You are on page 1of 12

Cost-Effectiveness Analyses of Lung Cancer

Screening Strategies Using Low-Dose


Computed Tomography: a Systematic
Review

Adam J. N. Raymakers, John Mayo,


Stephen Lam, J. Mark FitzGerald, David
G. T. Whitehurst & Larry D. Lynd

Applied Health Economics and Health


Policy

ISSN 1175-5652

Appl Health Econ Health Policy


DOI 10.1007/s40258-016-0226-5

1 23
Your article is protected by copyright and
all rights are held exclusively by Springer
International Publishing Switzerland. This e-
offprint is for personal use only and shall not
be self-archived in electronic repositories. If
you wish to self-archive your article, please
use the accepted manuscript version for
posting on your own website. You may
further deposit the accepted manuscript
version in any repository, provided it is only
made publicly available 12 months after
official publication or later and provided
acknowledgement is given to the original
source of publication and a link is inserted
to the published article on Springer's
website. The link must be accompanied by
the following text: "The final publication is
available at link.springer.com”.

1 23
Author's personal copy
Appl Health Econ Health Policy
DOI 10.1007/s40258-016-0226-5

SYSTEMATIC REVIEW

Cost-Effectiveness Analyses of Lung Cancer Screening Strategies


Using Low-Dose Computed Tomography: a Systematic Review
Adam J. N. Raymakers1,7 • John Mayo2 • Stephen Lam3,4 • J. Mark FitzGerald4 •

David G. T. Whitehurst5,6 • Larry D. Lynd1,7

 Springer International Publishing Switzerland 2016

Abstract 2014. Our search returned 393 unique results. After remov-
Background Lung cancer screening with low-dose com- ing studies that did not meet our inclusion criteria, 13 studies
puted tomography (LDCT) has been shown to deliver appre- remained. Costs are presented in 2014 US dollars (US$).
ciable reductions in mortality in high-risk patients. However, Results The results from the economic evaluations iden-
in an era of constrained medical resources, the cost-effec- tified in this review were varied. All identified studies
tiveness of such a program needs to be demonstrated. reported outcomes using either additional survival (life-
Objective The aim of this study was to systematically years gained) or quality-adjusted life-years (QALYs
review the literature analyzing the cost-effectiveness of gained). Results ranged from US$18,452 to US$66,480 per
lung cancer screening using LDCT. LYG and US$27,756 to US$243,077 per QALY gained for
Methods We searched MEDLINE, EMBASE, EBM repeated screening. The results of cost-effectiveness anal-
Reviews—Health Technology Assessment, the National yses were sensitive to several key model parameters,
Health Service Economic Evaluation Database (NHS-EED), including the prevalence of lung cancer, cost of LDCT for
and the Cochrane Database of Systematic Reviews. Due to screening, the proportion of lung cancer detected as
technological progress in CT, we limited our search to localized disease, lead time bias, and, if included, the
studies published between January 2000 and December characteristics of a smoking cessation program.
Conclusions The cost-effectiveness of a lung cancer
screening program using LDCT remains to be conclusively
& Adam J. N. Raymakers resolved. It is expected that its cost-effectiveness will lar-
raymaker@mail.ubc.ca gely depend on identifying an appropriate group of high-
1
risk subjects.
Collaboration for Outcomes Research and Evaluation
(CORE), Faculty of Pharmaceutical Sciences, University of
British Columbia, Room 4102–2405 Wesbrook Mall,
Vancouver, BC, Canada
2
Key Points for Decision Makers
Department of Radiology, University of British Columbia,
Vancouver, BC, Canada
3
Results from cost-effectiveness analyses of lung
Department of Integrative Oncology, British Columbia cancer screening with low-dose computed
Cancer Agency, Vancouver, BC, Canada
tomography are varied.
4
Division of Respiratory Medicine, Faculty of Medicine,
University of British Columbia, Vancouver, BC, Canada Smoking cessation programs appear to be an
5
Faculty of Health Sciences, Simon Fraser University, important component of a lung cancer screening
Burnaby, BC, Canada strategy.
6
Centre for Clinical Epidemiology and Evaluation, Vancouver Improvements in methods to properly identify high-
Coastal Health Research Institute, Vancouver, BC, Canada
risk patients will impact cost-effectiveness of
7
Centre for Health Evaluation and Outcomes Sciences screening strategies.
(CHEOS), St Paul’s Hospital, Vancouver, BC, Canada
Author's personal copy
A. J. N. Raymakers et al.

