- Email: [email protected]
Healthcare costs in the Danish randomised controlled lung cancer CT-screening trial: A registry study
Lung Cancer 83 (2014) 347–355
Contents lists available at ScienceDirect
Lung Cancer journal homepage: www.elsevier.com/locate/lungcan
Healthcare costs in the Danish randomised controlled lung cancer CT-screening trial: A registry study夽 Jakob F. Rasmussen a,∗ , Volkert Siersma a , Jesper H. Pedersen b , Bruno Heleno a , Zaigham Saghir b , John Brodersen a a The Research Unit for General Practice and Section of General Practice, Department of Public Health, University of Copenhagen, Oester Farimagsgade 5, 1014 Copenhagen K, Denmark b Department of Cardiothoracic Surgery, RT 2152, Rigshospitalet, University of Copenhagen, Blegdamsvej 9, 2100 Copenhagen, Denmark
a r t i c l e
i n f o
Article history: Received 1 October 2013 Received in revised form 12 December 2013 Accepted 16 December 2013 Keywords: Lung cancer Mass screening Cancer screening test Scan spiral CT Healthcare cost Cancer early detection
a b s t r a c t Objectives: Low dose computerised tomography (CT) screening for lung cancer can reduce lung-cancerspecific mortality. The objective of this study was to analyse healthcare costs and healthcare utilisation of participants in the Danish lung cancer CT-screening trial (DLCST). Materials and methods: This registry study was nested in a randomised controlled trial (DLCST). 4104 participants, current or former heavy smokers, aged 50–70 years were randomised to five annual low dose CT scans or usual care during 2004–2010. Total healthcare costs and healthcare utilisation data for both the primary and the secondary healthcare sector were retrieved from public registries from randomisation – September 2011 and compared between (1) the CT-screening group and the control group and, (2) the control group and each of the true-positive, false-positive and true-negative groups. Results: The median annual costs per participant were significantly higher in the CT-screening group (Euros [EUR] 1342, interquartile range [IQR] 750–2980) compared with the control group (EUR 1190, IQR 590–2692) (p < 0.0001). When the cost of the CT-screening programme was excluded, there was no longer a statistically significant difference between the CT-screening group (EUR 1155, IQR 567–2798) and the control group (p = 0.52). Analyses according to the diagnostic groups showed that annual costs were 10.57 (95% CI 7.09–15.75) times higher for the true-positive and 1.67 (95% CI 1.20–2.32) times higher for the false-positive group compared with the control group. Conclusion: Low dose lung cancer CT screening increases healthcare costs compared with no screening; this difference was attributable to the costs of the CT-screening programme. Overall healthcare costs were higher for the true-positive and false-positive groups than for the control group, also when excluding the cost of the CT-screening programme. This increase was outweighed by the larger true-negative group showing no significant differences in costs compared with the control group. © 2013 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Lung cancer is the primary cause of cancer-related deaths in the world [1,2]. The National Lung Screening Trial (NLST) has shown a relative risk reduction in lung-cancer-specific mortality of 20% and 6.7% in all-cause mortality using low dose computerised tomography (CT) screening [3]. The Danish Lung Cancer Screening Trial (DLCST) and five other European randomised controlled lung cancer CT-screening trials are currently being evaluated [4–10]. Data pooling of the European trials is required to show whether the lung-cancer-specific mortality reduction found in NLST is also
夽 ClinicalTrials.gov: NCT00496977. ∗ Corresponding author. Tel.: +45 3532 7171; fax: +45 3532 7131. E-mail address: [email protected] (J.F. Rasmussen). 0169-5002/$ – see front matter © 2013 Elsevier Ireland Ltd. All rights reserved. http://dx.doi.org/10.1016/j.lungcan.2013.12.005
a consistent result in the European trials. The final data analyses on mortality and overdiagnosis in the European trials are planned in 2015 and therefore cost-effectiveness analyses cannot be performed at present [5]. A recently published Cochrane review concluded that even though a lung-cancer-specific mortality reduction was found in the NLST more data are needed on cost effectiveness, overdiagnosis and false-positive results before recommendations of lung cancer CT-screening programmes should be made [11]. It is currently unclear whether lung cancer CT screening is cost effective [12]. The false positive results will contribute to the total costs of CT screening and the false-positive rate differs across the CT-screening trials (the NLST = 23%, the Dutch-Belgian Lung Cancer Screening Trial (NELSON) = 13% and the DLCST = 3%) [3,9,13]. A study analysing the healthcare costs in the NLST found that: “The high false-positive rate of LDCT screening, as reported in the
348
J.F. Rasmussen et al. / Lung Cancer 83 (2014) 347–355
NLST, will clearly contribute to these expenditures, because a falsepositive screening result leads to unnecessary follow-up tests and procedures, and their respective costs” [12]. However, the study included only the lung-related costs of the CT screening and followup procedures [12]. In mammography screening, it has been shown that both breast-related and non-breast-related outpatient visits increased significantly for those women receiving a false-positive result [14]. Moreover, the diagnostic follow-up of false-positive screening mammography results leads to more concern and more healthcare utilisation in general [14–17]. Thus, it is important to know whether participants in lung cancer CT screening seek more medical attention not only related to lung cancer screening. Consequently, the aim of this study was to analyse all healthcare costs in the DLCST population, in both the primary and the secondary healthcare sector, during the screening period. The costs were compared between (1) the screening group and the control group and, (2) the control group and each of the true-positive, falsepositive and true-negative groups. 2. Methods 2.1. Study population The DLCST ran during 2004–2010; the study design is described in detail elsewhere [9,18]. Briefly, 4104 current and former heavy smokers, aged 50–70 years, were randomised to CT screening or no screening with 2052 participants in each group (Fig. 1). The participants in the screening group were offered five annual low dose CT scans. Both groups made annual visits to a screening clinic and received lung function tests, smoking counselling, and completed questionnaires [18,19]. 2.2. DLCST budget The DLCST received a grant of EUR 2.33 million from the Ministry of Health and Prevention. This grant covered the expenses of the CT-screening programme (EUR 238/CT scan) including recruitment of participants, the 9800 CT scans and salary to the staff at the screening clinic during the five screening rounds. The CT-screening scans were not billed to the Danish healthcare system and thus not registered in the public registries. The costs of additional CT scans and all procedures in the follow-up of those with abnormal findings were not included in the grant and were thus billed to the Danish healthcare system and recorded in the public registries. 2.3. Registries and outcomes Denmark has a publicly financed healthcare system and all healthcare costs and services are recorded in public registries. Residents in Denmark have a unique personal identification number and every event in the public healthcare system is linked to this number. Thirteen outcomes related to specific healthcare costs or use, were analysed and categorised as the following: A. Total costs in primary and secondary healthcare sector (2 outcomes) 1. Total costs including costs of CT-screening programme (EUR 238/CT-scan) 2. Total costs excluding costs of CT-screening programme B. Primary healthcare sector (6 outcomes) 1. Costs of general practitioners 2. Procedures at general practitioners 3. Contacts to general practitioners.
Fig. 1. Design and diagnostic results of the Danish Lung Cancer Screening Trial.
4. Costs of other specialised medical doctors 5. Costs of psychologists 6. Costs of physiotherapists C. Secondary healthcare sector (5 outcomes) 1. Hospitalisation (days) 2. Outpatient visits 3. Emergency room contacts (including out-of-hours contacts in primary sector) 4. Surgical procedures 5. Non-surgical procedures Appendix Table 3 presents an overview of registries and data extraction. 2.4. Data sets and observation time 1. Screening group and control group: data on the 13 outcomes were cumulated over a total observation period from date of randomisation (2004–2006) to either September 2011 (66–84 months), death or migration and data were compared between screening group and control group. These outcomes were annualised into outcome per year to adjust for different observation times. The participants with an observation time < 12 months of the total observation time of 66–84 months were excluded
J.F. Rasmussen et al. / Lung Cancer 83 (2014) 347–355
from the analyses to avoid inflation of the healthcare costs when annualising the outcomes. 2. Diagnostic groups and control group: in the CT-screening group there were four possible diagnostic groups: (a) true-negative, (b) false-positive, (c) true-positive and (d) false-negative. As there was only one participant with a false-negative result, this person was excluded from the present study [9]. The healthcare costs of each of the three remaining diagnostic groups were compared with the healthcare costs of the control group. The participants in the screening group could receive different diagnoses during the five screening rounds. Therefore, the period of comparison for the screening group and the control group was defined as: the period from the date of CT screening (screening group) or visit to the clinic (control group) to the next date of CT screening or visit to the clinic. If a participant missed a round, the period following the previous round was defined as 365 days from the previous CT screening or visit to the clinic; the period following the missing round was set to missing. The fifth observation period was defined as 365 days from the date of the last CTscreening round or last visit to the clinic. The mean observation time for the first four periods was 377 days. The outcomes in each period were annualised to outcome per year to adjust for different observation times.
2.5. Statistical analysis Screening group versus control group: The differences in baseline characteristics between the DLCST randomisation groups were tested by chi-squared tests for categorical variables or t-tests for continuous variables. The differences in healthcare costs between participants in the screening group and the control group were analysed with Wilcoxon signed-rank test. Diagnostic groups versus control group: Since the diagnostic groups were not randomised, it was necessary to use a multivariable analysis. Some outcomes rarely used by the population include many zero observations (e.g. surgical procedures). Consequently, it is more appropriate to analyse the risk of having, for example, a surgical procedure at all in the observation period and thus analyse the prevalence of use. Other outcomes frequently used do not have many zero observations (e.g. total costs). In this aspect it is more relevant to analyse the quantity of use. Therefore, a multivariable analysis of the outcomes followed a two-part model where the three diagnostic groups in the period following a specific screening round were compared with the control group:
1. In part one, the prevalence of use of selected healthcare services or costs >0 in the period was analysed [20]. This prevalence was analysed in a Poisson regression approach so the regression parameters were equivalent to the logarithm of the relative risk of ever using the service in the observation period [21]. This prevalence part of the model was the most relevant for the healthcare services rarely used by the population (services in the secondary sector). 2. In part two, the quantity of healthcare costs or services was analysed. Only the participants who used the services or had costs >0 were included, and the quantity of use was analysed in a generalised linear model using a Gamma distribution and a logarithmic link function. The parameters from this model were interpreted as the logarithm of a multiplicative factor of how much more the service was used (or how much more cost) in one of the diagnostic groups compared with the use/costs in the control group. This quantity part of the model was the most relevant for more frequently used healthcare services mainly in the primary sector and healthcare costs.
