An aggregate score to predict the risk of large pleural effusion after pulmonary lobectomy^†

Abstract

OBJECTIVES

The volume of pleural effusion is one of the determinants of chest drain removal following pulmonary resection. Recent research suggests that values up to 400 ml/day are safe. The objective of this study was to develop an aggregate risk score to identify patients at higher risk of developing a large pleural effusion (LPE) (>400 ml/day) on postoperative day 2 (POD2) after pulmonary lobectomy.

METHODS

An observational study on 229 consecutive patients was conducted prospectively in two European centres (June 2012–September 2013). All patients underwent pulmonary lobectomy for lung cancer (thoracotomy: 131, video-assisted thoracic surgery: 98) and managed by single chest tube connected to an electronic-regulated suction device. Exclusion criteria were chest wall or diaphragm resection and postoperative-assisted mechanical ventilation. To build the aggregate score, variables were initially screened by univariable analysis, and then used in stepwise logistic regression analysis (validated by bootstrap). The scoring system was developed by proportional weighing of the significant predictor estimates, and patients were grouped in classes of incremental risk according to their total score.

RESULTS

The incidence of LPE on POD2 was 23% (53 of 229 patients). The independent risk factors associated with LPE on POD2 were age greater than 70 years (P = 0.01, bootstrap frequency 71%), a lower lobectomy (P = 0.03, bootstrap frequency 59%) and presence of COPD (P = 0.02, bootstrap frequency 63%). Each predictor received a weighted score of 1, and patients were grouped into three risk classes showing an incremental risk of LPE (P < 0.001): Class A (Score 0) 5 LPE in 66 patients, 7.5%; Class B (Score 1) 19 LPE in 88 patients, 22%; Class C (Score >1) 29 LPE in 75 patients, 39%.

CONCLUSIONS

The aggregate score is a reliable tool for identifying high-risk patients for LPE and assists in the selection of patients that can safely proceed to chest drain removal early after surgery.

INTRODUCTION

The amount of pleural effusion represents, along with the presence of air leak, one of the parameters taken into account to plan removal of chest tubes after major anatomical pulmonary resections. Some evidence from the literature and clinical experiences have shown that traditionally restrictive criteria can be safely relaxed and thresholds up to 400–500 ml per day have been proposed to allow for a safe removal of the chest tube [1–4].

The objective of this analysis was to develop an aggregate risk score to assist surgeons to identify those patients at higher risk of developing a large pleural effusion (LPE) (>400 ml/day) on POD2 after pulmonary lobectomy. The rationale is to provide a tool to assist selecting those patients that can safely proceed to chest tube removal early after operation.

METHODS

This is an observational analysis on prospectively collected data of 229 consecutive patients submitted to pulmonary lobectomy for non-small-cell lung cancer in two European centres (June 2012–September 2013) and managed with a single chest tube connected with an electronic chest drainage system featuring regulated suction (Thopaz, Medela Healthcare, Inc., Switzerland). All patients gave their consent to include their clinical data in our institutional database for clinical and scientific purposes and the Institutional Review Board of both hospitals approved the study.

As a rule, operations were performed through a muscle-sparing, nerve-sparing thoracotomy (131 cases) or video-assisted approach (98 cases) by qualified thoracic surgeons. Excluded were those patients with chest wall or diaphragm resection or requiring postoperative mechanical ventilation, conditions that could represent major confounders for chest tube management and pleural fluid output. Operability exclusion criteria were a predicted postoperative forced expiratory volume in 1 s (FEV1) and predicted postoperative carbon monoxide lung diffusion capacity <30% in addition to a VO₂ peak <10 ml/kg/min [5, 6].

Postoperative management focused on early mobilization, antithrombotic and antibiotic prophylaxis, and physical and respiratory rehabilitation. Postoperative chest pain was assessed at least twice daily and controlled through a systemic continuous infusion of analgesic drugs. Therapy was titrated to achieve a visual analogue score below than 5 (in a scale ranging from 0 to 10) during the first 48–72 h.

Chest tube removal criteria were as follows: pleural effusion less than 400 ml/day and absence of air leak defined as an airflow <30 ml/min for at least 8 h without any spike on the airflow graph recorded by the digital drainage system.