1 Background benefits that will also impact cost-effectiveness estimates.


Overdiagnosis is a concern with LDCT screening due to its
Lung cancer is the most common cause of cancer death high sensitivity for early-stage lung cancer. Accordingly, a
worldwide, with an estimated 1.8 million new cases and potential drawback of using LDCT to image the lungs is
1.6 million deaths in 2012 [1]. Over the past four decades, that more surgeries than clinically necessary may be
clinical interventions have had only a minimal effect on undertaken, which could be associated with increased
reducing death from lung cancer [2]. The 5-year survival costs, patient anxiety, and potential adverse events. The
rate for patients with lung cancer is currently less than alternative diagnostic capacity of LDCT and non-lung
18 % because most cases are not found until patients cancer health risks associated with smoking can lead to the
become symptomatic with advanced and incurable disease detection of incidental findings in screened subjects.
[3]. The National Lung Screening Trial (NLST) in the US Detection of non-lung cancer changes such as coronary
recently showed that lung cancer screening with low-dose artery calcifications or emphysema has advantages and
computed tomography (LDCT) in high-risk current and disadvantages in that it might improve outcomes by iden-
former smokers can reduce death from lung cancer by tifying these diseases but with the potential for overtreat-
20 % compared with chest X-ray, which provides an ment, complications, and increase in overall costs.
alternative strategy to improve lung cancer diagnoses and The objective of this study was to systematically review
outcomes [4]. The recommendation by the US Preventive and critically evaluate published studies that explore the
Services Task Force (USPSTF) to screen high-risk smokers cost-effectiveness of lung cancer screening using LDCT.
with LDCT [5, 6] and the recent decision by the Centers for We highlight the complexity of LDCT screening for lung
Medicare and Medicaid Services to fund LDCT screening cancer from an economic evaluation perspective and offer
under the Medicare program [7] means that LDCT suggestions to improve the quality of future studies.
screening will be implemented at the population level in
the US and likely in other countries.
A recent US study estimated that a lung cancer 2 Methods
screening program could potentially increase health care
expenditures by US$1.4 to US$2.1 billion depending on the A search strategy was developed using the following
program uptake rate [8]. Given limited healthcare resour- databases: MEDLINE, Excerpta Medica Database
ces, careful analysis is required prior to implementing (EMBASE), the National Health Service (NHS) Centre for
LDCT screening programs. Economic evaluation is a tool Reviews and Dissemination Health Technology Assess-
that can be used to simultaneously analyze both the costs ment database, National Health Service Economic Evalu-
and consequences of health programs, facilitating com- ation Database (NHS-EED), and the Cochrane Database of
parison of alternative programs that might not otherwise be Systematic Reviews. The search included the following
directly comparable. The results of these analyses allow subject headings (and included all relevant subheadings):
decision makers to make informed choices given scarce ‘cost-benefit analysis’, ‘health care costs’, ‘quality-adjusted
healthcare resources [9]. The simultaneous evaluation of life years’, ‘lung neoplasms’, ‘mass screening’, ‘models
costs and outcomes is particularly relevant for screening (economic)’, ‘tomography (X-ray computed)’, and ‘to-
studies. Screening for any disease can be expensive and the mography (spiral computed)’. Additional keywords inclu-
potential benefits tend to be derived from earlier identifi- ded in the search were ‘cost-effectiveness’, ‘lung cancer’,
cation and treatment initiation, which can generate addi- ‘economic evaluation’, and ‘computed tomography’. Two
tional associated costs. Therefore, in order to properly authors (AR and DW) independently reviewed and com-
assess a screening program, consequences (i.e. benefits) pared the results of the search. Inconsistencies in the results
should be shown in terms of length and/or quality of life were resolved through discussion. Any further uncertainty
that may be gained from the screening program, as opposed as to inclusion was discussed within the study team. Ref-
to costs averted. erences provided by the included studies were also sear-
There are several key factors in LDCT screening for ched using the same study selection criteria (Fig. 1). The
lung cancer that can impact estimates of cost-effectiveness, search strategy is detailed in the ‘‘Appendix’’.
such as lead-time, overdiagnosis, and incidental findings.
Lead time is particularly important for lung cancer 2.1 Study Selection
screening; it is the time difference between screening-de-
tected lung cancer and symptomatic presentation of lung Articles were chosen for inclusion if the authors conducted
cancer. A screening strategy that generates substantial lead a cost-effectiveness analysis evaluating the use of LDCT as
time could potentially generate significant therapeutic a screening tool for lung cancer. Due to substantial
Author's personal copy
Cost-Effectiveness of Lung Cancer Screening

Fig. 1 Study selection process,


with reasons for exclusion.
CRD-HTA Centre for Reviews
and Dissemination Health
Technology Assessment, NHS-
EED National Health Service
Economic Evaluation Database

advances in CT, primarily the introduction of multidetector allowed for, if ICERs were calculated, and if sensitivity
row technology, the search was limited to include studies analysis was performed. This review did not explicitly use
published between January 2000 and December 2014. a quality assessment metric but used these established
criteria to facilitate consideration of key study components.
2.2 Study Abstraction Such an approach has been used in previous review studies
[12].
The following parameters were extracted from each study
and entered into a standardized data abstraction form:
LDCT cost, LDCT sensitivity and specificity, key model 3 Results
parameters (perspective of the analysis, discount rate,
currency year, study location, and patient population), The search strategy yielded 393 unique results and 13
whether sensitivity analysis was performed, model char- studies met the inclusion criteria [13–25]. Ten of the
acteristics specific to lung cancer screening (i.e. consider- included studies were from the US, and one each from
ation of lead time bias and incidental findings analysis), Australia, Israel, and Japan. The process by which studies
and the calculated incremental cost-effectiveness ratio were included (along with reasons for exclusion) is out-
(ICER). To facilitate comparison between published stud- lined in Fig. 1.
ies all costs were adjusted to 2014 US dollars [10, 11].
We followed the criteria proposed by Drummond et al. 3.1 At-Risk Patient Population
(chapter 2) [9] for assessing economic evaluations. These
criteria assess cost-effectiveness analyses in terms of their The risk of lung cancer in the population to be screened
attributes, including whether or not comparators were well- will be a critical component impacting the cost-effective-
described, whether the data used to populate the model ness of such a program. Studies included in this review
were from reliable sources, whether differential timing was focused predominantly on high-risk groups (defined as
Author's personal copy
A. J. N. Raymakers et al.

