[Evaluation of the lung cancer screening program]
García Pérez L, Cantero Muñoz P, Hernández Yumar A, Puñal Riobóo J, Paz Valiñas L, Toledo Chávarri A, et al
Record ID 32018005345
Spanish
Original Title:
Evaluación del programa de cribado de cáncer de pulmón
Authors' objectives:
The main objective was to assess the comparative efficacy, safety and efficiency, and to
analyse the ethical, social, and organisational aspects, as well as the possible environmental impact and research needs of screening in populations at high risk of developing LC. To this end, 8 research questions were established. The degree of compliance with the essential requirements for implementing a population screening programme in the NHS was also assessed.
The target population consisted of adults, without LC at the time of diagnosis
(confirmed or suspected), smokers or ex-smokers or with other risk factors (occupational
or environmental exposure to carcinogens such as radon, asbestos or fine particles, COPD,
idiopathic pulmonary fibrosis or with a family history of lung cancer).
Research questions
1. What is the risk-benefit balance of screening for LC by LDCT in high-risk individuals
compared to no screening or routine diagnosis?
2. What is the risk-benefit balance of adding molecular biomarkers to LDCT for LC
screening in high-risk individuals compared to LDCT screening?
3. What is the cost-effectiveness of the LC screening programme?
4. What would be the budgetary impact for the NHS of implementing an LC screening
programme in the at-risk population in Spain?
5. What is the risk-benefit balance of the organisational variants of systematic LC
screening with LDCT (intervals, invitation) in high-risk individuals??
6. What are the best strategies for informing the target population about an LC screening
programme to optimise informed participation?
7. What are the potential ethical implications of implementing the screening programme?
8. What are the potential environmental impacts of implementing the screening
programme?
Authors' results and conclusions:
Efficacy and safety results of LDCT screening
Bibliographic search: Nine randomised trials (DANTE, DLCST, ITALUNG, LSS, LUSI,
MILD, NELSON, NLST and UKLS) were included. All included active and former smokers,
and one (UKLS) also included people with other risk factors. In addition, 13 ongoing research studies were located. The last search was conducted on 31 December 2021.
Characteristics of the studies included: the UKLS study was a feasibility study which,
despite meeting the inclusion criteria, did not provide data on the variables of interest
for assessing clinical effectiveness and was therefore not further detailed. The remaining
8 studies (90,836 randomised patients) differed in terms of the screening strategies used.
Six studies (DANTE, DLCST, ITALUNG, LUSI, MILD and NELSON) compared LDCT
screening versus no screening and were conducted in Europe (Italy, Denmark, Germany,
Netherlands and Belgium) and the remaining 2 (LSS and NLST) compared BCTD
screening versus chest radiography (CXR) and were conducted in the USA.
The sample size ranged between 3000 and 4000, except in NELSON and NLST
which were close to 16 000 and 53 500, respectively. The duration of the screening phase
ranged from 1 to 6 years, and the expected follow-up period ranged from 5 to 10 years. The
screening interval was 1 year in all studies except MILD and NELSON. MILD screened
annually or every 2 years and in NELSON, the screening interval was extended from 1 to
2 years and then to 2.5 years.
The studies included men and women who smoked before starting the study (at least
20 or 30 packs/year) or who had quit smoking less than 10 years before (15 years in the
NLST). DANTE only included men and NELSON added women later in the study. The
age range for inclusion of participants was set between ≥49 years and 75 years; MILD was
the only study that did not set an upper age limit.
The screening participation rate (adherence to screening) ranged from 81% to 96%
in the intervention groups. Three studies (DLCST, LUSI, MILD) reported contamination
between 1% and 7% (non-screening comparator), NLST 4% (Rx comparator) and the
remaining studies provided no information.