349
The analyses of both the prevalence part and the quantity part were multivariably adjusted for the following variables: sex, age, age-squared, screening-round number, days of observation, employment, living alone, socioeconomic group, smoking status (former/current smoker), pack years, pack years-squared and region of residence. Because the estimation did not converge for the full adjustment, possibly because the outcome prevalences were low, the outcome of psychologist costs was not adjusted for age-squared, employment, living alone and days of observation; the outcome of physiotherapist costs was not adjusted for agesquared, pack years-squared, region of residence and employment; and the outcome of emergency room contacts was not adjusted for days of observation. The outcome “Total costs including costs of CT-screening programme” was excluded, as the main focus in the multivariable analyses was to analyse the costs in the year directly following the CT-screening result. In both part one and two the repeated observation (up to five rounds) of the same participants was adjusted for using Generalised Estimating Equations (GEE). A combined multiplicative effect of being in one of the diagnostic groups compared with being in the control group was calculated by multiplying the relative risk from the prevalence part and the factor from the quantity part of the model. This multiplicative effect is referred to as the cumulative effect. Statistical significance was assessed at a 5% level when the screening group and the control group were compared. In the multivariable analyses p values were calculated individually for the prevalence part and for the quantity part of the model and we adjusted for multiple testing using the method of Benjamini–Hochberg [22]. Statistical Analysis Software 9.2 (SAS Institute, Cary, North Carolina) was used to analyse the data.
2.6. Ethics The DLCST has been approved by the Danish Scientific Ethical Committee: number of approval KA-02045. All participants signed an informed consent form. Both the DLCST (approval number 200553-1083) and this project (approval number 2011-41-5852) have been approved by the Danish Data Protection Agency. ClinicalTrials.gov: NCT00496977.
3. Results 3.1. Participation Overall participation in the DLCST was 95.5% in the screening group and 93.0% in the control group (Fig. 1). No significant differences in baseline socioeconomic characteristics or smoking habits were found between the two groups (Table 1). During the screening trial, 148 (3.6%) participants died and 27 (0.6%) emigrated; 31 participants had a total observation time <12 months and were excluded. The data on all 4104 participants in the DLCST were retrieved from the registries and none was lost to follow-up.
3.2. Screening group versus control group The annual total costs differed statistically significantly between the screening group and the control group when the cost of the screening programme was included. When this cost was excluded, the differences in total costs were not statistically significant. There were no statistically significant differences in the other outcomes (Table 2). In the screening group the rate of the positive tests was lower in the incidence rounds compared to the prevalence round (Fig. 1).
350
J.F. Rasmussen et al. / Lung Cancer 83 (2014) 347–355
Table 1 Baseline Characteristics in DLCST. Characteristics Sex, n (%) Age, mean (SD) Social group, n (%)
Men Women I II III IV V Employed, social group uncertain Outside the labour market
n/n
Screening group n = 2052
Control group n = 2052
2052/2052
1147 (55.9) 905 (44.1)
1120 (54.6) 932 (45.4)
2052/2052 2041/2039
57.3 (4.8) 155 (7.6) 402 (19.7) 378 (18.5) 545 (26.7) 265 (13.0) 182 (8.9) 114 (5.6)
57.3 (4.8) 141 (6.9) 410 (20.1) 378 (18.5) 551 (27.0) 282 (13.8) 168 (8.2) 109 (5.4)
School education, n (%)
7–9 years in school 10 years in school 12–13 years in school Other
2047/2047
698 (34.1) 775 (37.9) 553 (27.0) 21 (1.0)
715 (34.9) 790 (38.6) 532 (26.0) 10 (0.5)
Employment status, n (%)
Employed Studying Job seeking Retired
2043/2045
1366 (66.9) 9 (0.4) 113 (5.5) 555 (27.2)
1324 (64.7) 12 (0.6) 104 (5.1) 605 (29.6)
Region of residence, n (%)
Capital region Region Zealand Region of Southern Denmark Region of Central Denmark Region of Northern Denmark
2037/2044
1644 (80.7) 353 (17.3) 29 (1.4) 8 (0.4) 3 (0.2)
1653 (80.9) 349 (17.1) 28 (1.4) 11 (0.5) 3 (0.2)
Living alone, n (%)
No Yes
2039/2034
1457(71.5) 582(28.5)
1453(71.4) 581(28.6)
Smoking status, n (%)
Current smoker Former smoker
2050/2051
1544 (75.3) 506 (24.7)
1579 (77.0) 472 (23.0)
Smoking history, mean (SD)
Pack years
2051/2048
36.4 (13.4)
35.9 (13.4)
SD: standard deviation.
3.3. Diagnostic groups versus control group The mean total annual healthcare costs were statistically significantly higher for the true-positives (EUR 28 254) and false-positives (EUR 4464) groups compared with the control group (EUR 2673) with a factor of 10.57 (95% CI 7.09–15.75) and 1.67 (95% CI 1.20–2.32) respectively. The true-negative group had lower total annual healthcare costs with a factor of 0.96 (95% CI 0.86–1.09)
but did not show any statistically significant difference (Fig. 2 and Appendix Table 2). In the six primary healthcare outcomes the true-positive group had a statistically significant higher healthcare use in one outcome and no significant differences of the higher healthcare use in the outcomes of General practitioner procedures, General practitioner costs and Psychologist costs were found. The two remaining outcomes both revealed lower primary healthcare use: one showed a
Table 2 Comparison of annual healthcare costs and utilisation per participant between the screening group and the control group in DLCST. Outcome
Screening group Median
Total costs incl. cost of CT-screening programme Total costs excl. cost of CT-screening programme Hospitalisation days Outpatient visits Surgical procedures Non-surgical procedures Emergency room contactsa General practitioner costs General practitioner procedures General practitioner contacts Other specialist MD costs Psychologist costs Physiotherapist costs
Control group
Range
IQR
Median
Min
Max
Q1
Q3
1342
110
206508
750
2980
1155
22
206359
567
0 1 0 3 1 246 26 19 107 0 0
0 0 0 0 0 12 1 1 0 0 0
83 59 16 167 49 3253 221 165 1281 299 3322
0 0 0 1 0 153 15 11 37 0 0
Range
p-Value*
IQR
Min
Max
Q1
Q3
1190
11
63314
590
2692
<.0001
2798
1190
11
63314
590
2692
0.5179
1 3 1 8 2 378 41 30 211 0 4
0 1 0 3 1 253 26 20 107 0 0
0 0 0 0 0 4 0 0 0 0 0
84 74 12 956 367 5235 795 678 1639 496 3155
0 1 0 1 0 162 16 12 43 0 0
1 3 1 8 2 387 41 31 220 0 3
0.1881 0.5860 0.5861 0.9290 0.6099 0.2171 0.2468 0.1847 0.1449 0.3541 0.6010
Costs are presented in Euros, EUR. (Converted from DKK using the 15 November 2013 spot rate DKK 745.88 = EUR 100). IQR: interquartile range; Q1: quartile 1; Q3: quartile 3. a Out-of-hours contacts to general practitioners are included in emergency room contacts. * Statistical significance was assessed at a 5% level.
J.F. Rasmussen et al. / Lung Cancer 83 (2014) 347–355
351
Fig. 2. Prevalence and quantity of healthcare costs and utilisation in the diagnostic groups compared with the control group.
significant difference and one did not show any significant difference. A statistically significant higher healthcare use was found in four of five outcomes in the secondary healthcare sector (Table 3, Fig. 2, Appendix Tables 1 and 2). The false-positive group had statistically significant higher healthcare use in one of the six primary healthcare outcomes and no significant differences of the higher healthcare use in the outcomes of General practitioner contacts, General practitioner procedures, General practitioner costs and Physiotherapist costs were found. A statistically significant higher healthcare use was found in four of the five outcomes in the secondary healthcare sector (Table 3, Fig. 2, Appendix Tables 1 and 2). The true-negative group had lower healthcare use in all outcomes in the primary healthcare sector but none showed a significant difference. In the secondary healthcare sector a statistically significant lower healthcare use was found in the outcomes Outpatient visits and Non-surgical procedures (Table 3, Fig. 2, Appendix Tables 1 and 2).
4. Discussion This is the first study to present the total healthcare costs, in both the primary and secondary healthcare sector, of the participants in a randomised lung cancer CT-screening trial. The total costs of the screening group were significantly higher than the costs of the control group when the costs of the CT-screening programme were included. However, when the costs of the CT-screening programme were excluded, this difference disappeared. Therefore, the excess costs in the screening group were attributable to the costs of the screening programme itself and not by a greater healthcare use. The higher healthcare costs for the true-positive and the false-positive groups, presented as the cumulative effect, were found in both the secondary and primary sector (Table 3). Those diagnosed with lung cancer and those receiving an abnormal CT scan later confirmed to be false-positive may have had greater needs and concerns; accordingly, they consulted their general practitioners and other primary healthcare professionals more. The true-negative group may have
Table 3 Cumulative effect of prevalence and quantity of healthcare costs and utilisation in the diagnostic groups relative to the control group in the DLCST. Outcome
Total costs in primary and secondary healthcare sector Secondary healthcare sector
Primary healthcare sector
*
Total costs excl. cost of CT-screening programme Hospitalisation days Outpatient visits Surgical procedures Non-surgical procedures Emergency room contacts* General practitioner costs General practitioner procedures General practitioner contacts Other Specialist MD costs Psychologist costs Physiotherapist costs
Out-of-hours contacts to general practitioners are included in emergency room contacts; MD: Medical Doctor.
