European Journal of Cardio-Thoracic Surgery Advance Access published March 29, 2015
ORIGINAL ARTICLE
European Journal of Cardio-Thoracic Surgery (2015) 1–6 doi:10.1093/ejcts/ezv107
The mortality after surgery in primary lung cancer: results from the Danish Lung Cancer Registry† Anders Greena,*, Jacob Haugeb, Maria Iachinac and Erik Jakobsenb a
b c
Odense Patient data Explorative Network (OPEN), Odense University Hospital and Institute of Clinical Research, University of Southern Denmark, Odense, Denmark The Danish Lung Cancer Registry, Department of Thoracic Surgery, Odense University Hospital, Odense, Denmark Research Unit of Clinical Epidemiology, Institute of Clinical Research, University of Southern Denmark, Odense, Denmark
* Corresponding author. Odense Patient data Explorative Network (OPEN), Odense University Hospital and University of Southern Denmark, J. B. Winsløws Vej 9A, 3rd floor, DK-5000 Odense C, Denmark. Tel: +45-4088-7775; e-mail: [email protected]; [email protected] (A. Green). Received 25 November 2014; received in revised form 11 February 2015; accepted 19 February 2015
Abstract OBJECTIVES: The study has been performed to investigate the mortality within the first year after resection in patients with primary lung cancer, together with associated prognostic factors including gender, age, tumour stage, comorbidity, alcohol abuse, type of surgery and post-surgical complications. METHODS: All patients (n = 3363) from the nationwide Danish Lung Cancer Registry with first resection performed between 1 January 2007 and 31 December 2011 were analysed by Kaplan–Meier techniques and Cox-regression analysis concerning death within the first year after resection. Covariates included gender, age, comorbidity (Charlson comorbidity index), perioperative stage, type of resection, registered complications to surgery and alcohol abuse. RESULTS: The cumulative deaths after 30 days, 90 days, 180 days and 360 days were 72 (2.1%), 154 (4.6%), 239 (7.1%) and 478 (14.2%), respectively. Low stage, female gender, young age, no comorbidity, no postoperative complications, no alcohol abuse and lobectomy as type of resection were favourable for survival. CONCLUSIONS: Our results demonstrate that resection in primary lung cancer impacts mortality far beyond the initial 30 days after resection, which is conventionally considered a time window of relevance for the adverse outcome of surgery. Increased efforts should be made for optimizing the selection of patients suited for resection and for identifying patients at increased risk of death after resection. Furthermore, patients should be monitored more closely and more frequently, in particular those patients with high risk of death after resection. Keywords: Lung cancer • Resection • Mortality • Risk factors
INTRODUCTION Primary lung cancer is one of the most common types of cancers today. In Denmark alone, more than 4500 people are diagnosed every year. Despite improvements in recent years, the prognosis in lung cancer is very poor, with crude 5-year survival as low as 11–13% [1]. Resection, when applied to the right subset of patients, is the type of intervention associated with the best survival. Conventionally, survival after surgery is estimated from either the in-hospital mortality or the 30-day mortality and it is generally accepted that deaths occurring within 30 days after surgery are attributable to complications following surgery [2]. Thus, the 30-day mortality may serve as a benchmark for the overall quality of surgical intervention and appropriate selection † Presented at the 20th European Conference on General Thoracic Surgery, Essen, Germany, 10–13 June 2012.
of patients for surgery. A large number of single institution studies on postoperative mortality have been published, some of them recently [3, 4], but nationwide, population-based estimates on the 30-day mortality are limited. It is also in a national setting largely unknown whether complications after surgery and any associated excess mortality are limited to the 30-day post-surgical period or extend beyond 30 days [5–7]. The criteria for assessing the eligibility of patients for surgery have not been well-established and uniformly agreed upon, albeit it is known that factors such as age, gender, comorbidity all influence mortality [8–11]. The Danish Lung Cancer Registry (DLCR) provides unique opportunities for clinical and epidemiological research in lung cancer [12]. The present study has been performed to investigate the mortality within the first year after resection in patients with primary lung cancer, together with associated prognostic factors including gender, age, tumour stage, comorbidity, alcohol abuse, type of surgery and post-surgical complications.
© The Author 2015. Published by Oxford University Press on behalf of the European Association for Cardio-Thoracic Surgery. All rights reserved.
THORACIC
Cite this article as: Green A, Hauge J, Iachina M, Jakobsen E. The mortality after surgery in primary lung cancer: results from the Danish Lung Cancer Registry. Eur J Cardiothorac Surg 2015; doi:10.1093/ejcts/ezv107.
2
A. Green et al. / European Journal of Cardio-Thoracic Surgery
MATERIALS AND METHODS The Danish Lung Cancer Registry The DLCR has recorded all cases of primary lung cancer in Denmark, since its initiation in 2000. The database today contains information on more than 50 000 cases of lung cancer, with information on patient age, gender, tumour stage, type of surgery if applicable and many other data. Data are reported to the registry from the participating departments after completing diagnostic evaluation and specific treatments. Supplementary data on the patient’s vital status are obtained from the Danish Civil Registration System [13]. Pathology information related to all registered patients with lung cancer case is routinely retrieved from The Danish Pathology Register and data on comorbidity from the Danish National Patient Register [14, 15]. All information from these registers may be linked at patient level by means of the unique Personal Identification Number (PIN) assigned to every Danish citizen. The present analysis includes the 3363 patients with non-smallcell lung cancer (NSCLC) who had their first resection for primary lung cancer between 1 January 2007 and 31 December 2011. Patients with small-cell lung cancer (SCLC) were excluded.
Diagnostic procedures and treatment options Diagnostic procedures in suspected lung cancer are primarily performed to verify the presence of the disease and to establish the type of lung cancer, subsequently based on histology to the two main categories, SCLC (15%) and NSCLC (85%). Furthermore, clinical staging of the lung cancer, the performance of the patient and relevant comorbidities will be evaluated. For patients eligible for resection, further investigations may be performed to find the most appropriate treatment. Lung cancer patients with NSCLC are predominantly treated with surgical resection or chemo- and/or radiotherapy whereas patients with SCLC almost exclusively will be offered chemo- and/or radiotherapy. In Denmark, 20% of the NSCLC patients are eligible for surgery at the time of diagnosis. Clinical stage is the most important factor when deciding the choice of treatment, but additional risk factors considered include age, lung function and performance, abuse of alcohol or drugs as well as comorbidity. Resection is performed using a standard thoracotomy or a video-assisted minimally invasive procedure (VATS). In Denmark, 60% of the resections are performed using VATS procedures. The goal is to perform surgery with a macro- and microscopically radical resection, and to remove as little lung parenchyma as possible without compromising this need for a radical resection. Lobectomy and wedge resection are therefore the type of resections preferred. Pneumonectomy is performed when a stage requires a more extensive resection, and a smaller resection such as sub-segmental resection is performed for patients unsuited for larger resections. Less than 5% of operations will be exploratory eventually since the staging performed during operation prohibits any kind of resection.