Assessment of risk of bias and quality of evidence: the risk of bias assessed with the
Cochrane RoB tool was considered low in four studies (DLCST, ITALUNG, LUSI and
NELSON) and high in the remaining four, due to uncertainty about the generation of
the randomisation sequence (MILD, NLST), blinding (DANTE, MILD and NLST) or the
independence of the results (LSS). At the outcome level, the risk of bias for mortality
(overall and LC-specific), consequences of FPs and FNs and overdiagnosis was considered
low in three studies (DLCST, ITALUNG and NELSON). For adverse events, the only
study that provided information on adverse events was DANTE, which had a high risk of
bias. In studies where the risk of bias at the study level was considered high, no specific
assessment was made at the clinical outcome level. Depending on the outcome assessed,
the quality of the evidence ranged from low to high.
Clinical Results
Mortality
Regarding overall mortality, a meta-analysis was performed with data from 6 studies
comparing LDCT screening versus no screening: 3 at low risk of bias (DLCST, ITALUNG
and NELSON) and 3 at high risk (DANTE, MILD and LUSI). We used a random
effects model and considered the longest follow-up period ranging from 8 to 11 years. No
statistically significant differences in favour of screening were found both in the analysis
of studies with low risk of bias (IRR: 0.93, 95% CI: 0.69-1.26, p=0.43, I2: 51.4%) and in the
pooling of all studies (IRR: 0.95, 95% CI: 0.88-1.03, p=0.16, I2: 0%). To compare the effect
of LDCT screening versus CXR, we used data from 2 studies with high risk of bias (LSS
and NLST). Neither were significant differences observed between the two groups (IRR:
0.97, IC 95%: 0.92-1.02, p=0.17, I2
: 3.2%).
In summary, no significant reduction in overall mortality was observed after screening
for LC with LDCT, with little or no difference compared with no screening (high quality
evidence; GRADE profiles).
In terms of LC-mortality, a meta-analysis was performed with data from 6 studies
comparing LDCT screening versus no screening: 3 at low risk of bias (DLCST, ITALUNG
and NELSON) and 3 at high risk (DANTE, MILD and LUSI). We used a random effects
model and considered the longest follow-up period ranging from 8 to 11 years. The pooled
analysis of studies with low risk of bias showed no statistically significant difference
between the groups (IRR: 0.80; 95% CI: 0.60-1.06; p=0.07, I2: 0%), and the pooled analysis
of studies with low and high risk of bias showed a statistically significant effect in favour
of LDCT screening (IRR: 0.81; 95% CI: 0.72-0.91; p=0.004, I2: 0%). To compare the effect
of LDCT screening versus CXR, we used data from 2 studies with high risk of bias (LSS
and NLST) for the longest observation period (5 and 12 years from randomisation). The
sensitivity analysis showed similar results to the comparison of LDCT screening versus no
screening. (IRR: 0.89, IC 95%: 0.82-0.96, p=0.01, I2: 0%).
Sub-group analysis: subgroup analysis by tobacco exposure or screening strategy was
not possible. Stratification variables (potential effect modifiers) included: age of the CT
scanner and size of the centre (DANTE, DLCST, ITALUNG, MILD, LUSI, NELSON,
LSS and NLST), the presence of EPOC (DLSCT), gender (DANTE, LUSI, NELSON and
NLST), age (NELSON) and duration of the screening interval (MILD: annual or biennial
screening), and were analysed according to the type of comparator (unscreened or Rx).
The homogeneity or interaction test did not show statistical significance in any of the
subgroup analyses, either for overall mortality or for specific mortality.
In summary, the result of the meta-analysis indicated that LC screening with LDCT
could reduce LC-mortality compared to no screening (moderate quality evidence; GRADE
profiles).
Mortality benefit: The absolute effect estimate was a reduction of 5 deaths in overall
mortality per 1000 persons screened (95%CI -3 to 12) and 5 deaths per 1000 persons
(95%CI 3 to 8) in LC-mortality. In conclusion, LC screening with LDCT can reduce specific
mortality compared to no screening and makes little or no difference to overall mortality.
It is estimated that the number of people needed to be screened to prevent one death from
LC (PCN) would be 166 over 10 years.
