The term “failure to rescue” (FTR) was first described in 1992 by Silber et al. (1) and refers to the death of a patient after presenting a major postoperative complication. This metric has been proposed as an alternative parameter to postoperative morbidity and mortality rates to assess quality of care and to evaluate the performance of a surgical service or a hospital (2). For this reason, some institutions have already introduced FTR as a complementary indicator of surgical quality.

According to Silber et al. (1), while patientś features determine the occurrence of postoperative complications, hospital characteristics are associated to FTR. Thus, several studies (3–5) have related high rates of FTR to hospital factors such as a low volume of surgeries or high nurse-to-patient ratios. Additionally, according to Farjah et al. (6) the variation in mortality rates after pulmonary resection for lung cancer among hospitals is more related to the ability to rescue complicated patients than to the occurrence of complications.

On the other hand, although FTR is closely related to the ability to detect and treat complications early based on hospital features, some studies suggest that there are some patientś intrinsic factors that may increase the risk of FTR after complex surgeries with high rates of complications (7–11). Our hypothesis is that FTR is derived from a combination of patientś intrinsic factors and hospital features such as hospital volume.

Since FTR has been poorly investigated in thoracic surgery (2, 6) and no studies have been based on data obtained from a prospective nationally representative registry, we aimed to determine the incidence of FTR after anatomical lung resections and to investigate risk factors associated with FTR in a prospective nationwide multicenter setting.

A total of 3,533 patients were recruited. Globally, 2,435 patients who had no complications and 737 patients who experienced minor complications were excluded from the analysis. The final sample consisted of 361 patients (10.2%) who had any major complication. Of these, 59 could not be rescued. Therefore, incidence of FTR was 16.3%.

Univariate analysis of patient characteristics, surgical features, and hospital volume for the dependent variable of FTR following pulmonary anatomical lung resections are shown in Table 1.

Table 2 shows the results of the stepwise multivariable logistic regression analysis (dependent variable: rescue failure). In the final model, variables associated with FTR were: ppoDLCO% (OR, 0.98; 95% CI, 0.96–1; p = 0.067), cardiac comorbidity (OR, 2.1; 95% CI, 1.1–4; p = 0.024), extended resection (OR, 2.26; 95% CI, 0.94–5.41; p = 0.067), pneumonectomy (OR, 2.53; 95 CI, 1.07–6.03; p = 0.036) and hospital volume < 120 cases per year (OR, 2.53; CI 95%, 1.26–5.07; p = 0.009).

FTR rates among patients with cardiac disease history reached 19% and extended resection was associated with a 28.9% FTR rate. While 35.9% of patients who experienced any major postoperative complications following pneumonectomy died. Moreover, FTR rates in hospitals with low volume (<120 cases per year) was 20.3% against 10.1% in high-volume hospitals (≥120 cases annually).

The model showed acceptable levels of sensitivity and specificity: the area under the curve (95% CI) was 0.72 (0.64–0.79) (Figure 1) with a good degree of calibration (Hosmer-Lemeshow test value (p = 0.739).

The analyses of the nationwide prospective Spanish registry found that the incidence of mortality among patients who experienced any major postoperative complication following anatomical lung resection reaches 16.3%. This study shows similar failure to rescue rates after lung surgery to those previously reported in single institutional series (15).

In the current study, we identified several patient factors that combined with hospital volume may predict failure to rescue after anatomical lung resection. Prior research examining rescue failure rates in thoracic surgery focused more on characteristics of the institution rather than patient features (6, 16–19). In this regard, we found that FTR rates were 2-fold higher in low-volume hospitals in comparison to high-volume centres. Farjah et al. (6) also reported an important variation in FTR rates across hospitals. In their multicentre study based on the Society of Thoracic Surgeons General Thoracic Surgery Database, the risk of dying after major complications increased by 3-fold (20% vs. 6.9%, p < 0.001) across hospitals. Similarly, Tran et al. (19) reported that high-volume centre status was associated with reduced odds of mortality after re-intervention compared to low volume centres. Furthermore, Sanaiha et al. (20) found that high hospital lobectomy volume and minimally invasive approach decreased the odds of mortality after cardiovascular complications (myocardial infarction, cardiac arrest or pulmonary embolism). Additionally, minimally invasively approached patients at high-volume institutions had the lowest odds of all-cause mortality (OR 0.27) and myocardial infarction (OR 0.57). Several factors may contribute to such vastly different outcomes. According to Ghaferi et al. (21) teaching status, bed size and increased nurse-to-patient ratios may influence outcomes in patients undergoing pancreatectomy. Meanwhile, Ward et al. (22) suggested that improved intensive care services, rapid response teams and availability of personnel may have an important role in FTR outcomes. The combination of these factors at high-volume centres, may allow for an earlier detection and treatment of complications, thus rescuing patients from mortality; while, low-volume hospitals may lack these capabilities, which ultimately may explain the observed differences in outcomes.

However, although the hospital components are important to understand how to improve on these shortcomings, our purpose was to determine whether any inherent patient factors in combination with hospital volume were associated with mortality after postoperative complications. By identifying the high-risk patients, our efforts could then be focused on how to potentially prevent the occurrence of complications, how to rescue these patients after an adverse event or eventually referring high-risk patients to high-volume centres.

Our data demonstrate that ppoDLCO%, cardiac comorbidity, extended resection, and pneumonectomy in combination to hospital volume are the most important factors associated to FTR. These findings are consistent with prior studies illustrating the association of these factors with postoperative mortality after pulmonary resection. We previously reported in our institution (23) a 30-day mortality rate after pneumonectomy of 8.4% that reached 18.5% at six months after intervention. Among determinant factors of these results, we found age, laterality of the procedure and the occurrence of postoperative cardiorespiratory complications. Additionally, Dartevelle et al. (24) reported a mortality rate of 4% in T4 lung cancer patients undergoing extended resection after analysing a series of 388 cases operated in the last 30 years. Therefore, patients who may need a pneumonectomy or an extended resection should be thoughtfully selected, since some centres experience a high perioperative mortality rate, or referred to high-volume centres to guarantee optimal postoperative outcomes.

Several limitations need to be considered in this study. First, calculation of FTR was based on in-hospital deaths among patients who experience a major complication within 30 days after the operation, or later if the patient was still in the hospital. However, since a significant number of patients die after discharge within 90 days after pulmonary resection as a consequence of any surgical related complication (25), it can be considered that FTR based on 90-day mortality could be a better indicator of postoperative surgical care. Second, our data are based on a multicentre voluntary registry, so that, although the details of the data audit have been previously reported (12), bias related to patient selection and quality of data could have influenced our findings. Furthermore, our results should be confirmed in additional patient cohorts from different multi-institutional databases.

16.3% of patients who presented major complications after anatomical lung resection in the GEVATS series did not survive to discharge. ppoDLCO%, cardiac comorbidity, pneumonectomy, extended resection, and annual surgical volume were the risk factors most closely related to FTR. Based on these results, complex thoracic surgical pathology should be concentrated in high-volume centres in order to obtain the best results in this type of potentially high-risk procedures.