Diagnostic groups in screening group True-negative
False-positive
True-positive
0.96 0.98 0.89 1.00 0.95 1.05 0.96 0.97 0.95 0.96 0.78 0.78
1.66 2.05 1.49 1.51 1.63 1.03 1.08 1.10 1.07 0.98 1.56 1.87
10.61 9.99 6.27 4.07 6.03 0.76 1.21 1.23 1.26 0.52 3.14 0.41
352
J.F. Rasmussen et al. / Lung Cancer 83 (2014) 347–355
felt reassured that they were healthy because of the normal CT screening and therefore had less or no extra use of healthcare compared with the control group. Such a feeling of reassurance has been seen in mammography screening [23]. Hence, the higher healthcare costs in the true-positive and false-positive groups are outweighed by the tendency of lower costs in the larger true-negative group, and no differences in total costs between the screening group and the control group are found when costs of the CT programme are excluded. This study has limitations. Firstly, the indirect medical costs, such as days off work due to the CT screening and follow-up, were not included in this study. Secondly, some worried screening participants might have consulted private hospital services, which are not registered in the public registries. However, a previous study showed only a minor contamination of the control group in the DLCST; therefore, this bias is most likely small in the present study [24]. Thirdly, the cost of the low dose CT-screening scan in the DLCST (EUR 238/scan) was higher (as cost of entire CT-screening programme was included) than the cost of a low dose CT scan in a standard diagnostic setting (EUR 186). Therefore, a post hoc sensitivity analysis using the cost of EUR 186 was performed and this also showed a significantly higher total healthcare cost for the screening group. Finally, the three outcomes: psychologist- and physiotherapist costs and emergency room contacts showed a low prevalence and were therefore not adjusted for all covariates. However, they were included to represent healthcare costs not directly related to lung cancer and to show whether lung cancer CT-screening might have an effect on these costs. The cost of preventing one lung cancer death in CT screening in the NLST was estimated to be EUR 177 659 and the costs of one quality-adjusted life year gained in the NLST was estimated to be EUR 34 632 [12,25]. There are limitations in the methods used in these studies as only the costs directly related to the screening and follow-up procedures are included. If the healthcare costs in the primary sector had been included, costs would most likely have been higher, as our data shows that the increased costs in the true-positive and false-positive groups are not limited to the secondary healthcare sector. Furthermore, indirect costs should also be included in cost-effectiveness studies in order to gain the most comprehensive cost-estimate of LDCT screening. Across countries, the overall costs of a lung-cancer screening programme will be very sensitive to the rate of false positives. The average falsepositive rate per round in the DLCST (3%) was low compared with the NELSON trial (13%) and the NLST (23%) (Appendix Table 4) [3,9,13]. The marked difference between the false-positive rate in the NLST and the DLCST could be explained both by the differences in nodule criteria for a positive CT scan result and by the management algorithm used for suspicious nodules [3,18]. The nodule-size cut-offs for follow-up were 4–10 mm in the NLST and 5–15 mm in the DLCST. In the DLCST, CT scans were reviewed in a single centre by two radiologists using restrictive management of suspicious nodules. In the NLST, multiple radiologists in 33 centres had identical criteria for a positive CT scan result, but the interpretation most likely varied. Despite the low false-positive rate in the DLCST, there was only one patient with an interval lung cancer diagnosed in 9800 CT scans, indicating that virtually no lung cancers were missed [9]. In comparison, 44 individuals in the NLST were diagnosed with interval lung cancers in 75 126 CT scans (5.7/9800) [3]. Thus, the cost-neutrality of the DLCST after excluding the cost of the CT-screening programme needs to be analysed in different contexts and in the remaining CT-screening trials with higher false-positive rates and other healthcare systems. In addition to the clinical relevance of maintaining a low false-positive rate, we have also presented evidence for the economical relevance.
5. Conclusion In this study low dose lung cancer CT screening increases national healthcare costs compared with no screening; this increase is attributable to the costs of the CT-screening programme. Overall healthcare costs were higher for the true-positive and falsepositive groups than for the control group, also when excluding the cost of the CT-screening programme. This increase was outweighed by the large true-negative group showing no significant differences in costs compared with the control group. Conflicts of interest statement All authors have completed the ICMJE uniform disclosure form at www.icmje.org/coi disclosure.pdf and declare: no support from any organisation for the submitted work; JHP has received a grant funding the DLCST during 2004–2010 from the Danish Ministry of Health and Prevention and JFR received funding from this grant during the work with this present project; JHP participated in a single meeting in an advisory board for Roche Diagnostics on global issues of lung cancer screening; no other relationships or activities that could appear to have influenced the submitted work. Contributors The study was devised by JFR, JHP and JB. BH, VS and ZS provided comments. Data collection and data analysis was done by JFR, VS, JHP, ZS and JB. JFR drafted the manuscript and JHP, JB, VS, BH and ZS contributed to revisions of the manuscript and approved the final version to be published. All authors, external and internal, had full access to all of the data (including statistical reports and Tables) in the study and can take responsibility for the integrity of the data and the accuracy of the data analysis. The authors did not receive any writing assistance. Funding The Danish Lung Cancer Screening Trial and JFR were funded by the Danish Ministry of Health and Prevention (grant number 0900814). The funder had no role in study design, data collection, analysis, interpretation, writing process or in the decision to submit the article for publication. Acknowledgements We thank data manager Willy Karlslund for the comprehensive technical work generating the databases. Appendix A. Appendix Fig. A1. Tables A1–A5. Reference List (1) Benjamini Y, Hochberg Y. Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. Journal of the Royal Statistical Society Series B (Methodological) 1995; 57(1):289–300. (2) van Klaveren RJ, Oudkerk M, Prokop M, Scholten ET, Nackaerts K, Vernhout R et al. Management of lung nodules detected by volume CT scanning. N Engl J Med 2009; 361(23):2221–2229. (3) Aberle DR, Adams AM, Berg CD, Black WC, Clapp JD, Fagerstrom RM et al. Reduced lung-cancer mortality with lowdose computed tomographic screening. N Engl J Med 2011; 365(5):395–409.
J.F. Rasmussen et al. / Lung Cancer 83 (2014) 347–355
353
Table A1 Relative risk (prevalence) and factor (quantity) of healthcare use in the secondary sector in the diagnostic groups compared with control group. Outcomes
Groups
Relative risk (RR)
RR lower 95% CI
RR upper 95% CI
RR p-Value
Factor (F)
F lower 95% CI
F upper 95% CI
Factor p-Value
Cumulative effect†
Hospitalisation (days)
True-negative False-positive True-positive Control True-negative False-positive True-positive Control
0.95 1.46 6.21 1.00 0.95 1.20 1.66 1.00
0.86 1.13 5.34 1.00 0.91 1.10 1.54 1.00
1.05 1.89 7.23 1.00 0.98 1.31 1.79 1.00
0.34 0.004* <0.001* – 0.003* <0.001* <0.001* –
1.03 1.41 1.61 1.00 0.94 1.24 3.78 1.00
0.87 0.87 1.19 1.00 0.87 1.01 2.97 1.00
1.21 2.28 2.16 1.00 1.02 1.50 4.81 1.00
0.77 0.170 0.002* – 0.150 0.036 <0.001* –
0.98 2.05 9.99 1.00 0.89 1.49 6.27 1.00
Emergency room contacts
True-negative False-positive True-positive Control
0.99 0.97 1.16 1.00
0.93 0.79 0.79 1.00
1.06 1.18 1.70 1.00
0.81 0.74 0.44 –
1.06 1.06 0.65 1.00
0.90 0.77 0.45 1.00
1.24 1.47 0.94 1.00
0.48 0.71 0.021 –
1.05 1.03 0.76 1.00
Surgical procedures
True-negative False-positive True-positive Control
0.96 1.37 3.78 1.00
0.89 1.13 3.30 1.00
1.04 1.68 4.33 1.00
0.34 0.001* <0.001* –
1.04 1.10 1.08 1.00
0.96 0.89 0.85 1.00
1.13 1.36 1.37 1.00
0.29 0.39 0.55 –
1.00 1.51 4.07 1.00
Non-surgical procedures
True-negative False-positive True-positive Control
0.95 1.26 1.66 1.00
0.92 1.16 1.53 1.00
0.99 1.36 1.79 1.00
0.006* <0.001* <0.001* –
1.00 1.29 3.64 1.00
0.91 1.04 3.13 1.00
1.09 1.62 4.24 1.00
0.97 0.023 <0.001* –
0.95 1.63 6.03 1.00
Outpatient visits
CI: confidence interval. * After adjustment for multiple testing by using the method of Benjamini–Hochberg the level of statistical significance was assessed at a level of 0.013. † Cumulative effect = Relative risk × factor.
Table A2 Relative risk (prevalence) and factor (quantity) of healthcare use in the primary sector in the diagnostic groups compared with control group. Outcomes
Groups
Relative risk (RR)
RR lower 95% CI
RR upper 95% CI
RR p-Value
F upper 95% CI
Factor p-Value
General practitioner contacts
True-negative False-positive True-positive Control
0.99 0.99 1.05 1.00
0.98 0.96 1.01 1.00
1.00 1.02 1.08 1.00
0.150 0.68 0.008* –
0.96 1.07 1.20 1.00
0.92 0.99 1.05 1.00
1.01 1.17 1.38 1.00
0.093 0.092 0.009* –
0.95 1.07 1.26 1.00
General practitioner procedures
True-negative False-positive True-positive Control
0.99 1.00 1.04 1.00
0.98 0.97 1.01 1.00
1.00 1.03 1.08 1.00
0.20 0.75 0.016 –
0.97 1.10 1.18 1.00
0.92 1.01 1.03 1.00
1.02 1.20 1.35 1.00
0.27 0.023 0.016 –
0.97 1.10 1.23 1.00
General practitioner costs
True-negative False-positive True-positive Control
0.99 1.00 1.04 1.00
0.98 0.97 1.01 1.00
1.00 1.03 1.08 1.00
0.21 0.75 0.016 –
0.97 1.09 1.16 1.00
0.93 0.99 1.02 1.00
1.01 1.19 1.32 1.00
0.180 0.091 0.028 –
0.96 1.08 1.21 1.00
Other Specialist MD costs
True-negative False-positive True-positive Control
0.97 0.93 0.79 1.00
0.93 0.84 0.61 1.00
1.00 1.02 1.03 1.00
0.056 0.130 0.083 –
0.99 1.05 0.65 1.00
0.94 0.89 0.41 1.00
1.05 1.25 1.04 1.00
0.82 0.53 0.070 –
0.96 0.98 0.52 1.00
Psychologist costs
True-negative False-positive True-positive Control
0.70 0.82 3.40 1.00
0.50 0.28 0.97 1.00
0.98 2.39 11.83 1.00
0.035 0.72 0.055 –
1.12 1.90 0.92 1.00
0.93 1.24 0.49 1.00
1.35 2.92 1.76 1.00
0.24 0.003* 0.81 –
0.78 1.56 3.14 1.00
Physiotherapist costs
True-negative False-positive True-positive Control
0.98 1.20 0.83 1.00
0.85 0.90 0.47 1.00
1.12 1.59 1.45 1.00
0.77 0.21 0.50 –
0.80 1.56 0.49 1.00
0.57 0.94 0.32 1.00
1.12 2.58 0.76 1.00
0.190 0.084 0.001* –
0.78 1.87 0.41 1.00
Total Costs excl. cost of CT-screening programme
True-negative False-positive True-positive Control
1.00 1.00 1.00 1.00
0.99 0.99 1.00 1.00
1.00 1.01 1.01 1.00
0.22 0.47 0.30 –
0.96 1.67 10.57 1.00
0.86 1.20 7.09 1.00
1.09 2.32 15.75 1.00
0.55 0.002* <0.001* –
0.96 1.66 10.61 1.00
Factor (F)
F lower 95% CI
Cumulative effect†
Total costs exclusive cost of CT screening programme includes costs in both primary and secondary healthcare sector; CI: confidence interval; MD: Medical Doctor. * After adjustment for multiple testing by using the method of Benjamini–Hochberg the level of statistical significance was assessed at a level of 0.013. † Cumulative effect = Relative risk × factor.