being either 50 or 60 years of age or older with a signifi- decrease smoking rates among those undergoing screening
cant smoking history (i.e. C20–30 pack-years)). Of the 13 (high-risk patients), thereby reducing mortality from lung
studies identified in this review, ten [14–20, 22, 23, 25] cancer and other diseases [26]. The potential for this added
used similar definitions for their patient population, which benefit (and cost) was considered by McMahon et al. [20],
aligned closely with those used in NLST and Early Lung who included a smoking cessation program costing
Cancer Action Project (ELCAP) studies (Table 1). The US$352 per person. This program increased smoking ces-
remaining three studies screened unique and typically sation by 4–30 % depending on the patient population,
younger populations [13, 21, 24]. Beinfeld et al. [13] did above a background smoking cessation rate of 3 % [20].
not focus on a high-risk population but rather evaluated the Villanti et al. [25] used a background rate of 2.5 % and an
cost-effectiveness of a single whole-body screening for increase depending on the intensity of the smoking cessa-
eight malignancies (including lung cancer) in subjects tion program (1.5–3.04 times the background rate), and
greater than 50 years of age. The youngest age included in included costs for corresponding smoking cessation inter-
a screening program amongst included studies was ventions (counselling, nicotine replacement therapy,
observed in Okamoto et al. [21], who screened individuals pharmacotherapy).
40 years of age and older.
3.3 Low-Dose Computed Tomography Attributes
3.2 Modeling Approach
The sensitivity and specificity associated with LDCT is an
The modeling approach in all studies was similar, typically important component of the screening program and will
using decision analytic modeling to evaluate an LDCT- influence cost-effectiveness estimation. Reported sensitiv-
based screening program versus no screening. The majority ities of LDCT to detect nodules were dependent on the
of studies evaluated annual screening [14–18, 20, 22, 23, nodule size and ranged from 0.63 [20] to 0.94 [16, 27]
25], while four evaluated a one-time screening program (Table 2). The specificity of LDCT was reported less fre-
[13, 19, 21, 24]. All but one study [21] performed sensi- quently [13, 16, 17], ranging from 0.79 [13] to 0.97 [17]. In
tivity analysis on the model structure and/or key parame- the study by Shmueli et al. [24], no allowance was made
ters, and only one study conducted probabilistic sensitivity for false negative results in screening, which the authors
analysis [24]. Only ten of the studies reported the per- justify by stating that there is ‘‘… no practical observa-
spective from which the analysis was conducted [13–17, tional way to confirm false negative cases’’ (p. 923).
20, 22–25]. Lung cancer treatment costs were associated Depending on the location of the malignant nodule (e.g.
with the stage at which the cancer was diagnosed. Lung perihilar, vessel bifurcation), it is reasonable to expect that
cancer detected by LDCT screening was typically associ- LDCT will occasionally be falsely negative. In this cir-
ated with a ‘stage-shift’, meaning that the distribution of cumstance, subjects would be diagnosed when symptoms
LDCT-discovered lung cancers was ‘shifted’ toward an present, and the costs associated with screening will be
earlier stage of lung cancer. For example, in the study by purely additional. Therefore, failure to account for false
Mahadevia et al. [16], the authors employed a 50 % stage negative cases would overestimate the benefit conferred by
shift whereby the distribution of lung cancer cases (be- LDCT screening. The extent to which this scenario exists
tween localized and advanced disease) in the screened reflects an unquantified issue that will impact cost-effec-
group would be 50 % toward earlier disease compared with tiveness estimations.
the unscreened group.
Approximately half of the studies identified in this 3.4 Outcome Measures
review made some allowance for lead-time bias. In anal-
yses where lead-time was considered, it was assumed to be The studies identified in this review reported outcomes in
either 1 year [17], 2 years [24, 25], or 3 years [23]. The terms of life-years gained or quality-adjusted life-years
variation in the estimate for this model parameter reflects (QALYs) gained. More recent studies tended toward using
the uncertainty in what the actual lead-time for lung cancer QALYs as an outcome measure, consistent with most
is, and how it may be dependent on the patient population, guidelines [28, 29]. Prior to the study by Black et al. [14],
lung cancer histology, and the subjects’ response to no study contained an original estimate of utility values
treatment. elicited from patients, nor did any use patients’ responses
A major consideration in the cost-effectiveness of lung to preference-based utility instruments for patients under-
cancer screening is smoking cessation. Four studies inclu- going lung cancer screening—all used previously pub-
ded smoking cessation programs with annual screening lished utility estimates. Black et al. [14] produced utility
[20, 22, 23, 25]. A smoking cessation program coupled weights from a subset of 11,696 of the NLST patients who
with a lung cancer screening strategy has the potential to answered the Short-Form 36 (SF-36). Utility weights for
Table 1 Summary of identified studies
Author Year Country Patient population Screening Modelling approach Time Discount Indirect Perspective Sensitivity
frequency horizon rate costs (reported) analysis

Beinfeld 2002 US 50? year old males One time Decision tree Lifetime Not No ‘Quasai’ Yes
[13] applied societal
Black [14] 2014 US 55–74 years and a smoking history of 30? pack years Annual Trial based Lifetime 3% Yes Societal Yes
(NLST) economic
evaluation
Cost-Effectiveness of Lung Cancer Screening

