The role of comorbidity in the management and prognosis in nonsmall cell lung cancer: a population-based study

Abstract

Background: Coexisting disease constitutes a challenge for the provision of optimal cancer care. The influence of comorbidity on lung cancer management and prognosis remains incompletely understood. We assessed the influence of comorbidity on treatment intensity and prognosis in a population-based setting in patients with nonsmall cell lung cancer.

Material and methods: Our study was based on information available in Lung Cancer Data Base Sweden (LcBaSe), a database generated by record linkage between the National Lung Cancer Register (NLCR) and several other population-based registers in Sweden. The NLCR includes data on clinical characteristics on 95% of all patients with lung cancer in Sweden since 2002. Comorbidity was assessed using the Charlson Comorbidity Index. Logistic regression and time to event analysis was used to address the association between comorbidity and treatment and prognosis.

Results: In adjusted analyses encompassing 19,587 patients with a NSCLC diagnosis and WHO Performance Status 0–2 between 2002 and 2011, those with stage-IA–IIB disease and severe comorbidity were less likely to be offered surgery (OR: 0.45; 95% CI: 0.36–0.57). In late-stage disease (IIIB–IV), severe comorbidity was also associated with lower chemotherapy treatment intensity (OR: 0.76; 95% CI: 0.65–0.89). In patients with early, but not late-stage disease, severe comorbidity in adjusted analyses was associated with an increased all-cause mortality, while lung cancer-specific mortality was largely unaffected by comorbidity burden.

Conclusions: Comorbidity contributes to the poor prognosis in NSCLC patients. Routinely published lung cancer survival statistics not considering coexisting disease conveys a too pessimistic picture of prognosis. Optimized management of comorbid conditions pre- and post-NSCLC-specific treatment is likely to improve outcomes.

Introduction

Lung cancer is the fifth most common type of malignancy and the most common cancer-related cause of death in Sweden [1]. Since only about one-fifth of nonsmall cell lung cancer (NSCLC) patients is diagnosed with potentially curable early-stage disease, the overall prognosis is generally poor with survival rates being strongly associated with stage at diagnosis [2].

The presence of coexisting disease constitutes a challenge for the provision of optimal cancer care, particularly in elderly patients. Accurate assessment and consideration of comorbidity in clinical decision making is likely to improve several dimensions of patient outcomes, including a reduction of therapeutic risks related to medical conditions beyond the cancer site and extent of malignant disease [3]. Because of a high median age at diagnosis [4] and that smoking is a shared major risk factor with many other chronic diseases, the prevalence of coexisting conditions can be expected to be particularly high in in lung cancer patients [5]. However, the influence of comorbidity on lung cancer management and prognosis remains incompletely understood. While it has been suggested that comorbidity is prognostically less relevant in malignancies with poor survival, [6] there is evidence of not only a lower treatment intensity and poorer survival in men and women with lung cancer with severe comorbidity, but also an increased perioperative mortality in operable NSCLC patients [7–15]

The aim of our study was twofold: firstly, to compare the prevalence of comorbid conditions between NSCLC patients and randomly selected lung cancer-free controls in a population-based setting; secondly, to examine the influence of comorbidity on patterns of treatment and prognosis in patients with NSCLC.

Material and methods

Data collection

Our study was based on information available in Lung Cancer Data Base Sweden (LcBaSe), a research database generated by means of record linkage between the National Lung Cancer Register (NLCR) and several other population-based registers in Sweden. The NLCR includes 95% of all patients diagnosed with lung cancer in Sweden since 2002 with information on gender, date of diagnosis, age at diagnosis, smoking status, histopathology, clinical stage at diagnosis and initial treatment [16]. Primary treatment is registered as surgery, stereotactic radiotherapy (available from 2007 and onwards), chemo/RT (available from 2007 and onwards), chemotherapy and radiotherapy. Information is also available on WHO performance status (PS), based on a physician assessment of functional activity at time of diagnosis. PS is based on a scale ranging from 0 (fully active) to five (dead), with a gradually decreasing capability of self-care [17]. For each individual with lung cancer, five controls were randomly sampled among lung cancer-free men and women in the Swedish population, matched for birth year, sex and county of residence.