Morbidity
In terms of adverse events, only one study with high risk of bias provided data on adverse events
(DANTE). It compared LDCT screening versus no screening and provided information on
adverse events after surgery and adverse events of severity ≥3 after surgery. Results were
presented for the longest observation period since randomisation (maximum 8 years). The
analysis showed a significant difference in the incidence of adverse events after surgery for
suspicious findings against LDCT screening (OR: 3.48, IC 95%: 1.41–8.62, p=0.004), and in
adverse events with a severity ≥3 (OR: 4.25, IC 95%: 0.92-19.69, p=0.04). It was estimated
that for every 81 people screened there would be one adverse event following surgery, and
for every 184 people screened there would be one adverse event of severity ≥3.
Conclusion regarding harm related to adverse events: LC screening with LDCT may
increase adverse events compared to non-screening (low quality evidence; GRADE
profiles).
Screening risks
The studies did not provide the necessary data to assess the consequences of false negative
results.
With regard to the consequences of false positive results, 3 studies with low risk of bias
(DLCST, ITALUNG and NELSON) and 3 studies with high risk of bias (DANTE, MILD
and LUSI) provided information on this issue. All of the studies compared LDCT screening
versus no screening and the need for an invasive procedure was only recorded in the
intervention group, except in DANTE which subjected all participants to a CXR and sputum
cytology 3 days before the start of the study, so it is unclear whether the difference between
the groups is due to LDCT screening alone. Given the heterogeneity between studies in the
presentation of results, for this outcome we did not provide an overall estimate, but rather a
range (minimum - maximum) of effect estimates from individual studies.
Between 0.1% and 1.5% of those invited for screening underwent an invasive
diagnostic test as a result of a FP result, and 0.1-1.3% of those with a benign finding
underwent surgery. Overall, between 0.1% and 1.5% of those screened experienced a
consequence of a FP result, i.e. between 1 and 15 out of every 1000 participants invited for
LC screening will undergo an invasive diagnostic evaluation or resection that would not
have been performed without screening and with a subsequent benign result.
Analysis by subgroups: a subgroup analysis of the impact of VET was not possible as
no usable data was available.
Conclusion on harm related to the consequences of the FP outcomes: LC screening
with LDCT causes harm arising from the consequences of FP outcomes compared to no
screening (high quality of evidence; GRADE profiles).
To assess overdiagnosis we used data from 6 studies comparing LDCT screening
versus no screening: 3 with low risk of bias (DLCST, ITALUNG and NELSON) and 3 with
high risk of bias (DANTE, LUSI, MILD). Data from 2 studies with high risk of bias were
used for comparison with Rx (LSS and NLST).
The risk of overdiagnosis was assessed in two different ways: 1) overdiagnosis related
to people invited for screening and 2) overdiagnosis related to people diagnosed with LC
during the screening phase. Due to the heterogeneity found between studies, it was not
possible to make an overall estimate for this outcome. In order to present the data in a
transparent way, the proportion of overdiagnoses for both reference values is presented as
a range (minimum-maximum) of the point estimates of the individual studies.
Overdiagnosis of those invited for screening: among the 6 studies comparing LDCT
screening with no screening, ITALUNG was the only study with fewer cases of LC
diagnosed in the intervention group than in the control group at overall follow-up. Therefore,
overdiagnosis could not be detected in this study. Neither was overdiagnosis found in the
biennial screening of the MILD study. The risk of overdiagnosis was highest in DANTE
and DLCST, at 2.2% and 2.1%, respectively. In LUSI, NELSON and the annual screening
of the MILD study, the risk of overdiagnosis was 0.9%, 0.6% and 1.4%, respectively. In the
2 studies comparing LDCT versus Rx screening, the risk of overdiagnosis was calculated
to be 1.2% and 0.1% in LSS and NLST.
Overdiagnosis of people diagnosed with LC during the screening phase: data were
available from 5 studies to estimate the risk of overdiagnosis in the presence of a lung
cancer diagnosis, including 4 studies comparing LDCT screening and no screening (DLCST,
ITALUNG, LUSI and NELSON). The result of the DLCST study was particularly striking,
as it estimated a risk of overdiagnosis of 63.2% (using the total number of lung cancer
diagnoses in the intervention group as the denominator). In LUSI and NELSON the risk
of overdiagnosis was 28.6% and 16.2%, respectively. In ITALUNG no overdiagnosis could
be detected. In NLST an overdiagnosis risk of 2.8% was calculated in comparison with
chest X-ray screening.
