Learning curve in minimally invasive mitral valve surgery: a single-center experience

From October 2014 to January 2019, a total number of 204 patients underwent minimally invasive mitral valve surgery (MIMVS) at our institute met the inclusion criteria. The operations are performed by a single surgeon who were specialized in adult cardiac surgery before starting the minimally invasive program. We excluded patients with more than moderate aortic regurgitation, concomitant aortic valve disease requiring operation, a history of right chest operation or chest irradiation, severe aortoiliac stenotic diseases and prior cardiac surgery. Preoperative, operative and postoperative data was collected prospectively and analysed.

The research was approved by the ethical board of the University of Medicine and Pharmacy at Ho Chi Minh City, number 141/DHYD-HDDD, on April 11th 2018.

All operations are performed by a single surgeon who is experienced in mitral valve surgery via conventional sternotomy.

The patient was placed on the supine position with a cushion under the right scapula to facilitate exposing the two atria. Single lumen endotracheal tube intubation was used. Cardiopulmonary bypass was set up with cannulation of the femoral artery and vein through a small incision in the right groin. A 4–5 cm skin incision was made parallel to the anterior axillary line. A video camera was inserted through a 5 mm port in the third right intercostal space. The transthoracic Chitwood aortic cross-clamp was inserted and aortic clamping was performed. Two liters of Custodiol HTK solution was delivered antegradely into the aortic root through a long metal cardioplegic needle and repeated every 120 min if necessary. A left atriotomy is performed and a sump suction is inserted to keep the surgical field cleared. A left atrial retractor was used to expose the mitral valve. We then assessed the mitral valve for the feasibility of repairing, with the minimally invasive approach, nearly all mitral plasty techniques could be applied, in our center, we regularly used the plication technique, artificial chordae, loop technique [8], a rigid annuloplasty ring was implanted to support the repair. If the valve was severely damaged and repair was not achievable, mitral valve replacement would be performed. Transesophageal Echocardiography (TEE) was used to control the result of the operation.

In case of tricuspid valve repair, the annuloplasty with a tricuspid rigid ring (Carpentier – Edwards ring, Edwards Lifesiences, Irvine, CA, U.S) was used. Concomitant Maze procedure was performed with a long monopolar radiofrequency malleable probe (Fig. 1).

Based on Wu et al.’s criteria for technical failure [6], we lightly modified the definitions to match our population. Technical failure was defined with the presence of one or more of the events below:

Continuous variables are expressed as mean ± SD and categorical variables are expressed as proportion. Early mortality and complications together with mid-term results, including mortality, recurrent rate and reoperation rate are collected and analysed.

For the learning curve, according to Holzhey et al. [5], we divided patients into groups of 50 successive cases to analyse the difference of the following variables between groups using the χ2 test:

The trend of these variables over the years was also evaluated.

The cumulative sum (CUSUM) failure analysis was used to determine the number of cases needed to overcome the learning curve. This method has been described recently to generate the learning curves for new techniques [6, 9]. We applied the statistical principles from Rogers et al. [10]. CUSUM was defined as Sn = ∑(Xi − p0), where Xi = 0 for success and Xi = 1 for failure, p0 is the target reference. In this study, p0 was chosen at 0.1, meaning the acceptable composite failure rate was 10%, and a p1 = 0.2 was also as an unacceptable complication rate according to previous publications [5, 6]. Based on p1 and p2, control boundaries were calculated and drawn (false-positive α = 0.05, false-negative β = 0,05). When the curve crosses the upper boundary, the surgeon needs to revise his technique and result. When the curve crossed the lower boundary, the complication rate of the surgeon is equal or below the acceptable rate, meaning the learning curve is achieved. We analysed the CUSUM analysis for the whole group as well as the replacement and repair group separately.

We analysed the trends of the following variables over the years (Figs. 2, 3, 4 and 5):

Applying Holzhey’s method [5], we divided patients into groups of 50 consecutive operations to assess the trends of these variations, the concomitant procedure rate and composite technical failure rate Table 6.

This statistical analysis showed a gradual decrease in the important variables toward a positive results trend in MIMVS as the accumulated number of cases increased in time (Figs. 6 and 7).

The overall CUSUM learning curve showed that at the level of 75 cases, the curve started to contact the blue lower boundary and crossed the boundary at the level of 100 cases. However, the learning process seemed to be faster for mitral valve replacement. In this case, CUSUM curve contacted the lower boundary at 40 cases and passed the boundary at 60 cases. On the contrary, the learning curve of mitral valve repair appeared to be longer, it contacted the blue boundary at the 75 cases but fluctuated significantly. Unfortunately, the number of cases were not enough to assess the point where the curve passed the blue line, but at least at 90 cases, the curve had not passed the lower boundary.

This study only reported a single surgeon’s experience in one center and the number of cases were still limited. This was not a comparative design, as a result, a definitive conclusion could not be made.