Using the personal identity number assigned to all Swedish permanent residents and available in all registers, additional individual level information was retrieved from the National Patient Register, the Swedish Cancer Register, a Social database and the Cause of Death Register. Since 1987, the National Patient Register records information on hospital admissions and discharges diagnosis coded according to the International Classification of Diseases (ICD) from all hospitals in Sweden. Law mandates reporting of all newly diagnosed tumors to the Swedish Cancer Register (SCR). With a very high completeness for solid tumors, information in the SCR includes information on tumor site, date of diagnosis, histological type and stage at diagnosis [16]. Data on educational level was retrieved from the longitudinal integration database for health insurance and the labor market (LISA), a nationwide social database with individual level information available on highest achieved education for all residents aged 16 and older [18]. Information on date and underlying cause of death was obtained from the Cause of Death Register, coded according to the ICD. The number of nonreported events is low, estimated at less than 1% of all deaths [19].

Comorbidity

Information on coexisting medical conditions was obtained from the National Patient Register from which a main diagnosis and up to seven secondary discharge diagnoses from in-hospital stays up to 15 years before the date of the lung cancer diagnosis were retrieved. Malignancies other than lung cancer were identified in the Swedish Cancer Register. The Charlson comorbidity index (CCI) was derived for each lung cancer patient by summing the weighted scores for all comorbidities. Originally suggested by Charlson et al. as a predictor of one-year mortality [20], the CCI has become one of the most commonly used tools to assess the burden of comorbidity. The index is based on 19 disease groups categorized according to their estimated influence on mortality with a specific weight assigned to each disease category (1, 2, 3 and 6) [21]. The CCI has been validated in several studies, most of which have shown that the index represents at good predictor of one-year mortality [20]. For the purpose of the present study, all 19 medical conditions except metastatic carcinoma were identified 15 years prior to diagnosis of NSCLC. Since almost all (88%) metastastic carcinomas were registered in the Patient Register 0–6 months before the diagnosis of lung cancer (and were considered to be related to that event), we restricted the inclusion of metastatic carcinomas in the CCI to those recorded 15 years to 6 months before the date of the lung cancer diagnosis. Patients were categorized into three comorbidity groups: no (CCI 0), mild (CCI 1-2) and severe comorbidity (CCI 3+).

Study population

We identified a total of 32,676 men and women with a lung cancer diagnosis in the NLCR between 2002 and 2011. Among these, 24,840 (76.0%) were diagnosed with a nonsmall cell lung cancer (NSCLC) with pathology confirmed histology of adenocarcinoma, squamous cell carcinoma, large cell carcinoma, or adenosquamous carcinoma. Patients were excluded from the study population if the diagnosis was based on clinical assessment only (n = 22) or unknown (n = 115). The final eligible cohort encompassed 24,703 NSCLC patients divided into clinical subgroups (stage IA–IIB, IIIA and IIIB–IV) (Tables 1 and 2). The main analyses were restricted to 19,587 NSCLC patients with PS 0–2, a patient group with a functional status generally considered eligible for treatment (Tables 3 and 4, Figures 1 and 2).

Figure 1. Cumulative- and cause-specific survival by Charlson comorbidity index (CCI).

Figure 2. Cumulative probability of death by lung cancer or other causes by Charlson comorbidity index (CCI).

Table 1. Charlon comorbidity index (CCI) and medical conditions within CCI among patients diagnosed with NSCLC and lung cancer-free controls.