354
J.F. Rasmussen et al. / Lung Cancer 83 (2014) 347–355
Table A3 Overview of outcomes and registries used for data extraction. Outcomes
Registries NPR
Total costs Hospitalisation (days) Outpatient visits Emergency room contacts
x x x
Surgical procedures Non-surgical procedures General practitioner contacts
x x
NHI
DRG-DAGS
Description of outcomes
x
x
Total healthcare costs in primary and secondary sector excluding costs in the psychiatric secondary sector Days in hospital, not as outpatient Number of visits in outpatient clinic Number of out-of-hours contacts to general practitioners and all contacts to emergency rooms Number of surgical procedures in hospital Number of non-surgical techniques and procedures in hospital Number of contacts to general practitioners including ordinary consultation, home visit, telephone- and email consultation Number of surgical and non-surgical procedures at general practitioners Healthcare costs of general practitioners Healthcare costs of specialised medical doctors (excl. general practitioners) in the primary sector Healthcare costs of psychologists in the primary sector Healthcare costs of physiotherapists and chiropractors in the primary sector
x
x
General practitioner procedures
x
General practitioner costs Other specialist MD costs
x x
Psychologist costs Physiotherapy costs
x x
NPR: National Patient Registry; NHI: National Health Insurance Registry; DRG-DAGS: Diagnostic Related Groups-Diagnostic Outpatient Group System Registries; MD: Medical Doctor. Data on costs and utilisation of the primary healthcare sector were extracted from the National Health Insurance Service Registry. Data on utilisation of the secondary healthcare sector were extracted from the National Patient Registry. The costs of all somatic services of the secondary healthcare sector during 2006–2010 were retrieved from the Diagnostic Related Groups and Danish Outpatient Group System Registries. From the National Patient Registry and the National Health Insurance Service Registry, the data were extracted from the date of randomisation (2004–2006) until 15 September 2011. From the Diagnostic Related Groups and Danish Outpatient Group System registries, data were extracted and available only during 1 January 2006–31 December 2010. The total healthcare costs included psychiatric costs from only the primary and not the secondary sector. In the analyses the outcome Total costs was divided into the outcomes Total costs including costs of the CT-screening programme and Total costs excluding costs of the CT-screening programme and therefore only 12 outcomes are presented above.
Table A4 False-positive rates in the Dutch–Belgian lung cancer screening trial (NELSON), the National Lung Screening Trial (NLST) and the Danish Lung Cancer Screening Trial (DLCST). Trial
Number screened
NELSON2
7557 7289
Round of screening
Lung nodules over study threshold
Baseline Year 1
1570 570
Lung cancer nodules 70 54
Nodules not lung cancer
False-positive rate (nodules not lung cancer/no. screened)
1500 516
0.1985 0.0708
Average false-positive rate
0.1346 NLST
3
26 309 24 715 24 102
Baseline Year 1 Year 2
7191 6901 4054
270 168 211
6921 6733 3843
0.2631 0.2724 0.1594
2047 1976 1944 1982 1851
Baseline Year 1 Year 2 Year 3 Year 4
179 45 52 44 51
17 11 13 12 16
162 34 39 32 35
0.0791 0.0172 0.0201 0.0161 0.0189
0.2316 DLCST
4
0.0303
Table A5 Diagnostic results and participation during five screening rounds in the DLCST. Group
Round 0
1
2
3
4
Screening group
True negative False positive True positive False negative Missing
1868 162 17 0 5
1931 34 11 0 71
1892 39 13 0 32
1937 32 12 1 0
1800 35 16 0 131
Control group
Control Missing
2052 0
1953 99
1877 76
1838 39
1820 18
J.F. Rasmussen et al. / Lung Cancer 83 (2014) 347–355
Fig. A1. Mean annual healthcare costs per participant in the control group and the diagnostic groups in the DLCST during five screening rounds.
(4) Saghir Z, Dirksen A, Ashraf H, Bach KS, Brodersen J, Clementsen PF et al. CT screening for lung cancer brings forward early disease. The randomised Danish Lung Cancer Screening Trial: status after five annual screening rounds with low-dose CT. Thorax 2012; 67(4):296–301. References [1] Ott JJ, Ullrich A, Mascarenhas M, Stevens GA. Global cancer incidence and mortality caused by behavior and infection. J Public Health (Oxf) 2011;33(June (2)):223–33. [2] Siegel R, Naishadham D, Jemal A. Cancer statistics, 2012. CA Cancer J Clin 2012;62(January (1)):10–29. [3] Aberle DR, Adams AM, Berg CD, Black WC, Clapp JD, Fagerstrom RM, et al. Reduced lung-cancer mortality with low-dose computed tomographic screening. N Engl J Med 2011;365(August (5)):395–409. [4] Becker N, Motsch E, Gross ML, Eigentopf A, Heussel CP, Dienemann H, et al. Randomized study on early detection of lung cancer with MSCT in Germany: study design and results of the first screening round. J Cancer Res Clin Oncol 2012;138(September (9)):1475–86. [5] Field JK, van KR, Pedersen JH, Pastorino U, Paci E, Becker N, et al. European randomized lung cancer screening trials: post NLST. J Surg Oncol 2013;(July), http://dx.doi.org/10.1002/jso.23383.
355
[6] Infante M, Cavuto S, Lutman FR, Brambilla G, Chiesa G, Ceresoli G, et al. A randomized study of lung cancer screening with spiral computed tomography: three-year results from the DANTE trial. Am J Respir Crit Care Med 2009;180(September (5)):445–53. [7] Lopes PA, Picozzi G, Mascalchi M, Maria CF, Carrozzi L, Comin C, et al. Design, recruitment and baseline results of the ITALUNG trial for lung cancer screening with low-dose CT. Lung Cancer 2009;64(April (1)):34–40. [8] Pastorino U, Rossi M, Rosato V, Marchiano A, Sverzellati N, Morosi C, et al. Annual or biennial CT screening versus observation in heavy smokers: 5-year results of the MILD trial. Eur J Cancer Prev 2012;21(May (3)):308–15. [9] Saghir Z, Dirksen A, Ashraf H, Bach KS, Brodersen J, Clementsen PF, et al. CT screening for lung cancer brings forward early disease. The randomised Danish Lung Cancer Screening Trial: status after five annual screening rounds with low-dose CT. Thorax 2012;67(April (4)):296–301. [10] van Iersel CA, de Koning HJ, Draisma G, Mali WP, Scholten ET, Nackaerts K, et al. Risk-based selection from the general population in a screening trial: selection criteria, recruitment and power for the Dutch-Belgian randomised lung cancer multi-slice CT screening trial (NELSON). Int J Cancer 2007;120(February (4)):868–74. [11] Manser R, Lethaby A, Irving LB, Stone C, Byrnes G, Abramson MJ, et al. Screening for lung cancer. Cochrane Database Syst Rev 2013;6, http://dx.doi.org/10.1002/14651858.CD001991.pub3. [12] Goulart BH, Bensink ME, Mummy DG, Ramsey SD. Lung cancer screening with low-dose computed tomography: costs, national expenditures, and costeffectiveness. J Natl Compr Canc Netw 2012;10(February (2)):267–75. [13] van Klaveren RJ, Oudkerk M, Prokop M, Scholten ET, Nackaerts K, Vernhout R, et al. Management of lung nodules detected by volume CT scanning. N Engl J Med 2009;361(December (23)):2221–9. [14] Barton MB, Moore S, Polk S, Shtatland E, Elmore JG, Fletcher SW. Increased patient concern after false-positive mammograms: clinician documentation and subsequent ambulatory visits. J Gen Intern Med 2001;16(March (3)):150–6. [15] Lafata JE, Simpkins J, Lamerato L, Poisson L, Divine G, Johnson CC. The economic impact of false-positive cancer screens. Cancer Epidemiol Biomarkers Prev 2004;13(December (12)):2126–32. [16] Lidbrink E, Elfving J, Frisell J, Jonsson E. Neglected aspects of false positive findings of mammography in breast cancer screening: analysis of false positive cases from the Stockholm trial. BMJ 1996;312(February (7026)):273–6. [17] Brodersen J, Siersma V. Long-term psychosocial consequences of false-positive screening mammography. Ann Fam Med 2013;11(2):106–15. [18] Pedersen JH, Ashraf H, Dirksen A, Bach K, Hansen H, Toennesen P, et al. The Danish randomized lung cancer CT screening trial – overall design and results of the prevalence round. J Thorac Oncol 2009;4(May (5)):608–14. [19] Brodersen J, Thorsen H, Kreiner S. Consequences of screening in lung cancer: development and dimensionality of a questionnaire. Value Health 2010;13(August (5)):601–12. [20] Diehr P, Yanez D, Ash A, Hornbrook M, Lin DY. Methods for analyzing health care utilization and costs. Annu Rev Public Health 1999;20:125–44. [21] Zou G. A modified Poisson regression approach to prospective studies with binary data. Am J Epidemiol 2004;159(April (7)):702–6. [22] Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc B 1995;57(January (1)):289–300. [23] Brodersen J, Siersma V, Ryle M. Breast cancer screening: “reassuring” the worried well? Scand J Public Health 2011;39(May (3)):326–32. [24] Saghir Z, Ashraf H, Dirksen A, Brodersen J, Pedersen JH. Contamination during 4 years of annual CT screening in the Danish Lung Cancer Screening Trial (DLCST). Lung Cancer 2011;71(March (3)):323–7. [25] Villanti AC, Jiang Y, Abrams DB, Pyenson BS. A cost-utility analysis of lung cancer screening and the additional benefits of incorporating smoking cessation interventions. PLoS ONE 2013;8(8):e71379.
