Real-time monitoring of a video-assisted thoracoscopic surgery lobectomy programme using a specific cardiopulmonary...

To implement internal monitoring using a risk-adjusted model specific for video-assisted thoracoscopic surgery (VATS) lobectomy.

Retrospective analysis on prospectively collected data of 348 patients submitted to VATS lobectomy (August 2012-August 2014). Baseline and surgical variables were tested for a possible association with postoperative cardiopulmonary complications. Logistic regression and bootstrap resampling analyses were used to develop the risk-adjusted model to obtain the predicted morbidity of 50 consecutive patients (September 2014-November 2014). A risk-adjusted control chart was constructed to track down practice variation during this period. Patients were ordered by date of operation and assigned a score represented by the individual predicted morbidity: the plotted line goes up in case of absence of complications and goes down by the predicted morbidity minus 1 in case of complications. Over time, if outcomes are as expected based on the risk-adjusted model, the plotted line will tend to be close to zero.

Cardiopulmonary complications and in-hospital/30-day mortality rates were 14% (47 cases) and 1.8% (6 cases), respectively. Age (P = 0.006, coefficient 0.55, bootstrap frequency 76%) and predicted postoperative forced expiratory volume in 1 s (ppoFEV1) (P < 0.001, coefficient -0.38, bootstrap frequency 98%) remained independently associated with cardiopulmonary morbidity after logistic regression and bootstrap analyses. The following risk logit model for cardiopulmonary morbidity after VATS lobectomy was generated: -3.17 -0.038XppoFEV1 +0.55Xage. The risk-adjusted control chart showed a downward trend indicating a worse than expected performance in the audited period.

The present analysis offers a methodological template for VATS lobectomy that helps to evaluate the surgical programme. It aims to give a real-time monitoring with the possibility to confront the performance of the centre with the population-specific expectancies. Moreover, being reactive with time, this method allows early detection of underperformance and implementation of critical change in clinical practice.