2.1 Data collection

For the purpose of this population-based cohort study we used data from Lung Cancer Data Base Sweden, a research database generated by record linkages between the National Lung Cancer Register and several other Swedish population-based registers. The setup of Lung Cancer Data Base Sweden has previously been described in detail [4, 16, 17].

In short, the National Lung Cancer Register contains individual level information on smoking history, performance status, diagnostics, tumor characteristics (tumor stage and histopathology) and first line treatment. Smoking history is recorded based on self-report and categorized as smoker (current smoker), former smoker (no smoking during the last year) and non-smoker (never smoked on a regular basis). Performance status (PS) is based on the World Health Organization (WHO) performance score as assessed by the treating physician [18].

By means of record linkage, information in the National Lung Cancer Register was supplemented with individual level information available in the National Patient Register, the Swedish Cancer Register, the social database Longitudinal Integration Database for Health Insurance and Labour Market Studies and the Cause of Death Register.

The National Patient Register contains information on diagnosis made in hospital in- and outpatient settings and was used to estimate the Charlson Comorbidity Index (CCI) [19] categorized into three groups; no (CCI 0), mild (CCI 1–2), and severe comorbidity (CCI ≥ 3). The Swedish Cancer Register, to which reporting is mandated by law, includes data on all cancer diagnoses as reported by clinicians and pathologists.

The continuously updated Longitudinal Integration Database for Health and Insurance and Labour Market Studies database includes information on several different indicators of socioeconomic position [20]. For the purpose of the present study socioeconomic status was assessed based on highest achieved educational level collapsed into three groups: low ≤9 years of schooling, middle 10–12 years and high ≥13 years, corresponding to mandatory school, high school and post-high school (college and university).

Date and cause of death was available in the Cause of Death Register that includes data on main and contributing causes of death [21]. The completeness of the Cause of Death Register is high and is commonly used to assess cause-specific survival [22].

The study population was divided into three groups according to region of birth; “Sweden”, “Nordic” including residents born in the other Nordic countries (Denmark, Finland, Iceland and Norway), and “Non-Nordic” including all non-Nordic countries of origin. Because of small numbers, further subgrouping of Non-Nordic immigrants based on country or geographic region of origin was not possible.

2.2 Statistical methods

Demographic characteristics, clinical characteristics and patterns of management (diagnostic and treatment intensity) in all patients diagnosed with non-small cell lung cancer in Sweden 2002–2016 are presented. Comparison between the groups were made using the Chi square test. Univariable and multivariable logistic regression models were used to examine possible differences in management, including treatment modalities, based on region of birth. Data were stratified by stage at diagnosis and with analyses restricted to patients with PS 0–2 adjusted for age, sex, educational level, smoking history, stage at diagnosis, CCI, PS, histopathology, health care region in Sweden and year of diagnosis. The selection of patients with PS 0–2 was made to include patients eligible for oncological treatment. In a final step, cause-specific and overall mortality were compared between region of birth using the Kaplan-Meier method and Cox regression models with Hazard Ratios and 95% confidence intervals. Missing data were excluded in all the analyses.

Survival time was calculated as the time interval between the date of lung cancer diagnosis and date of death, and patient were censored at emigration or end of follow-up (December 31, 2016), whichever came first. In the cause-specific analyses, death due other causes were censored. All tests were two-sided, and a 5% level was considered statistically significant. Statistical analyses were performed using R version 3.5.0.

2.3 Research ethics

The study was approved by the Research Ethics Board in Stockholm (2016/1137-32; 2017/2026-32; 2017-445-32).