Study variables Vital status for each patient was obtained from the Danish Civil Registration System. Age at the time of surgery and gender were available from the PIN. DLCR provided data concerning surgical staging of the lung cancer, type and method of surgery,
complications registered, lung function, time from the start of the diagnostic evaluation until surgery and alcohol abuse. During surgery, tumours were classified via the tumour, nodes metastasis system, with subsequent transformation to stages following the International Association for the Study of Lung Cancer. For the purpose of this study, perioperative stages I and II (n = 2704) were assigned to one group whereas stages III and IV (n = 625 of which 105 were stage IV) were assigned to a separate group. Resection in lung cancer was categorized as pneumonectomy, lobectomy and wedge- or segment-resection. For the purpose of this study, type and method of surgery have been categorized as 0 for the wedge and segment resection, 1 for the lobectomy and 2 for the pneumonectomy. Patients undergoing explorative surgery (n = 170, corresponding with 4.8%) were excluded from analysis. Complications in DLCR were reported by the surgical departments on a mandatory basis using the following 14 predefined categories: (i) arrhythmia, (ii) acute myocardial infarction, (iii) pneumonia, (iv) atelectasis, (v) air leakage, (vi) empyema, (vii) bronco-pleura fistula, (viii) respirator treatment, with recording of days of respirator treatment, (ix) pneumothorax, (x) neurological complications, (xi) resurgery due to bleeding, (xii) resurgery due to other causes, (xiii) infection and (xiv) other complications. Only complications occurring in the period from operation to discharge were included. For the purpose of the present analysis, patients were grouped according to whether they had no, one or more than one complication after resection. Classification of alcohol abuse in DLCR followed the definitions established by the Danish National Board of Health according to which the consumption of more than 14 units for women and 21 units for men per week is considered to be abuse. Young age has, for the purpose of this study, been defined at age being lower than the mean age for our entire patient population, 65.9 years. Co-morbidity for each patient up to 10 years before the diagnosis of lung cancer diagnosis was assessed on the basis of a search for each patient in the Danish National Patient Register, established in 1977. This register contains data on all patient contacts, including coding of diagnosis and procedures, at Danish hospitals. For the classification of comorbidity, we used the Charlson comorbidity index (CCI) [16, 17], with grouping according to increasing level of comorbidity: CCI score = 0; CCI score = 1 and CCI score >1. For the present analysis, all hospital contacts associated with any cancer diagnosis and registered within 150 days before the date of lung cancer diagnosis were excluded to avoid any false contributions to CCI from diagnoses of other cancers turning out eventually to be primary lung cancer.
Statistical analysis Overall survival was illustrated by Kaplan–Meier plots by the variables such as postoperative stage, CCI, complications and type of surgery. To estimate the effect on mortality of the variables of interest by different time intervals (from 0 to 30 days, from 31 to 90 days, from 91 to 180 and from 181 to 360 days), we constructed the time-varying covariates for each variable of interest. For age, the new variables comprised (i) age30, where the hazard ratio (HR) for this variable will indicate the effect of age on 0 to 30-day survival; (ii) age90, where the HR for this variable will indicate the effect of age on 31- to 90-day survival; (iii) age180, where the HR for this variable will indicate the effect of age on 91- to 180-day survival and (iv) age360, where the HR for this variable will
indicate the effect of age on 181- to 360-day survival. In a similar way, time-varying covariates were defined for the other variables of interest (gender, stage, CCI, alcohol abuse, type of operation and complications). Then, we implemented an extended Cox model [18]. After fitting the model with the four time-varying covariates for each variable, we tested the heterogeneity between the intervals, variable by variable using a Wald test. Using the Wald test, we also tested that the HR for the wedge and segment resection to lobectomy is the same as the HR from lobectomy to pneumonectomy for all time intervals. Proportions and HRs are presented with 95% confidence limits in parentheses. The statistical analysis was performed using the STATA version 12.
Ethical aspects DLCR has been approved as a national clinical database by the Danish National Board of Health ( j.nr. 7-201-03-15/1). Permission for the present study was granted by the Danish Data Protection Agency ( j.nr. 2008-58-0035) and the National Board of Health ( j.nr. 7-505-29-1863/1).
RESULTS The clinical characteristics and number of deaths by time interval after resection are given in Table 1 for the 3363 patients from DLCR that underwent resection for primary lung cancer in
3
Denmark during the period 1 January 2007 through 31 December 2011. The majority of the patients had stage less than stage III [n = 2704, 81% (80%; 83%) of those classified], no alcohol abuse [n = 3021, 90% (89%; 92%), no postoperative complications [n = 2412, 72% (70%; 73%)] and had lobectomy as the type of operation [n = 2658, 79% (78%; 81%) of those classified]. A total of 478 patients died within 360 days after surgery corresponding with a cumulative mortality rate at 2% (2%; 3%) after 30 days, increasing to 5% (4%; 5%), 7% (6%; 8%) and 14% (13%; 15%) after 90, 180 and 360 days, respectively. Figure 1 shows the cumulative survival up to 500 days after resection for the study population by stage, CCI, postoperative complication status and type of resection. Low stage, no comorbidity, no postoperative complications and lobectomy as type of resection were characterized by best survival. For the groupings by CCI complications, gradients were seen with worsening survival with increasing level of comorbidity and increasing number of complications, respectively. In general, the survival curves tended to follow linear trends, however for the group with high number of complications and for patients with pneumonectomy as type of resection, there was an initial sharp decline in survival within the first 50 and 30 days, respectively. The full model in the detailed statistical analysis suggested that gender, age, abuse and CCI all had effects that may be considered constant across the intervals of the follow-up time since resection. For the other variables, HRs were estimated for each follow-up interval separately (Table 2). For demographic variables, females had a statistically significant lower mortality than males, HR = 0.79 (0.68; 0.93) and older age (defined as age above the mean age
Table 1: Overview of the study population by variables and deaths during the first year after resection
All patients Gender Male Female Age Younger than mean Older than mean Stage Unclassified 0, I, II III, IV Charlson comorbidity index 0 1 >1 Alcohol abuse Yes No Type of resection Not classified Wedge and segment resection Lobectomy Pneumonectomy Complications 0 1 >1
Number of patients
0–30 day mortality
31–90 day mortality
91–180 day mortality
181–360 day mortality
360-day survivors
3363
72
82
85
239
2885
1683 (50) 1680 (50)
50 (69) 22 (31)
48 (59) 34 (41)
52 (61) 33 (39)
134 (56) 105 (44)
1399 (48) 1486 (52)
1539 (46) 1824 (54)
20 (28) 52 (72)
18 (22) 64 (78)
33 (39) 52 (61)
104 (44) 135 (56)
1364 (47) 1521 (53)
34 (1) 2704 (80) 625 (19)
0 (0) 51 (71) 21 (29)
1 (1) 55 (67) 26 (32)
1 (1) 47 (55) 37 (44)
2 (1) 148 (62) 89 (37)
30 (1) 2403 (83) 452 (16)
1700 (51) 745 (22) 918 (27)
26 (36) 21 (29) 25 (35)
31 (38) 17 (21) 34 (41)
41 (48) 17 (20) 27 (32)
113 (47) 62 (26) 64 (27)
1489 (52) 628 (22) 768 (27)
342 (10) 3021 (90)
13 (18) 59 (82)
12 (15) 70 (85)
16 (19) 69 (81)
29 (12) 210 (88)
272 (9) 2613 (91)
4 (0) 451 (13) 2658 (79) 250 (7)
0 (0.0) 8 (11) 46 (64) 18 (25)
0 (0.0) 14 (17) 57 (70) 11 (13)
0 (0.0) 14 (16) 60 (71) 11 (13)
0 (0.0) 36 (15) 176 (74) 27 (11)
4 (0) 379 (13) 2319 (80) 183 (6)
2412 (72) 628 (19) 323 (10)
25 (35) 17 (24) 30 (42)
36 (44) 20 (24) 26 (32)
48 (57) 25 (29) 12 (14)
161 (67) 45 (19) 33 (14)
2142 (74) 521 (18) 222 (8)
Data are numbers indicating deaths occurring in the follow-up intervals (days) from date of resection (percentage of all patients).
