Impact of new cancer medicine – real-world evidence from Danish register study of lung cancer patients
Impact of new cancer medicine – real-world evidence from Danish register study of lung cancer...
Rudolfsen, Jan H.; Hjortsø, Mads D.; Pedersen, Mikkel H.; Pilgaard, Trine; Pøhl, Mette
2023-03-04 00:00:00
ACTA ONCOLOGICA https://doi.org/10.1080/0284186X.2023.2185104 ORIGINAL ARTICLE Impact of new cancer medicine – real-world evidence from Danish register study of lung cancer patients a b a b c Jan H. Rudolfsen , Mads D. Hjortsø , Mikkel H. Pedersen , Trine Pilgaard and Mette Pøhl a b c Incentive, Holte, Denmark; Pfizer Denmark, Ballerup, Denmark; Department of Oncology, Copenhagen University Hospital, Rigshospitalet, Copenhagen, Denmark ABSTRACT ARTICLE HISTORY Received 6 October 2022 Background: Lung cancer (LC) is the leading cause of cancer deaths worldwide. Several new treatments Accepted 17 February 2023 have become available in recent decades, but little research exists on the impact of these on productivity, early retirement and survival for LC patients and their spouses. This study evaluates the effect of new KEYWORDS medicines on productivity, early retirement and survival for LC patients and their spouses. Lung cancer; real-world Methods: Data from the period 1 January 2004–31 December 2018 were collected from complete evidence; new medication; Danish registers. LC cases diagnosed before approval of first targeted therapy (19 June 2006, Before register data; survival; early patients) were compared with those who received at least one new cancer treatment, diagnosed after retirement this date (After patients). Subgroup analyses based on cancer stage, and epidermal growth factor receptor (EGFR) or anaplastic lymphoma kinase (ALK) mutation were conducted. Linear regression and cox regression were used to estimate the outcomes including productivity, unemployment, early retirement, and mortality. Spouses of Before and After patients were compared on earnings, sick leave, early retirement, and healthcare utilisation. Results: The study population comprised of 4,350 patient (2,175 After/2,175 Before). Patients who received new treatments had a significantly reduced risk of death (Hazard ratio ¼ 0.76, Confidence interval: 0.71 0.82) and reduced risk of early retirement (Hazard ratio: 0.54, Confidence interval: 0.38 0.79). No significant differences in earnings, unemployment, or sick leave were found. Spouses of Before patients had a higher cost of healthcare services after diagnosis compared to spouses of After patients. For productivity, early retirement and sick leave, no significant differences were found between the spouse groups. Conclusion: Patients who received innovative new treatments had reduced risk of death and reduced risk of early retirement. Spouses of LC patients who received new treatments had lower healthcare costs in the years following diagnosis. All findings indicate that recipients of new treatments had reduced burden of illness. Introduction employed after treatment, compared to the general popula- tion [5]. Moreover, less than half of LC survivors employed at Lung cancer (LC) is the leading cause of cancer deaths time of diagnosis return to work after treatment [6], with worldwide, with tobacco smoking being the primary cause ‘Fatigue’ being reported as the primary reason [7]. of illness. The illness and treatment of cancer leads to dis- Returning to work or maintaining normal physical func- rupted function of the primary organs, leading to severe tioning after treatment is essential for self-esteem and qual- reduction in health-related quality of life (HRQoL) [1]. For ity of life. From a healthcare sector and societal perspective, patients diagnosed 1997–2007, the five-year survival rate in there is a need for updated information on the real-world Europe was only 13% [2]. Subsequently, new innovative tar- impact that these new treatments have on patients’ work geted cancer treatments, for example, inhibitors of epidermal productivity, life expectancy and the possible impact on growth factor receptor (EGFR), anaplastic lymphoma kinase close relatives. (ALK), and immune checkpoint inhibitors (ICIs), designed to In this study, we analyse productivity, risk of early retire- block the action of immune checkpoints have improved ment, unemployment and overall mortality for LC patients treatment efficacy, reduced toxicity and improved clinical who received new innovative LC treatments in Denmark outcomes [3]. between 19 June 2006 and 31 December 2018. We matched Cancer is considered a chronic disease in the workforce and compared these LC patients with those diagnosed [4]. LC survivors have 2–3 times lower probability of being before the introduction of new targeted treatments – that is, CONTACT Mads D. Hjortsø mads.hjortso@pfizer.com Lautrupvang 8, Ballerup, 2750, Denmark Supplemental data for this article is available online at https://doi.org/10.1080/0284186X.2023.2185104. 2023 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The terms on which this article has been published allow the posting of the Accepted Manuscript in a repository by the author(s) or with their consent. 2 J. H. RUDOLFSEN ET AL. patients diagnosed with LC between 1 January 2004 and 18 Study populations June 2006. Moreover, we identified spouses of LC patients as LC patients diagnosed between 1 January 2004 and 31 informal caregivers. We considered the productivity, sick December 2018 were included. On 19 June 2006, the new leave, risk of early retirement and health service utilisation targeted medication erlotinib became available for Danish LC following their partner’s LC diagnosis. patients. Patients diagnosed after this date who were treated with any of the predefined new targeted cancer medicines were defined as the study’s After population. New targeted Material and methods medications were identified by treatment codes in the NPR. The study was designed as a retrospective cohort study These treatments were (treatment code in parenthesis) using the national Danish registers, covering the entire erlotinib (BWHA404), bevacizumab (BOHJ19B1), gefitinib Danish population (5.7 million). LC patients were identified (ML01EB01), crizotinib (BWHA413), afatinib (BWHA417), ceriti- as individuals registered with International Statistical nib (ML01ED02), Osimertinib (BWHA434), alectinib (BWHA440), brigatinib (ML01ED04), dacomitinib (ML01EB07), Classification of Diseases and Related Health Problems (ICD- lorlatinib (BWHA448), pembrolizumab (BOHJ19J3) and nivolu- 10) code DC34 (including all subgroups) as primary or sec- mab (BOHJ19H2). Patients diagnosed before 19 June 2006, ondary diagnosis in the Danish National Patient Register were regarded as the Before population (control). (NPR) [8]. All Danish residents have a unique 10-digit per- In addition, six subgroup analyses (SG 1–6) were per- sonal identification number, making it possible to merge per- formed. The first two subgroup analyses used later dates for son-level data from all Danish registers. The population was introduction of new targeted therapies (SG 1 and 2). SG 3 validated by cross-checking registrations in the Danish only considered Stage I–III LC patients, while SG 4 considered Cancer Register (DCR) [9]. All patients diagnosed after 1 Stage IV LC patients. SG 5 compared ALK-positive LC patients January 2004 and before 31 December 2018 were included. in the After group with Stage IV LC patients in the Before group. Lastly, LC patients with EGFR-mutations in the After group were compared with Stage IV LC patients in the Lung cancer patients Before group (SG 6). LC cases who received at least one of the new medications SG 1 focussed on the introduction of crizotinib on 26 in the year of or in the year following date of diagnosis was November 2012, using this date to separate between Before identified in the NPR. Characteristics of tumour size, lymph and After populations. SG 2 focussed on the additional new node involvement and metastatic spread (TNM classification ICIs pembrolizumab and nivolumab used to treat locally of malignant tumours) from DCR was used to determine LC advanced or metastatic non-small cell LC (NSCLC). In this stage at time of diagnosis. Date of birth, marital status, subgroup analysis, 1 January 2017 was used to distinguish region of residence and date of death (if applicable) were between Before and After populations. collected from the Central Person Register (CPR) [10]. Highest In SG 3 and 4, Stages I–III and Stage IV patients were con- achieved education of the patients before diagnosis was col- sidered separately. Cancer stage (I–IV) were determined by lected from the Population Education Register (PER) [11]. TNM codes from the DCR. During the study period, there were three versions of the TNM classifications. We applied Patients income and labour market affiliation were gathered the staging algorithm used by Norwegian Cancer Register to from the Income Register [12] and the DREAM database (The determine stage. There is extensive cooperation between the Danish Register of sickness absence, compensation benefits Nordic Cancer registers, thus the algorithm should be applic- and social transfer payments) [13], respectively. Tumour path- able to Danish data. ology was defined using the Systemised Nomenclature of In SG 5, ALK-positive LC patients were analysed. ALK-posi- Medicine (SNOMED) codes from The Danish National