta-analysis, we focused particularly on the following complications: the infection (wet ear), wound gap, visible scar, nausea and vomiting. No significant difference was found between both techniques with regard to infection and wound gap, but there was a highly significant decrease in the rates of visible scar, nausea and vomiting as well as the overall complications rate in favour of ET. Postoperatively, the wet ear results from a severe middle ear infection [37] and the wound gap following suture removal from early loose stitches [27] rather than the procedure itself. In the present study, a meta-analysis of cosmetic results through the presence or absence of a visible scar revealed that the endoscope was definitely preferred over the microscope. For ET, the transcanal approach to the middle ear and smaller incision with minimum tissue dissection for harvesting a graft lead to early wound healing and less scarring on the graft donor site [16, 18, 26]. Besides, avoiding the post-auricular route reduces the incidence of auricular displacement and asymmetry of the pinna yielding better cosmetic outcomes [12, 14, 19, 22]. Similar to meta-analytic results of visible scar, the rates of nausea and vomiting were significantly lower after ET than after MT. Nausea and vomiting are unpleasant events and are associated with patient discomfort and dissatisfaction during postoperative recovery [38]. These two adverse events require the administration of various treatment modalities and consequently can expand recovery room time, increase nursing care requests and the duration of hospital stay, and can further increase total healthcare costs [39].

In concordance with our results regarding complications, Lee et al. [35] reported that wound problems of ET were significantly lower than those of MT, but there was no significant difference between ET and MT regarding wet ear.

Our meta-analysis possesses several strengths. Our findings were comparable to those presented in the previously published meta-analyses [34‐36]. Unlike the preceding meta-analytic publications about the same topic, the present study included more randomized controlled trials and other comparative studies that the search had yielded. Aiming at a better assessment of efficacy and safety, we also added more parameters for comparison. As much as we could, we held some variable risk factors constant, such as the source of the graft tissue and the graft placement method, to reduce clinical heterogeneity.

Admittedly, our meta-analysis has a few limitations. There was a noticeable variance in the other risk factors influencing surgical outcomes (e.g. the age of patients, and the size or location of TM perforation). This variance resulted in raising concerns about clinical heterogeneity. Furthermore, publication bias with highly significant heterogeneity could limit the integrity of our analytic results regarding operative time and ABGs improvement. Nevertheless, this study provided results that may be beneficial to decision-making and outcome prediction in patients receiving ET.