	Cases	%	Controls	%	p value
Charlson Comorbidity Index (CCI)
CCI 0	13,756	55.7	128,759	69.6	–
CCI 1–2	8011	32.4	42,902	23.2	–
CCI 3+	2936	11.9	13,294	7.2	<.001
Medical condition
Myocardial infarction	2006	8.1	11,427	6.2	<.001
Congestive heart failure	1719	7.0	9446	5.1	<.001
Peripheral vascular disease	1409	5.7	4027	2.2	<.001
Cerebrovascular disease	2215	9.0	14,554	7.9	<.001
Dementia	117	0.5	2389	1.3	<.001
Chronic pulmonary disease	3117	12.6	7125	3.9	<.001
Connective tissue disease-rheumatic disease	615	2.5	2891	1.6	<.001
Peptic ulcer disease	732	3.0	3265	1.8	<.001
Mild liver disease	295	1.2	942	0.5	<.001
Diabetes without complications	1836	7.4	10,155	5.5	<.001
Diabetes with complications	326	1.3	1840	1.0	<.001
Paraplegia and hemiplegia	108	0.4	647	0.4	.036
Renal disease	325	1.3	1949	1.1	<.001
Cancer	3251	13.2	17,664	9.6	<.001
Moderate or severe liver disease	73	0.3	233	0.1	<.001
Metastatic carcinoma^a	158	0.6	1258	0.7	.49
AIDS/HIV	6	0.0	24	0.0	.158

^a 15 years to 6 months prior index date for the index cases.

Table 2. Demographic and clinical characteristics among patients diagnosed with NSCLC by Charlson comorbidity index (CCI).

	CCI 0		CCI 1–2		CCI 3+			Total
	n	%	n	%	n	%	p value	n	%
Sex
Male	6776	51.4	4541	34.4	1872	14.2	–	13,189	100
Female	6980	60.6	3470	30.1	1064	9.2	<.001	11,514	100
Age at diagnosis
0–59	3259	76.3	833	19.5	177	4.1	–	4269	100
60–69	5079	60.9	2497	29.9	770	9.2	–	8346	100
70–79	3981	47.1	3182	37.7	1287	15.2	–	8450	100
80–89	1394	39.9	1425	40.8	676	19.3	–	3495	100
90+	43	30.1	74	51.7	26	18.2	<.001	143	100
Level of education
Low	5896	51.6	4010	35.1	1528	13.4	–	11,434	100
Middle	5483	58.3	2897	30.8	1032	11.0	–	9412	100
High	2129	62.5	963	28.3	312	9.2	<.001	3404	100
Smoking history
Smoker	6761	59.8	3420	30.3	1123	9.9	–	11,304	100
Former smoker	5000	49.3	3613	35.6	1522	15.0	–	10,135	100
Never smoker	1689	63.4	754	28.3	219	8.2	<.001	2662	100
Calender period
2002–2006	6490	57.4	3626	32.1	1191	10.5	–	11,307	100
2007–2011	7266	54.2	4385	32.7	1745	13.0	<.001	13,396	100
Cause of death
Alive	2248	59.0	1194	31.4	365	9.6	–	3807	100
Lung cancer	10,562	57.2	5858	31.7	2037	11.0	–	18,457	100
Other Causes	946	38.8	959	39.3	534	21.9	<.001	2439	100
WHO PS
0	3162	67.0	1244	26.4	311	6.6	–	4717	100
1	5511	58.3	2946	31.2	996	10.5	–	9453	100
2	2718	50.2	1939	35.8	760	14.0	–	5417	100
3	1388	45.5	1135	37.2	529	17.3	–	3052	100
4	536	46.5	416	36.1	201	17.4	<.001	1153	100
Histopathology
Adenocarcinoma	7955	59.3	4078	30.4	1388	10.3	–	13,421	100
Squamous cell	3260	48.3	2436	36.1	1054	15.6	–	6750	100
Large cell	2381	56.1	1398	33.0	462	10.9	–	4241	100
Adenosquamous	160	55.0	99	34.0	32	11.0	<.001	291	100
Stage at diagnosis
IA–IIB	2827	48.9	2112	36.5	845	14.6	–	5784	100
IIIA	1115	53.7	693	33.3	270	13.0	–	2078	100
IIIB–IV	9589	58.6	5026	30.7	1743	10.7	<.001	16,358	100
Multidisciplinary conf.
No	5107	55.3	3017	32.7	1116	12.1		9240	100
Yes	7933	56.4	4522	32.1	1620	11.5	.202	14,075	100

Table 3. Treatment among patients diagnosed with NSCLC by Charlson comorbidity index (CCI), binary logistic regression with odds ratios and/or 95% confidence intervals (CI).