tive patients were identified if observed with SNOMED code Pathology Register and Databank [14]. T08 or T25-T29 (including sublevels) in combination with F2911 within 30 days of diagnosis. The first ALK-positive LC patient was identified in 2007, but ALK-specific treatment Spouses of lung cancer patients was not available before 2012. ALK-positive LC patients was Spouses, defined as a person married to or in a domestic compared with Stage IV LC patients in the Before group. relationship with LC patients at the time of diagnosis were Similarly, in SG 6, patients observed with SNOMED code identified in the CPR. For spouses, date of birth, region of T08 or T25-T29 (including sublevels) in combination with one residence from the CPR [10], highest achieved education of the following codes: FE13CA, FE13CB, FE13CC, FE13CD, from the PER [11], income from the Income Register [12] and FE13CK, FE13CP, F29616, FE13C3, FE13C5 within 30 days of labour market affiliation from the DREAM database [13] was diagnosis were classified as EGFR-positive patients. EGFR also collected. patients diagnosed during the After period was compared Healthcare utilisation for spouses was measured as con- with Stage IV LC patients in the Before period. tacts and cost from the NPR for hospital care, from the Two analyses focussed on the effect on Spouses of LC National Health Insurance Register [15] for primary care, and patients. One analysis where spouses of Before LC patients the Register of Medicinal Product Statistics for costs of pre- are matched and compared with spouses of After LC patients scription drugs. (Matched 1:1). And one analysis of spouses of Before LC ACTA ONCOLOGICA 3 patients are matched and compared with Spouses of After subgroup analysis, GLM (generalized linear model) distances LC patients and individuals not associated with cancer was used to achieve successful matching. Matching covari- (Matched 1:1:3). The reason for conducting two analyses, as ates was age, sex, region of residency, cancer stage, educa- opposed to one, was to retain as many spouses as reason- tion, and income in t-1. ably possible in the analysis. Spouses of Before patients was matched with spouses of After patients using the same method but evaluated with GLM distances. Covariates were age, sex, region of residence, Outcomes for lung cancer patients income in t-1 and education. Additionally, an analysis was For LC patients, the following outcomes were analysed; prod- performed where spouses of Before and After patients and uctivity (proxied by earnings), unemployment, risk of early persons not associated with cancer was matched on the retirement, and risk of death. same criteria, using exact matching (1:1:3, Before spouses: Productivity was measured as average salary (active After spouses: No Cancer) income) of Before/After LC patients in the year before diag- Successful matching was achieved if the following three nosis (t-1) and five years after diagnosis (t5). Earnings were criteria were fulfilled: (1) <15% deviance in variance ratio, (2) adjusted to a 2018 price level according to the consumer <±0.1 difference in standardised mean, and (3) at least 80% price index as calculated by Statistics Denmark. Transfer pay- of cases in the smallest group was retained. ments was not included. Unemployment was measured as average number of Statistical analysis weeks with unemployment each year from t-1 to t5 in the Before and After groups. Earnings, unemployment, sick leave and utilisation of health- Early retirement is granted on a municipality level if a care services, was analysed using used ordinary least square physician evaluates a person as being permanently unfit to (OLS) regression with the treatment variable (Before/After) work. Due to a policy change for early retirement in 2013, interacted with index period (t-1 to t5) to estimate the aver- we censored cases on 1 January 2013 for this analysis. age value of the respective outcomes by period. The follow-up time for overall mortality ranged from date Early retirement and all-cause mortality was analysed of diagnosis until 31 December 2019. using Cox proportional hazard regression (Cox model) to estimate the difference in risk of event between the groups. Outcomes for spouses of lung cancer patients The hazard ratios produced from the Cox model was adjusted for age, sex, cancer stage (where applicable), educa- For spouses of LC patients, the following outcomes were tion and region of residence. In analysis of early retirement, analysed; productivity (proxied by earnings), risk of early cases were censored at death, emigration or when turning retirement, incidence of long-term sick leave and health service 65 years old. The latter corresponding to criteria for age utilisation (proxied by cost). related retirement until 2008 in the study period. Productivity and risk of early retirement was evaluated To avoid immortal time, survival was analysed from day of same way as with the LC patients. first treatment onwards. First treatment was defined as treat- Use of health services was measured as the sum of the ment with antibodies or immunomodulatory treatment (pro- costs of healthcare services consumed in primary and hos- cedure code BOHJ, including subgroups), or treatments pital sector as well as cost of prescription drugs. The costs conditional to special medical treatment principles, for were adjusted to 2018 price level according to the Consumer example cytostatic agents (procedure codes BWH, including Price Index. Fees and Diagnosis Related Group (DRG)-tariffs subgroups). were applied as unit costs for primary healthcare and hos- Statistical significance was set to 5% in all models and pital sector resource use. Market prices were applied as unit was not adjusted for multiple testing and should be inter- costs for prescription drugs. preted as such. Long-term sick leave was defined as sick leave lasting lon- All analyses were performed in the statistical software R, ger than four weeks. As all shorter sick leave periods were version 4.1.0 (www.r-project.org). discarded, we consider the fifth week of sick leave as the first week of long-term sick leave. Results Balancing the samples Figure 1 presents a flow chart of the merging of data sets. A total of 50,275 eligible LC cases were identified during the Because the After sample only comprised those LC patients study period. who received one of the innovative LC medicines in focus, A total of 803 cases were not identified in the CPR. Table we matched these patients with Before patients to ensure 1A in the Supplementary appendix presents summary statis- comparable populations. Matching was done with optimal tics on the remaining 49,472 patients and their spouses. The pairwise propensity score matching, minimising the distance patients had a median age of 69 years, with 51% being between each patient in the After sample with a suitable males. In the Before period, 5,753 cases were to be matched match in the Before sample. Distance was evaluated by Mahalanobis distances. In the case of Stage I–III and EGFR with 2,430 After cases. A total of 38,518 patients were 4 J. H. RUDOLFSEN ET AL. Figure 1. Flow chart from identification of patients to matching of data sets. classified as ‘Other’, as they were diagnosed in the After cancer incidence between 2004 and 2018. Males had an 18% period, but did not receive any of the new innovative increase in incidence, while females had a 54% increase. treatments. More than half of the cases (51%) had Stage IV cancer at For spouses, 25,541 spouses were identified, with median time of diagnosis. ALK-positive and EGFR-positive mutations age of 67 (Supplementary Table 1A). Of these spouses, 1,643 were not registered before 2006 but had a steady increase in were spouses of LC patients in the After population. A total cases until 2012, when the incidence became stable. of 3,414 were spouses of Before LC patients (3,272 spouses of LC Patients diagnosed between 1 January 2004 and 18 Results main analysis June 2006). For LC patients in SG 2, 397 spouses were identi- fied. The remaining 20,087 spouses were spouses of ‘Other’ Table 2 present summary characteristics of the matched sam- LC patients. Summary statistics for all matched subsets are ple used in the main analysis. The matched sample was on provided in Supplementary Appendix A. average 1 year younger than the overall population (pre- Table 1 display LC incidence by stage and histologic sub- sented in Supplementary Table 1A) with a higher proportion type during the study period. There was a 26.5% increase in of females, but otherwise comparable. ACTA ONCOLOGICA 5 Table 1. Annual incidence by gender, stage, pathology and total. Year 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Male 2,051 2,091 2,121 2,228 2,223 2,259 2,296 2,267 2,371 2,236 2,403 2,302 2,335 2,386 2,340 Female 1,723 1,793 1,867 1,988 2,013 2,028 2,199 2,207 2,193 2,293 2,249 2,324 2,345 2,391 2,434 Total 3,774 3,884 3,988 4,216 4,236 4,287 4,495 4,474 4,564 4,529 4,652 4,626 4,680 4,777 4,774 Stage I 161 153 163 173 184 194 257 276 354 352 404 453 485 529 597 Stage II 442 464 378 459 431 454 547 572 614 647 695 673 752 748 741 Stage IIIa 237 282 256 288 306 324 339 335 354 307 372 380 400 368 299 Stage IIIb 632 637 664 667 639 595 631 645 581 629 573 593 563 617 671 Stage IV 1,903 1,955 2,101 2,229 2,338 2,402 2,498 2,385 2,394 2,344 2,357 2,270 2,195 2,229 2,212 Unknown stage 351 339 