THORACIC
A. Green et al. / European Journal of Cardio-Thoracic Surgery
4
A. Green et al. / European Journal of Cardio-Thoracic Surgery
Figure 1: Kaplan–Meier plots (with 95% CI) of survival by selected variables. Note truncation of the Y axis.
Table 2: Summary of the statistical analysis of the effect on mortality by variable and interval of the follow-up time from resection
Gender Age Stage30 Stage90 Stage180 Stage360 CCI Abuse30 Lobectomy vs wedge and segment resection Pneumonectomy vs wedge and segment resection Optype90 Optype180 Optype360 Compl30 Compl90 Compl180 Compl360
Hazard ratio
95% CI
0.79 1.47 1.36 2.14 3.71 3.10 1.19 1.58 0.75 2.55 0.98 1.05 1.12 3.03 2.39 1.53 1.18
0.68 1.24 0.78 1.34 2.41 2.37 1.08 1.26 0.33 1.01 0.60 0.66 0.85 2.26 1.84 1.15 0.95
0.93 1.73 2.39 3.43 5.73 4.04 1.30 1.99 1.68 6.39 1.62 1.67 1.48 4.06 3.11 2.03 1.47
P-values for departure from unity
P-values for Wald test
0.005 0.000 0.279 0.002 0.000 0.000 0.000 0.000 0.483 0.047 0.945 0.843 0.426 0.000 0.000 0.004 0.133
0.558 0.094 0.049
0.167 0.267 0.007
0.001
Numbers appearing in labels reflect the end of the time interval concerned. CCI: Charlson co-morbidity index; Abuse: alcohol abuse; Optype: type of resection; Compl: complications.
(65.9 years) of the entire study population) was associated with increased mortality, HR = 1.47 (1.24; 1.73). Among the clinically related variables, high level of CCI was associated with increased mortality, HR = 1.19 (1.08; 1.30). High stage had no statistically significant effect on 30-day mortality [HR = 1.36, (0.78; 2.39)] but was associated with increased mortality during the time period from 31 to 360 days after resection with
HR 2.14 (1.34; 3.43) for the 31 to 90 days after resection, 3.71 (2.41; 5.73) for the 91–180 days after resection and 3.10 (2.37; 4.04) the 181–360 days after resection, respectively. Wedge and segment resection had no significant effect on 30-day mortality compared with lobectomy [HR = 0.75 (0.33; 1.68)], whereas pneumonectomy was associated with increased 30-day mortality compared with lobectomy [HR = 2.55 (1.01; 6.39)]. The wedge and
segment resection and pneumonectomy had a proportional and non-significant effect on 31- to 360-day mortality with hazard rations at 0.98 (0.60; 1.62) for the 31–90 days after resection, 1.05 (0.66; 1.67) for the 91–180 days after resection and 1.12 (0.85; 1.48) for the 181–360 days after resection. For high level of complications, there was an initial highly increased mortality and even though there was a tendency to decreasing excess mortality with increasing follow-up interval from HR at 3.03 (2.26; 4.06) for the time period from 0 to 30 days after resection to HR at 1.18 (0.95; 1.47) for the time period from 181 to 360 days after resection, the effect remained statistically significant throughout the full period of 360 days after resection.
DISCUSSION This study makes use of one of the largest population-based lung cancer registries with high completeness of registration and data validity particularly concerning the inclusion of surgical data [12]. Since the unique features of the Danish Civil Registration System makes it possible to ensure the complete follow-up of the vital status of all patients [13], we believe that our results are valid for resection for primary lung cancer in Denmark. The results underline that excess mortality has an impact far beyond the initial 30 days after resection, and that many factors, including advanced disease stage and the fact of experiencing complications after resection have a negative effect on survival at least through the first year after resection. Also type of resection, in particular pneumonectomy, impacts survival negatively. Our results are clinically plausible. Thus, high age, advanced stage, high level of comorbidity, abuse of alcohol, the experience of complications after resection and male gender are all associated with increased mortality that persists at least throughout the first year after resection. The findings also suggest that the excessive mortality associated with experiencing complications may taper off as follow-up time increases. It is not possible from our study to investigate further the mechanisms underlying the excessive mortality. Based on a sample of death certificates obtained for this patient population, almost all deaths recorded are attributed to lung cancer regardless of time elapsed since resection. Accordingly, additional information on treatment and comorbidities developing after resection will be needed for a better understanding of the explanation behind the excessive mortality. Recent studies on 30- and 90-day mortality after lung cancer surgery have shown similar results. A US single institutional retrospective analysis from 2010 reported data on 1845 patient, but comparability is hampered by the inclusion of patients with benign disease [19]. Damhuis et al. [20] presented data from a regional Dutch cancer registry with a 30- and 90-day mortality of 3.6 and 6.8%, respectively, and a similar doubling in cumulative mortality has been reported in another US investigation [21]. All studies have underlined the importance of surveillance of groups of risk patients. Two large national reports have reported a doubling in cumulative mortality with increasing follow-up time after resection [22, 23]. The US National Cancer Data Base, covering more than 100000 major resections (lobectomies and pneumonectomies) between 2007 and 2011 reported that hospital volume was significantly associated with 30- and 90-day increased mortality. Further, in line with our results it was found that old age, male sex, high stage, pneumonectomy and comorbidity was associated with 30- and 90-day increased mortality [22]. The UK study included 10 991 patients operated between 2004 and 2010 and reported that age was the
5
most important predictor of death within 90 days and furthermore that performance status, lung function, stage and procedure type were associated with death [23]. Whereas we consider our Danish data to be virtually complete, the two much larger national registries from the USA and the UK are hampered with incomplete data. Thus, the US data capture only 80% of all new lung cancers and there are missing follow-up data for a considerable proportion of the patients [22]. Similarly, the UK reports cover only 52% of the expected number of new cases and with key information missing for a major part of the patients [23]. Even though the 30-day mortality in these two national inventories was on par with or tends to be somewhat larger than Danish results for both the cumulative 30-day mortality and the cumulative 90-day mortality (2.1% in Denmark vs 3.0% in the UK and 2.8% in the USA and 4.6% in Denmark vs 5.9 in the UK and 5.4 in the USA), respectively. Our study cannot elucidate the association between readmission after surgery and subsequent death. Hu et al. [24] found that a positive association between re-admission and 30- and 90-day cumulative mortality and that patient comorbidity, type of operation and socioeconomic factors are associated with readmission. It is under debate to which extent hospital volume is associated with early postoperative death. The notion is supported by the two national reports from the UK and the USA mentioned above [22, 23] whereas a comparable study from France could not confirm these findings, but found that surgeon volume was important [25]. This study confirms that the type of operation as well as the occurrence of complications represents important prognostic factors of mortality after surgery. In particular, pneumonectomy and high levels of complications are associated with a significant excess mortality. But our study adds the important information that an initial sharp decline in survival is observed within the first 50 days. The findings presented here have immediate implications for clinical practice and the further improvement of patient safety in the management of primary lung cancer. First, it is necessary to supplement the conventional 30-day postoperative mortality indicator with mortality indicators covering a much longer follow-up interval. Secondly, because of the long-term impact of mortality, patients should be monitored more closely and frequently after resection to optimize the chances of identifying patients at increased risk of death after resection. This applies in particular to resected patients with advanced disease, older patients and patients experiencing complications after resection. Finally, further studies addressing the mechanisms behind the excessive mortality after resection in primary lung cancer may provide the basis for refining the tools for the stratification of patients concerning their risk of postoperative complications and other adverse events, including premature death.