Chirikos 2002 US 60?, smoking history of 10? pack years (ELCAP) Annual Not reported 15 years 7.5 % No National Yes
[15] (costs payer
only)
Mahadevia 2003 US Current, quitting, and former heavy smokers aged 60 Annual Markov model 20 years 3% No Societal Yes
[16]
Manser 2005 Australia Males 60–64 years and smoking history of 40 Annual Markov model 15 years 3% No 3rd party Yes
[17] cigarettes a day for 40 years (high risk) payer
Marshall 2001 US 60–74 years (ELCAP) Annual Decision tree 5 years 3% No Not reported Yes
[18]
Marshall 2001 US 60–74 years; high and low risk One time Decision Tree 5 years 3% No Not reported Yes
[19]
McMahon 2011 US 50–74 years and 20? pack years of smoking history Annual Patient-level Lifetime 3% Yes Societal Yes
[20] simulation model
Okamato 2000 Japan Mass screening; 40? years and smoking index greater One time Deterministic Lifetime Not No Not reported No
[21] than 600 mathematical reported
Author's personal copy

model
Pyenson 2012 US Smokers and former smokers aged 50–64 with Annual Actuarial analysis 15 years Not No Commercial Yes
[22] minimum of 30 pack years applied payer
Pyenson 2014 US 55–80 years and a smoking history of 30? pack years Annual Actuarial analysis Lifetime Not No Medicare Yes
[23] (and smoked within past 15 years) applied
Shmueli 2013 Israel Moderate to heavy smokers aged 45? One time decision tree Lifetime 3% No Health Yes
[24] system
Villantini 2013 US 50–64 years and 30? pack years Annual Actuarial analysis 15 years Not No Commercial Yes
[25] applied payer
NLST National Lung Cancer Screening Trial, ELCAP Early Lung Cancer Action Project, US United States
Author's personal copy
A. J. N. Raymakers et al.

Table 2 Reported sensitivity, specificity, and costs of CT


Author CT sensitivity/specificity Currency (year) CT cost [2014 US$]

Repeated screening
Black [14] 0.94; 0.73a USD (2009) 285 [314.49]
Chirikos [15] NR USD (2000) 291 [400.06]
-416 [571.91]b
Mahadevia [16] 0.94; 0.79 USD (2001) 300 [401.02]
Manser [17] 0.86; 0.97 AUSD (2002) 280 [275.69]
Marshall [18] NR USD (1999) 150 [213.15]
McMahon [20] 0.63–1 (depending on nodule size); NR USD (2006) 283 [332.32]
Pyenson [22] NR USD (2012) 247 [254.68]
Pyenson [23] NR USD (2014) 241
Villantini [25] NR USD (2012) 210 [216.53]
One-time screening
Beinfeld [13] 0.83; 0.83 USD (2001) 900 [1203.06]e
c
Marshall [19] NR USD (2001) 150 [200.51]
Okamato [21] 0.90; NR USD/Japanese Yen (NR) 30 [42.63]
Shmueli [24] NRd USD (2011) 74 [77.88]
a
Not reported but assumed based on the results of the NLST (27)
b
Varied based on usage of contrast material
c
Reported a 100 % cancer detection rate
d
No allowance was made for false negative results
e
Whole body scan

two studies [20, 25] were obtained from Hanmer et al. [30], values were used, the difference in cost-effectiveness esti-
who reported utility scores based on the US Medication mates relates to the modeling approach taken by the authors.
Expenditure Panel Survey (MEPS). This survey, conducted For example, McMahon et al. [20] and Villanti et al. [25] are
in 2001, collected SF-12 (version 1) and EQ-5D (a pref- US-based studies with substantially different ICER esti-
erence-based measure of health-related quality of life) mates (US$169,097 and US$29,118 per QALY, respec-
scores from a representative sample of the non-institu- tively). Both studies used utility weights from the same
tionalized population over the age of 18 years. Three study [30] and LDCT costs that were not remarkably dif-
studies [16–18] used a systematic review by Earle et al. ferent (US$332 and US$217, respectively). In a sensitivity
[31] as a source for utility values. Shmueli et al. [24] used analysis conducted by Villanti et al. [25], the cost of LDCT
utility values obtained from a panel of family physicians was increased 150 %, and the resulting ICER only increased
who were asked to consider a series of hypothetical sce- to US$45,989 per QALY gained (from a base-case estimate
narios and, for each scenario, complete the EQ-5D-3L of US$29,118 per QALY). Two aspects of this study are
instrument. worth noting; first, the time horizon used in this analysis was
15 years, not lifetime, and, second, the population was
3.5 Model Results males and females 50–64 years of age. In their analysis,
McMahon et al. [20] used a lifetime time horizon with
Cost-effectiveness estimates varied substantially between people aged 50–74 years. It should be noted that all studies
studies. For life-years gained (i.e. using survival as the which reported ICERs in excess of US$100,000 per QALY
measure of outcome), ICERs ranged from US$8441 [19] to or life-year gained were completed before or did not
US$201,847 [13] per life-year gained for one-time screen- incorporate the results of the NLST [13, 16, 17, 20].
ing, and US$18,452 [23] and US$66,480 [15] for repeated Overall, the model results (Table 3) appeared to be
screening. For studies using QALYs as the outcome mea- particularly sensitive to the prevalence of lung cancer, the
sure, results ranged from US$27,756 [18] to US$243,077 cost of the LDCT screening test, the proportion of lung
[20] per QALY gained (Table 3). For one-time screening, cancer detected as localized disease, and lead time bias. In
only one study reported an ICER in terms of QALYs addition, for those studies that included a smoking cessa-
(US$1541/QALY) [24]. One study did not report an incre- tion program, the model results were particularly sensitive
mental ratio [21]. In instances where similar parameter to the characteristics of this program.
Author's personal copy
Cost-Effectiveness of Lung Cancer Screening

