Poor preoperative patient-reported quality of life is associated with complications following pulmonary lobectomy for lung cancer†

Abstract

OBJECTIVES: To assess whether quality of life (QOL) was associated with cardiopulmonary complications following pulmonary lobectomy for lung cancer.

METHODS: Retrospective analysis of 200 consecutive patients who had pulmonary lobectomy for lung cancer (September 2014–October 2015). QOL was assessed by the self-administration of the European Organisation for Research and Treatment of Cancer QLQ-C30 questionnaire within 2 weeks before the operation. The individual QOL scales were tested for a possible association with cardiopulmonary complications along with other objective baseline and surgical parameters by univariable and multivariable analyses.

RESULTS: Forty-three patients (21.5%) developed postoperative cardiopulmonary complications; 4 of them died within 30 days (2%). Univariable analysis showed that, compared to patients without complications, those with complications reported a lower global health status (GHS) [59.1; standard deviation (SD) 27.2 vs 69.6; SD 20.6, P = 0.02], were older (71.2; SD 8.4 vs 67.7; SD 9.4, P = 0.03), had lower values of forced expiratory volume in one second (FEV1) (83.9; SD 27.2 vs 91.4; SD 20.9), P = 0.06) and carbon monoxide lung diffusion capacity (DLCO) (67.9; SD 20.9 vs 74.2; SD 17.6, P = 0.02) and higher performance score (0.76; SD 0.63 vs 0.53; SD 0.64, P = 0.02). Stepwise logistic regression analysis showed that factors independently associated with cardiopulmonary complications were age [odds ratio (OR) 1.04, 95% CI 1.0–1.09, P = 0.02] and patient-reported GHS [OR 0.98, 95% confidence interval (CI) 0.96–0.99, P = 0.006], whereas other objective parameters (i.e. FEV1, DLCO) were not. The best cut-off value for GHS to discriminate patients with complications after surgery was 50 (c-index 0.65, 95% CI 0.58–0.72).

CONCLUSIONS: A poor GHS perceived by the patient was associated with postoperative cardiopulmonary morbidity. Patient perceptions and values should be included in the risk stratification process to tailor cancer treatment.

INTRODUCTION

Patient-reported quality of life (QOL) has been shown to correlate poorly with functional parameters commonly used to assess fitness before lung resection [1–3]. Furthermore, the trajectory of QOL after lung resection for NSCLC does not always follow the cardiorespiratory impairment expected in high-risk classes [4].

Despite the efforts made to decrease the incidence of adverse events after major resection for lung cancer by improving patient selection, none of the risk models is characterized by 100% accuracy. Unknown or unmeasured subjective factors may play a role in determining the postoperative course or even the long-term prognosis. For instance, a recent paper has investigated the impact of the visually observed physical status on estimates of surgical risk [5, 6], demonstrating that surgeons differentiated relative risk of lobectomy based on clinical vignettes.

Nevertheless, current risk-stratification algorithms do not include patient-reported outcomes (PROs) as measures of subjective well-being [7].

Investigating the impact of the subjective perception of well-being on surgical outcome may help refine risk stratification by providing a PRO as a global measure of many potential factors.

The objective of this analysis was therefore to assess whether QOL is associated with cardiopulmonary complications following pulmonary lobectomy.

METHODS

This retrospective analysis was performed on a prospectively maintained database. Two hundred consecutive patients undergoing pulmonary lobectomy for lung cancer (September 2014–October 2015) completed a preoperative QOL questionnaire and their responses were analysed. All cancer patients were discussed at multidisciplinary tumour board meetings. Operability exclusion criteria were in accordance with current guidelines [7].

All patients were operated by qualified general thoracic surgeons either by a video-assisted thoracoscopic technique (VATS; n = 175) or by a muscle-sparing thoracotomy (n = 25), depending upon the surgical indications (stage, size and location of the tumour). All patients had a systematic mediastinal lymph node dissection along with the lung cancer resection.

Following the operation, the patients were extubated in the operating room and transferred to an intensive care unit for constant monitoring where they spent the first postoperative night. They were subsequently transferred to a dedicated general thoracic surgical ward unless clinical conditions dictated otherwise.

Postoperative care followed standardized pathways of care and included early as possible mobilization and oral food intake, intense chest physiotherapy and rehabilitation, deep venous thrombosis prophylaxis and chest pain control using a combination of patient-controlled analgesia and paravertebral infusion of local anaesthetic.

Quality of life assessment

QOL was assessed by the administration of the European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire (EORTC QLQ-C30), Version 3, within 2 weeks before the operation. The EORTC QLQ-C30 is an internationally validated cancer-specific QOL questionnaire [8]. It is composed of multi-item and single-item scales measuring global health status (GHS; akin to overall QOL), five functional scales analysing physical, role, emotional, cognitive and social functioning and nine symptom scales analysing fatigue, nausea and vomiting, pain, dyspnoea, insomnia, appetite loss, constipation, diarrhoea and financial difficulties.