377 362 304 281 192 235 240 227 215 235 249 256 206 ALK positive 0 0 0 <5 <5 <5 6 7192027 20333825 EGFR positive <5 <5 11 8 13 17 46 83 115 152 149 160 164 161 164 Table 2. Summary statistics, matched lung cancer sample, main analysis. Characteristic Overall, N¼ 4,350 Before, N¼ 2,175 After, N¼ 2,175 Age 68 (61, 74) 68 (61, 74) 68 (62, 74) Income one year before diagnosis 0 (0, 14,537) 0 (0, 14,731) 0 (0, 14,102) Sex Male 1,904 (44%) 952 (44%) 952 (44%) Female 2,446 (56%) 1,223 (56%) 1,223 (56%) Education Primary or no education 2,063 (47%) 1,096 (50%) 967 (44%) Short higher education 2,126 (49%) 1,008 (46%) 1,118 (51%) Long higher education 82 (1.9%) 34 (1.6%) 48 (2.2%) Unknown education 79 (1.8%) 37 (1.7%) 42 (1.9%) Stage Stage I 84 (1.9%) 42 (1.9%) 42 (1.9%) Stage II 212 (4.9%) 106 (4.9%) 106 (4.9%) Stage IIIa 326 (7.5%) 163 (7.5%) 163 (7.5%) Stage IIIb 624 (14%) 312 (14%) 312 (14%) Stage IV 3,024 (70%) 1,512 (70%) 1,512 (70%) Unknown stage 80 (1.8%) 40 (1.8%) 40 (1.8%) Region Capital 902 (21%) 541 (25%) 361 (17%) Central Jutland 1,247 (29%) 533 (25%) 714 (33%) Northern Jutland 429 (9.9%) 206 (9.5%) 223 (10%) Zealand 877 (20%) 441 (20%) 436 (20%) Southern Denmark 895 (21%) 454 (21%) 441 (20%) Median with the inter quartile range in brackets, or number of observations with ratio in brackets. Out of 2175 Before patients 2130 died during the study Spouses of After patients had significantly lower health- period, while the equivalent numbers for After patients was care costs in the years following diagnosis (t-1–t5). In 1953 out of 2175. Six Before patients died in the period Figure 2, the costs are displayed. Results of analysis with between test and diagnosis. All the After patients received spouses of Before and After patients are to the left, and any of the new innovative LC medicine, while 1213 Before results of analysis with spouses of Before and After patients patient received standard medical treatment. In adjusted as well as persons without cancer to the right. analysis, After patients had significantly reduced risk of death Healthcare costs generated by spouses of After patients (HR ¼ 0.76, 0.710.82) compared to the matched Before seems unaffected by the LC diagnosis. For spouses of Before group. patients, however, the annual healthcare costs increased For productivity, no significant differences were found slightly about $95 (700 DKK) from t0 to t1 and plateaued between Before and After populations. Average income in around this new level. In the analysis including individuals the year before diagnosis was $9600 (DKK 72,000) and not associated with cancer, a similar pattern was seen. There decreased to about $4000 (DKK 30,000) five years after diag- were no significant differences in healthcare costs between nosis. No significant differences were found in unemploy- persons without cancer and spouses of After LC patients, ment (LC patients) or long-term sick leave (spouses) between until t4 and t5 (Figure 2). Before and After groups (data not shown). Out of 649 (20,5%) at risk Before cases, 133 (20,5%) entered early retirement before 1 January 2013 versus 40 out Subgroup analyses of 334 (12%) at-risk After cases. Patients in the After sample Results from risk of death and early retirement for subgroup had significantly reduced risk of early retirement (HR: 0.54, analyses are presented in Table 3. All After subgroups had 0.380.79). For spouses, we found no significant differences reduced risk of death compared with their respective Before (HR 1.06, 0.234.94). However, this result should be seen in populations (see Table 3). The largest reduction in risk of context of fewer than five spouses of After LC patients entered early retirement. death was found in ALK-positive patients (SG 5) (HR ¼ 0.37, 6 J. H. RUDOLFSEN ET AL. (A) (B) Healthcare costs Healthcare costs 1,200 1,200 1,000 1,000 800 800 600 600 400 400 200 200 0 0 -1 01 23 45 -1 0 1 2 3 4 5 After Before After Before NoCancer After Before NoCancer Healthcare costs by period Time t-1 t0 t1 t2 t3 t4 t5 t-1 t0 t1 t2 t3 t4 t5 After 802 804 815 823 829 791 802 834 841 845 853 869 821 818 CI 745-859 747-861 756-874 762-885 762-897 717-865 718-886 762-907 768-914 770-920 775-931 784-954 727-915 712-925 848 876 948 995 990 957 1003 933 970 1037 1083 1065 988 1021 Before 791-905 819-933 891-1006 936-1053 931-1049 897-1017 942-1064 860-1006 897-1043 963-1110 1008-1157 989-1140 911-1065 943-1100 CI NoCancer 789 828 862 909 934 973 976 747-831 786-870 820-905 866-952 891-978 929-1017 931-1021 CI After: 45 After: 13 After: -17 After: -56 After: -65 After: -152 After: -158 -134 -201 -47 -72 -171 -161 -166 (151 - -60) (119 - -92) (89- -123) (50 – -162) (42 - -172) (-44 - -259) (-49 - -266) Difference Before: 144 Before: 142 Before: 175 Before: 174 Before: 130 Before: 15 Before: 45 42 – -135 16- -161 --45 - -222 -82 - -260 -71 - -250 -76 - -257 -110 - -292 (272 – 16) (270 – 14) (303 – 45) (303 – 45) (260 – 0) (146 - -115) (177 - -86) Figure 2. Healthcare costs (Eur) generated by spouses of LC patients. A: Analysis of spouses of Before and After patients (N ¼ N ¼ 1292). B: SpouseBefore SpouseAfter Analysis of spouses of Before and After patients, as well as persons not associated with cancer (N ¼ N ¼ 1139, N ¼ 3415). For panel (A) SpouseBefore SpouseAfter NoCancer difference was tested between Before and After patients, for panel (B) difference was tested between After – Control, and Before-Control. Table 3. Results from subgroup analyses. Before/After Subgroup number n Lung cancer cases Early retirement Mortality 1 1,709/572 Before/After, new treatment No event in After HR ¼ 0.63 (0.560.72) 2 4,023/1,352 Before/After subgroup No event in After HR ¼ 0.56 (0.510.62) 3 1,219/650 Stages I–III HR: 0.81 HR ¼ 0.82 (0.45–1.5) (0.730.92) 4 1,512/1,512 Stage IV HR: 0.57 HR: 0.73 (0.37–0.87) (0.670.80) 5 171/57 ALK positive Too few events in ALK HR ¼ 0.37 (0.250.55) 6 599/215 EGFR positive HR: 0.47 HR ¼ 0.59 (0.13–1.69) (0.490.72) Before as reference for hazard ratio in all analyses. After populations defined as patients treated with erlotinib, bevacizumab, gefiti- nib, crizotinib, afatinib, ceritinib, Osimertinib, alectinib, brigatinib, dacomitinib, lorlatinib, pembrolizumab or nivolumab. Before popu- a b lation diagnosed before 19 June 2006. After received nivolumab, pembrolizumab or durvalumab. Treated Before/After 26 November 2012, Definition of After population same as in main analysis. Stage I–III LC patients diagnosed with Stage I–III cancer according to TNM codes. Stage IV patients diagnosed with Stage IV according to TNM codes. ALK-positive patients (SNOMED code T08 or T25–T29 (including sublevels) in combination with F2911 within 30 days of diagnosis) in the After population compared with Stage IV LC patients in the Before population. LC patients with EGFR-mutations (SNOMED code T08 or T25-T29 (including sublevels) in combination with one of the following codes: FE13CA, FE13CB, FE13CC, FE13CD, FE13CK, FE13CP, F29616, FE13C3, FE13C5 within 30 days of diagnosis) in the After group were compared with Stage IV LC patients in the Before group. 0.250.55) corresponding to a 63% reduction in hazard of requirements in 2013. In analysis of SG 5 and 6, there were death, while the smallest reduction was found in the analysis fewer than five events. As a result, there is high uncertainty of Stages I–III patients (SG 3) (HR ¼ 0.82, 0.730.92). associated with these estimates. For SG 1 and 2, analysis of early retirement was not In SG4, there was a significant reduced risk of early retire- applicable due to the reform in early retirement ment (HR ¼ 0.57, 0.370.87) corresponding to a 43% ACTA ONCOLOGICA 7 reduction in hazard. For SG 3, HR indicated a reduced hazard diagnosis, compared to spouses of Before LC patients. of early retirement in the After group however, not a statis- Severity of diagnosis has been shown to have a positive cor- tically significant reduction. relation with the burden on informal caregivers [17]. Patients diagnosed after 26 November 2012 who received Unobserved variables such as identification of biomarkers, new LC medicines (SG2) and patients with ALK-positive can- new surgical techniques, improved radiotherapeutic treat- cer (SG5) both earned more in the year before diagnosis, ment and methods for earlier diagnosis [18] as well as compared with their matched Before samples. In SG2, After changes in the patient population could be responsible for patients had significantly higher unemployment rate two some of the reduced risk of mortality or early retirement. years after diagnosis compared with their matched Before Particularly, incidence of LC for women is increasing [19]. During the study period, the ratio of female patients group. No other significant differences were found in changed from 45.5% in 2004 to 51% in 2018. Such changes unemployment or sick leave for any subgroup (results avail- able from corresponding author). in the patient population underline the importance of real- world studies, due to tendency of selection of specific patient populations used in randomised controlled trials, that Discussion might not be representative of the real-world LC population. When comparing LC patients before and after introduction of new innovative medicine, it was found that LC patients Strength and weaknesses receiving new innovative medicines had reduced hazard of The analysis is based on national registers, spanning 17 years death