Contents lists available at ScienceDirect
Lung Cancer journal homepage: www.elsevier.com/locate/lungcan
Healthcare costs in the Danish randomised controlled lung cancer CT-screening trial: A registry study夽 Jakob F. Rasmussen a,∗ , Volkert Siersma a , Jesper H. Pedersen b , Bruno Heleno a , Zaigham Saghir b , John Brodersen a a The Research Unit for General Practice and Section of General Practice, Department of Public Health, University of Copenhagen, Oester Farimagsgade 5, 1014 Copenhagen K, Denmark b Department of Cardiothoracic Surgery, RT 2152, Rigshospitalet, University of Copenhagen, Blegdamsvej 9, 2100 Copenhagen, Denmark
a r t i c l e
i n f o
Article history: Received 1 October 2013 Received in revised form 12 December 2013 Accepted 16 December 2013 Keywords: Lung cancer Mass screening Cancer screening test Scan spiral CT Healthcare cost Cancer early detection
a b s t r a c t Objectives: Low dose computerised tomography (CT) screening for lung cancer can reduce lung-cancerspecific mortality. The objective of this study was to analyse healthcare costs and healthcare utilisation of participants in the Danish lung cancer CT-screening trial (DLCST). Materials and methods: This registry study was nested in a randomised controlled trial (DLCST). 4104 participants, current or former heavy smokers, aged 50–70 years were randomised to five annual low dose CT scans or usual care during 2004–2010. Total healthcare costs and healthcare utilisation data for both the primary and the secondary healthcare sector were retrieved from public registries from randomisation – September 2011 and compared between (1) the CT-screening group and the control group and, (2) the control group and each of the true-positive, false-positive and true-negative groups. Results: The median annual costs per participant were significantly higher in the CT-screening group (Euros [EUR] 1342, interquartile range [IQR] 750–2980) compared with the control group (EUR 1190, IQR 590–2692) (p < 0.0001). When the cost of the CT-screening programme was excluded, there was no longer a statistically significant difference between the CT-screening group (EUR 1155, IQR 567–2798) and the control group (p = 0.52). Analyses according to the diagnostic groups showed that annual costs were 10.57 (95% CI 7.09–15.75) times higher for the true-positive and 1.67 (95% CI 1.20–2.32) times higher for the false-positive group compared with the control group. Conclusion: Low dose lung cancer CT screening increases healthcare costs compared with no screening; this difference was attributable to the costs of the CT-screening programme. Overall healthcare costs were higher for the true-positive and false-positive groups than for the control group, also when excluding the cost of the CT-screening programme. This increase was outweighed by the larger true-negative group showing no significant differences in costs compared with the control group. © 2013 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Lung cancer is the primary cause of cancer-related deaths in the world [1,2]. The National Lung Screening Trial (NLST) has shown a relative risk reduction in lung-cancer-specific mortality of 20% and 6.7% in all-cause mortality using low dose computerised tomography (CT) screening [3]. The Danish Lung Cancer Screening Trial (DLCST) and five other European randomised controlled lung cancer CT-screening trials are currently being evaluated [4–10]. Data pooling of the European trials is required to show whether the lung-cancer-specific mortality reduction found in NLST is also
夽 ClinicalTrials.gov: NCT00496977. ∗ Corresponding author. Tel.: +45 3532 7171; fax: +45 3532 7131. E-mail address: [email protected] (J.F. Rasmussen). 0169-5002/$ – see front matter © 2013 Elsevier Ireland Ltd. All rights reserved. http://dx.doi.org/10.1016/j.lungcan.2013.12.005
a consistent result in the European trials. The final data analyses on mortality and overdiagnosis in the European trials are planned in 2015 and therefore cost-effectiveness analyses cannot be performed at present [5]. A recently published Cochrane review concluded that even though a lung-cancer-specific mortality reduction was found in the NLST more data are needed on cost effectiveness, overdiagnosis and false-positive results before recommendations of lung cancer CT-screening programmes should be made [11]. It is currently unclear whether lung cancer CT screening is cost effective [12]. The false positive results will contribute to the total costs of CT screening and the false-positive rate differs across the CT-screening trials (the NLST = 23%, the Dutch-Belgian Lung Cancer Screening Trial (NELSON) = 13% and the DLCST = 3%) [3,9,13]. A study analysing the healthcare costs in the NLST found that: “The high false-positive rate of LDCT screening, as reported in the
348
J.F. Rasmussen et al. / Lung Cancer 83 (2014) 347–355
NLST, will clearly contribute to these expenditures, because a falsepositive screening result leads to unnecessary follow-up tests and procedures, and their respective costs” [12]. However, the study included only the lung-related costs of the CT screening and followup procedures [12]. In mammography screening, it has been shown that both breast-related and non-breast-related outpatient visits increased significantly for those women receiving a false-positive result [14]. Moreover, the diagnostic follow-up of false-positive screening mammography results leads to more concern and more healthcare utilisation in general [14–17]. Thus, it is important to know whether participants in lung cancer CT screening seek more medical attention not only related to lung cancer screening. Consequently, the aim of this study was to analyse all healthcare costs in the DLCST population, in both the primary and the secondary healthcare sector, during the screening period. The costs were compared between (1) the screening group and the control group and, (2) the control group and each of the true-positive, falsepositive and true-negative groups. 2. Methods 2.1. Study population The DLCST ran during 2004–2010; the study design is described in detail elsewhere [9,18]. Briefly, 4104 current and former heavy smokers, aged 50–70 years, were randomised to CT screening or no screening with 2052 participants in each group (Fig. 1). The participants in the screening group were offered five annual low dose CT scans. Both groups made annual visits to a screening clinic and received lung function tests, smoking counselling, and completed questionnaires [18,19]. 2.2. DLCST budget The DLCST received a grant of EUR 2.33 million from the Ministry of Health and Prevention. This grant covered the expenses of the CT-screening programme (EUR 238/CT scan) including recruitment of participants, the 9800 CT scans and salary to the staff at the screening clinic during the five screening rounds. The CT-screening scans were not billed to the Danish healthcare system and thus not registered in the public registries. The costs of additional CT scans and all procedures in the follow-up of those with abnormal findings were not included in the grant and were thus billed to the Danish healthcare system and recorded in the public registries. 2.3. Registries and outcomes Denmark has a publicly financed healthcare system and all healthcare costs and services are recorded in public registries. Residents in Denmark have a unique personal identification number and every event in the public healthcare system is linked to this number. Thirteen outcomes related to specific healthcare costs or use, were analysed and categorised as the following: A. Total costs in primary and secondary healthcare sector (2 outcomes) 1. Total costs including costs of CT-screening programme (EUR 238/CT-scan) 2. Total costs excluding costs of CT-screening programme B. Primary healthcare sector (6 outcomes) 1. Costs of general practitioners 2. Procedures at general practitioners 3. Contacts to general practitioners.
Fig. 1. Design and diagnostic results of the Danish Lung Cancer Screening Trial.
4. Costs of other specialised medical doctors 5. Costs of psychologists 6. Costs of physiotherapists C. Secondary healthcare sector (5 outcomes) 1. Hospitalisation (days) 2. Outpatient visits 3. Emergency room contacts (including out-of-hours contacts in primary sector) 4. Surgical procedures 5. Non-surgical procedures Appendix Table 3 presents an overview of registries and data extraction. 2.4. Data sets and observation time 1. Screening group and control group: data on the 13 outcomes were cumulated over a total observation period from date of randomisation (2004–2006) to either September 2011 (66–84 months), death or migration and data were compared between screening group and control group. These outcomes were annualised into outcome per year to adjust for different observation times. The participants with an observation time < 12 months of the total observation time of 66–84 months were excluded
J.F. Rasmussen et al. / Lung Cancer 83 (2014) 347–355
from the analyses to avoid inflation of the healthcare costs when annualising the outcomes. 2. Diagnostic groups and control group: in the CT-screening group there were four possible diagnostic groups: (a) true-negative, (b) false-positive, (c) true-positive and (d) false-negative. As there was only one participant with a false-negative result, this person was excluded from the present study [9]. The healthcare costs of each of the three remaining diagnostic groups were compared with the healthcare costs of the control group. The participants in the screening group could receive different diagnoses during the five screening rounds. Therefore, the period of comparison for the screening group and the control group was defined as: the period from the date of CT screening (screening group) or visit to the clinic (control group) to the next date of CT screening or visit to the clinic. If a participant missed a round, the period following the previous round was defined as 365 days from the previous CT screening or visit to the clinic; the period following the missing round was set to missing. The fifth observation period was defined as 365 days from the date of the last CTscreening round or last visit to the clinic. The mean observation time for the first four periods was 377 days. The outcomes in each period were annualised to outcome per year to adjust for different observation times.
2.5. Statistical analysis Screening group versus control group: The differences in baseline characteristics between the DLCST randomisation groups were tested by chi-squared tests for categorical variables or t-tests for continuous variables. The differences in healthcare costs between participants in the screening group and the control group were analysed with Wilcoxon signed-rank test. Diagnostic groups versus control group: Since the diagnostic groups were not randomised, it was necessary to use a multivariable analysis. Some outcomes rarely used by the population include many zero observations (e.g. surgical procedures). Consequently, it is more appropriate to analyse the risk of having, for example, a surgical procedure at all in the observation period and thus analyse the prevalence of use. Other outcomes frequently used do not have many zero observations (e.g. total costs). In this aspect it is more relevant to analyse the quantity of use. Therefore, a multivariable analysis of the outcomes followed a two-part model where the three diagnostic groups in the period following a specific screening round were compared with the control group:
1. In part one, the prevalence of use of selected healthcare services or costs >0 in the period was analysed [20]. This prevalence was analysed in a Poisson regression approach so the regression parameters were equivalent to the logarithm of the relative risk of ever using the service in the observation period [21]. This prevalence part of the model was the most relevant for the healthcare services rarely used by the population (services in the secondary sector). 2. In part two, the quantity of healthcare costs or services was analysed. Only the participants who used the services or had costs >0 were included, and the quantity of use was analysed in a generalised linear model using a Gamma distribution and a logarithmic link function. The parameters from this model were interpreted as the logarithm of a multiplicative factor of how much more the service was used (or how much more cost) in one of the diagnostic groups compared with the use/costs in the control group. This quantity part of the model was the most relevant for more frequently used healthcare services mainly in the primary sector and healthcare costs.