The scores range from 0 to 100 after linear transformation of the raw scores. A high score for a functional scale represents a high level of functioning (healthier), whereas a high score for a symptom scale represents a high level of symptoms/problems.

This instrument has been extensively tested for reliability and validity [9, 10].

Statistical analysis

The principle end point of the analysis was the incidence of postoperative cardiopulmonary complications occurring within 30 days from the operation or over a longer period if the patient was still in the hospital. Cardiopulmonary complications were defined according to joint Society of Thoracic Surgeons-European Society of Thoracic Surgeons standard definitions [11] and included the following: respiratory failure requiring mechanical ventilation longer than 24 h or reintubation, acute respiratory distress syndrome (ARDS), pulmonary embolism, pulmonary oedema, pneumonia, atelectasis requiring bronchoscopy, atrial fibrillation needing medical treatment or cardioversion, acute myocardial ischaemia, acute cardiac failure, stroke and acute kidney failure. Several baseline and surgical variables, including the EORTC QLQ-C30 scales, were tested for a possible association with cardiopulmonary complications using univariable analysis. Normal distribution of the variables was tested using the Shapiro–Wilk normality test. Numerical variables with normal distribution were compared using the unpaired Student t-test, whereas those without normal distribution were compared by the Mann–Whitney test. Categorical variables were compared using the χ² test. The Fisher’s exact test was used when the frequency in one or more cells was < 10.

In addition to the EORTC QOL scales, the following variables were tested: age; sex; body mass index (BMI); forced expiratory volume in one second (FEV1) expressed as percentage of normal for age sex and height (FEV1%); carbon monoxide lung diffusion capacity expressed as percentage of normal for age, sex and height; history of coronary artery disease, cerebrovascular disease (CVD), diabetes or chronic kidney disease (CKD); Eastern Cooperative Oncology Group performance score; surgical approach (VATS versus open surgery) and pathological tumour stage (pT1 versus pT > 1). Those variables that were significant (P < 0.05) using univariable analysis were then used as independent predictors in a stepwise logistic regression with backward elimination. Variables with P < 0.05 were retained in the final model and their reliability was tested by bootstrap analysis with 1000 samples. In the bootstrap analysis, repeated samples with the same number of subjects as the original database were generated with replacement and the logistic regression was repeated in each of these simulated samples. The variables occurring in more than 50% of the samples were judged to be stable and retained in the final model [12, 13].

Receiver operating characteristic (ROC) analysis was used to determine the best cut-off value of significant QOL scales associated with complications. The best cut-off was determined by identifying the point closest to the top left part of the plot of the ROC curve.

The analysis was performed using the STATA 12.0 (Stata Corp., College Station, TX, USA) statistical software.

RESULTS

The baseline objective characteristics of the patients included in this study are shown in Table 1. Their preoperative QOL scales are reported in Table 2. Forty-three patients (21.5%) developed cardiopulmonary complications within 30 days from operation; 4 of them died in-hospital or within 30 days (2%).

Cardiopulmonary complications in order of frequency were pneumonia, 23; atrial fibrillation, 18; atelectasis requiring bronchoscopy, 11; respiratory failure, 9; ARDS, 5; pulmonary oedema, 3; myocardial infarct, 3. Twenty-five patients had more than one complication.

Objective variables significantly associated with complications after univariable analysis were age (P = 0.03), FEV1 (P = 0.06), carbon monoxide lung diffusion capacity (DLCO) (P = 0.02), performance score (P = 0.02) and previous history of CVD (P = 0.08) (Table 3).

Univariable analysis showed that patients with any complications had poorer GHS (P = 0.02) and lower physical functioning (P = 0.09) and role functioning (P = 0.09) values. They also had greater appetite loss (P = 0.047) and dyspnoea (P = 0.07) compared with cases with no complications. No other QOL scales were associated with outcome (Table 4).

Stepwise logistic regression analysis showed that the only factors that remained independently associated with cardiopulmonary complications were age (P = 0.02) and GHS scale (P = 0.006), whereas other objective parameters (i.e. FEV1, DLCO) were not (Table 5).

To control for the effect of surgical factors such as intraoperative complications needing conversion, we performed an additional analysis including this variable as an independent predictor along with age and GHS. This test showed that age (P = 0.015) and GHS (P = 0.008) retained their independent significant association with cardiopulmonary complications even after controlling for the effect of conversion.

Of the 178 patients with GHS > 50, 33 (18%) developed complications (P = 0.001) (Table 6).