ACKNOWLEDGEMENTS We thank Peter Gustav, academic data manager, for skilled data management, particularly concerning the algorithm to calculate the Charlson’s comorbidity index from the Danish National Patient Register. Conflict of interest: none declared.
REFERENCES [1] Hakulinen T, Engholm G, Gislum M, Storm HH, Klint A, Tryggvadóttir L et al. Trends in the survival of patients diagnosed with cancers in the
THORACIC
A. Green et al. / European Journal of Cardio-Thoracic Surgery
6
[2]
[3]
[4]
[5] [6]
[7]
[8]
[9]
[10]
[11]
[12]
A. Green et al. / European Journal of Cardio-Thoracic Surgery respiratory system in the nordic countries 1964-2003 followed up to the end of 2006. Acta Oncol 2010;49:608–23. Strand T, Rostad H, Møller B, Norstein J. Survival after resection for primary lung cancer: a population based study of 3211 resected patients. Thorax 2006;61:710–5. Schneider L, Farrokhyar F, Schieman C, Hanna WC, Shargall Y, Finley CJ. The burden of death following discharge after lobectomy. Eur J Cardiothorac Surg 2014; doi:10.1093/ejcts/ezu427. Schneider L, Farrokhyar F, Schieman C, Shargall Y, D’Souza J, Camposilvan I et al. Pneumonectomy: the burden of death after discharge and predictors of surgical mortality. Ann Thorac Surg 2014;98:1976–81. Nagai K, Yoshida J, Nishimura M. Postoperative mortality in lung cancer patients. Ann Thorac Cardiovasc Surg 2007;13:373–7. Jakobsen E, Palshof T, Osterlind K, Pilegaard H. Data from a national lung cancer registry contributes to improve outcome and quality of surgery: Danish results. Eur J Cardiothorac Surg 2009;35:348–52. Damhuis R, Coonar A, Plaisier P, Dankers M, Bekkers J, Linklater K et al. A case-mix model for monitoring of postoperative mortality after surgery for lung cancer. Lung Cancer 2006;51:123–9. Harichand-Herdt S, Ramalingam SS. Gender-associated differences in lung cancer: clinical characteristics and treatment outcomes in women. Semin Oncol 2009;36:572–80. Sigel K, Bonomi M, Packer S, Wisnivesky J. Effect of age on survival of clinical stage in non-small-cell lung cancer. Ann Surg Oncol 2009;16: 1912–7. Allen MS, Darling GE, Pechet TTV, Mitchell JD, Herndon JE, Landreneau RJ et al. Morbidity and mortality of major pulmonary resections in patients with early-stage lung cancer: initial results of the randomized, prospective ACOSOG z0030 trial. Ann Thorac Surg 2006;81: 1013–20. Janssen-Heijnen MLG, Smulders S, Lemmens VEPP, Smeenk FWJM, van Geffen HJAA, Coebergh JWW. Effect of comorbidity on the treatment and prognosis of elderly patients with non-small cell lung cancer. Thorax 2004; 59:602–7. Jakobsen E, Green A, Østerlind K, Riis T, Iachina M, Palshof T. Nationwide quality improvement in lung cancer care: the role of the Danish Lung Cancer Group and Registry. J Thorac Oncol 2013;8:1238–47.
[13] Pedersen CB. The Danish civil registration system. Scand J Public Health 2011;39(Suppl. 7):22–5. [14] Bjerregaard B, Larsen OB. The Danish pathology register. Scand J Public Health 2011;39(Suppl. 7):72–4. [15] Lynge E, Sandegaard JL, Rebolj M. The Danish national patient register. Scand J Public Health 2011;39(Suppl. 7):30–3. [16] Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987;40:373–83. [17] Thygesen SK, Christiansen CF, Christensen S, Lash TL, Sørensen HT. The predictive value of ICD-10 diagnostic coding used to assess Charlson comorbidity index conditions in the population-based Danish Registry of Patients. BMC Med Res Methodol 2011;11:83 doi:10.1186/1471-228811-83. [18] Fisher LD, Lin DY. Time-dependent covariates in the Cox proportionalhazards regression model. Annu Rev Public Health 1999;20:145–57. [19] Bryant AS, Rudemiller K, Cerfolio RJ. The 30- versus 90-day operative mortality after pulmonary resection. Ann Thorac Surg 2010;89:1717–23. [20] Damhuis RA, Wijnhoven BP, Plaisier PW, Kirkels WJ, Kranse R, van Lanschot JJ. Comparison of 30-day, 90-day and in-hospital postoperative mortality for eight different cancer types. Br J Surg 2012;99:1149–54. [21] McMillan RR, Berger A, Sima CS, Lou F, Dycoco J, Rusch V et al. Thirtyday mortality underestimates the risk of early death after major resections for thoracic malignancies. Ann Thorac Surg 2014;98:1769–75. [22] Pezzi CM, Mallin K, Mendez AS, Greer Gay E, Putnam JB Jr. Ninety-day mortality after resection for lung cancer is nearly double 30-day mortality. J Thorac Cardiovasc Surg 2014;148(5):2269–77. [23] Powell HA, Tata LJ, Baldwin DR, Stanley RA, Khakwani A, Hubbard RB. Early mortality after surgical resection for lung cancer: an analysis of the English National Lung cancer audit. Thorax 2013;68:826–34. [24] Hu Y, McMurry TL, Isbell JM, Stukenborg GJ, Kozower BD. Readmission after lung cancer resection is associated with a 6-fold increase in 90-day postoperative mortality. J Thorac Cardiovasc Surg 2014;97:973–9. [25] Falcoz PE, Puyraveau M, Rivera C, Bernard A, Massard G, Mauny F et al. The impact of hospital and surgeon volume on the 30-day mortality of lung cancer surgery: a nation-based reappraisal. J Thorac Cardiovasc Surg 2014;148:841–8.