349
The analyses of both the prevalence part and the quantity part were multivariably adjusted for the following variables: sex, age, age-squared, screening-round number, days of observation, employment, living alone, socioeconomic group, smoking status (former/current smoker), pack years, pack years-squared and region of residence. Because the estimation did not converge for the full adjustment, possibly because the outcome prevalences were low, the outcome of psychologist costs was not adjusted for age-squared, employment, living alone and days of observation; the outcome of physiotherapist costs was not adjusted for agesquared, pack years-squared, region of residence and employment; and the outcome of emergency room contacts was not adjusted for days of observation. The outcome “Total costs including costs of CT-screening programme” was excluded, as the main focus in the multivariable analyses was to analyse the costs in the year directly following the CT-screening result. In both part one and two the repeated observation (up to five rounds) of the same participants was adjusted for using Generalised Estimating Equations (GEE). A combined multiplicative effect of being in one of the diagnostic groups compared with being in the control group was calculated by multiplying the relative risk from the prevalence part and the factor from the quantity part of the model. This multiplicative effect is referred to as the cumulative effect. Statistical significance was assessed at a 5% level when the screening group and the control group were compared. In the multivariable analyses p values were calculated individually for the prevalence part and for the quantity part of the model and we adjusted for multiple testing using the method of Benjamini–Hochberg [22]. Statistical Analysis Software 9.2 (SAS Institute, Cary, North Carolina) was used to analyse the data.
2.6. Ethics The DLCST has been approved by the Danish Scientific Ethical Committee: number of approval KA-02045. All participants signed an informed consent form. Both the DLCST (approval number 200553-1083) and this project (approval number 2011-41-5852) have been approved by the Danish Data Protection Agency. ClinicalTrials.gov: NCT00496977.
3. Results 3.1. Participation Overall participation in the DLCST was 95.5% in the screening group and 93.0% in the control group (Fig. 1). No significant differences in baseline socioeconomic characteristics or smoking habits were found between the two groups (Table 1). During the screening trial, 148 (3.6%) participants died and 27 (0.6%) emigrated; 31 participants had a total observation time <12 months and were excluded. The data on all 4104 participants in the DLCST were retrieved from the registries and none was lost to follow-up.
3.2. Screening group versus control group The annual total costs differed statistically significantly between the screening group and the control group when the cost of the screening programme was included. When this cost was excluded, the differences in total costs were not statistically significant. There were no statistically significant differences in the other outcomes (Table 2). In the screening group the rate of the positive tests was lower in the incidence rounds compared to the prevalence round (Fig. 1).
350
J.F. Rasmussen et al. / Lung Cancer 83 (2014) 347–355
Table 1 Baseline Characteristics in DLCST. Characteristics Sex, n (%) Age, mean (SD) Social group, n (%)
Men Women I II III IV V Employed, social group uncertain Outside the labour market
n/n
Screening group n = 2052
Control group n = 2052
2052/2052
1147 (55.9) 905 (44.1)
1120 (54.6) 932 (45.4)
2052/2052 2041/2039
57.3 (4.8) 155 (7.6) 402 (19.7) 378 (18.5) 545 (26.7) 265 (13.0) 182 (8.9) 114 (5.6)
57.3 (4.8) 141 (6.9) 410 (20.1) 378 (18.5) 551 (27.0) 282 (13.8) 168 (8.2) 109 (5.4)
School education, n (%)
7–9 years in school 10 years in school 12–13 years in school Other
2047/2047
698 (34.1) 775 (37.9) 553 (27.0) 21 (1.0)
715 (34.9) 790 (38.6) 532 (26.0) 10 (0.5)
Employment status, n (%)
Employed Studying Job seeking Retired
2043/2045
1366 (66.9) 9 (0.4) 113 (5.5) 555 (27.2)
1324 (64.7) 12 (0.6) 104 (5.1) 605 (29.6)
Region of residence, n (%)
Capital region Region Zealand Region of Southern Denmark Region of Central Denmark Region of Northern Denmark
2037/2044
1644 (80.7) 353 (17.3) 29 (1.4) 8 (0.4) 3 (0.2)
1653 (80.9) 349 (17.1) 28 (1.4) 11 (0.5) 3 (0.2)
Living alone, n (%)
No Yes
2039/2034
1457(71.5) 582(28.5)
1453(71.4) 581(28.6)
Smoking status, n (%)
Current smoker Former smoker
2050/2051
1544 (75.3) 506 (24.7)
1579 (77.0) 472 (23.0)
Smoking history, mean (SD)
Pack years
2051/2048
36.4 (13.4)
35.9 (13.4)
SD: standard deviation.
3.3. Diagnostic groups versus control group The mean total annual healthcare costs were statistically significantly higher for the true-positives (EUR 28 254) and false-positives (EUR 4464) groups compared with the control group (EUR 2673) with a factor of 10.57 (95% CI 7.09–15.75) and 1.67 (95% CI 1.20–2.32) respectively. The true-negative group had lower total annual healthcare costs with a factor of 0.96 (95% CI 0.86–1.09)
but did not show any statistically significant difference (Fig. 2 and Appendix Table 2). In the six primary healthcare outcomes the true-positive group had a statistically significant higher healthcare use in one outcome and no significant differences of the higher healthcare use in the outcomes of General practitioner procedures, General practitioner costs and Psychologist costs were found. The two remaining outcomes both revealed lower primary healthcare use: one showed a
Table 2 Comparison of annual healthcare costs and utilisation per participant between the screening group and the control group in DLCST. Outcome
Screening group Median
Total costs incl. cost of CT-screening programme Total costs excl. cost of CT-screening programme Hospitalisation days Outpatient visits Surgical procedures Non-surgical procedures Emergency room contactsa General practitioner costs General practitioner procedures General practitioner contacts Other specialist MD costs Psychologist costs Physiotherapist costs
Control group
Range
IQR
Median
Min
Max
Q1
Q3
1342
110
206508
750
2980
1155
22
206359
567
0 1 0 3 1 246 26 19 107 0 0
0 0 0 0 0 12 1 1 0 0 0
83 59 16 167 49 3253 221 165 1281 299 3322
0 0 0 1 0 153 15 11 37 0 0
Range
p-Value*
IQR
Min
Max
Q1
Q3
1190
11
63314
590
2692
<.0001
2798
1190
11
63314
590
2692
0.5179
1 3 1 8 2 378 41 30 211 0 4
0 1 0 3 1 253 26 20 107 0 0
0 0 0 0 0 4 0 0 0 0 0
84 74 12 956 367 5235 795 678 1639 496 3155
0 1 0 1 0 162 16 12 43 0 0
1 3 1 8 2 387 41 31 220 0 3
0.1881 0.5860 0.5861 0.9290 0.6099 0.2171 0.2468 0.1847 0.1449 0.3541 0.6010
Costs are presented in Euros, EUR. (Converted from DKK using the 15 November 2013 spot rate DKK 745.88 = EUR 100). IQR: interquartile range; Q1: quartile 1; Q3: quartile 3. a Out-of-hours contacts to general practitioners are included in emergency room contacts. * Statistical significance was assessed at a 5% level.
J.F. Rasmussen et al. / Lung Cancer 83 (2014) 347–355
351
Fig. 2. Prevalence and quantity of healthcare costs and utilisation in the diagnostic groups compared with the control group.
significant difference and one did not show any significant difference. A statistically significant higher healthcare use was found in four of five outcomes in the secondary healthcare sector (Table 3, Fig. 2, Appendix Tables 1 and 2). The false-positive group had statistically significant higher healthcare use in one of the six primary healthcare outcomes and no significant differences of the higher healthcare use in the outcomes of General practitioner contacts, General practitioner procedures, General practitioner costs and Physiotherapist costs were found. A statistically significant higher healthcare use was found in four of the five outcomes in the secondary healthcare sector (Table 3, Fig. 2, Appendix Tables 1 and 2). The true-negative group had lower healthcare use in all outcomes in the primary healthcare sector but none showed a significant difference. In the secondary healthcare sector a statistically significant lower healthcare use was found in the outcomes Outpatient visits and Non-surgical procedures (Table 3, Fig. 2, Appendix Tables 1 and 2).
4. Discussion This is the first study to present the total healthcare costs, in both the primary and secondary healthcare sector, of the participants in a randomised lung cancer CT-screening trial. The total costs of the screening group were significantly higher than the costs of the control group when the costs of the CT-screening programme were included. However, when the costs of the CT-screening programme were excluded, this difference disappeared. Therefore, the excess costs in the screening group were attributable to the costs of the screening programme itself and not by a greater healthcare use. The higher healthcare costs for the true-positive and the false-positive groups, presented as the cumulative effect, were found in both the secondary and primary sector (Table 3). Those diagnosed with lung cancer and those receiving an abnormal CT scan later confirmed to be false-positive may have had greater needs and concerns; accordingly, they consulted their general practitioners and other primary healthcare professionals more. The true-negative group may have
Table 3 Cumulative effect of prevalence and quantity of healthcare costs and utilisation in the diagnostic groups relative to the control group in the DLCST. Outcome
Total costs in primary and secondary healthcare sector Secondary healthcare sector
Primary healthcare sector
*
Total costs excl. cost of CT-screening programme Hospitalisation days Outpatient visits Surgical procedures Non-surgical procedures Emergency room contacts* General practitioner costs General practitioner procedures General practitioner contacts Other Specialist MD costs Psychologist costs Physiotherapist costs
Out-of-hours contacts to general practitioners are included in emergency room contacts; MD: Medical Doctor.