Table 3 Results of identified economic evaluations


Author Currency (year) Outcome metric DCost [2014 US$] DEffect ICER [2014 US$]

Repeated screening
Black [14] USD (2009) LY and QALY 1631 [1799.77] 0.0316 LY and 52000 [57380.69] per LY and 81000
0.0201 QALYs [89381.46] per QALY
Chirikos [15] USD (2000) LY 76500 [105170.25] 1.582 48357 [66479.97]
Mahadevia [16] USD (2001) QALY 4600 [6148.99] 0.039 116300 [155462.54]
Manser [17] AUSD (2002) QALY 1649 [1623.64] 0.016a 105090 [103473.51]
a
Marshall [18] USD (1999) QALY 960 [1364.15] 0.0491a 19533 [27756.11]
McMahon [20] USD (2006) QALY 1778–3637 0.009–0.022 144000–207000 [169097.
[2087.88–4270.88] 26–243077.31]
Pyenson [22] USD (2012) LY NR NR 18862 [19448.76]
Pyenson [23] USD (2014) LY 16053a 0.87a 18452
a a
Villantini [25] USD (2012) QALY 1546 [1594.09] 0.055 28240 [29118.50]
One-time screening
Beinfeld [13] USD (2001) LY 2513 [3359.22] 0.016 151000 [201847.32]
Marshall [19] USD (1999) LY 260a [369.46] 0.0438a 5940 [8440.65]
Okamato [21] USD/Japanese Life saved NR NR NRb
Yen (NR)
Shmueli [24] USD (2011) QALY 86 [90.51] 0.0591 1464 [1540.78]
USD United States Dollars, LY life years, LYG life years gained, QALY quality-adjusted life-years, AUSD Australian dollars, NR not reported
a
Calculated based on results reported in the study
b
Reported only graphically

4 Discussion screening program, including the frequency of screening,


patients’ age, and lung cancer risk. Most of these issues
This systematic review reports on the published economic have been addressed by subsequent cost-effectiveness
evaluations of lung cancer screening incorporating LDCT, analyses, which are described in the current paper,
and is the first review to do so after publication of the although patients’ age and the definition of ‘high risk’
results of the NLST. In addition, we sought to identify the varied across studies.
key components of an economic evaluation for lung cancer Multiple identified analyses had issues with model
screening and provide insight into areas of improvement transparency and proper use and understanding of termi-
for future studies. The results of this review suggest that nology used in economic evaluation. For example, earlier
whether LDCT screening for lung cancer is cost effective studies did not present model parameters as clearly as later
remains unclear and published cost-effectiveness estimates studies [15, 18, 19, 21], and sensitivity and specificity
are heavily contingent on the modeling approach and val- values were absent from several studies [15, 18, 19, 22–
ues of several key parameters. The identified analyses were 24]. Clarity regarding the cost-year, discount rate, and
particularly sensitive to the at-risk population to be perspective of the analysis also varied. Several studies
screened, the cost of the LDCT screening test, the pro- claimed to conduct their analysis from a ‘societal’ per-
portion of lung cancer detected as localized disease, lead- spective but it was unclear what societal costs were
time bias, and, if included, the characteristics of a smoking included [13, 16, 20]. In addition, Pyenson et al. [23] did
cessation program. This review reiterates the importance of not discount costs or outcomes, stating that 2014 costs were
economic evaluation prior to the implementation of pop- used throughout their analysis, precluding the necessity of
ulation-based screening programs. discounting future costs. This reflects a lack of under-
A previously conducted review by Black et al. [33] standing regarding the concept of time preference and why
reported that in the absence of a proper trial to evaluate discounting is important in the estimation of cost-effec-
LDCT as a method for lung cancer screening, no decision tiveness. Only two studies included indirect costs; one
about adoption could be made due to the paucity of good- estimated lost wages for patients receiving a lung cancer
quality data demonstrating the efficacy for LDCT in this diagnosis [25], and the other included losses due to travel
capacity. The report suggested that several factors would time associated with screening for both patients and care-
be very important to the cost-effectiveness of a lung cancer givers [14]. While sensitivity analyses were conducted in
Author's personal copy
A. J. N. Raymakers et al.