Incomplete fissures were developed by using mechanical staplers. A fissureless technique (division of fissure as the last step of the surgical procedure) was used for all video-assisted thoracic surgery (VATS) lobectomies. In open cases, a traditional approach was used to isolate the hilar structures inside the fissure. The presence of air leak was tested at the end of operation and if present every attempt was made to minimize it by applying additional sutures. No sealants, pleural tents or buttressing material were used in this series. A single 24-French chest tube was positioned at the end of the procedure. The chest tube was immediately connected to the digital chest drainage system (Thopaz), which was set at different levels of regulated pressure according to institutional protocols or concomitant clinical trials. Measurements of air leak were continuously recorded. The data were stored in the drainage system and subsequently downloaded into a computer by using the appropriate software and used for the analysis. Measurements of pleural fluid were recorded at morning and evening rounds.

Statistical analysis

The normal distribution of numeric variables was assessed by the Shapiro–Wilk test. Continuous variables were compared by means of the unpaired Student's t-test or Mann–Whitney U-test, as appropriate. Categorical variables were compared by means of Fisher's exact test or χ² test.

Variables with a P-value <0.1 at univariable analysis were entered as independent predictors into a stepwise logistic regression analysis with backward selection to verify their independent association with a pleural effusion larger than 400 ml on POD2 (dependent variable). Variables were retained in to the model if their P-values were <0.1. The following factors were initially screened by univariable analysis and then tested in the multivariable regression analysis: age, body mass index (BMI), gender, FEV₁, forced expiratory volume in 1 s to forced vital capacity ratio (FEV₁/FVC), carbon monoxide lung diffusion capacity, presence of moderate to severe chronic obstructive pulmonary disease (COPD defined as FEV₁ <80% and FEV₁/FVC ratio <0.7), side and site of resection, presence of pleural adhesions (dense adhesions occupying more than 30% of the entire lung or the entire residual lobe). For the purpose of this analysis, significant (P < 0.05) numeric variables were tested for a threshold effect and dichotomized by using receiver operating characteristics (ROC) analysis.

The point closest to the top left part of the ROC plot was used to identify the best cut-off.

Predictor reliability was assessed by using a bootstrap resampling technique with 1000 samples [7 –9].

Bootstrap consisted of generating 1000 simulated populations from the distribution of the original dataset. Each simulation results in a new sample of the same size as the original data set (229 patients), which is generated through a process of random selection (with replacement) of individuals from the original sample.

At each step of the simulation, every individual from the original data set is again eligible to be selected, irrespective of whether he has already been selected. Therefore, in each bootstrap sample, some of the original individuals may not be represented and others may be represented more than once. This random sampling with replacement is repeated to generate 1000 new simulated populations, ensuring accurate statistics without assumptions by combining and analysing the information generated from these many datasets. Multivariable analysis was repeated in each of these 1000 simulated samples and those variables that resulted significant in more than 50% of the samples were judged stable or reliable [7–9].

The scoring system was developed by proportional weighing of the significant predictors and based on their regression coefficients. One point was assigned to the predictor with the smallest coefficients. Since all the resulting significant predictors in the final model had coefficients of similar value, they were all assigned a value of 1 point (see Results). An aggregate risk score was generated for each patient by summing each estimate [10]. Finally, patients were grouped in classes of incremental risk according to their total score.

All the statistical tests were two-tailed, with a significance level of 0.05, and were performed on the statistical software Stata 12.0 (Stata Corp, College Station, TX, USA).

RESULTS

The characteristics of the patients in the study are given in Table 1.

The mean values of pleural effusion measured on POD1 and POD2 were 370 ml [standard deviation (SD) 267] and 291 ml (SD 193) (P < 0.0001), respectively.

Eighty-two patients (36%) had a LPE, defined as an effusion larger than 400 ml, on POD1. These patients developed a significantly larger effusion on POD2 compared with those with smaller effusion on POD1 (358 ml, SD 194 vs 255 ml, SD 189, P = 0.0001). Thirty of 82 patients (37%) with a LPE on POD1 developed an LPE on POD2.

Patients with LPE on POD2 had a longer mean chest tube duration (6.5 vs 4.5 days, P = 0.0001) and mean postoperative stay (6.9 vs 5.5 days, P = 0.001) compared with those without LPE.