Diagnostic groups in screening group True-negative
False-positive
True-positive
0.96 0.98 0.89 1.00 0.95 1.05 0.96 0.97 0.95 0.96 0.78 0.78
1.66 2.05 1.49 1.51 1.63 1.03 1.08 1.10 1.07 0.98 1.56 1.87
10.61 9.99 6.27 4.07 6.03 0.76 1.21 1.23 1.26 0.52 3.14 0.41
352
J.F. Rasmussen et al. / Lung Cancer 83 (2014) 347–355
felt reassured that they were healthy because of the normal CT screening and therefore had less or no extra use of healthcare compared with the control group. Such a feeling of reassurance has been seen in mammography screening [23]. Hence, the higher healthcare costs in the true-positive and false-positive groups are outweighed by the tendency of lower costs in the larger true-negative group, and no differences in total costs between the screening group and the control group are found when costs of the CT programme are excluded. This study has limitations. Firstly, the indirect medical costs, such as days off work due to the CT screening and follow-up, were not included in this study. Secondly, some worried screening participants might have consulted private hospital services, which are not registered in the public registries. However, a previous study showed only a minor contamination of the control group in the DLCST; therefore, this bias is most likely small in the present study [24]. Thirdly, the cost of the low dose CT-screening scan in the DLCST (EUR 238/scan) was higher (as cost of entire CT-screening programme was included) than the cost of a low dose CT scan in a standard diagnostic setting (EUR 186). Therefore, a post hoc sensitivity analysis using the cost of EUR 186 was performed and this also showed a significantly higher total healthcare cost for the screening group. Finally, the three outcomes: psychologist- and physiotherapist costs and emergency room contacts showed a low prevalence and were therefore not adjusted for all covariates. However, they were included to represent healthcare costs not directly related to lung cancer and to show whether lung cancer CT-screening might have an effect on these costs. The cost of preventing one lung cancer death in CT screening in the NLST was estimated to be EUR 177 659 and the costs of one quality-adjusted life year gained in the NLST was estimated to be EUR 34 632 [12,25]. There are limitations in the methods used in these studies as only the costs directly related to the screening and follow-up procedures are included. If the healthcare costs in the primary sector had been included, costs would most likely have been higher, as our data shows that the increased costs in the true-positive and false-positive groups are not limited to the secondary healthcare sector. Furthermore, indirect costs should also be included in cost-effectiveness studies in order to gain the most comprehensive cost-estimate of LDCT screening. Across countries, the overall costs of a lung-cancer screening programme will be very sensitive to the rate of false positives. The average falsepositive rate per round in the DLCST (3%) was low compared with the NELSON trial (13%) and the NLST (23%) (Appendix Table 4) [3,9,13]. The marked difference between the false-positive rate in the NLST and the DLCST could be explained both by the differences in nodule criteria for a positive CT scan result and by the management algorithm used for suspicious nodules [3,18]. The nodule-size cut-offs for follow-up were 4–10 mm in the NLST and 5–15 mm in the DLCST. In the DLCST, CT scans were reviewed in a single centre by two radiologists using restrictive management of suspicious nodules. In the NLST, multiple radiologists in 33 centres had identical criteria for a positive CT scan result, but the interpretation most likely varied. Despite the low false-positive rate in the DLCST, there was only one patient with an interval lung cancer diagnosed in 9800 CT scans, indicating that virtually no lung cancers were missed [9]. In comparison, 44 individuals in the NLST were diagnosed with interval lung cancers in 75 126 CT scans (5.7/9800) [3]. Thus, the cost-neutrality of the DLCST after excluding the cost of the CT-screening programme needs to be analysed in different contexts and in the remaining CT-screening trials with higher false-positive rates and other healthcare systems. In addition to the clinical relevance of maintaining a low false-positive rate, we have also presented evidence for the economical relevance.
5. Conclusion In this study low dose lung cancer CT screening increases national healthcare costs compared with no screening; this increase is attributable to the costs of the CT-screening programme. Overall healthcare costs were higher for the true-positive and falsepositive groups than for the control group, also when excluding the cost of the CT-screening programme. This increase was outweighed by the large true-negative group showing no significant differences in costs compared with the control group. Conflicts of interest statement All authors have completed the ICMJE uniform disclosure form at www.icmje.org/coi disclosure.pdf and declare: no support from any organisation for the submitted work; JHP has received a grant funding the DLCST during 2004–2010 from the Danish Ministry of Health and Prevention and JFR received funding from this grant during the work with this present project; JHP participated in a single meeting in an advisory board for Roche Diagnostics on global issues of lung cancer screening; no other relationships or activities that could appear to have influenced the submitted work. Contributors The study was devised by JFR, JHP and JB. BH, VS and ZS provided comments. Data collection and data analysis was done by JFR, VS, JHP, ZS and JB. JFR drafted the manuscript and JHP, JB, VS, BH and ZS contributed to revisions of the manuscript and approved the final version to be published. All authors, external and internal, had full access to all of the data (including statistical reports and Tables) in the study and can take responsibility for the integrity of the data and the accuracy of the data analysis. The authors did not receive any writing assistance. Funding The Danish Lung Cancer Screening Trial and JFR were funded by the Danish Ministry of Health and Prevention (grant number 0900814). The funder had no role in study design, data collection, analysis, interpretation, writing process or in the decision to submit the article for publication. Acknowledgements We thank data manager Willy Karlslund for the comprehensive technical work generating the databases. Appendix A. Appendix Fig. A1. Tables A1–A5. Reference List (1) Benjamini Y, Hochberg Y. Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. Journal of the Royal Statistical Society Series B (Methodological) 1995; 57(1):289–300. (2) van Klaveren RJ, Oudkerk M, Prokop M, Scholten ET, Nackaerts K, Vernhout R et al. Management of lung nodules detected by volume CT scanning. N Engl J Med 2009; 361(23):2221–2229. (3) Aberle DR, Adams AM, Berg CD, Black WC, Clapp JD, Fagerstrom RM et al. Reduced lung-cancer mortality with lowdose computed tomographic screening. N Engl J Med 2011; 365(5):395–409.
J.F. Rasmussen et al. / Lung Cancer 83 (2014) 347–355
353
Table A1 Relative risk (prevalence) and factor (quantity) of healthcare use in the secondary sector in the diagnostic groups compared with control group. Outcomes
Groups
Relative risk (RR)
RR lower 95% CI
RR upper 95% CI
RR p-Value
Factor (F)
F lower 95% CI
F upper 95% CI
Factor p-Value
Cumulative effect†
Hospitalisation (days)
True-negative False-positive True-positive Control True-negative False-positive True-positive Control
0.95 1.46 6.21 1.00 0.95 1.20 1.66 1.00
0.86 1.13 5.34 1.00 0.91 1.10 1.54 1.00
1.05 1.89 7.23 1.00 0.98 1.31 1.79 1.00
0.34 0.004* <0.001* – 0.003* <0.001* <0.001* –
1.03 1.41 1.61 1.00 0.94 1.24 3.78 1.00
0.87 0.87 1.19 1.00 0.87 1.01 2.97 1.00
1.21 2.28 2.16 1.00 1.02 1.50 4.81 1.00
0.77 0.170 0.002* – 0.150 0.036 <0.001* –
0.98 2.05 9.99 1.00 0.89 1.49 6.27 1.00
Emergency room contacts
True-negative False-positive True-positive Control
0.99 0.97 1.16 1.00
0.93 0.79 0.79 1.00
1.06 1.18 1.70 1.00
0.81 0.74 0.44 –
1.06 1.06 0.65 1.00
0.90 0.77 0.45 1.00
1.24 1.47 0.94 1.00
0.48 0.71 0.021 –
1.05 1.03 0.76 1.00
Surgical procedures
True-negative False-positive True-positive Control
0.96 1.37 3.78 1.00
0.89 1.13 3.30 1.00
1.04 1.68 4.33 1.00
0.34 0.001* <0.001* –
1.04 1.10 1.08 1.00
0.96 0.89 0.85 1.00
1.13 1.36 1.37 1.00
0.29 0.39 0.55 –
1.00 1.51 4.07 1.00
Non-surgical procedures
True-negative False-positive True-positive Control
0.95 1.26 1.66 1.00
0.92 1.16 1.53 1.00
0.99 1.36 1.79 1.00
0.006* <0.001* <0.001* –
1.00 1.29 3.64 1.00
0.91 1.04 3.13 1.00
1.09 1.62 4.24 1.00
0.97 0.023 <0.001* –
0.95 1.63 6.03 1.00
Outpatient visits
CI: confidence interval. * After adjustment for multiple testing by using the method of Benjamini–Hochberg the level of statistical significance was assessed at a level of 0.013. † Cumulative effect = Relative risk × factor.
Table A2 Relative risk (prevalence) and factor (quantity) of healthcare use in the primary sector in the diagnostic groups compared with control group. Outcomes
Groups
Relative risk (RR)
RR lower 95% CI
RR upper 95% CI
RR p-Value
F upper 95% CI
Factor p-Value
General practitioner contacts
True-negative False-positive True-positive Control
0.99 0.99 1.05 1.00
0.98 0.96 1.01 1.00
1.00 1.02 1.08 1.00
0.150 0.68 0.008* –
0.96 1.07 1.20 1.00
0.92 0.99 1.05 1.00
1.01 1.17 1.38 1.00
0.093 0.092 0.009* –
0.95 1.07 1.26 1.00
General practitioner procedures
True-negative False-positive True-positive Control
0.99 1.00 1.04 1.00
0.98 0.97 1.01 1.00
1.00 1.03 1.08 1.00
0.20 0.75 0.016 –
0.97 1.10 1.18 1.00
0.92 1.01 1.03 1.00
1.02 1.20 1.35 1.00
0.27 0.023 0.016 –
0.97 1.10 1.23 1.00
General practitioner costs
True-negative False-positive True-positive Control
0.99 1.00 1.04 1.00
0.98 0.97 1.01 1.00
1.00 1.03 1.08 1.00
0.21 0.75 0.016 –
0.97 1.09 1.16 1.00
0.93 0.99 1.02 1.00
1.01 1.19 1.32 1.00
0.180 0.091 0.028 –
0.96 1.08 1.21 1.00
Other Specialist MD costs
True-negative False-positive True-positive Control
0.97 0.93 0.79 1.00
0.93 0.84 0.61 1.00
1.00 1.02 1.03 1.00
0.056 0.130 0.083 –
0.99 1.05 0.65 1.00
0.94 0.89 0.41 1.00
1.05 1.25 1.04 1.00
0.82 0.53 0.070 –
0.96 0.98 0.52 1.00
Psychologist costs
True-negative False-positive True-positive Control
0.70 0.82 3.40 1.00
0.50 0.28 0.97 1.00
0.98 2.39 11.83 1.00
0.035 0.72 0.055 –
1.12 1.90 0.92 1.00
0.93 1.24 0.49 1.00
1.35 2.92 1.76 1.00
0.24 0.003* 0.81 –
0.78 1.56 3.14 1.00
Physiotherapist costs
True-negative False-positive True-positive Control
0.98 1.20 0.83 1.00
0.85 0.90 0.47 1.00
1.12 1.59 1.45 1.00
0.77 0.21 0.50 –
0.80 1.56 0.49 1.00
0.57 0.94 0.32 1.00
1.12 2.58 0.76 1.00
0.190 0.084 0.001* –
0.78 1.87 0.41 1.00
Total Costs excl. cost of CT-screening programme
True-negative False-positive True-positive Control
1.00 1.00 1.00 1.00
0.99 0.99 1.00 1.00
1.00 1.01 1.01 1.00
0.22 0.47 0.30 –
0.96 1.67 10.57 1.00
0.86 1.20 7.09 1.00
1.09 2.32 15.75 1.00
0.55 0.002* <0.001* –
0.96 1.66 10.61 1.00
Factor (F)
F lower 95% CI
Cumulative effect†
Total costs exclusive cost of CT screening programme includes costs in both primary and secondary healthcare sector; CI: confidence interval; MD: Medical Doctor. * After adjustment for multiple testing by using the method of Benjamini–Hochberg the level of statistical significance was assessed at a level of 0.013. † Cumulative effect = Relative risk × factor.