overall, and in all subgroups. This is in line with with reliable, routinely collected data before and after date expectations as new treatments are most often approved of diagnosis. based on a proven survival benefit compared to former Misclassification of data might occur. The authors could standard treatment in clinical randomised trials. Note, no new systemic treatments targeting early-stage LC and locally not validate the coding practice used in the inclusion/exclu- advanced LC (stage I–III) were introduced during the study. sion criteria. However, the Danish registers are of high qual- ity and are routinely used in population-based inquiries such The reduced hazard of death is therefore likely due to newer as the present. Furthermore, the sample size reduces the operation techniques and radiotherapy deliveries which have impact of random variation on the effect estimates. increased survival in this subgroup. The reduced hazard in For some matches there are several years between dates advanced stages is therefore probable due to advancements of diagnosis. Other developments in LC treatment, Small Cell in systemic treatment for subgroups of patients. In early and LC vs Non-Small Cell LC, performance status and other clinic- locally advanced stages (I–III), LC patients had a 18% reduc- ally covariates were not accounted for when matching. tion in hazard of death, while ALK patients who were treated Therefore, there could be differences in patients’ health at with new targeted medicine during this time reduced their baseline not captured by the stage indicator or distribution hazard with 63%. of Non-Small/Small Cell LC can vary in the Before and After LC patients who received the new innovative LC medi- population. This can have affected the results. cines had a significantly reduced hazard of early retirement. The matching criteria were met for all sets, except for the The subgroup analyses showed a significant hazard reduction ALK-positive group. Due to the relatively small sample of for Stage IV LC patients. This finding coincides well with the ALK-positive cases, we chose to breach the variance ratio cri- reduced toxicity from new innovative medicines. terion for age to retain all but two ALK-positive cases. For the other outcomes, there were sporadic significant As there are multiple outcomes and populations and sub- differences in subgroups. However, without a consistent pat- populations there is a risk of Type 1 error. The reader should tern of differences these results might be a result of natural be aware of this and consider any result in context of all variation, rather than an effect of the treatment. It should be results. noted that higher survival rate in the After population likely causes survival bias. In other words, when more patients sur- vive into retirement age, they are registered with zero Conclusion income, thus reducing the average earnings in the After population. The new innovative LC medicines introduced in Denmark The combination of reduced risk for both early retirement during the study period have had a positive effect on the and death is interesting. A previous study found that LC patients’ life. This is evident in survival, risk of early retire- patients did not live long enough to retire early [16]. ment and reduced healthcare utilisation for spouses of LC However, LC patients in advanced settings who received patients. The reduced risk of early retirement demonstrates new innovative treatments both lived longer and had an improvement of the low labour market participation in LC reduced risk of early retirement. With fatigue being the pri- patients following diagnosis and treatment. Spouses of those mary cause of exit from the labour market [7], it could sug- who received the new treatments generated lower health- gest that reduced toxicity through personalised medicine care costs, possible due to a reduced burden of illness for reduces fatigue symptoms in the patients. the LC patients. Spouses of After LC patients had significantly lower While significant improvements in treatment have been healthcare costs in the years following their partners LC achieved in the last 20 years, the prognosis for LC patients is 8 J. H. RUDOLFSEN ET AL. Europe 1999–2007: results from the EUROCARE-5 study. Eur J still not satisfactory, and continued effort to improve clinical Cancer. 2015;51(15):2242–2253. outcomes are necessary. [3] The global challenge of cancer. Nat Cancer. 2020;1(1):1–2. [4] Lawless GD. The working patient with cancer: implications for payers and employers. Am Health Drug Benefits. 2009;2(4): Ethics approval and consent to participate 168–173. The study was register-based and complied with the regulations and [5] Vayr F, Savall F, Bigay-Game L, et al. Lung cancer survivors and instructions set up by Statistics Denmark. No additional ethics commit- employment: a systematic review. Lung Cancer. 2019;131:31–39. tee approval is required for register-based research according to [6] Rashid H, Eichler M, Hechtner M, et al. Returning to work in lung Danish law. cancer survivors—a multi-center cross-sectional study in Germany. Support Care Cancer. 2021;29(7):3753–3765. [7] Kim YA, Yun YH, Chang YJ, et al. Employment status and work- Disclosure statement related difficulties in lung cancer survivors compared with the general population. Ann Surg. 2014;259(3):569–575. The study was financed by Pfizer Denmark. Jan Håkon Rudolfsen and [8] Lynge E, Sandegaard JL, Rebolj M. The Danish national patient Mikkel H. Pedersen are employees at Incentive, which is a paid vendor register. Scand J Public Health. 2011;39(7_suppl):30–33. of Pfizer Denmark. Mads D. Hjortsø and Trine Pilgaard are employee of [9] Gjerstorff ML. The Danish cancer registry. Scand J Public Health. Pfizer Denmark. Mette Pøhl was paid by Pfizer Denmark for her work as 2011;39(7_suppl):42–45. an oncology specialist in the project group. [10] Pedersen CB. The Danish civil registration system. Scand J Public Health. 2011;39(7_suppl):22–25. [11] Jensen VM, Rasmussen AW. Danish education registers. Scand J Funding Public Health. 2011;39(7_suppl):91–94. This study was supported by Pfizer Denmark. [12] Baadsgaard M, Quitzau J. Danish registers on personal income and transfer payments. Scand J Public Health. 2011;39(7_suppl): 103–105. ORCID [13] Hjollund NH, Larsen FB, Andersen JH. Register-based follow-up of social benefits and other transfer payments: accuracy and degree Mikkel H. Pedersen http://orcid.org/0000-0002-3568-767X of completeness in a Danish interdepartmental administrative Trine Pilgaard http://orcid.org/0000-0002-1771-956X database compared with a population-based survey. Scand J Public Health. 2007;35(5):497–502. [14] Erichsen R, Lash TL, Hamilton-Dutoit SJ, et al. Existing data sour- Data availability statement ces for clinical epidemiology: the Danish national pathology The data that support the findings of this study are available from registry and data bank. Clin Epidemiol. 2010;2:51–56. Statistics Denmark’s Research Service, but restrictions apply to the avail- [15] Andersen JS, Olivarius NDF, Krasnik A. The Danish national ability of these data, which were used under licence for the current health service register. Scand J Public Health. 2011; study, and so are not publicly available. Additional analyses are however Jul39(7_suppl):34–37. available from the authors upon reasonable request and with permission [16] Taskila-Åbrandt T, Pukkala E, Martikainen R, et al. Employment of Statistics Denmark’s Research Service. status of Finnish cancer patients in 1997. Psychooncology. 2005; 14(3):221–226. [17] Alshammari B, Noble H, McAneney H, et al. Factors associated References with burden in caregivers of patients with end-stage kidney dis- ease (A systematic review). Healthcare. 2021;9(9):1212. [1] Novello S, Kaiser R, Mellemgaard A, et al. Analysis of patient- [18] Lemjabbar-Alaoui H, Hassan OU, Yang YW, et al. Lung cancer: reported outcomes from the LUME-Lung 1 trial: a randomised, biology and treatment options. Biochim Biophys Acta BBA - Rev double-blind, placebo-controlled, phase III study of second-line Cancer. 2015;1856(2):189–210. nintedanib in patients with advanced non-small cell lung cancer. [19] Egleston BL, Meireles SI, Flieder DB, et al. Population-based Eur J Cancer. 2015;51(3):317–326. trends in lung cancer incidence in women. Semin Oncol. 2009; [2] Francisci S, Minicozzi P, Pierannunzio D, EUROCARE-5 Working Group, et al. Survival patterns in lung and pleural cancer in 36(6):506–515.
http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.pngActa OncologicaTaylor & Francishttp://www.deepdyve.com/lp/taylor-francis/impact-of-new-cancer-medicine-real-world-evidence-from-danish-register-7nK5AzrtWW
Impact of new cancer medicine – real-world evidence from Danish register study of lung cancer patients
Impact of new cancer medicine – real-world evidence from Danish register study of lung cancer patients
Abstract
Abstract Background Lung cancer (LC) is the leading cause of cancer deaths worldwide. Several new treatments have become available in recent decades, but little research exists on the impact of these on productivity, early retirement and survival for LC patients and their spouses. This study evaluates the effect of new medicines on productivity, early retirement and survival for LC patients and their spouses. Methods Data from the period 1 January 2004–31 December 2018 were collected...