Patients operated on through VATS had similar volumes of pleural effusion measured on POD1 and POD2 compared with those submitted to thoracotomy (POD1: 370 ml, SD 277 vs 370 ml, SD 254, P = 1.00. POD2: 277 ml, SD 170 vs 302 ml, SD 209, P = 0.30). The proportion of patients with an LPE on POD2 was similar in the two groups (thoracotomy 34 of 131, 26% vs VATS 19 of 98, 19%, P = 0.24).

Fifty-three of 229 (23%) patients had an LPE on POD2. Table 2 shows the results of the univariable comparison between patients with and without LPE on POD2. In particular, LPE on POD2 occurred more frequently in patients older than 70 (P = 0.003), in males (P = 0.037), in those with lower FEV1 (P = 0.005), moderate to severe COPD (P = 0.016) and in those submitted to lower lobectomies (P = 0.022).

A less negative regulated pleural pressure (−8 vs −20 cmH₂O) set on the electronic chest drainage system was associated with a larger effusion measured on POD2 (324 vs 262 ml, P = 0.02) and with an increased proportion of LPE on POD2 (29 of 97, 30% vs 23 of 126, 18%, P = 0.04).Seventy-four patients had an air leak evident on POD2. The presence of an air leak on POD2 was not associated with an increased frequency of LPE (LPE and air leak on POD2 20 of 74–27% vs LPE and no air leak on POD2 33 of 155–21%, P = 0.3).

Table 3 shows the results of the stepwise logistic regression analysis (Hosmer–Lemeshow goodness- of-fit test = 4, P = 0.6; c-index 0.7).

An age older than 70 (P = 0.011), a lower lobectomy (P = 0.030) and presence of COPD (P = 0.018) were independent risk factors associated with LPE on POD2. Since the individual predictors had regression coefficients of similar magnitude, they were assigned the same weighted score of one point.

The individual points were summed for each patient to obtain an effusion risk score ranging from 0 to 3 (66 patients had a score of 0, 88 had a score of 1, 64 had a score of 2 and 11 a score of 3).

Three classes of risk were identified showing an incremental risk of LPE on POD2 (P < 0.001): Class A (Score 0) 7.5%, Class B (Score 1) 22% and Class C (Score >1) 39% (Fig. 1).

DISCUSSION

Physiology background

Pleural fluid turnover is fully regulated at parietal level through lymphatic absorption. The daily physiological pleural fluid filtration is estimated as 350 ml/day. Pleural lymphatics act as an efficient negative feedback system to regulate pleural fluid dynamics as they can markedly increase draining flow (20–30-fold) in response to increased filtration. Pleural turnover physiology studies revealed greater lymphatic drainage over the caudal and diaphragmatic surfaces and in the mediastinal regions [4, 11–14].

Clinical background and rationale for the study

Recent studies reported safe removal of chest tubes up to 400–500 ml (24 h) after major lung resection [1–4]. However, no clinical guidelines have been published in the literature on chest tube management and more precisely on the safe timing of chest tube removal. Current clinical practice remains mostly based on empirical observations or traditions. This paper provides a new aggregate risk score to predict the incidence of LPE after pulmonary lobectomy and assist in implementing safe protocols of chest tube removal.

Main findings

Twenty-three percent of patients had an LPE (>400 ml/day) on POD2. We found that the presence of COPD, an age older than 70 years and a lower lobectomy were independent factors associated with LPE. Patients without any of these factors had a risk of LPE of 7.5% while those with two or more of these factors had a risk of LPE of 39%.

Previous studies reported similar findings on the greater amount of fluid drained following lower lobectomies [15]. One study on early removal of the chest tube on the day after a VATS lobectomy, independent of the drainage volume, suggested careful surveillance in patients undergoing right lower lobectomies due to increased observed postoperative morbidity [16]. The greater amount of pleural drainage after a lower lobectomy can be due to the existence of a larger residual space after removing a large amount of lung tissue.

The patients with an age greater than 70 years should be considered to be at a high risk for developing LPE following lung resection. Ageing is a major risk factor for microvascular dysfunction and hyperpermeability. The potential mechanisms by which ageing leads to barrier dysfunction relate to disruption of adherent junctions of the endothelial cells and associated proteins of the interstitial matrix, causing an increase in microvascular permeability through the paracellular pathway. Other factors may contribute such as the increased age-dependent oxidative stress, inflammatory mediators and apoptotic signalling. The hyperpermeability may also result in an increase in protein escape from microvessels [17].