ORIGINAL ARTICLE
European Journal of Cardio-Thoracic Surgery (2015) 1–6 doi:10.1093/ejcts/ezv107
The mortality after surgery in primary lung cancer: results from the Danish Lung Cancer Registry† Anders Greena,*, Jacob Haugeb, Maria Iachinac and Erik Jakobsenb a
b c
Odense Patient data Explorative Network (OPEN), Odense University Hospital and Institute of Clinical Research, University of Southern Denmark, Odense, Denmark The Danish Lung Cancer Registry, Department of Thoracic Surgery, Odense University Hospital, Odense, Denmark Research Unit of Clinical Epidemiology, Institute of Clinical Research, University of Southern Denmark, Odense, Denmark
* Corresponding author. Odense Patient data Explorative Network (OPEN), Odense University Hospital and University of Southern Denmark, J. B. Winsløws Vej 9A, 3rd floor, DK-5000 Odense C, Denmark. Tel: +45-4088-7775; e-mail: [email protected]; [email protected] (A. Green). Received 25 November 2014; received in revised form 11 February 2015; accepted 19 February 2015
Abstract OBJECTIVES: The study has been performed to investigate the mortality within the first year after resection in patients with primary lung cancer, together with associated prognostic factors including gender, age, tumour stage, comorbidity, alcohol abuse, type of surgery and post-surgical complications. METHODS: All patients (n = 3363) from the nationwide Danish Lung Cancer Registry with first resection performed between 1 January 2007 and 31 December 2011 were analysed by Kaplan–Meier techniques and Cox-regression analysis concerning death within the first year after resection. Covariates included gender, age, comorbidity (Charlson comorbidity index), perioperative stage, type of resection, registered complications to surgery and alcohol abuse. RESULTS: The cumulative deaths after 30 days, 90 days, 180 days and 360 days were 72 (2.1%), 154 (4.6%), 239 (7.1%) and 478 (14.2%), respectively. Low stage, female gender, young age, no comorbidity, no postoperative complications, no alcohol abuse and lobectomy as type of resection were favourable for survival. CONCLUSIONS: Our results demonstrate that resection in primary lung cancer impacts mortality far beyond the initial 30 days after resection, which is conventionally considered a time window of relevance for the adverse outcome of surgery. Increased efforts should be made for optimizing the selection of patients suited for resection and for identifying patients at increased risk of death after resection. Furthermore, patients should be monitored more closely and more frequently, in particular those patients with high risk of death after resection. Keywords: Lung cancer • Resection • Mortality • Risk factors
INTRODUCTION Primary lung cancer is one of the most common types of cancers today. In Denmark alone, more than 4500 people are diagnosed every year. Despite improvements in recent years, the prognosis in lung cancer is very poor, with crude 5-year survival as low as 11–13% [1]. Resection, when applied to the right subset of patients, is the type of intervention associated with the best survival. Conventionally, survival after surgery is estimated from either the in-hospital mortality or the 30-day mortality and it is generally accepted that deaths occurring within 30 days after surgery are attributable to complications following surgery [2]. Thus, the 30-day mortality may serve as a benchmark for the overall quality of surgical intervention and appropriate selection † Presented at the 20th European Conference on General Thoracic Surgery, Essen, Germany, 10–13 June 2012.
of patients for surgery. A large number of single institution studies on postoperative mortality have been published, some of them recently [3, 4], but nationwide, population-based estimates on the 30-day mortality are limited. It is also in a national setting largely unknown whether complications after surgery and any associated excess mortality are limited to the 30-day post-surgical period or extend beyond 30 days [5–7]. The criteria for assessing the eligibility of patients for surgery have not been well-established and uniformly agreed upon, albeit it is known that factors such as age, gender, comorbidity all influence mortality [8–11]. The Danish Lung Cancer Registry (DLCR) provides unique opportunities for clinical and epidemiological research in lung cancer [12]. The present study has been performed to investigate the mortality within the first year after resection in patients with primary lung cancer, together with associated prognostic factors including gender, age, tumour stage, comorbidity, alcohol abuse, type of surgery and post-surgical complications.
© The Author 2015. Published by Oxford University Press on behalf of the European Association for Cardio-Thoracic Surgery. All rights reserved.
THORACIC
Cite this article as: Green A, Hauge J, Iachina M, Jakobsen E. The mortality after surgery in primary lung cancer: results from the Danish Lung Cancer Registry. Eur J Cardiothorac Surg 2015; doi:10.1093/ejcts/ezv107.
2
A. Green et al. / European Journal of Cardio-Thoracic Surgery
MATERIALS AND METHODS The Danish Lung Cancer Registry The DLCR has recorded all cases of primary lung cancer in Denmark, since its initiation in 2000. The database today contains information on more than 50 000 cases of lung cancer, with information on patient age, gender, tumour stage, type of surgery if applicable and many other data. Data are reported to the registry from the participating departments after completing diagnostic evaluation and specific treatments. Supplementary data on the patient’s vital status are obtained from the Danish Civil Registration System [13]. Pathology information related to all registered patients with lung cancer case is routinely retrieved from The Danish Pathology Register and data on comorbidity from the Danish National Patient Register [14, 15]. All information from these registers may be linked at patient level by means of the unique Personal Identification Number (PIN) assigned to every Danish citizen. The present analysis includes the 3363 patients with non-smallcell lung cancer (NSCLC) who had their first resection for primary lung cancer between 1 January 2007 and 31 December 2011. Patients with small-cell lung cancer (SCLC) were excluded.
Diagnostic procedures and treatment options Diagnostic procedures in suspected lung cancer are primarily performed to verify the presence of the disease and to establish the type of lung cancer, subsequently based on histology to the two main categories, SCLC (15%) and NSCLC (85%). Furthermore, clinical staging of the lung cancer, the performance of the patient and relevant comorbidities will be evaluated. For patients eligible for resection, further investigations may be performed to find the most appropriate treatment. Lung cancer patients with NSCLC are predominantly treated with surgical resection or chemo- and/or radiotherapy whereas patients with SCLC almost exclusively will be offered chemo- and/or radiotherapy. In Denmark, 20% of the NSCLC patients are eligible for surgery at the time of diagnosis. Clinical stage is the most important factor when deciding the choice of treatment, but additional risk factors considered include age, lung function and performance, abuse of alcohol or drugs as well as comorbidity. Resection is performed using a standard thoracotomy or a video-assisted minimally invasive procedure (VATS). In Denmark, 60% of the resections are performed using VATS procedures. The goal is to perform surgery with a macro- and microscopically radical resection, and to remove as little lung parenchyma as possible without compromising this need for a radical resection. Lobectomy and wedge resection are therefore the type of resections preferred. Pneumonectomy is performed when a stage requires a more extensive resection, and a smaller resection such as sub-segmental resection is performed for patients unsuited for larger resections. Less than 5% of operations will be exploratory eventually since the staging performed during operation prohibits any kind of resection.