most studies, probabilistic sensitivity analysis was only PLCOm2012 model selected fewer patients for screening,
performed in one study [24]. This is a major limitation yet identified 12.4 % more lung cancers [39]. This selec-
given the uncertainty surrounding many model parameters tion method for screening could increase both efficiency of
and the insights that could be gained through the use of LDCT screening and cost-effectiveness. The use of more
probabilistic methods to inform future research [34]. Two sophisticated risk prediction models to select smokers for
studies were comprehensive in that while they modeled the LDCT screening versus NLST or USPSTF criteria should
benefits of LDCT screening, the potential harm (in terms of be validated in a prospective trial to compare these
aggregate radiation) was also taken into account [17, 20]. parameters (number of lung cancers identified, differences
Manser [17] highlighted a potential ‘worst-case’ scenario, in the number needed to screen to detect one lung cancer,
where under the most pessimistic of published conditions, and the positive predictive value). The evidence for the
for example an LDCT sensitivity of 0.65 and cost of appropriate tumour size and growth rate that are predictive
US$552, LDCT screening could result in a loss of life- of lung cancer has also been improving [40]. Finally, the
years in their hypothetical cohort, pointing to the impor- prevalence of lung cancer in the screened population and
tance of a programmatic approach to screening imple- the frequency of follow-up of screened subjects can greatly
mentation versus ad hoc or opportunistic screening. influence the cost of the program. Reduction in the fre-
Only one study incorporated the possibility of incidental quency of follow-up LDCT scans using larger nodule size
findings affecting cost-effectiveness estimates in the base- to define a positive screen has also been proposed [41] but
case analysis. Black et al. [14] assigned a cost of US$500 prospective cost-effectiveness analyses are needed [42].
for a patient with a clinically significant incidental finding. Lung cancer screening offers an excellent opportunity
It is possible that there could be additional benefits from for physicians to promote smoking cessation in their
non-lung cancer findings in annual screenings for patients patients [26]. Smokers who participated in lung cancer
who would be at high risk for other diseases given their age screening had a three- to fivefold higher smoking cessation
and smoking history. For example, in their study of 4073 rate compared with smokers in the general population,
patients undergoing lung cancer screening, Kucharzyk which is generally between 2 and 3 % [43–45]. The ben-
et al. [35] found that 19 % of patients had ‘actionable’ efits of smoking cessation that could be combined with a
incidental findings. In addition, given the associated risk of lung cancer screening program need to be included in
chronic obstructive pulmonary disease (COPD) in smokers, future cost-effectiveness analyses, or at a minimum as a
as technology advances LDCT could be used as a diag- sensitivity analyses.
nostic tool for COPD, which suffers from underdiagnosis, The NLST showed that lung cancer screening may
particularly in milder disease states [36]. Villanti et al. [25] confer a benefit of approximately 20 % in reduced lung
report that annual scans may also be beneficial for cancer mortality in a high-risk patient population. While
improving risk identification for coronary heart disease the results of the NLST should have a substantial impact on
(CHD) [37]. the cost-effectiveness of a lung cancer screening program,
Several factors may have implications for future cost- clinical effectiveness does not (necessarily) equate with
effectiveness analyses. Technological advances in multi- cost-effectiveness and, in an era of constrained healthcare
detector CT scanner technology will likely decrease resources and emerging expensive medical technologies,
examination time and radiation dose, which has the economic evaluation can help us make efficient resource
potential to improve the cost-effectiveness of LDCT allocation decisions. The studies identified in this review
screening programs. In addition to technological advances, highlight that while there may be survival and quality-of-
improved selection criteria for patients can have a signifi- life gains that accrue from lung cancer screening (partic-
cant impact on cost-effectiveness. In the NLST, the 20 % ularly in high-risk populations), there is considerable
of participants with the lowest risk (using age 55–74 years variation in ICERs. Studies conducted prior to the com-
and C30 pack-years as inclusion criteria) accounted for pletion and availability of results from the NLST typically
only 1 % of prevented lung cancer deaths [4]. Tammemägi had higher ICERs, whereas more recent results show more
et al. [38] compared the NLST selection criteria with a favourable ICERs, due to both lower incremental costs and
modified version of the Prostate, Lung, Colorectal and better effects.
Ovarian (PLCO) Cancer Screening Trial criteria
(PLCOm2012). Compared with NLST, they found the 4.1 Limitations
PLCOm2012 showed increased sensitivity and positive pre-
dictive value. More recently, Tammemägi et al. [39] found This systematic review only included studies that were
that the PLCOm2012 risk model for selecting patients for published from the year 2000 onward since, for techno-
screening had greater sensitivity, specificity, and positive logical reasons, LDCT was inadequate for screening prior
predictive value than the USPSTF criteria. Moreover, the to that time. In addition, our review was limited to studies
Author's personal copy
Cost-Effectiveness of Lung Cancer Screening

published in English as personnel to review non-English or ‘‘cost benefit analysis’’ /