354
J.F. Rasmussen et al. / Lung Cancer 83 (2014) 347–355
Table A3 Overview of outcomes and registries used for data extraction. Outcomes
Registries NPR
Total costs Hospitalisation (days) Outpatient visits Emergency room contacts
x x x
Surgical procedures Non-surgical procedures General practitioner contacts
x x
NHI
DRG-DAGS
Description of outcomes
x
x
Total healthcare costs in primary and secondary sector excluding costs in the psychiatric secondary sector Days in hospital, not as outpatient Number of visits in outpatient clinic Number of out-of-hours contacts to general practitioners and all contacts to emergency rooms Number of surgical procedures in hospital Number of non-surgical techniques and procedures in hospital Number of contacts to general practitioners including ordinary consultation, home visit, telephone- and email consultation Number of surgical and non-surgical procedures at general practitioners Healthcare costs of general practitioners Healthcare costs of specialised medical doctors (excl. general practitioners) in the primary sector Healthcare costs of psychologists in the primary sector Healthcare costs of physiotherapists and chiropractors in the primary sector
x
x
General practitioner procedures
x
General practitioner costs Other specialist MD costs
x x
Psychologist costs Physiotherapy costs
x x
NPR: National Patient Registry; NHI: National Health Insurance Registry; DRG-DAGS: Diagnostic Related Groups-Diagnostic Outpatient Group System Registries; MD: Medical Doctor. Data on costs and utilisation of the primary healthcare sector were extracted from the National Health Insurance Service Registry. Data on utilisation of the secondary healthcare sector were extracted from the National Patient Registry. The costs of all somatic services of the secondary healthcare sector during 2006–2010 were retrieved from the Diagnostic Related Groups and Danish Outpatient Group System Registries. From the National Patient Registry and the National Health Insurance Service Registry, the data were extracted from the date of randomisation (2004–2006) until 15 September 2011. From the Diagnostic Related Groups and Danish Outpatient Group System registries, data were extracted and available only during 1 January 2006–31 December 2010. The total healthcare costs included psychiatric costs from only the primary and not the secondary sector. In the analyses the outcome Total costs was divided into the outcomes Total costs including costs of the CT-screening programme and Total costs excluding costs of the CT-screening programme and therefore only 12 outcomes are presented above.
Table A4 False-positive rates in the Dutch–Belgian lung cancer screening trial (NELSON), the National Lung Screening Trial (NLST) and the Danish Lung Cancer Screening Trial (DLCST). Trial
Number screened
NELSON2
7557 7289
Round of screening
Lung nodules over study threshold
Baseline Year 1
1570 570
Lung cancer nodules 70 54
Nodules not lung cancer
False-positive rate (nodules not lung cancer/no. screened)
1500 516
0.1985 0.0708
Average false-positive rate
0.1346 NLST
3
26 309 24 715 24 102
Baseline Year 1 Year 2
7191 6901 4054
270 168 211
6921 6733 3843
0.2631 0.2724 0.1594
2047 1976 1944 1982 1851
Baseline Year 1 Year 2 Year 3 Year 4
179 45 52 44 51
17 11 13 12 16
162 34 39 32 35
0.0791 0.0172 0.0201 0.0161 0.0189
0.2316 DLCST
4
0.0303
Table A5 Diagnostic results and participation during five screening rounds in the DLCST. Group
Round 0
1
2
3
4
Screening group
True negative False positive True positive False negative Missing
1868 162 17 0 5
1931 34 11 0 71
1892 39 13 0 32
1937 32 12 1 0
1800 35 16 0 131
Control group
Control Missing
2052 0
1953 99
1877 76
1838 39
1820 18
J.F. Rasmussen et al. / Lung Cancer 83 (2014) 347–355
Fig. A1. Mean annual healthcare costs per participant in the control group and the diagnostic groups in the DLCST during five screening rounds.
(4) Saghir Z, Dirksen A, Ashraf H, Bach KS, Brodersen J, Clementsen PF et al. CT screening for lung cancer brings forward early disease. The randomised Danish Lung Cancer Screening Trial: status after five annual screening rounds with low-dose CT. Thorax 2012; 67(4):296–301. References [1] Ott JJ, Ullrich A, Mascarenhas M, Stevens GA. Global cancer incidence and mortality caused by behavior and infection. J Public Health (Oxf) 2011;33(June (2)):223–33. [2] Siegel R, Naishadham D, Jemal A. Cancer statistics, 2012. CA Cancer J Clin 2012;62(January (1)):10–29. [3] Aberle DR, Adams AM, Berg CD, Black WC, Clapp JD, Fagerstrom RM, et al. Reduced lung-cancer mortality with low-dose computed tomographic screening. N Engl J Med 2011;365(August (5)):395–409. [4] Becker N, Motsch E, Gross ML, Eigentopf A, Heussel CP, Dienemann H, et al. Randomized study on early detection of lung cancer with MSCT in Germany: study design and results of the first screening round. J Cancer Res Clin Oncol 2012;138(September (9)):1475–86. [5] Field JK, van KR, Pedersen JH, Pastorino U, Paci E, Becker N, et al. European randomized lung cancer screening trials: post NLST. J Surg Oncol 2013;(July), http://dx.doi.org/10.1002/jso.23383.
355
[6] Infante M, Cavuto S, Lutman FR, Brambilla G, Chiesa G, Ceresoli G, et al. A randomized study of lung cancer screening with spiral computed tomography: three-year results from the DANTE trial. Am J Respir Crit Care Med 2009;180(September (5)):445–53. [7] Lopes PA, Picozzi G, Mascalchi M, Maria CF, Carrozzi L, Comin C, et al. Design, recruitment and baseline results of the ITALUNG trial for lung cancer screening with low-dose CT. Lung Cancer 2009;64(April (1)):34–40. [8] Pastorino U, Rossi M, Rosato V, Marchiano A, Sverzellati N, Morosi C, et al. Annual or biennial CT screening versus observation in heavy smokers: 5-year results of the MILD trial. Eur J Cancer Prev 2012;21(May (3)):308–15. [9] Saghir Z, Dirksen A, Ashraf H, Bach KS, Brodersen J, Clementsen PF, et al. CT screening for lung cancer brings forward early disease. The randomised Danish Lung Cancer Screening Trial: status after five annual screening rounds with low-dose CT. Thorax 2012;67(April (4)):296–301. [10] van Iersel CA, de Koning HJ, Draisma G, Mali WP, Scholten ET, Nackaerts K, et al. Risk-based selection from the general population in a screening trial: selection criteria, recruitment and power for the Dutch-Belgian randomised lung cancer multi-slice CT screening trial (NELSON). Int J Cancer 2007;120(February (4)):868–74. [11] Manser R, Lethaby A, Irving LB, Stone C, Byrnes G, Abramson MJ, et al. Screening for lung cancer. Cochrane Database Syst Rev 2013;6, http://dx.doi.org/10.1002/14651858.CD001991.pub3. [12] Goulart BH, Bensink ME, Mummy DG, Ramsey SD. Lung cancer screening with low-dose computed tomography: costs, national expenditures, and costeffectiveness. J Natl Compr Canc Netw 2012;10(February (2)):267–75. [13] van Klaveren RJ, Oudkerk M, Prokop M, Scholten ET, Nackaerts K, Vernhout R, et al. Management of lung nodules detected by volume CT scanning. N Engl J Med 2009;361(December (23)):2221–9. [14] Barton MB, Moore S, Polk S, Shtatland E, Elmore JG, Fletcher SW. Increased patient concern after false-positive mammograms: clinician documentation and subsequent ambulatory visits. J Gen Intern Med 2001;16(March (3)):150–6. [15] Lafata JE, Simpkins J, Lamerato L, Poisson L, Divine G, Johnson CC. The economic impact of false-positive cancer screens. Cancer Epidemiol Biomarkers Prev 2004;13(December (12)):2126–32. [16] Lidbrink E, Elfving J, Frisell J, Jonsson E. Neglected aspects of false positive findings of mammography in breast cancer screening: analysis of false positive cases from the Stockholm trial. BMJ 1996;312(February (7026)):273–6. [17] Brodersen J, Siersma V. Long-term psychosocial consequences of false-positive screening mammography. Ann Fam Med 2013;11(2):106–15. [18] Pedersen JH, Ashraf H, Dirksen A, Bach K, Hansen H, Toennesen P, et al. The Danish randomized lung cancer CT screening trial – overall design and results of the prevalence round. J Thorac Oncol 2009;4(May (5)):608–14. [19] Brodersen J, Thorsen H, Kreiner S. Consequences of screening in lung cancer: development and dimensionality of a questionnaire. Value Health 2010;13(August (5)):601–12. [20] Diehr P, Yanez D, Ash A, Hornbrook M, Lin DY. Methods for analyzing health care utilization and costs. Annu Rev Public Health 1999;20:125–44. [21] Zou G. A modified Poisson regression approach to prospective studies with binary data. Am J Epidemiol 2004;159(April (7)):702–6. [22] Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc B 1995;57(January (1)):289–300. [23] Brodersen J, Siersma V, Ryle M. Breast cancer screening: “reassuring” the worried well? Scand J Public Health 2011;39(May (3)):326–32. [24] Saghir Z, Ashraf H, Dirksen A, Brodersen J, Pedersen JH. Contamination during 4 years of annual CT screening in the Danish Lung Cancer Screening Trial (DLCST). Lung Cancer 2011;71(March (3)):323–7. [25] Villanti AC, Jiang Y, Abrams DB, Pyenson BS. A cost-utility analysis of lung cancer screening and the additional benefits of incorporating smoking cessation interventions. PLoS ONE 2013;8(8):e71379.