COPD is a condition causing important lung damage. Unfortunately, no scientific data are available to explain the increased amount of pleural drainage after lung resection in COPD patients. Three co-factors may favour the increase in the lung microvascular filtration and increased pleural effusion in COPD. It may be related to an increased capillary pressure due to the decrease in the vascular bed, overdistension and increased blood flow (overperfusion) of the remaining lung after a lung resection.

Limitations

The main limitation of the study is the lack of data on the time course of the rate of fluid removal. Future research is needed to define how fluid drainage is decreasing in the two groups (large vs small effusions). The study includes only patients submitted to lobectomy, it may be interesting to analyse further data on sublobar resections.

Additionally, a validation of the findings should be performed in a population with a use of traditional suction device, as we analysed data only for the patients on regulated suction.

We arbitrarily selected the cut-off of LPE at 400 ml/day based on the most recent evidence from the literature and clinical experience. However, the clinical relevance of this threshold remains to be proven.

Clinical and research implications

The risk score could be useful to alert the surgeon and the patients of potential complications and increase awareness.

It may be a useful tool in clinical practice to implement safe protocols of chest tube removal, identifying those patients that can safely proceed to chest tube removal early after resection if air leak criteria are also met.

In the future, electronic drainage systems capable to monitor and record the pleural effusion through the chest tubes will be able to assist in analysing trends of fluid output.

Our finding of a larger effusion in those patients with a less negative regulated pressure set on the device, warrants future research. The use of novel digital systems featuring continuous recording of fluid output will be ideal to more precisely investigate whether different levels of regulated pleural pressure would be able to influence pleural fluid turnover after resection. A more negative intra-pleural pressure will cause a closer apposition of the visceral to the parietal pleura opposing intra-pleural flow of fluid. A more negative intra-pleural pressure set by Thopaz increases the intra-pleural flow resistance more than a less negative pressure. As a consequence, the total fluid drainage using −20 cm H₂O will be less.

CONCLUSION

In conclusion, we were able to develop an aggregate score associated with incremental risk of LPE. This score can be used to easily identify patients at increased risk of large pleural effusion after lung resection and tailor a safe chest tube management.

Conflict of interest: Alessandro Brunelli and Kostas Papagiannopoulos have a consultancy agreement with Medela Healthcare, Switzerland.

REFERENCES

APPENDIX. CONFERENCE DISCUSSION

Dr D. Waller(Leicester, UK): How do you explain the results? Why should COPD be associated with larger amounts of pleural fluid? I don't understand the mechanism.

Dr Hristova: The reasons are very complex. We need to go to the basic pathophysiology of the residual pleural space to explain what is happening after lung resection. One of the reasons could be due to the reduced elastic recoil of the lung in COPD patients. Also, those patients have increased capillary pressure and this leads to increased filtration. Of course, these are only theories and nothing has been proven in physiology studies because most of the data comes from animal studies rather than from humans. That is something that we need to look at.

Dr Waller: What do you think?

Dr Hristova: I think it is mainly the capillary pressure. It could also be inflammation, based on the presentation from the Canadian group. It's quite an interesting finding.

Dr. T Walles(Würzburg, Germany): If I have a patient with 300 ml coming out and he is 65 years old, has good lung function following an upper lobe lobectomy, then, according to your score, I can go ahead and remove the tubes?

Dr. Hristova: We created our algorithm for chest tube removal for ourselves in Leeds. We usually don't remove chest drains on the first day after a lobectomy. However, in selected cases with no risk factors, that is also a possibility. For postoperative day 2 we set the cutoff point at 400 ml and, in the absence of risk factors, we can remove the chest drain.

Dr Walles: Your data is on a retrospective basis; you analysed the patients and produced your score. Have you already evaluated it in a prospective manner to be able to say whether it is working? Do you plan to do other work because of what you have already seen?

Dr Hristova: This is a retrospective analysis on prospectively collected data, but, yes, that's also one of the things that we are planning to perform in the future.

Dr Walles: But you are not doing it now that you have your score and then you can really evaluate it in your patients?

Dr Hristova: We haven't started it yet but most probably in the future, yes, this will be the case.

Author notes