3.1 Demographics and clinical characteristics

We identified a total of 53,359 patients diagnosed with lung cancer between 2002 and 2016. Following exclusion of histopathological subtypes other than non-small cell lung cancer (n = 12,490) and missing data on educational level (n = 794), the final study population consisted of 40,075 men and women with non-small cell lung cancer. Demographic and clinical characteristics are presented in Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Demographic and clinical characteristics.

https://doi.org/10.1371/journal.pone.0278706.t001

In the population under study, 83.6% were born in Sweden, 7.4% in the other Nordic countries and 9.0% in Non-Nordic countries. There was a marked male predominance in the Non-Nordic cohort (Sweden 50.2%, Nordic 53.8% and Non-Nordic 66.3%). Compared to native born Swedes the prevalence of current smoking was higher in immigrants (Sweden 42.4%, Nordic 51.3% and Non-Nordic 48.5%).

Non-Nordic patients were younger at diagnosis (age 0–59 years: Sweden 13.8%, Nordic 15.3% and Non-Nordic 29.3%), had a better performance status (PS 0: Sweden 21.9%, Nordic 19.5% and Non-Nordic 26.8%) and a higher educational level (post-high school: Sweden 15.6%, Nordic 9.3% and Non-Nordic 22.2%). A histopathology of squamous cell carcinoma was slightly more common in Nordic immigrants (Sweden 25.3%, Nordic 30.1% and Non-Nordic 26.4%). There were no differences in comorbidity burden as assessed by CCI or stage at diagnosis.

3.2 Diagnostic intensity and multidisciplinary team assessment

Diagnostic work-up with bronchoscopy, endobronchial ultrasound (EBUS), computed tomography (CT) thorax, ultrasound (US) or CT abdomen, thoracocentesis, thoracoscopy, transthoracal biopsy and CT or MRI brain and, in adenocarcinomas, testing for epidermal growth factor receptor (EGFR) mutation were evenly performed across all groups (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Diagnostic intensity.

https://doi.org/10.1371/journal.pone.0278706.t002

Non-Nordic immigrants more often underwent positron emission tomography (PET) (Sweden 42.9%, Nordic 41.3% and Non-Nordic 50.5%) and were more often discussed in a multidisciplinary team setting (MDT) (Sweden 63.4%, Nordic 64.7% and Non-Nordic 70.1%). (Table 2). Following adjustments for age these differences were significant in Non-Nordic immigrants for both PET and MDT (aOR 1.22 (95% CI 1.13–1.33); aOR 1.29 (95% CI 1.18–1.42) and for MDT in Nordic immigrants (aOR 1.12 (95% CI 1.01–1.24). After adjustments for further potentially modifying factors (age, sex, level of education, CCI, stage, year of diagnosis, performance status, smoking history and histopathology) these differences remained (Non-Nordic immigrants PET aOR 1.32 (95% CI 1.19–1.45) and MDT aOR 1.30 (95% CI 1.17–1.44); Nordic immigrants MDT aOR 1.19 (95% CI 1.07–1.33)).

3.3 Treatment intensity

First line treatment modalities in patients with PS 0–2 and stage IA-IV NSCLC are presented in Table 3.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Treatment modality.

https://doi.org/10.1371/journal.pone.0278706.t003

3.4 Overall and cause-specific mortality

In stage IA-IIB and IIIA, but not in advanced stage IIIB-IV disease, overall and cause-specific survival was higher in Non-Nordic immigrants compared to Nordic immigrants and Swedish born patients. (Fig 1). In a fully adjusted model this was reflected in lower overall and cause-specific mortality in stage IA-IIB disease in Non-Nordic immigrants. In model adjusted for age a similar difference was observed for stage IIIA NSCLC. In advanced stage IIIB-IV lung cancer, there were no significant differences in mortality between native born and immigrant patients. (Table 4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Overall and cause-specific survival.

Cumulative overall and cause-specific survival in patients with performance status 0–2 diagnosed with non-small cell lung cancer in Sweden 2002–2016 by geographic region of origin.

https://doi.org/10.1371/journal.pone.0278706.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Overall and cause specific mortality.

https://doi.org/10.1371/journal.pone.0278706.t004

In contrast, both overall and cause-specific survival was lower in Nordic immigrants with stage IA-IIIA disease compared to Swedish born patients (Fig 1). Following adjustments, these findings remained significant in stage IA-IIB, but not in more advanced disease.