ACTA ONCOLOGICA https://doi.org/10.1080/0284186X.2023.2185104 ORIGINAL ARTICLE Impact of new cancer medicine – real-world evidence from Danish register study of lung cancer patients a b a b c Jan H. Rudolfsen , Mads D. Hjortsø , Mikkel H. Pedersen , Trine Pilgaard and Mette Pøhl a b c Incentive, Holte, Denmark; Pfizer Denmark, Ballerup, Denmark; Department of Oncology, Copenhagen University Hospital, Rigshospitalet, Copenhagen, Denmark ABSTRACT ARTICLE HISTORY Received 6 October 2022 Background: Lung cancer (LC) is the leading cause of cancer deaths worldwide. Several new treatments Accepted 17 February 2023 have become available in recent decades, but little research exists on the impact of these on productivity, early retirement and survival for LC patients and their spouses. This study evaluates the effect of new KEYWORDS medicines on productivity, early retirement and survival for LC patients and their spouses. Lung cancer; real-world Methods: Data from the period 1 January 2004–31 December 2018 were collected from complete evidence; new medication; Danish registers. LC cases diagnosed before approval of first targeted therapy (19 June 2006, Before register data; survival; early patients) were compared with those who received at least one new cancer treatment, diagnosed after retirement this date (After patients). Subgroup analyses based on cancer stage, and epidermal growth factor receptor (EGFR) or anaplastic lymphoma kinase (ALK) mutation were conducted. Linear regression and cox regression were used to estimate the outcomes including productivity, unemployment, early retirement, and mortality. Spouses of Before and After patients were compared on earnings, sick leave, early retirement, and healthcare utilisation. Results: The study population comprised of 4,350 patient (2,175 After/2,175 Before). Patients who received new treatments had a significantly reduced risk of death (Hazard ratio ¼ 0.76, Confidence interval: 0.71 0.82) and reduced risk of early retirement (Hazard ratio: 0.54, Confidence interval: 0.38 0.79). No significant differences in earnings, unemployment, or sick leave were found. Spouses of Before patients had a higher cost of healthcare services after diagnosis compared to spouses of After patients. For productivity, early retirement and sick leave, no significant differences were found between the spouse groups. Conclusion: Patients who received innovative new treatments had reduced risk of death and reduced risk of early retirement. Spouses of LC patients who received new treatments had lower healthcare costs in the years following diagnosis. All findings indicate that recipients of new treatments had reduced burden of illness. Introduction employed after treatment, compared to the general popula- tion [5]. Moreover, less than half of LC survivors employed at Lung cancer (LC) is the leading cause of cancer deaths time of diagnosis return to work after treatment [6], with worldwide, with tobacco smoking being the primary cause ‘Fatigue’ being reported as the primary reason [7]. of illness. The illness and treatment of cancer leads to dis- Returning to work or maintaining normal physical func- rupted function of the primary organs, leading to severe tioning after treatment is essential for self-esteem and qual- reduction in health-related quality of life (HRQoL) [1]. For ity of life. From a healthcare sector and societal perspective, patients diagnosed 1997–2007, the five-year survival rate in there is a need for updated information on the real-world Europe was only 13% [2]. Subsequently, new innovative tar- impact that these new treatments have on patients’ work geted cancer treatments, for example, inhibitors of epidermal productivity, life expectancy and the possible impact on growth factor receptor (EGFR), anaplastic lymphoma kinase close relatives. (ALK), and immune checkpoint inhibitors (ICIs), designed to In this study, we analyse productivity, risk of early retire- block the action of immune checkpoints have improved ment, unemployment and overall mortality for LC patients treatment efficacy, reduced toxicity and improved clinical who received new innovative LC treatments in Denmark outcomes [3]. between 19 June 2006 and 31 December 2018. We matched Cancer is considered a chronic disease in the workforce and compared these LC patients with those diagnosed [4]. LC survivors have 2–3 times lower probability of being before the introduction of new targeted treatments – that is, CONTACT Mads D. Hjortsø mads.hjortso@pfizer.com Lautrupvang 8, Ballerup, 2750, Denmark Supplemental data for this article is available online at https://doi.org/10.1080/0284186X.2023.2185104. 2023 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The terms on which this article has been published allow the posting of the Accepted Manuscript in a repository by the author(s) or with their consent. 2 J. H. RUDOLFSEN ET AL. patients diagnosed with LC between 1 January 2004 and 18 Study populations June 2006. Moreover, we identified spouses of LC patients as LC patients diagnosed between 1 January 2004 and 31 informal caregivers. We considered the productivity, sick December 2018 were included. On 19 June 2006, the new leave, risk of early retirement and health service utilisation targeted medication erlotinib became available for Danish LC following their partner’s LC diagnosis. patients. Patients diagnosed after this date who were treated with any of the predefined new targeted cancer medicines were defined as the study’s After population. New targeted Material and methods medications were identified by treatment codes in the NPR. The study was designed as a retrospective cohort study These treatments were (treatment code in parenthesis) using the national Danish registers, covering the entire erlotinib (BWHA404), bevacizumab (BOHJ19B1), gefitinib Danish population (5.7 million). LC patients were identified (ML01EB01), crizotinib (BWHA413), afatinib (BWHA417), ceriti- as individuals registered with International Statistical nib (ML01ED02), Osimertinib (BWHA434), alectinib (BWHA440), brigatinib (ML01ED04), dacomitinib (ML01EB07), Classification of Diseases and Related Health Problems (ICD- lorlatinib (BWHA448), pembrolizumab (BOHJ19J3) and nivolu- 10) code DC34 (including all subgroups) as primary or sec- mab (BOHJ19H2). Patients diagnosed before 19 June 2006, ondary diagnosis in the Danish National Patient Register were regarded as the Before population (control). (NPR) [8]. All Danish residents have a unique 10-digit per- In addition, six subgroup analyses (SG 1–6) were per- sonal identification number, making it possible to merge per- formed. The first two subgroup analyses used later dates for son-level data from all Danish registers. The population was introduction of new targeted therapies (SG 1 and 2). SG 3 validated by cross-checking registrations in the Danish only considered Stage I–III LC patients, while SG 4 considered Cancer Register (DCR) [9]. All patients diagnosed after 1 Stage IV LC patients. SG 5 compared ALK-positive LC patients January 2004 and before 31 December 2018 were included. in the After group with Stage IV LC patients in the Before group. Lastly, LC patients with EGFR-mutations in the After group were compared with Stage IV LC patients in the Lung cancer patients Before group (SG 6). LC cases who received at least one of the new medications SG 1 focussed on the introduction of crizotinib on 26 in the year of or in the year following date of diagnosis was November 2012, using this date to separate between Before identified in the NPR. Characteristics of tumour size, lymph and After populations. SG 2 focussed on the additional new node involvement and metastatic spread (TNM classification ICIs pembrolizumab and nivolumab used to treat locally of malignant tumours) from DCR was used to determine LC advanced or metastatic non-small cell LC (NSCLC). In this stage at time of diagnosis. Date of birth, marital status, subgroup analysis, 1 January 2017 was used to distinguish region of residence and date of death (if applicable) were between Before and After populations. collected from the Central Person Register (CPR) [10]. Highest In SG 3 and 4, Stages I–III and Stage IV patients were con- achieved education of the patients before diagnosis was col- sidered separately. Cancer stage (I–IV) were determined by lected from the Population Education Register (PER) [11]. TNM codes from the DCR. During the study period, there were three versions of the TNM classifications. We applied Patients income and labour market affiliation were gathered the staging algorithm used by Norwegian Cancer Register to from the Income Register [12] and the DREAM database (The determine stage. There is extensive cooperation between the Danish Register of sickness absence, compensation benefits Nordic Cancer registers, thus the algorithm should be applic- and social transfer payments) [13], respectively. Tumour path- able to Danish data. ology was defined using the Systemised Nomenclature of In SG 5, ALK-positive LC patients were analysed. ALK-posi- Medicine (SNOMED) codes from The Danish National tive patients were identified if observed with SNOMED code Pathology Register and Databank [14]. T08 or T25-T29 (including sublevels) in combination with F2911 within 30 days of diagnosis. The first ALK-positive LC patient was identified in 2007, but ALK-specific treatment Spouses of lung cancer patients was not available before 2012. ALK-positive LC patients was Spouses, defined as a person married to or in a domestic compared with Stage IV LC patients in the Before group. relationship with LC patients at the time of diagnosis were Similarly, in SG 6, patients observed