	CCI 0				CCI 1-2				CCI 3+				Total
	n	%	aOR**	95% CI	n	%	aOR**	95% CI	n	%	aOR**	95% CI	n	%
Stage IA–IIB, who PS 0–2	–	–	–	–	–	–	–	–	–	–	–	–	–	–
Planned surgery	2178	80.7	1.00	ref.	1334	69.0	0.69	0.58–0.83	413	56.4	0.46	0.36–0.57	3925	73.2
Stereotactic radiotherapy*	57	14.8	1.00	ref.	112	32.4	1.94	1.22–3.10	84	46.7	2.51	1.46–4.32	253	27.8
Stage IIIA, who PS 0–2
Planned surgery	207	20.5	1.00	ref.	84	13.8	0.92	0.68–1.24	21	9.8	0.74	0.44–1.23	312	17.0
Chemo/RT*	278	48.1	1.00	ref.	126	36.2	0.74	0.55–0.99	53	39.3	1.12	0.73–1.71	457	43.1
Radiotherapy	489	48.3	1.00	ref.	268	44.0	0.80	0.64–0.99	105	49.5	1.12	0.82–1.53	862	47.0
Stage IIIB–IV, who PS 0–2
Chemotherapy	6176	82.5	1.00	ref.	2668	76.6	0.97	0.87–1.08	754	69.7	0.76	0.65–0.89	9598	79.6
Radiotherapy	1531	20.5	1.00	ref.	741	21.3	1.03	0.92–1.15	224	20.7	1.03	0.87–1.23	2496	20.7

*Only available from 2007 and onwards.

**Adjusted odds ratios; adjusted for sex, age at diagnosis, year of diagnosis, level of education, smoking history, stage at diagnosis, performance status and histology.

Table 4. Overall and cause-specific mortality expressed as crude and adjusted hazard ratios (HR) and 95% confidence intervals (CI) by stage at diagnosis and performance status 0–2.

	Overall mortality				Cause-specific mortality
	HR	95% CI	aHR*	95% CI	HR	95% CI	aHR*	95% CI
Stage IA–IIB, who PS 0–2
CCI 0	1.00	ref.	1.00	ref.	1.00	ref.	1.00	ref.
CCI 1–2	1.32	1.21–1.43	1.17	0.96–1.43	1.13	1.03–1.24	0.90	0.71–1.13
CCI 3+	1.77	1.59–1.96	1.50	1.18–1.90	1.33	1.17–1.52	1.08	0.81–1.44
Stage IIIA, who PS 0–2
CCI 0	1.00	ref.	1.00	ref.	1.00	ref.	1.00	ref.
CCI 1–2	1.29	1.15–1.44	1.12	0.95–1.32	1.20	1.07–1.35	1.08	0.91–1.29
CCI 3+	1.44	1.23–1.70	1.07	0.85–1.34	1.23	1.03–1.48	0.91	0.71–1.17
Stage IIIB–IV, who PS 0–2
CCI 0	1.00	ref.	1.00	ref.	1.00	ref.	1.00	ref.
CCI 1–2	1.09	1.05–1.14	0.97	0.93–1.02	1.06	1.01–1.10	0.95	0.91–1.00
CCI 3+	1.19	1.11–1.27	0.97	0.91–1.04	1.09	1.02–1.17	0.91	0.85–0.98

* Adjusted hazard ratios; adjusted for treatment, sex, age at diagnosis, year of diagnosis, level of education, smoking history, stage at diagnosis, performance status and histology.