DISCUSSION

Main findings

We were able to show that the subjective perception of a poor GHS is associated with postoperative cardiopulmonary morbidity after pulmonary lobectomy.

Our finding suggests that the level of preoperative QOL represents important information that can be used along with other preoperative parameters to refine the evaluation of patients to be submitted to lung resection for NSCLC. To our knowledge, this is the first time that a patient-reported parameter supersedes the traditional objective risk factors of postoperative outcomes in a context using modern functional guidelines to select patients for operation.

Findings in the existing literature

The objective functional parameters measured before operation (pulmonary function tests and exercise test) have traditionally been considered surrogates of the patients’ health status, in particular when discussing with them the possible lung cancer treatment options. The increasing number of publications on PROs during the last decade has led surgeons to reconsider the importance and the different weights of objective and subjective outcomes in thoracic surgery.

Few trials have attempted to identify objective factors associated with perioperative changes in QOL. In particular, Handy et al. [14] demonstrated that patients with poor DLCO had worse preoperative physical functioning and QOL in addition to worse postoperative QOL, health and functioning and psychological/spiritual status. However, in that study, a generic questionnaire was administered precluding the evaluation of more specific cancer-related symptoms. In a previous paper, we were able to identify that some important QOL domains were correlated with the postoperative perceived health status using the SF-36 in 172 patients submitted to pulmonary resections [15].

Preoperative objective parameters should not be considered a surrogate of QOL inasmuch as the latter is a multidimensional concept encompassing social, emotional, cognitive, physical and functional well-being [16]. Other studies confirmed the presence of unaccounted-for factors in the standardized definition of high-risk surgical candidates, showing no significant changes in terms of QOL compared with their low-risk counterparts (i.e. older/younger, chronic obstructive pulmonary disease (COPD)/no COPD, low/high VO2Max) [4, 17–19].

The overall failure to predict how the patient will feel months after surgery questions the entire process of surgical patient selection. The most up-to-date guidelines in fact do not include QOL or other patient-reported outcome scores in their algorithms [7, 20]. In 2010, the British Thoracic Society [21] for the first time introduced in the flow-chart for risk assessment for surgery the acceptance of risk by the patient, which can be considered a complex process involving still unknown body-mind interactions.

Clinical inference

One explanation of our results can be found in the presence of unknown patient characteristics that are associated with the early postoperative outcome and that are better reflected in self-rated parameters than in objective measures. This explanation may refer to the concept of interoception, the individual’s superior ability to sense and incorporate even non-conceptualized sensations of bodily status into self-ratings of health [22, 23].

The same interpretation can also apply to the recently published papers in which physical and mental components of QOL were associated with overall and cancer-specific survival in patients with early-stage NSCLC [24, 25].

Other authors have tried to explain different and unexpected results after introducing patient-reported outcomes in clinical health-applied research, with the Wilson–Cleary model linking clinical variables with QOL [26]. Improving patient outcomes is in fact pivotal to identifying causal pathways that link different types of outcomes to each other, especially when patient-reported outcomes are included. The model implies that symptom status, functional health, general health perceptions and overall QOL are casually connected dimensions of QOL. According to this model, the General Health summary score, as the most reliable predictor of postoperative complications, can be interpreted as an expression of a multilevel continuum, with a domain of physiological variables as the starting point and unaccounted for personal and environmental factors playing an important role.

The emerging evidence of a genetic substrate influencing the subjective perception of physical and emotional well-being [27] may help in the future to explain the association between components of health-related QOL and the occurrence of postoperative complications independent of other objective risk factors.

We found that almost half of the patients with GHS < 50 developed complications compared to 18% of those with a higher preoperative GHS. This information has important clinical implications.

First, patient-reported GHS may be included in future flowcharts used to select patients for operation. This QOL domain is calculated from only two questions of the EORTC QLQ-C30 questionnaire: (i) How would you rate your overall health during the past week? and (ii) How would you rate your overall QOL during the past week? Both of these questions are scored using a scale of 1–7 where 1 represents very poor global health/QOL and 7, an excellent global health/QOL. The two scores are averaged (raw score) and then a linear transformation is performed using the following equation to yield a score ranging from 0 to 100: GHS = ([raw score−1]/6)) × 100. The inclusion of a shorter version of these questionnaires in the preoperative algorithms is warranted, especially after the publication of the new version of the EORTC Lung Cancer Module, which contains surgical items for the first time [28].

Moreover, the information that a low GHS score is associated with poor early outcome may warrant the proposal and analysis of rehabilitation and physical support programs. These programs can improve patient well-being and potentially reduce postoperative morbidity.