with SNOMED code identified in the CPR. For spouses, date of birth, region of T08 or T25-T29 (including sublevels) in combination with one residence from the CPR [10], highest achieved education of the following codes: FE13CA, FE13CB, FE13CC, FE13CD, from the PER [11], income from the Income Register [12] and FE13CK, FE13CP, F29616, FE13C3, FE13C5 within 30 days of labour market affiliation from the DREAM database [13] was diagnosis were classified as EGFR-positive patients. EGFR also collected. patients diagnosed during the After period was compared Healthcare utilisation for spouses was measured as con- with Stage IV LC patients in the Before period. tacts and cost from the NPR for hospital care, from the Two analyses focussed on the effect on Spouses of LC National Health Insurance Register [15] for primary care, and patients. One analysis where spouses of Before LC patients the Register of Medicinal Product Statistics for costs of pre- are matched and compared with spouses of After LC patients scription drugs. (Matched 1:1). And one analysis of spouses of Before LC ACTA ONCOLOGICA 3 patients are matched and compared with Spouses of After subgroup analysis, GLM (generalized linear model) distances LC patients and individuals not associated with cancer was used to achieve successful matching. Matching covari- (Matched 1:1:3). The reason for conducting two analyses, as ates was age, sex, region of residency, cancer stage, educa- opposed to one, was to retain as many spouses as reason- tion, and income in t-1. ably possible in the analysis. Spouses of Before patients was matched with spouses of After patients using the same method but evaluated with GLM distances. Covariates were age, sex, region of residence, Outcomes for lung cancer patients income in t-1 and education. Additionally, an analysis was For LC patients, the following outcomes were analysed; prod- performed where spouses of Before and After patients and uctivity (proxied by earnings), unemployment, risk of early persons not associated with cancer was matched on the retirement, and risk of death. same criteria, using exact matching (1:1:3, Before spouses: Productivity was measured as average salary (active After spouses: No Cancer) income) of Before/After LC patients in the year before diag- Successful matching was achieved if the following three nosis (t-1) and five years after diagnosis (t5). Earnings were criteria were fulfilled: (1) <15% deviance in variance ratio, (2) adjusted to a 2018 price level according to the consumer <±0.1 difference in standardised mean, and (3) at least 80% price index as calculated by Statistics Denmark. Transfer pay- of cases in the smallest group was retained. ments was not included. Unemployment was measured as average number of Statistical analysis weeks with unemployment each year from t-1 to t5 in the Before and After groups. Earnings, unemployment, sick leave and utilisation of health- Early retirement is granted on a municipality level if a care services, was analysed using used ordinary least square physician evaluates a person as being permanently unfit to (OLS) regression with the treatment variable (Before/After) work. Due to a policy change for early retirement in 2013, interacted with index period (t-1 to t5) to estimate the aver- we censored cases on 1 January 2013 for this analysis. age value of the respective outcomes by period. The follow-up time for overall mortality ranged from date Early retirement and all-cause mortality was analysed of diagnosis until 31 December 2019. using Cox proportional hazard regression (Cox model) to estimate the difference in risk of event between the groups. Outcomes for spouses of lung cancer patients The hazard ratios produced from the Cox model was adjusted for age, sex, cancer stage (where applicable), educa- For spouses of LC patients, the following outcomes were tion and region of residence. In analysis of early retirement, analysed; productivity (proxied by earnings), risk of early cases were censored at death, emigration or when turning retirement, incidence of long-term sick leave and health service 65 years old. The latter corresponding to criteria for age utilisation (proxied by cost). related retirement until 2008 in the study period. Productivity and risk of early retirement was evaluated To avoid immortal time, survival was analysed from day of same way as with the LC patients. first treatment onwards. First treatment was defined as treat- Use of health services was measured as the sum of the ment with antibodies or immunomodulatory treatment (pro- costs of healthcare services consumed in primary and hos- cedure code BOHJ, including subgroups), or treatments pital sector as well as cost of prescription drugs. The costs conditional to special medical treatment principles, for were adjusted to 2018 price level according to the Consumer example cytostatic agents (procedure codes BWH, including Price Index. Fees and Diagnosis Related Group (DRG)-tariffs subgroups). were applied as unit costs for primary healthcare and hos- Statistical significance was set to 5% in all models and pital sector resource use. Market prices were applied as unit was not adjusted for multiple testing and should be inter- costs for prescription drugs. preted as such. Long-term sick leave was defined as sick leave lasting lon- All analyses were performed in the statistical software R, ger than four weeks. As all shorter sick leave periods were version 4.1.0 (www.r-project.org). discarded, we consider the fifth week of sick leave as the first week of long-term sick leave. Results Balancing the samples Figure 1 presents a flow chart of the merging of data sets. A total of 50,275 eligible LC cases were identified during the Because the After sample only comprised those LC patients study period. who received one of the innovative LC medicines in focus, A total of 803 cases were not identified in the CPR. Table we matched these patients with Before patients to ensure 1A in the Supplementary appendix presents summary statis- comparable populations. Matching was done with optimal tics on the remaining 49,472 patients and their spouses. The pairwise propensity score matching, minimising the distance patients had a median age of 69 years, with 51% being between each patient in the After sample with a suitable males. In the Before period, 5,753 cases were to be matched match in the Before sample. Distance was evaluated by Mahalanobis distances. In the case of Stage I–III and EGFR with 2,430 After cases. A total of 38,518 patients were 4 J. H. RUDOLFSEN ET AL. Figure 1. Flow chart from identification of patients to matching of data sets. classified as ‘Other’, as they were diagnosed in the After cancer incidence between 2004 and 2018. Males had an 18% period, but did not receive any of the new innovative increase in incidence, while females had a 54% increase. treatments. More than half of the cases (51%) had Stage IV cancer at For spouses, 25,541 spouses were identified, with median time of diagnosis. ALK-positive and EGFR-positive mutations age of 67 (Supplementary Table 1A). Of these spouses, 1,643 were not registered before 2006 but had a steady increase in were spouses of LC patients in the After population. A total cases until 2012, when the incidence became stable. of 3,414 were spouses of Before LC patients (3,272 spouses of LC Patients diagnosed between 1 January 2004 and 18 Results main analysis June 2006). For LC patients in SG 2, 397 spouses were identi- fied. The remaining 20,087 spouses were spouses of ‘Other’ Table 2 present summary characteristics of the matched sam- LC patients. Summary statistics for all matched subsets are ple used in the main analysis. The matched sample was on provided in Supplementary Appendix A. average 1 year younger than the overall population (pre- Table 1 display LC incidence by stage and histologic sub- sented in Supplementary Table 1A) with a higher proportion type during the study period. There was a 26.5% increase in of females, but otherwise comparable. ACTA ONCOLOGICA 5 Table 1. Annual incidence by gender, stage, pathology and total. Year 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Male 2,051 2,091 2,121 2,228 2,223 2,259 2,296 2,267 2,371 2,236 2,403 2,302 2,335 2,386 2,340 Female 1,723 1,793 1,867 1,988 2,013 2,028 2,199 2,207 2,193 2,293 2,249 2,324 2,345 2,391 2,434 Total 3,774 3,884 3,988 4,216 4,236 4,287 4,495 4,474 4,564 4,529 4,652 4,626 4,680 4,777 4,774 Stage I 161 153 163 173 184 194 257 276 354 352 404 453 485 529 597 Stage II 442 464 378 459 431 454 547 572 614 647 695 673 752 748 741 Stage IIIa 237 282 256 288 306 324 339 335 354 307 372 380 400 368 299 Stage IIIb 632 637 664 667 639 595 631 645 581 629 573 593 563 617 671 Stage IV 1,903 1,955 2,101 2,229 2,338 2,402 2,498 2,385 2,394 2,344 2,357 2,270 2,195 2,229 2,212 Unknown stage 351 339 377 362 304 281 192 235 240 227 215 235 249 256 206 ALK positive 0 0 0 <5 <5 <5 6 7192027 20333825 EGFR positive <5 <5 11 8 13 17 46 83 115 152 149 160 164 161 164 Table 2. Summary statistics, matched lung cancer sample, main analysis. Characteristic Overall, N¼ 4,350 Before, N¼ 2,175 After, N¼ 2,175 Age 68 (61, 74) 68 (61, 74) 68 (62, 74) Income one year before diagnosis 0 (0, 14,537) 0 (0, 14,731) 0 (0, 14,102) Sex Male 1,904 (44%) 952 (44%) 952 (44%) Female 2,446 (56%) 1,223 (56%) 1,223 (56%) Education Primary or no education 2,063 (47%) 1,096 (50%) 967 (44%) Short higher education 2,126 (49%) 1,008 (46%) 1,118 (51%) Long higher education 82 (1.9%) 34 (1.6%) 48 (2.2%) Unknown education 79 (1.8%) 37 (1.7%) 42 (1.9%) Stage Stage I 84 (1.9%) 42 (1.9%) 42 (1.9%) Stage II 212 (4.9%) 106 (4.9%) 106 (4.9%) Stage IIIa 326 (7.5%) 