Study variables Vital status for each patient was obtained from the Danish Civil Registration System. Age at the time of surgery and gender were available from the PIN. DLCR provided data concerning surgical staging of the lung cancer, type and method of surgery,
complications registered, lung function, time from the start of the diagnostic evaluation until surgery and alcohol abuse. During surgery, tumours were classified via the tumour, nodes metastasis system, with subsequent transformation to stages following the International Association for the Study of Lung Cancer. For the purpose of this study, perioperative stages I and II (n = 2704) were assigned to one group whereas stages III and IV (n = 625 of which 105 were stage IV) were assigned to a separate group. Resection in lung cancer was categorized as pneumonectomy, lobectomy and wedge- or segment-resection. For the purpose of this study, type and method of surgery have been categorized as 0 for the wedge and segment resection, 1 for the lobectomy and 2 for the pneumonectomy. Patients undergoing explorative surgery (n = 170, corresponding with 4.8%) were excluded from analysis. Complications in DLCR were reported by the surgical departments on a mandatory basis using the following 14 predefined categories: (i) arrhythmia, (ii) acute myocardial infarction, (iii) pneumonia, (iv) atelectasis, (v) air leakage, (vi) empyema, (vii) bronco-pleura fistula, (viii) respirator treatment, with recording of days of respirator treatment, (ix) pneumothorax, (x) neurological complications, (xi) resurgery due to bleeding, (xii) resurgery due to other causes, (xiii) infection and (xiv) other complications. Only complications occurring in the period from operation to discharge were included. For the purpose of the present analysis, patients were grouped according to whether they had no, one or more than one complication after resection. Classification of alcohol abuse in DLCR followed the definitions established by the Danish National Board of Health according to which the consumption of more than 14 units for women and 21 units for men per week is considered to be abuse. Young age has, for the purpose of this study, been defined at age being lower than the mean age for our entire patient population, 65.9 years. Co-morbidity for each patient up to 10 years before the diagnosis of lung cancer diagnosis was assessed on the basis of a search for each patient in the Danish National Patient Register, established in 1977. This register contains data on all patient contacts, including coding of diagnosis and procedures, at Danish hospitals. For the classification of comorbidity, we used the Charlson comorbidity index (CCI) [16, 17], with grouping according to increasing level of comorbidity: CCI score = 0; CCI score = 1 and CCI score >1. For the present analysis, all hospital contacts associated with any cancer diagnosis and registered within 150 days before the date of lung cancer diagnosis were excluded to avoid any false contributions to CCI from diagnoses of other cancers turning out eventually to be primary lung cancer.
Statistical analysis Overall survival was illustrated by Kaplan–Meier plots by the variables such as postoperative stage, CCI, complications and type of surgery. To estimate the effect on mortality of the variables of interest by different time intervals (from 0 to 30 days, from 31 to 90 days, from 91 to 180 and from 181 to 360 days), we constructed the time-varying covariates for each variable of interest. For age, the new variables comprised (i) age30, where the hazard ratio (HR) for this variable will indicate the effect of age on 0 to 30-day survival; (ii) age90, where the HR for this variable will indicate the effect of age on 31- to 90-day survival; (iii) age180, where the HR for this variable will indicate the effect of age on 91- to 180-day survival and (iv) age360, where the HR for this variable will
indicate the effect of age on 181- to 360-day survival. In a similar way, time-varying covariates were defined for the other variables of interest (gender, stage, CCI, alcohol abuse, type of operation and complications). Then, we implemented an extended Cox model [18]. After fitting the model with the four time-varying covariates for each variable, we tested the heterogeneity between the intervals, variable by variable using a Wald test. Using the Wald test, we also tested that the HR for the wedge and segment resection to lobectomy is the same as the HR from lobectomy to pneumonectomy for all time intervals. Proportions and HRs are presented with 95% confidence limits in parentheses. The statistical analysis was performed using the STATA version 12.
Ethical aspects DLCR has been approved as a national clinical database by the Danish National Board of Health ( j.nr. 7-201-03-15/1). Permission for the present study was granted by the Danish Data Protection Agency ( j.nr. 2008-58-0035) and the National Board of Health ( j.nr. 7-505-29-1863/1).
RESULTS The clinical characteristics and number of deaths by time interval after resection are given in Table 1 for the 3363 patients from DLCR that underwent resection for primary lung cancer in
3
Denmark during the period 1 January 2007 through 31 December 2011. The majority of the patients had stage less than stage III [n = 2704, 81% (80%; 83%) of those classified], no alcohol abuse [n = 3021, 90% (89%; 92%), no postoperative complications [n = 2412, 72% (70%; 73%)] and had lobectomy as the type of operation [n = 2658, 79% (78%; 81%) of those classified]. A total of 478 patients died within 360 days after surgery corresponding with a cumulative mortality rate at 2% (2%; 3%) after 30 days, increasing to 5% (4%; 5%), 7% (6%; 8%) and 14% (13%; 15%) after 90, 180 and 360 days, respectively. Figure 1 shows the cumulative survival up to 500 days after resection for the study population by stage, CCI, postoperative complication status and type of resection. Low stage, no comorbidity, no postoperative complications and lobectomy as type of resection were characterized by best survival. For the groupings by CCI complications, gradients were seen with worsening survival with increasing level of comorbidity and increasing number of complications, respectively. In general, the survival curves tended to follow linear trends, however for the group with high number of complications and for patients with pneumonectomy as type of resection, there was an initial sharp decline in survival within the first 50 and 30 days, respectively. The full model in the detailed statistical analysis suggested that gender, age, abuse and CCI all had effects that may be considered constant across the intervals of the follow-up time since resection. For the other variables, HRs were estimated for each follow-up interval separately (Table 2). For demographic variables, females had a statistically significant lower mortality than males, HR = 0.79 (0.68; 0.93) and older age (defined as age above the mean age
Table 1: Overview of the study population by variables and deaths during the first year after resection
All patients Gender Male Female Age Younger than mean Older than mean Stage Unclassified 0, I, II III, IV Charlson comorbidity index 0 1 >1 Alcohol abuse Yes No Type of resection Not classified Wedge and segment resection Lobectomy Pneumonectomy Complications 0 1 >1
Number of patients
0–30 day mortality
31–90 day mortality
91–180 day mortality
181–360 day mortality
360-day survivors
3363
72
82
85
239
2885
1683 (50) 1680 (50)
50 (69) 22 (31)
48 (59) 34 (41)
52 (61) 33 (39)
134 (56) 105 (44)
1399 (48) 1486 (52)
1539 (46) 1824 (54)
20 (28) 52 (72)
18 (22) 64 (78)
33 (39) 52 (61)
104 (44) 135 (56)
1364 (47) 1521 (53)
34 (1) 2704 (80) 625 (19)
0 (0) 51 (71) 21 (29)
1 (1) 55 (67) 26 (32)
1 (1) 47 (55) 37 (44)
2 (1) 148 (62) 89 (37)
30 (1) 2403 (83) 452 (16)
1700 (51) 745 (22) 918 (27)
26 (36) 21 (29) 25 (35)
31 (38) 17 (21) 34 (41)
41 (48) 17 (20) 27 (32)
113 (47) 62 (26) 64 (27)
1489 (52) 628 (22) 768 (27)
342 (10) 3021 (90)
13 (18) 59 (82)
12 (15) 70 (85)
16 (19) 69 (81)
29 (12) 210 (88)
272 (9) 2613 (91)
4 (0) 451 (13) 2658 (79) 250 (7)
0 (0.0) 8 (11) 46 (64) 18 (25)
0 (0.0) 14 (17) 57 (70) 11 (13)
0 (0.0) 14 (16) 60 (71) 11 (13)
0 (0.0) 36 (15) 176 (74) 27 (11)
4 (0) 379 (13) 2319 (80) 183 (6)
2412 (72) 628 (19) 323 (10)
25 (35) 17 (24) 30 (42)
36 (44) 20 (24) 26 (32)
48 (57) 25 (29) 12 (14)
161 (67) 45 (19) 33 (14)
2142 (74) 521 (18) 222 (8)
Data are numbers indicating deaths occurring in the follow-up intervals (days) from date of resection (percentage of all patients).