Statistical methods

Distributions of the Charlson Comorbidity Index (CCI) by clinical and demographic subgroups were tested using the Pearson chi-square test. In binary logistic regression models, adjusted odds ratios (OR) with 95% confidence intervals (CI) were used to assess whether treatment differed by comorbidity burden by clinical subgroups and performance status 0–2. To evaluate comorbidity burden in relation to mortality, the outcomes of interest were lung cause-specific mortality (C34.9 in ICD 9/10) and all-cause mortality. Survival time was defined as the interval between the date of lung cancer diagnosis and the date of death, or emigration or end of follow-up (31 December 2011). Cause-specific and overall survival was assessed by the Kaplan–Meier method by clinical subgroups and performance status 0–2.

The relative risk of death of lung cancer and all causes was expressed as hazard ratios (HR) with 95% confidence internals (CI) using univariate and multivariable Cox regression models. The multivariable models were adjusted for sex, age at diagnosis, year of diagnosis, education, smoking history, stage at diagnosis, performance status, histology and initial treatment. The assumption of proportional hazards in the Cox regression models were not violated and was verified visually and tested based on weighted residuals. In addition, a cumulative incidence function was derived to address the different causes of deaths in patients with comorbidity burden and stage IA–IIB diagnosis.

All tests were two-sided and statistical significant was considered at a 5% level. Statistical analyzes were performed using R 9.2 [22].

This study was approved by the Regional Ethics Board in Stockholm (2012/4:6).

Results

Comorbidity burden in NSCLC patients and lungcancer-free controls

The burden of comorbidity expressed as the Charlson Comorbidity Index (CCI) was significantly higher in NSCLC patients compared to lung cancer-free controls. The proportion of NSCLC patients and controls with no records of comorbidity (CCI 0) in the Patient Register was 55.7% and 69.6%, respectively (Table 1).

All medical conditions included in the CCI except metastatic carcinoma and AIDS/HIV were more prevalent among patients with lung cancer. The most common recorded conditions in NSCLC patients were malignancies other than lung cancer (13.2%), chronic pulmonary disease (12.6%) and cerebrovascular disease (9.0%). The corresponding prevalence in controls was 9.6%, 3.9% and 7.9% (Table 1).

Comorbidity burden by demographic and clinical characteristics

Among patients with NSCLC, there were significant differences in comorbidity burden by all demographic and clinical characteristics. As expected, both high age at diagnosis and low educational level were associated with a higher comorbidity burden. Of special note was that a higher proportion of patients with early compared to late stage disease had severe comorbidity (14.6% and 10.7%, respectively) (Table 2).

A similar magnitude relative difference remained following exclusion of comorbidities registered during a six-month period prior to the date of lung cancer diagnosis and also in a multivariable analysis adjusted for age, sex, WHO performance score and smoking history (data not shown). Comorbidity was unrelated to the likelihood of clinical decision making in a multidisciplinary conference setting (MDC).

Comorbidity burden and clinical decision making in NSCLC patients with PS 0–2

Among patients with early stage-IA–IIB disease and performance status 0–2, surgery was significantly less likely to be offered to those with severe comorbidity (OR: 0.46; 95% CI: 0.36–0.57), while the opposite was observed for stereotactic radiotherapy (OR: 2.51; 95% CI: 1.46–4.32) (Table 3).

In patients with stage-IIIA NSCLC and PS 0–2, there was no clear association between comorbidity burden and the likelihood to undergo surgery (OR: 0.74; 95% CI: 0.44–1.23), or to receive chemo/RT (OR: 1.12; 95% CI: 0.73–1.71) or radiotherapy (OR: 1.12; 95% CI: 0.82–1.53). In patients with stage-IIIB–IV disease and performance status 0–2, chemotherapy, but not radiotherapy, was less likely to be offered to patients with severe comorbidity (chemotherapy OR: 0.76; 95% CI: 0.65–0.89, radiotherapy OR: 1.03; 95% CI: 0.87–1.23) (Table 3).