163 (7.5%) 163 (7.5%) Stage IIIb 624 (14%) 312 (14%) 312 (14%) Stage IV 3,024 (70%) 1,512 (70%) 1,512 (70%) Unknown stage 80 (1.8%) 40 (1.8%) 40 (1.8%) Region Capital 902 (21%) 541 (25%) 361 (17%) Central Jutland 1,247 (29%) 533 (25%) 714 (33%) Northern Jutland 429 (9.9%) 206 (9.5%) 223 (10%) Zealand 877 (20%) 441 (20%) 436 (20%) Southern Denmark 895 (21%) 454 (21%) 441 (20%) Median with the inter quartile range in brackets, or number of observations with ratio in brackets. Out of 2175 Before patients 2130 died during the study Spouses of After patients had significantly lower health- period, while the equivalent numbers for After patients was care costs in the years following diagnosis (t-1–t5). In 1953 out of 2175. Six Before patients died in the period Figure 2, the costs are displayed. Results of analysis with between test and diagnosis. All the After patients received spouses of Before and After patients are to the left, and any of the new innovative LC medicine, while 1213 Before results of analysis with spouses of Before and After patients patient received standard medical treatment. In adjusted as well as persons without cancer to the right. analysis, After patients had significantly reduced risk of death Healthcare costs generated by spouses of After patients (HR ¼ 0.76, 0.710.82) compared to the matched Before seems unaffected by the LC diagnosis. For spouses of Before group. patients, however, the annual healthcare costs increased For productivity, no significant differences were found slightly about $95 (700 DKK) from t0 to t1 and plateaued between Before and After populations. Average income in around this new level. In the analysis including individuals the year before diagnosis was $9600 (DKK 72,000) and not associated with cancer, a similar pattern was seen. There decreased to about $4000 (DKK 30,000) five years after diag- were no significant differences in healthcare costs between nosis. No significant differences were found in unemploy- persons without cancer and spouses of After LC patients, ment (LC patients) or long-term sick leave (spouses) between until t4 and t5 (Figure 2). Before and After groups (data not shown). Out of 649 (20,5%) at risk Before cases, 133 (20,5%) entered early retirement before 1 January 2013 versus 40 out Subgroup analyses of 334 (12%) at-risk After cases. Patients in the After sample Results from risk of death and early retirement for subgroup had significantly reduced risk of early retirement (HR: 0.54, analyses are presented in Table 3. All After subgroups had 0.380.79). For spouses, we found no significant differences reduced risk of death compared with their respective Before (HR 1.06, 0.234.94). However, this result should be seen in populations (see Table 3). The largest reduction in risk of context of fewer than five spouses of After LC patients entered early retirement. death was found in ALK-positive patients (SG 5) (HR ¼ 0.37, 6 J. H. RUDOLFSEN ET AL. (A) (B) Healthcare costs Healthcare costs 1,200 1,200 1,000 1,000 800 800 600 600 400 400 200 200 0 0 -1 01 23 45 -1 0 1 2 3 4 5 After Before After Before NoCancer After Before NoCancer Healthcare costs by period Time t-1 t0 t1 t2 t3 t4 t5 t-1 t0 t1 t2 t3 t4 t5 After 802 804 815 823 829 791 802 834 841 845 853 869 821 818 CI 745-859 747-861 756-874 762-885 762-897 717-865 718-886 762-907 768-914 770-920 775-931 784-954 727-915 712-925 848 876 948 995 990 957 1003 933 970 1037 1083 1065 988 1021 Before 791-905 819-933 891-1006 936-1053 931-1049 897-1017 942-1064 860-1006 897-1043 963-1110 1008-1157 989-1140 911-1065 943-1100 CI NoCancer 789 828 862 909 934 973 976 747-831 786-870 820-905 866-952 891-978 929-1017 931-1021 CI After: 45 After: 13 After: -17 After: -56 After: -65 After: -152 After: -158 -134 -201 -47 -72 -171 -161 -166 (151 - -60) (119 - -92) (89- -123) (50 – -162) (42 - -172) (-44 - -259) (-49 - -266) Difference Before: 144 Before: 142 Before: 175 Before: 174 Before: 130 Before: 15 Before: 45 42 – -135 16- -161 --45 - -222 -82 - -260 -71 - -250 -76 - -257 -110 - -292 (272 – 16) (270 – 14) (303 – 45) (303 – 45) (260 – 0) (146 - -115) (177 - -86) Figure 2. Healthcare costs (Eur) generated by spouses of LC patients. A: Analysis of spouses of Before and After patients (N ¼ N ¼ 1292). B: SpouseBefore SpouseAfter Analysis of spouses of Before and After patients, as well as persons not associated with cancer (N ¼ N ¼ 1139, N ¼ 3415). For panel (A) SpouseBefore SpouseAfter NoCancer difference was tested between Before and After patients, for panel (B) difference was tested between After – Control, and Before-Control. Table 3. Results from subgroup analyses. Before/After Subgroup number n Lung cancer cases Early retirement Mortality 1 1,709/572 Before/After, new treatment No event in After HR ¼ 0.63 (0.560.72) 2 4,023/1,352 Before/After subgroup No event in After HR ¼ 0.56 (0.510.62) 3 1,219/650 Stages I–III HR: 0.81 HR ¼ 0.82 (0.45–1.5) (0.730.92) 4 1,512/1,512 Stage IV HR: 0.57 HR: 0.73 (0.37–0.87) (0.670.80) 5 171/57 ALK positive Too few events in ALK HR ¼ 0.37 (0.250.55) 6 599/215 EGFR positive HR: 0.47 HR ¼ 0.59 (0.13–1.69) (0.490.72) Before as reference for hazard ratio in all analyses. After populations defined as patients treated with erlotinib, bevacizumab, gefiti- nib, crizotinib, afatinib, ceritinib, Osimertinib, alectinib, brigatinib, dacomitinib, lorlatinib, pembrolizumab or nivolumab. Before popu- a b lation diagnosed before 19 June 2006. After received nivolumab, pembrolizumab or durvalumab. Treated Before/After 26 November 2012, Definition of After population same as in main analysis. Stage I–III LC patients diagnosed with Stage I–III cancer according to TNM codes. Stage IV patients diagnosed with Stage IV according to TNM codes. ALK-positive patients (SNOMED code T08 or T25–T29 (including sublevels) in combination with F2911 within 30 days of diagnosis) in the After population compared with Stage IV LC patients in the Before population. LC patients with EGFR-mutations (SNOMED code T08 or T25-T29 (including sublevels) in combination with one of the following codes: FE13CA, FE13CB, FE13CC, FE13CD, FE13CK, FE13CP, F29616, FE13C3, FE13C5 within 30 days of diagnosis) in the After group were compared with Stage IV LC patients in the Before group. 0.250.55) corresponding to a 63% reduction in hazard of requirements in 2013. In analysis of SG 5 and 6, there were death, while the smallest reduction was found in the analysis fewer than five events. As a result, there is high uncertainty of Stages I–III patients (SG 3) (HR ¼ 0.82, 0.730.92). associated with these estimates. For SG 1 and 2, analysis of early retirement was not In SG4, there was a significant reduced risk of early retire- applicable due to the reform in early retirement ment (HR ¼ 0.57, 0.370.87) corresponding to a 43% ACTA ONCOLOGICA 7 reduction in hazard. For SG 3, HR indicated a reduced hazard diagnosis, compared to spouses of Before LC patients. of early retirement in the After group however, not a statis- Severity of diagnosis has been shown to have a positive cor- tically significant reduction. relation with the burden on informal caregivers [17]. Patients diagnosed after 26 November 2012 who received Unobserved variables such as identification of biomarkers, new LC medicines (SG2) and patients with ALK-positive can- new surgical techniques, improved radiotherapeutic treat- cer (SG5) both earned more in the year before diagnosis, ment and methods for earlier diagnosis [18] as well as compared with their matched Before samples. In SG2, After changes in the patient population could be responsible for patients had significantly higher unemployment rate two some of the reduced risk of mortality or early retirement. years after diagnosis compared with their matched Before Particularly, incidence of LC for women is increasing [19]. During the study period, the ratio of female patients group. No other significant differences were found in changed from 45.5% in 2004 to 51% in 2018. Such changes unemployment or sick leave for any subgroup (results avail- able from corresponding author). in the patient population underline the importance of real- world studies, due to tendency of selection of specific patient populations used in randomised controlled trials, that Discussion might not be representative of the real-world LC population. When comparing LC patients before and after introduction of new innovative medicine, it was found that LC patients Strength and weaknesses receiving new innovative medicines had reduced hazard of The analysis is based on national registers, spanning 17 years death overall, and in all subgroups. This is in line with with reliable, routinely collected data before and after date expectations as new treatments are most often approved of diagnosis. based on a proven survival benefit compared to former Misclassification of data might occur. The authors could standard treatment in clinical randomised trials. Note, no new systemic treatments targeting early-stage LC and locally not validate the coding practice used in the inclusion/exclu- advanced LC (stage I–III) were introduced during the study. sion criteria. However, the Danish registers are of high qual- ity and are routinely used in population-based inquiries such The reduced hazard of death is therefore likely due to newer as the present. Furthermore, the sample size reduces the operation techniques and radiotherapy deliveries which have impact of random variation on the effect estimates. increased survival in this subgroup. The reduced hazard in For some matches there are several years between dates advanced stages is therefore probable due to advancements of diagnosis. Other developments in LC