To date no composite measure includes both objective measures and patient-derived utilities. However, major initiatives are underway that highlight the crucial role of PROs in the standard set of outcomes, such as the International Consortium for Health Outcomes Measurement Standard Set for Lung Cancer or the US Patient-Centered Outcomes Research Institute. Our group developed a survival aggregate score, including objective and subjective patient-based parameters, to refine the prognostic stratification of patients with early-stage NSCLC after surgical treatment [19].

The inclusion of the PROs in large databases would be challenging to realize in clinical practice, but more efforts need to be done in the surgical field to increase the representativeness of these outcomes in everyday clinical practice, where they are largely missing [29]. To facilitate and support these measurements, researchers in the UK have developed a technical system for regularly collecting PROs on line, at repeated post-diagnostic time points, for linking and storing these with patients’ clinical data in cancer registries, and for electronically managing the related patient monitoring and communications [30].

LIMITATIONS

This study may have the following limitations: Because the study was performed in a single centre, generalization of the results to other settings or countries would need to be verified by independent studies. Differences in case-mix and economic, social and cultural characteristics of patients may have an impact on results.

The analysis was performed on patients with lung cancer considered suitable for lobectomy. Generalization to benign diseases or sublobar resections would need to be verified.

We excluded patients undergoing pneumonectomy because this operation has been consistently found associated with an increased risk of complications and higher mortality rates. The association between QOL measures and outcome after pneumonectomy warrants a separate analysis.

Most of the operations (88%) were performed using VATS. It would be interesting to verify whether similar results would be found analysing a population with a larger proportion of thoracotomies.

Admittedly, the discrimination ability of GHS is only moderate as indicated by the c-index value. This result may be due to other factors associated with complications, which may be unknown or unaccounted for, or to the imprecision of the QOL measurement instrument used in this study.

In fact, the results may be influenced by the instrument used to measure QOL. In this study, we used the EORTC QLQ-C30, which is a generic questionnaire specifically validated for cancer patients. The use of other QOL tools (i.e. SF36) may lead to different findings and warrant specific analyses allowing for comparison with the general population.

CONCLUSION

The subjective perception of poor GHS is associated with postoperative cardiopulmonary morbidity after pulmonary lobectomy. This finding warrants the adoption of a holistic approach during the surgical shared decision-making process. Patient perceptions and values should be included in the risk stratification process to tailor cancer treatment.

Conflict of interest: none declared.

REFERENCES

APPENDIX. CONFERENCE DISCUSSION

Dr J. Schirren(Wiesbaden, Germany): I think it is not a limitation that you speak only about lobectomies; I think this is a good structure for your study. But you should add to this that you have only 88% VATS lobectomies, and I would be very interested in the oncological status of the patients and how many nodes you dissected. That would help our understanding.

Dr Pompili: We did not include the postoperative oncological results in this analysis, because the objective of the analysis was to assess whether preoperative subjective parameters were associated with the postoperative course. So the oncological evaluation was not the main focus of this study. We didn’t put the nodal status into our univariable analysis at all.

Dr H.V. Kara(Istanbul, Turkey): I want to ask how much this has affected your decision-making in your clinic? Do you think it is possible to postpone surgery due to low or poor status? Can you comment on that?

Dr Pompili: I think the psychological rehabilitation programme can have the same drawbacks of physical rehabilitation. Cancer waiting time remains the main issue to better plan a rehabilitation programme. So we need to find the perfect timeframe between the first preoperative quality of life assessment and the postoperative psychological or physical rehabilitation programme.

Dr Kara: Ninety day results are also discussed. Are you planning as a future direction for this study to make a calculation for 90 days or do you have any preliminary data you want to share?

Dr Pompili: Of course this is good input and we would like to look at 90 days and correlate it to the health economics analysis as well, because it was recently demonstrated to have significant impact on health care costs.

Dr Kara: I noticed that you include both performance status and the GHS score in your study. Both of them are of significant difference in your univariate analysis but only the questionnaire turned out to be an independent predictive factor. Of course you are not referring to the surgeons. Our patients are doing a better job than the surgeons in evaluating their real condition and the risk of surgery. So what exactly is in this questionnaire that is missing from the performance status evaluation by surgeons? What do you think?

Dr Pompili: I think that our personal idea of health status is completely different from something defined by objective physical parameters, but relates to each of us as individuals. So the patient can describe better in quality of life surveys the impact of the disease on his or her daily lifestyle. Probably each person has a different daily lifestyle, which can be completely different and not captured by performance status, which is very generic, for example. So I am quite sure that it is crucial to answer the question on a quality of life questionnaire about the impact of the disease or its symptoms on your daily activities. That can, in turn, be different from that a doctor’s, for example, or a health care worker’s. The different impact of the disease on our individual lives can be something that is not captured by the performance status.

Author notes