Comorbidity burden and mortality in NSCLC patients with PS 0–2

Figure 1 illustrates the overall and the cancer-specific survival in patients with a PS 0–2, stratified by stage at diagnosis and comorbidity burden. In patients with stage-IA–IIB disease, both overall and cause-specific survival was lower among patients with severe comorbidity compared to those with no record of concomitant disease. The overall 5-year survival in patients with stage IA-IIB NSCLC with CCI O was 51% compared to 30.6% in those with CCI 3+.

In patients with stage IIIA and stage IIIB–IV, the overall and cause-specific survival was lower or equal in patients with severe comorbidity compared to that observed in other CCI groups. The probability of succumbing to causes of death other than lung cancer was elevated in patients with early-stage NSCLC and multiple coexisting diseases (Figure 2).

In Cox regression models adjusted for demographic, clinical and treatment factors, the risk of all-cause mortality was elevated in patients with stage-IA–IIB disease with a performance status 0–2 and severe comorbidity (HR: 1.50; 95% CI: 1.18–1.90) In stage IIIA and stage IIIB–IV and PS 0–2, there were no differences in all- cause mortality between comorbidity groups (Table 4).

In adjusted analyses, there was no evidence of differences in cause-specific mortality between comorbidity groups within the clinical subgroups with the exception of patients with late stage-IIIB–IV disease. In this subgroup, the risk of lung cancer-specific mortality was lower in patients with severe compared to those with no comorbidity (HR: 0.91; 95% CI: 0.85–0.98) (Table 4).

Discussion

Using information from a large, nationwide population-based cohort of patients with lung cancer, we report a considerable higher comorbidity burden in NSCLC patients compared to lung cancer-free, age-matched controls, both when assessed by the Charlson Comorbidity Index and by specific conditions. NSCLC patients with severe comorbidity were more likely to be men, former smokers, elderly, to have a low educational level and a poorer performance status compared to patients with no or mild comorbidity. As expected, several coexisting medical conditions were more prevalent in smokers compared to nonsmokers, such as peripheral vascular disease, chronic pulmonary disease and myocardial infarction [23–25]. This is likely to reflect a difference in smoking history between lung cancer patients and controls. However, because of the absence of information on smoking history among controls, the magnitude of this influence cannot be evaluated based on the data at hand. Further, our findings indicate that comorbidity influences clinical decision making; compared to NSCLC patients with no history of co-existing medical conditions, those with severe comorbidity were less likely to receive extensive treatment, such as surgery in early stage disease. In patients with early, but not late-stage disease, severe comorbidity in adjusted analyses was associated with an increased all-cause mortality, while lung cancer-specific mortality was largely unaffected by comorbidity burden. Further, our results indicate that routinely published lung cancer survival statistics conveys a too dismal picture of prognosis. Of special note is that we found a 20% point difference in estimated 5-year overall survival between stage-IA–IIB patients with CCI 0 and CCI 3+.

Our finding of a higher prevalence of severe comorbidity in patients with early compared to late-stage NSCLC remained also after exclusion of registered comorbidities during a six-month period prior to the date of lung cancer diagnosis and in multivariable analysis. Thus, this finding appear unlikely to be explained by a more extensive work-up prior to decision-making in stage-I–IIA disease but may reflect a higher likelihood incidental detection of small lung tumors in patients with long-standing comorbid conditions.

According to the Swedish national treatment guidelines for lung cancer, [26] all patients with NSCLC stage IA–IIB, and in some cases, stage-IIIA disease should be considered for surgery with curative intention if possible with regard to functional status. Stereotactic radiotherapy is recommended for patients with early-stage disease judged not to be candidates for surgery. We found that surgery was less likely to be offered to patients with early-stage disease, PS 0–2 and severe comorbidity. While no association was observed between comorbidity and treatment decisions in patients with NSCLC-stage IIIA, patients with late-stage IIIB–IV disease, severe comorbidity and performance status 0–2 were less likely to receive chemotherapy but not radiotherapy. Our findings corroborate results from previous smaller studies showing that the likelihood to receive treatment according to guidelines decreases with increasing number of comorbid conditions [5,27,28].