THORACIC
A. Green et al. / European Journal of Cardio-Thoracic Surgery
4
A. Green et al. / European Journal of Cardio-Thoracic Surgery
Figure 1: Kaplan–Meier plots (with 95% CI) of survival by selected variables. Note truncation of the Y axis.
Table 2: Summary of the statistical analysis of the effect on mortality by variable and interval of the follow-up time from resection
Gender Age Stage30 Stage90 Stage180 Stage360 CCI Abuse30 Lobectomy vs wedge and segment resection Pneumonectomy vs wedge and segment resection Optype90 Optype180 Optype360 Compl30 Compl90 Compl180 Compl360
Hazard ratio
95% CI
0.79 1.47 1.36 2.14 3.71 3.10 1.19 1.58 0.75 2.55 0.98 1.05 1.12 3.03 2.39 1.53 1.18
0.68 1.24 0.78 1.34 2.41 2.37 1.08 1.26 0.33 1.01 0.60 0.66 0.85 2.26 1.84 1.15 0.95
0.93 1.73 2.39 3.43 5.73 4.04 1.30 1.99 1.68 6.39 1.62 1.67 1.48 4.06 3.11 2.03 1.47
P-values for departure from unity
P-values for Wald test
0.005 0.000 0.279 0.002 0.000 0.000 0.000 0.000 0.483 0.047 0.945 0.843 0.426 0.000 0.000 0.004 0.133
0.558 0.094 0.049
0.167 0.267 0.007
0.001
Numbers appearing in labels reflect the end of the time interval concerned. CCI: Charlson co-morbidity index; Abuse: alcohol abuse; Optype: type of resection; Compl: complications.
(65.9 years) of the entire study population) was associated with increased mortality, HR = 1.47 (1.24; 1.73). Among the clinically related variables, high level of CCI was associated with increased mortality, HR = 1.19 (1.08; 1.30). High stage had no statistically significant effect on 30-day mortality [HR = 1.36, (0.78; 2.39)] but was associated with increased mortality during the time period from 31 to 360 days after resection with
HR 2.14 (1.34; 3.43) for the 31 to 90 days after resection, 3.71 (2.41; 5.73) for the 91–180 days after resection and 3.10 (2.37; 4.04) the 181–360 days after resection, respectively. Wedge and segment resection had no significant effect on 30-day mortality compared with lobectomy [HR = 0.75 (0.33; 1.68)], whereas pneumonectomy was associated with increased 30-day mortality compared with lobectomy [HR = 2.55 (1.01; 6.39)]. The wedge and
segment resection and pneumonectomy had a proportional and non-significant effect on 31- to 360-day mortality with hazard rations at 0.98 (0.60; 1.62) for the 31–90 days after resection, 1.05 (0.66; 1.67) for the 91–180 days after resection and 1.12 (0.85; 1.48) for the 181–360 days after resection. For high level of complications, there was an initial highly increased mortality and even though there was a tendency to decreasing excess mortality with increasing follow-up interval from HR at 3.03 (2.26; 4.06) for the time period from 0 to 30 days after resection to HR at 1.18 (0.95; 1.47) for the time period from 181 to 360 days after resection, the effect remained statistically significant throughout the full period of 360 days after resection.
DISCUSSION This study makes use of one of the largest population-based lung cancer registries with high completeness of registration and data validity particularly concerning the inclusion of surgical data [12]. Since the unique features of the Danish Civil Registration System makes it possible to ensure the complete follow-up of the vital status of all patients [13], we believe that our results are valid for resection for primary lung cancer in Denmark. The results underline that excess mortality has an impact far beyond the initial 30 days after resection, and that many factors, including advanced disease stage and the fact of experiencing complications after resection have a negative effect on survival at least through the first year after resection. Also type of resection, in particular pneumonectomy, impacts survival negatively. Our results are clinically plausible. Thus, high age, advanced stage, high level of comorbidity, abuse of alcohol, the experience of complications after resection and male gender are all associated with increased mortality that persists at least throughout the first year after resection. The findings also suggest that the excessive mortality associated with experiencing complications may taper off as follow-up time increases. It is not possible from our study to investigate further the mechanisms underlying the excessive mortality. Based on a sample of death certificates obtained for this patient population, almost all deaths recorded are attributed to lung cancer regardless of time elapsed since resection. Accordingly, additional information on treatment and comorbidities developing after resection will be needed for a better understanding of the explanation behind the excessive mortality. Recent studies on 30- and 90-day mortality after lung cancer surgery have shown similar results. A US single institutional retrospective analysis from 2010 reported data on 1845 patient, but comparability is hampered by the inclusion of patients with benign disease [19]. Damhuis et al. [20] presented data from a regional Dutch cancer registry with a 30- and 90-day mortality of 3.6 and 6.8%, respectively, and a similar doubling in cumulative mortality has been reported in another US investigation [21]. All studies have underlined the importance of surveillance of groups of risk patients. Two large national reports have reported a doubling in cumulative mortality with increasing follow-up time after resection [22, 23]. The US National Cancer Data Base, covering more than 100000 major resections (lobectomies and pneumonectomies) between 2007 and 2011 reported that hospital volume was significantly associated with 30- and 90-day increased mortality. Further, in line with our results it was found that old age, male sex, high stage, pneumonectomy and comorbidity was associated with 30- and 90-day increased mortality [22]. The UK study included 10 991 patients operated between 2004 and 2010 and reported that age was the
5
most important predictor of death within 90 days and furthermore that performance status, lung function, stage and procedure type were associated with death [23]. Whereas we consider our Danish data to be virtually complete, the two much larger national registries from the USA and the UK are hampered with incomplete data. Thus, the US data capture only 80% of all new lung cancers and there are missing follow-up data for a considerable proportion of the patients [22]. Similarly, the UK reports cover only 52% of the expected number of new cases and with key information missing for a major part of the patients [23]. Even though the 30-day mortality in these two national inventories was on par with or tends to be somewhat larger than Danish results for both the cumulative 30-day mortality and the cumulative 90-day mortality (2.1% in Denmark vs 3.0% in the UK and 2.8% in the USA and 4.6% in Denmark vs 5.9 in the UK and 5.4 in the USA), respectively. Our study cannot elucidate the association between readmission after surgery and subsequent death. Hu et al. [24] found that a positive association between re-admission and 30- and 90-day cumulative mortality and that patient comorbidity, type of operation and socioeconomic factors are associated with readmission. It is under debate to which extent hospital volume is associated with early postoperative death. The notion is supported by the two national reports from the UK and the USA mentioned above [22, 23] whereas a comparable study from France could not confirm these findings, but found that surgeon volume was important [25]. This study confirms that the type of operation as well as the occurrence of complications represents important prognostic factors of mortality after surgery. In particular, pneumonectomy and high levels of complications are associated with a significant excess mortality. But our study adds the important information that an initial sharp decline in survival is observed within the first 50 days. The findings presented here have immediate implications for clinical practice and the further improvement of patient safety in the management of primary lung cancer. First, it is necessary to supplement the conventional 30-day postoperative mortality indicator with mortality indicators covering a much longer follow-up interval. Secondly, because of the long-term impact of mortality, patients should be monitored more closely and frequently after resection to optimize the chances of identifying patients at increased risk of death after resection. This applies in particular to resected patients with advanced disease, older patients and patients experiencing complications after resection. Finally, further studies addressing the mechanisms behind the excessive mortality after resection in primary lung cancer may provide the basis for refining the tools for the stratification of patients concerning their risk of postoperative complications and other adverse events, including premature death.