results were not available. However, our search returned or ‘‘cost-effectiveness analysis’’/
very few non-English results (Fig. 1), therefore we feel that or economic aspect/
our results are robust. Lastly, without access to the com- or economic evaluation/
plete models from identified studies it is difficult to fully or health economics/
comprehend the discrepancy in the magnitude of model 4. computed tomography.mp. or computer asssisted
results (ICERs) beyond stated model parameters and tomography /
assumptions. 5. spiral computer assisted tomography/
6. 4 or 5/
7. 1, 2, and 3/
5 Conclusions 8. 6 and 7/
Evidence from recent trials has shown that significant
reductions in mortality can be achieved with screening a
high-risk population for lung cancer with LDCT; however,
References
whether or not such a program is cost effective is still
contentious. The impact of incidental findings, only con- 1. Stewart BW, International Agency for Research on Cancer,
sidered by one study in this review, will undoubtedly affect World Health Organization. World cancer report 2014. Lyon:
cost-effectiveness estimates in this population with ele- International Agency for Research in Cancer; 2014.
vated cardiovascular, COPD, and cancer risk. The cost- 2. Siegel R, Naishadham D, Jemal A. Cancer statistics, 2013. CA
Cancer J Clin. 2013;63(1):11–30.
effectiveness of a lung cancer screening program might 3. Coleman M, Forman D, Bryant H, Butler J, Rachet B, Maringe C,
also be enhanced with improved risk stratification, et al. Cancer survival in Australia, Canada, Denmark, Norway,
knowledge of nodule size predictive of lung cancer, Sweden, and the UK, 1995–2007 (the International Cancer
advances of CT technology, and inclusion of a smoking Benchmarking Partnership): an analysis of population-based
cancer registry data. Lancet. 2011;377(9760):127–38.
cessation program alongside a screening strategy. 4. Team National Lung Screening Trial Research, Aberle DR,
Adams AM, Berg CD, Black WC, Clapp JD, et al. Reduced lung-
Author contributions Adam Raymakers generated the original cancer mortality with low-dose computed tomographic screening.
idea for this study, conducted the systematic review, and drafted and N Engl J Med. 2011;365(5):395–409.
submitted the manuscript. John Mayo and Mark FitzGerald helped 5. Moyer VA. Screening for lung cancer: US Preventive Services
with the conceptualization of the study and editing of the manuscript. Task Force Recommendation Statement. Ann Intern Med. 2014;
Both authors added valuable clinical input to the study. Stephen Lam 160(5):330–8.
reviewed and edited the manuscript and provided critical insight into 6. Humphrey L, Deffebach M, Pappas M, Baumann C, Artis K,
the subject area. David Whitehurst was the second reviewer in the Mitchell JP, et al. Screening for lung cancer: systematic review to
systematic review, and assisted with drafting and editing the manu- update the U.S. Preventive Services Task Force Recommenda-
script, and manuscript submission. Larry Lynd helped with concep- tion. Rockville (MD): Agency for Healthcare Research and
tualization of the study, and drafting and editing the manuscript. All Quality (US); 2013. Available at: http://www.ncbi.nlm.nih.gov/
authors approved the final version of the manuscript. books/NBK154610/. Accessed 24 Mar 2015
7. Centers for Medicare and Medicaid Services (CMS). Proposed
Compliance with Ethical Standards decision memo for screening for lung cancer with low dose
computed tomography (LDCT) (CAG-00439N). 2014. Available
Conflicts of interest Adam Raymakers, John Mayo, Stephen Lam, at: https://www.cms.gov/medicare-coverage-database/details/
Mark FitzGerald, David Whitehurst, and Larry Lynd declare no nca-decision-memo.aspx?NCAId=274. Accessed 24 Mar 2015.
competing interests with this study. 8. Goulart BHL, Bensink ME, Mummy DG, Ramsey SD. Lung
cancer screening with low-dose computed tomography: costs,
Funding No specific funding was used for this study. national expenditures, and cost-effectiveness. J Natl Compr
Cancer Netw. 2012;10(2):267–75.
9. Drummond M, Sculpher M, Torrance G, O’Brien B, Stoddart G.
Methods for the economic evaluation of health care programmes.
Appendix: EMBASE search strategy, searched 3rd ed. New York: Oxford University Press; 2005.
via OvidSP 10. Organization for Economic Co-Operation and Development.
Purchasing power parities and exchange rates. Available at:
http://stats.oecd.org/Index.aspx?DataSetCode=SNA_TABLE4.
1. lung cancer/
Accessed 23 Mar 2015.
2. mass screening/ 11. United States Bureau of Labour Statistics. Consumer Price Index
or screening test/ (CPI). Available at: http://www.bls.gov/cpi/cpid1502.pdf. Acces-
or cancer screening / sed 23 Mar 2015.
12. Raymakers AJN, Mayo J, Marra CA, FitzGerald M. Diagnostic
or screening
strategies incorporating computed tomography angiography for
3. economic evaluation.mp or economics/ pulmonary embolism: a systematic review of cost-effectiveness
or ‘‘health care cost’’/ analyses. J Thorac Imaging. 2014;29(4):209–16.
Author's personal copy
A. J. N. Raymakers et al.