treatment, Small Cell in systemic treatment for subgroups of patients. In early and LC vs Non-Small Cell LC, performance status and other clinic- locally advanced stages (I–III), LC patients had a 18% reduc- ally covariates were not accounted for when matching. tion in hazard of death, while ALK patients who were treated Therefore, there could be differences in patients’ health at with new targeted medicine during this time reduced their baseline not captured by the stage indicator or distribution hazard with 63%. of Non-Small/Small Cell LC can vary in the Before and After LC patients who received the new innovative LC medi- population. This can have affected the results. cines had a significantly reduced hazard of early retirement. The matching criteria were met for all sets, except for the The subgroup analyses showed a significant hazard reduction ALK-positive group. Due to the relatively small sample of for Stage IV LC patients. This finding coincides well with the ALK-positive cases, we chose to breach the variance ratio cri- reduced toxicity from new innovative medicines. terion for age to retain all but two ALK-positive cases. For the other outcomes, there were sporadic significant As there are multiple outcomes and populations and sub- differences in subgroups. However, without a consistent pat- populations there is a risk of Type 1 error. The reader should tern of differences these results might be a result of natural be aware of this and consider any result in context of all variation, rather than an effect of the treatment. It should be results. noted that higher survival rate in the After population likely causes survival bias. In other words, when more patients sur- vive into retirement age, they are registered with zero Conclusion income, thus reducing the average earnings in the After population. The new innovative LC medicines introduced in Denmark The combination of reduced risk for both early retirement during the study period have had a positive effect on the and death is interesting. A previous study found that LC patients’ life. This is evident in survival, risk of early retire- patients did not live long enough to retire early [16]. ment and reduced healthcare utilisation for spouses of LC However, LC patients in advanced settings who received patients. The reduced risk of early retirement demonstrates new innovative treatments both lived longer and had an improvement of the low labour market participation in LC reduced risk of early retirement. With fatigue being the pri- patients following diagnosis and treatment. Spouses of those mary cause of exit from the labour market [7], it could sug- who received the new treatments generated lower health- gest that reduced toxicity through personalised medicine care costs, possible due to a reduced burden of illness for reduces fatigue symptoms in the patients. the LC patients. Spouses of After LC patients had significantly lower While significant improvements in treatment have been healthcare costs in the years following their partners LC achieved in the last 20 years, the prognosis for LC patients is 8 J. H. RUDOLFSEN ET AL. Europe 1999–2007: results from the EUROCARE-5 study. Eur J still not satisfactory, and continued effort to improve clinical Cancer. 2015;51(15):2242–2253. outcomes are necessary. [3] The global challenge of cancer. Nat Cancer. 2020;1(1):1–2. [4] Lawless GD. The working patient with cancer: implications for payers and employers. Am Health Drug Benefits. 2009;2(4): Ethics approval and consent to participate 168–173. The study was register-based and complied with the regulations and [5] Vayr F, Savall F, Bigay-Game L, et al. Lung cancer survivors and instructions set up by Statistics Denmark. No additional ethics commit- employment: a systematic review. Lung Cancer. 2019;131:31–39. tee approval is required for register-based research according to [6] Rashid H, Eichler M, Hechtner M, et al. Returning to work in lung Danish law. cancer survivors—a multi-center cross-sectional study in Germany. Support Care Cancer. 2021;29(7):3753–3765. [7] Kim YA, Yun YH, Chang YJ, et al. Employment status and work- Disclosure statement related difficulties in lung cancer survivors compared with the general population. Ann Surg. 2014;259(3):569–575. The study was financed by Pfizer Denmark. Jan Håkon Rudolfsen and [8] Lynge E, Sandegaard JL, Rebolj M. The Danish national patient Mikkel H. Pedersen are employees at Incentive, which is a paid vendor register. Scand J Public Health. 2011;39(7_suppl):30–33. of Pfizer Denmark. Mads D. Hjortsø and Trine Pilgaard are employee of [9] Gjerstorff ML. The Danish cancer registry. Scand J Public Health. Pfizer Denmark. Mette Pøhl was paid by Pfizer Denmark for her work as 2011;39(7_suppl):42–45. an oncology specialist in the project group. [10] Pedersen CB. The Danish civil registration system. Scand J Public Health. 2011;39(7_suppl):22–25. [11] Jensen VM, Rasmussen AW. Danish education registers. Scand J Funding Public Health. 2011;39(7_suppl):91–94. This study was supported by Pfizer Denmark. [12] Baadsgaard M, Quitzau J. Danish registers on personal income and transfer payments. Scand J Public Health. 2011;39(7_suppl): 103–105. ORCID [13] Hjollund NH, Larsen FB, Andersen JH. Register-based follow-up of social benefits and other transfer payments: accuracy and degree Mikkel H. Pedersen http://orcid.org/0000-0002-3568-767X of completeness in a Danish interdepartmental administrative Trine Pilgaard http://orcid.org/0000-0002-1771-956X database compared with a population-based survey. Scand J Public Health. 2007;35(5):497–502. [14] Erichsen R, Lash TL, Hamilton-Dutoit SJ, et al. Existing data sour- Data availability statement ces for clinical epidemiology: the Danish national pathology The data that support the findings of this study are available from registry and data bank. Clin Epidemiol. 2010;2:51–56. Statistics Denmark’s Research Service, but restrictions apply to the avail- [15] Andersen JS, Olivarius NDF, Krasnik A. The Danish national ability of these data, which were used under licence for the current health service register. Scand J Public Health. 2011; study, and so are not publicly available. Additional analyses are however Jul39(7_suppl):34–37. available from the authors upon reasonable request and with permission [16] Taskila-Åbrandt T, Pukkala E, Martikainen R, et al. Employment of Statistics Denmark’s Research Service. status of Finnish cancer patients in 1997. Psychooncology. 2005; 14(3):221–226. [17] Alshammari B, Noble H, McAneney H, et al. Factors associated References with burden in caregivers of patients with end-stage kidney dis- ease (A systematic review). Healthcare. 2021;9(9):1212. [1] Novello S, Kaiser R, Mellemgaard A, et al. Analysis of patient- [18] Lemjabbar-Alaoui H, Hassan OU, Yang YW, et al. Lung cancer: reported outcomes from the LUME-Lung 1 trial: a randomised, biology and treatment options. Biochim Biophys Acta BBA - Rev double-blind, placebo-controlled, phase III study of second-line Cancer. 2015;1856(2):189–210. nintedanib in patients with advanced non-small cell lung cancer. [19] Egleston BL, Meireles SI, Flieder DB, et al. Population-based Eur J Cancer. 2015;51(3):317–326. trends in lung cancer incidence in women. Semin Oncol. 2009; [2] Francisci S, Minicozzi P, Pierannunzio D, EUROCARE-5 Working Group, et al. Survival patterns in lung and pleural cancer in 36(6):506–515.
Journal
Acta Oncologica
– Taylor & Francis
Published: Mar 4, 2023
Keywords: Lung cancer; real-world evidence; new medication; register data; survival; early retirement
Recommended Articles
Loading...
References
Analysis of patient-reported outcomes from the LUME-Lung 1 trial: a randomised, double-blind, placebo-controlled, phase III study of second-line nintedanib in patients with advanced non-small cell lung cancer
Survival patterns in lung and pleural cancer in Europe 1999–2007: results from the EUROCARE-5 study
The global challenge of cancer
The working patient with cancer: implications for payers and employers
Lung cancer survivors and employment: a systematic review
Returning to work in lung cancer survivors—a multi-center cross-sectional study in Germany
Employment status and work-related difficulties in lung cancer survivors compared with the general population
The Danish national patient register
The Danish cancer registry
The Danish civil registration system
Danish education registers
Danish registers on personal income and transfer payments
Register-based follow-up of social benefits and other transfer payments: accuracy and degree of completeness in a Danish interdepartmental administrative database compared with a population-based survey
Existing data sources for clinical epidemiology: the Danish national pathology registry and data bank
The Danish national health service register
Employment status of Finnish cancer patients in 1997
Factors associated with burden in caregivers of patients with end-stage kidney disease (A systematic review)
Lung cancer: biology and treatment options
Population-based trends in lung cancer incidence in women
Share the Full Text of this Article for FREE
You can share this free article with as many people as you like with the url below! We hope you enjoy this feature!
Sign up for your 14-Day Free Trial Now!
Read and print from thousands of top scholarly journals.
To get new article updates from a journal on your personalized homepage, please log in first, or sign up for a DeepDyve account if you don’t already have one.
All DeepDyve websites use cookies to improve your online experience. They were placed on your computer when you launched this website. You can change your cookie settings through your browser.