ACKNOWLEDGEMENTS We thank Peter Gustav, academic data manager, for skilled data management, particularly concerning the algorithm to calculate the Charlson’s comorbidity index from the Danish National Patient Register. Conflict of interest: none declared.
REFERENCES [1] Hakulinen T, Engholm G, Gislum M, Storm HH, Klint A, Tryggvadóttir L et al. Trends in the survival of patients diagnosed with cancers in the
THORACIC
A. Green et al. / European Journal of Cardio-Thoracic Surgery
6
[2]
[3]
[4]
[5] [6]
[7]
[8]
[9]
[10]
[11]
[12]
A. Green et al. / European Journal of Cardio-Thoracic Surgery respiratory system in the nordic countries 1964-2003 followed up to the end of 2006. Acta Oncol 2010;49:608–23. Strand T, Rostad H, Møller B, Norstein J. Survival after resection for primary lung cancer: a population based study of 3211 resected patients. Thorax 2006;61:710–5. Schneider L, Farrokhyar F, Schieman C, Hanna WC, Shargall Y, Finley CJ. The burden of death following discharge after lobectomy. Eur J Cardiothorac Surg 2014; doi:10.1093/ejcts/ezu427. Schneider L, Farrokhyar F, Schieman C, Shargall Y, D’Souza J, Camposilvan I et al. Pneumonectomy: the burden of death after discharge and predictors of surgical mortality. Ann Thorac Surg 2014;98:1976–81. Nagai K, Yoshida J, Nishimura M. Postoperative mortality in lung cancer patients. Ann Thorac Cardiovasc Surg 2007;13:373–7. Jakobsen E, Palshof T, Osterlind K, Pilegaard H. Data from a national lung cancer registry contributes to improve outcome and quality of surgery: Danish results. Eur J Cardiothorac Surg 2009;35:348–52. Damhuis R, Coonar A, Plaisier P, Dankers M, Bekkers J, Linklater K et al. A case-mix model for monitoring of postoperative mortality after surgery for lung cancer. Lung Cancer 2006;51:123–9. Harichand-Herdt S, Ramalingam SS. Gender-associated differences in lung cancer: clinical characteristics and treatment outcomes in women. Semin Oncol 2009;36:572–80. Sigel K, Bonomi M, Packer S, Wisnivesky J. Effect of age on survival of clinical stage in non-small-cell lung cancer. Ann Surg Oncol 2009;16: 1912–7. Allen MS, Darling GE, Pechet TTV, Mitchell JD, Herndon JE, Landreneau RJ et al. Morbidity and mortality of major pulmonary resections in patients with early-stage lung cancer: initial results of the randomized, prospective ACOSOG z0030 trial. Ann Thorac Surg 2006;81: 1013–20. Janssen-Heijnen MLG, Smulders S, Lemmens VEPP, Smeenk FWJM, van Geffen HJAA, Coebergh JWW. Effect of comorbidity on the treatment and prognosis of elderly patients with non-small cell lung cancer. Thorax 2004; 59:602–7. Jakobsen E, Green A, Østerlind K, Riis T, Iachina M, Palshof T. Nationwide quality improvement in lung cancer care: the role of the Danish Lung Cancer Group and Registry. J Thorac Oncol 2013;8:1238–47.
[13] Pedersen CB. The Danish civil registration system. Scand J Public Health 2011;39(Suppl. 7):22–5. [14] Bjerregaard B, Larsen OB. The Danish pathology register. Scand J Public Health 2011;39(Suppl. 7):72–4. [15] Lynge E, Sandegaard JL, Rebolj M. The Danish national patient register. Scand J Public Health 2011;39(Suppl. 7):30–3. [16] Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987;40:373–83. [17] Thygesen SK, Christiansen CF, Christensen S, Lash TL, Sørensen HT. The predictive value of ICD-10 diagnostic coding used to assess Charlson comorbidity index conditions in the population-based Danish Registry of Patients. BMC Med Res Methodol 2011;11:83 doi:10.1186/1471-228811-83. [18] Fisher LD, Lin DY. Time-dependent covariates in the Cox proportionalhazards regression model. Annu Rev Public Health 1999;20:145–57. [19] Bryant AS, Rudemiller K, Cerfolio RJ. The 30- versus 90-day operative mortality after pulmonary resection. Ann Thorac Surg 2010;89:1717–23. [20] Damhuis RA, Wijnhoven BP, Plaisier PW, Kirkels WJ, Kranse R, van Lanschot JJ. Comparison of 30-day, 90-day and in-hospital postoperative mortality for eight different cancer types. Br J Surg 2012;99:1149–54. [21] McMillan RR, Berger A, Sima CS, Lou F, Dycoco J, Rusch V et al. Thirtyday mortality underestimates the risk of early death after major resections for thoracic malignancies. Ann Thorac Surg 2014;98:1769–75. [22] Pezzi CM, Mallin K, Mendez AS, Greer Gay E, Putnam JB Jr. Ninety-day mortality after resection for lung cancer is nearly double 30-day mortality. J Thorac Cardiovasc Surg 2014;148(5):2269–77. [23] Powell HA, Tata LJ, Baldwin DR, Stanley RA, Khakwani A, Hubbard RB. Early mortality after surgical resection for lung cancer: an analysis of the English National Lung cancer audit. Thorax 2013;68:826–34. [24] Hu Y, McMurry TL, Isbell JM, Stukenborg GJ, Kozower BD. Readmission after lung cancer resection is associated with a 6-fold increase in 90-day postoperative mortality. J Thorac Cardiovasc Surg 2014;97:973–9. [25] Falcoz PE, Puyraveau M, Rivera C, Bernard A, Massard G, Mauny F et al. The impact of hospital and surgeon volume on the 30-day mortality of lung cancer surgery: a nation-based reappraisal. J Thorac Cardiovasc Surg 2014;148:841–8.