13. Beinfeld MT, Wittenberg E, Gazelle GS. Cost-effectiveness of 30. Hanmer J, Lawrence WF, Anderson JP, Kaplan RM, Fryback
whole-body CT screening. Radiology. 2005;234(2):415–22. DG. Report of nationally representative values for the noninsti-
14. Black WC, Gareen IF, Soneji SS, Sicks JD, Keeler EB, Aberle tutionalized US adult population for 7 health-related quality-of-
DR, et al. Cost-effectiveness of CT screening in the national lung life scores. Med Decis Making. 2006;26(4):391–400.
screening trial. N Engl J Med. 2014;371(19):1793–802. 31. Earle CC, Chapman RH, Baker CS, Bell CM, Stone PW, Sand-
15. Chirikos TN, Hazelton T, Tockman M, Clark R. Cost-effective- berg EA, et al. Systematic overview of cost-utility assessments in
ness of screening for lung cancer. JAMA. 2003;289(18):2358. oncology. J Clin Oncol. 2000;18(18):3302–17.
16. Mahadevia PJ, Fleisher LA, Frick KD, Eng J, Goodman SN, 32. Manser R. Screening for lung cancer: a review. Curr Opin Pulm
Powe NR. Lung cancer screening with helical computed Med. 2004;10(4):266–71.
tomography in older adult smokers: a decision and cost-effec- 33. Black C, Bagust A, Boland A, Walker S, McLeod C, De Verteuil
tiveness analysis. J Am Med Assoc. 2003;289(3):313–22. R, et al. The clinical effectiveness and cost-effectiveness of
17. Manser R, Dalton A, Carter R, Byrnes G, Elwood M, Campbell computed tomography screening for lung cancer: systematic
DA. Cost-effectiveness analysis of screening for lung cancer with reviews. Health Technol Assess. 2006;10(3):iii–iv, ix–x, 1–90.
low dose spiral CT (computed tomography) in the Australian 34. Claxton K, Sculpher M, McCabe C, Briggs A, Akehurst R,
setting. Lung Cancer. 2005;48(2):171–85. Buxton M, et al. Probabilistic sensitivity analysis for NICE
18. Marshall D, Simpson KN, Earle CC, Chu CW. Economic deci- technology assessment: not an optional extra. Health Econ.
sion analysis model of screening for lung cancer. Eur J Cancer. 2005;14(4):339–47.
2001;37(14):1759–67. 35. Kucharczyk MJ, Menezes RJ, McGregor A, Paul NS, Roberts
19. Marshall D, Simpson KN, Earle CC, Chu C. Potential cost-ef- HC. Assessing the impact of incidental findings in a lung cancer
fectiveness of one-time screening for lung cancer (LC) in a high screening study by using low-dose computed tomography. Can
risk cohort. Lung Cancer. 2001;32(3):227–36. Assoc Radiol J. 2011;62(2):141–5.
20. McMahon PM, Kong CY, Bouzan C, Weinstein MC, Cipriano 36. Galbán CJ, Han MK, Boes JL, Chughtai KA, Meyer CR, Johnson
LE, Tramontano AC, et al. Cost-effectiveness of computed TD, et al. Computed tomography-based biomarker provides
tomography screening for lung cancer in the United States. unique signature for diagnosis of COPD phenotypes and disease
J Thorac Oncol. 2011;6(11):1841–8. progression. Nat Med. 2012;18(11):1711–5.
21. Okamoto N. Cost-effectiveness of lung cancer screening in Japan. 37. Polonsky TS. Coronary artery calcium score and risk classifica-
Cancer. 2000;89(11 Suppl):2489–93. tion for coronary heart disease prediction. JAMA. 2010;303(16):
22. Pyenson BS, Sander MS, Jiang Y, Kahn H, Mulshine JL. An 1610.
actuarial analysis shows that offering lung cancer screening as an 38. Tammemägi MC, Katki HA, Hocking WG, Church TR, Caporaso
insurance benefit would save lives at relatively low cost. Health N, Kvale PA, et al. Selection criteria for lung-cancer screening.
Aff (Millwood). 2012;31(4):770–9. N Engl J Med. 2013;368(8):728–36.
23. Pyenson BS, Henschke CI, Yankelevitz DF, Yip R, Dec E. 39. Tammemägi MC, Church TR, Hocking WG, Silvestri GA, Kvale
Offering lung cancer screening to high-risk medicare beneficia- PA, Riley TL, et al. Evaluation of the lung cancer risks at which
ries saves lives and is cost-effective: an actuarial analysis. Am to screen ever- and never-smokers: screening rules applied to the
Health Drug Benefits. 2014;7(5):272–81. PLCO and NLST cohorts. PLoS Med. 2014;11(12):e1001764.
24. Shmueli A, Fraifeld S, Peretz T, Gutfeld O, Gips M, Sosna J, 40. Horeweg N, van Rosmalen J, Heuvelmans MA, van der Aalst
et al. Cost-effectiveness of baseline low-dose computed tomog- CM, Vliegenthart R, Scholten ET, et al. Lung cancer probability
raphy screening for lung cancer: the Israeli experience. Value in patients with CT-detected pulmonary nodules: a prespecified
Health. 2013;16(6):922–31. analysis of data from the NELSON trial of low-dose CT
25. Villanti AC, Jiang Y, Abrams DB, Pyenson BS. A cost-utility screening. Lancet Oncol. 2014;15(12):1332–41.
analysis of lung cancer screening and the additional benefits of 41. Henschke CI. Definition of a positive test result in computed
incorporating smoking cessation interventions. PLoS One. tomography screening for lung cancer: a cohort study. Ann Intern
2013;8(8):e71379. Med. 2013;158(4):246.
26. Taylor KL, Cox LS, Zincke N, Mehta L, McGuire C, Gelmann E. 42. Lam S, McWilliams A, Mayo J, Tammemagi M. Computed
Lung cancer screening as a teachable moment for smoking ces- tomography screening for lung cancer: What is a positive screen?
sation. Lung Cancer. 2007;56(1):125–34. Ann Intern Med. 2013;158(4):289–90.
27. Aberle DR, DeMello S, Berg CD, Black WC, Brewer B, Church 43. Stead LF, Bergson G, Lancaster T. Physician advice for smoking
TR, et al. Results of the two incidence screenings in the National cessation. In: Cochrane Database of Systematic Reviews. Wiley;
Lung Screening Trial. N Engl J Med. 2013;369(10):920–31. 1996. Available at: http://onlinelibrary.wiley.com/doi/10.1002/
28. NICE. The social care guidance manual. Incorporating economic 14651858.CD000165.pub3/abstract. Accessed 25 Mar 2015
evaluation. Guidance and guidelines. Available at: http:// 44. Gomez MM, LoBiondo-Wood G. Lung cancer screening with
www.nice.org.uk/article/pmg10/chapter/incorporating-economic- low-dose CT: its effect on smoking behavior. J Adv Pract Oncol.
evaluation#measuring-and-valuing-effects. Accessed 8 Apr 2015. 2013;4(6):405–14.
29. Canadian Agency for Drugs and Technologies in Health. 45. Tammemägi MC, Berg CD, Riley TL, Cunningham CR, Taylor
Guidelines for the economic evaluation of health technologies KL. Impact of lung cancer screening results on smoking cessa-
Canada. Ottawa (ON): Canadian Agency for Drugs and Tech- tion. J Natl Cancer Inst. 2014;106(6):dju084.
nologies in Health; 2006. Available at: https://www.cadth.ca/
media/pdf/186_EconomicGuidelines_e.pdf. Accessed 8 Apr 2015.

You might also like