The Swedish lung cancer treatment guidelines recommend that management decisions should be made in a multidisciplinary conference (MDC) setting [26]. However, overall only 60% of all patients were discussed during a MDC. In line with results from a recently published American study [29], we found no association between burden of comorbidity and the likelihood of clinical decision making in a MDC setting.

Our finding of a low treatment intensity in patients with early-stage disease, low-performance status and severe comorbidity may possibly reflect suboptimized management of comorbid conditions that precludes surgery, which in turn is likely to negatively affect outcomes. This provide further support to the need to consult other specialists in order to optimize the patients clinical status prior to start of treatment and avoid preconceived notions based on age and health status that can lead to undertreatment.

Corroborating results from earlier studies [7–13], we found differences in all-cause mortality between comorbidity groups in early, but not late-stage NSCLC. In patients with early-stage disease, the probability of dying from other causes than lung cancer increased with multiple coexisting diseases. The observed differences in outcome by comorbidity burden may in part reflect well-justified management decisions. However, these findings may also reflect suboptimized management of coexisting conditions and nonjustified treatment decisions based on preconceptions and assumptions of a high comorbidity burden related to older age.

Taken together, our findings indicate that comorbidity strongly influences the poor prognosis in patients with NSCLC, not only as a possible reflection of lower treatment intensity, but also as an independent prognostic factor.

Strengths of our study included the population-based setting with clinical information retrieved from the NLCR, covering virtually all lung cancer patients in Sweden during the period under study [16]. Additional information from other nationwide registries was obtained by record linkage based on the individually unique personal identity numbers assigned to all residents of Sweden. The underlying cause of death was retrieved from the Swedish Cause of Death Register, with a very high completeness [30]. To estimate the burden of coexisting disease, we used the Charlson comorbidity index, a commonly used and validated tool to estimate comorbidity burden [20,21,31]. As opposed to most earlier studies that have addressed the influence of CCI on overall- and cause-specific mortality in cancer patients, we were able to adjust for a large number of prognostic factors including sex, age at diagnosis, education, smoking history, stage, performance status and histology.

Several limitations need mentioning. By relying on discharge diagnosis available in the National Patient Register [32], we were unable to include information on coexisting disease not requiring in-hospital care. Thus, the comorbidity burden is likely to have been underestimated. Similar to earlier studies in this area, no information was available on the severity and duration of coexisting disease. Despite the large data set at hand, low numbers precluded the assessment of the influence of specific comorbid conditions on mortality. While complete, some records in the cause of death records may have been misclassified with regard to underlying cause of death.

Taken together, our findings indicate that comorbidity represents an important contributing factor to the poor overall prognosis observed in NSCLC patients, partly by lowering treatment intensity, but also as an independent factor. We suggest that when the expected prognosis is discussed with the patient and next to kin, not only tumor stage-based survival curves, but also expected survival incorporating comorbidity should be taken into consideration. Our results provide further support of the importance to avoid preconceived notions based on health status and possibly also age, leading to under treatment. Optimized management of coexisting disease in a multidisciplinary setting pre- and post-NSCLC-specific treatment may increase the number of lung cancer patients eligible for treatment according to guidelines and help improve outcomes.

Acknowledgments

The project was made possible by the continuous reporting by Swedish clinicians to the National Lung Cancer Register of Sweden and the work by the NLCR steering group: Gunnar Wagenius (chairman), Stefan Bergström, Bengt Bergman, Annelie Behndig, Mikael Johansson, Per Fransson, Kristina Lamberg Lundström, Anna Öjdahl-Bodén, Hanna Carstens, Karl-Gustaf Kölbeck, Andreas Hallqvist, Mona Gilleryd, Anders Vikström, Magnus Kentsson, Lars Ek and Sven-Börje Ewers.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This project was supported by a grant from the Swedish Cancer Society (15-0804) and from the Regional Research Council Uppsala-Örebro (RFR-654111).