4.1 Summary of results

In this retrospective nationwide cohort study of patients with NSCLC immigrants from Non-Nordic countries were younger at diagnosis, had a higher educational level and a better performance status compared to Nordic immigrants and Swedish born patients reflecting the presence of a ‘healthy migrant effect’. We found only minor differences in patterns of care that were attenuated after adjustment for age and other potential confounding factors. If anything, immigrants from Non-Nordic countries were more often offered PET, were discussed in a multidisciplinary team setting and had lower mortality in early stage disease. Taken together, we found no evidence that patients with an immigrant background are disadvantaged when management decisions are made.

4.2 Comparisons to previous studies

In Sweden, almost one fifth of the total population of 10.3 million inhabitants had an immigrant background in 2019, i.e. born outside of Sweden [15]. Since 1955, Sweden has a tax-financed universal health care system aiming to provide all citizens access to health care at low out-of-pocket cost. Immigrants with a permanent residence permit obtain the same rights to health care as native Swedes and are offered free of charge interpreter services.

Our observation of special characteristics of the immigrant population, i.e. the healthy migrant effect, corroborate findings in previous studies [9–11, 23]. Younger age at time of a lung cancer diagnosis in immigrants may reflect differences in lifetime smoking history, lifestyle and environmental and occupational exposures [7]. Ethnic differences in genomic features, for example EGFR-expression in European and East Asian populations, and the interaction between genes and other factors may influence not only age at onset, but also outcome [11, 24, 25].

Our findings broadly confirm results of earlier studies conducted in Denmark, Sweden and Norway that had a primary focus on incidence and survival [7, 23, 26–28]. Compared to the native Norwegian population, Hjerkind et al found a higher incidence of lung cancer in immigrants from other Nordic countries and Eastern Europe [23]. Similar to our findings, the Norwegian study found no evidence of more advanced stage disease in the immigrant population [29]. To the best of our knowledge, only one previous study has compared patterns of lung cancer management between native born and immigrants [7]. Similar to the results of that study, we found that treatment according to guidelines was readily offered also to patients with an immigrant background. A high use of PET-DT in Non-Nordic immigrants may reflect that a majority reside in urban areas with large hospitals where this diagnostic modality is readily available. Koyi et al and Norredam et al found no significant differences in lung cancer mortality between native born and immigrant patients in Sweden and Denmark [7, 28]. Similar to our findings, a lower lung cancer mortality in immigrants has been reported from Norway [27]. Based on the results of the adjusted analyses in our study, we conclude that the better prognosis in Non-Nordic immigrants most likely reflect younger age at diagnosis. We have no clear explanation of our findings of poorer outcomes in Nordic immigrants with early-stage lung cancer. One possible explanation is a predominance of Finnish immigrants that have documented high mortality rates associated with socioeconomic factors, alcohol-related morbidity and cardiovascular disease. [30, 31].

4.3 Strengths and limitations

Strengths of our study included the use of data available in high quality population-based registers containing detailed individual level information on not only patient and tumor characteristics, but also management and outcomes. Because of the nationwide setting, the size of the dataset and the high completeness of the National Lung Cancer Register, selection bias was not an issue.

Limitations included the absence of detailed information on genetics, language skills, lifestyle factors, health awareness, health care seeking behavior and ability to navigate the health care system. Data on region of origin, but not specific country of origin were made available from Statistics Sweden. Due to small numbers, further subgrouping of Non-Nordic immigrants based on geographic region of origin was not possible. Information on smoking history was self-reported and is therefore subject to misclassification due to misreporting or recall bias. The assessment of comorbidity burden was based on in- and outpatient hospital care records while diagnoses made in primary health care setting was unavailable. For this reason, the comorbidity burden is likely to have been underestimated. Furthermore, no information was available to correctly characterize patients that died shortly after diagnosis or start of treatment. We were also unable to account for unregistered return of immigrants to their country of origin after a lung cancer diagnosis resulting in loss of follow-up and overestimation of survival (i.e. the so called salmon effect) [27].