Conclusion on harm related to over-diagnosis: the diagnosis of LC requires histological
or cytological confirmation. It can be assumed that almost all cases with a diagnosis of
LC were also addressed. Any diagnostic procedure and treatment/therapy carries the risk
of side effects and complications. Studies estimated that out of 1000 people invited for
LC screening, between 0 and 22 were diagnosed with a LC that would not have caused
symptoms for the rest of their life and in the presence of a LC diagnosis the risk of
overdiagnosis ranges from 0% to 63%. No relationship was observed between the risk of
overdiagnosis and screening strategies.
In conclusion, LC screening with LDCT causes harm compared to non-screening in
terms of overdiagnosis, due to invasive diagnostic procedures and the resulting treatment,
including complications and associated side effects (high quality evidence; GRADE
profiles appendix 1.6).
Health-related quality of life
None of the included studies provided information on HRQoL, or were not useful for the
assessment of benefit.
Efficacy and safety results of biomarker screening
Bibliographic search: 1 randomised trial (ECLS) comparing LC screening with LDCT
followed by an antibody test (EarlyCDT Lung test) versus no screening was included. In
addition, 2 ongoing research studies were located. The last search was conducted on 31
December 2021.
Characteristics of the study included: the ECLS study was a feasibility trial that evaluated
the effectiveness of an antibody test in the early detection of LC and stratified people
according to their risk of developing LC. Patients assigned to the intervention group
were tested at baseline and, depending on the result, a different follow-up was carried out
[Test +: X-ray and CT scan without contrast; if the image shows evidence of LC, LDCT
is scheduled every 6 months; Test -: they are similar to controls (recommendations for
smoking cessation and clinical surveillance)].
The study was conducted in the UK and included men and women with a 2-year risk
of developing LC of at least 2%, defined as (former) smokers of at least 20 packs/year or
with a lesser smoking history, but with a family history of first-degree LC. The sample size
was 12,208 people, aged 50-75 years (mean age 60.5). The duration of the screening phase was 2 years and the interval between rounds was 6 months.
The screening participation rate (adherence to screening) was 90.2% in the intervention
group (test and LDCT) and 89.6% in the control group (no screening).
Assessment of risk of bias and quality of evidence: the risk of bias assessed with the
Cochrane RoB tool was considered high due to uncertainty in blinding and concealment
of group allocation. At the outcome level, the risk of bias assessment for both overall and
specific mortality was considered high.
Clinical results
The ECLS study only provided information relevant to the outcome of overall and
LC-mortality.
Mortality
Analysis of the data showed no significant difference in the reduction of overall mortality
between the intervention and control groups (HR: 0.8, 95% CI: 0.61-1.07, p=0.11), suggesting that LC screening with antibody test followed by an imaging test may have little or no difference in overall mortality compared to no screening (low quality evidence; GRADE profiles).
There was also no significant difference in the reduction of LC-mortality between the
intervention and control groups (HR: 0.71, 95% CI: 0.38-1.34, p=0.28), suggesting that LC
screening with antibody test followed by an imaging test may have little or no difference
in LC-mortality compared to non-screening (low quality evidence; GRADE profiles
appendix 1.6).
Mortality benefit: the absolute effect estimate was 2 fewer deaths per 1000 persons
(95% CI -1 to 7) in overall mortality and 1 fewer death per 1000 persons (95% CI -1 to
2) in LC-mortality over 2 years. In conclusion, it is considered that LC screening with an
antibody test followed by an imaging test may have little or no difference in overall or
LC-mortality compared to no screening.
Benefits and risks of screening with biomarkers and LDCT
In the absence of quality evidence, it is not possible to establish whether the use of
biomarkers as an adjunct to LDCT within the LC screening process in at-risk groups
would result in extra benefit or less harm compared to LC screening with LDCT alone or
compared to routine clinical practice.
Cost-effectiveness analysis
The economic evaluation follows the recommendations of the Spanish guideline for
economic evaluations. A mathematical model was built following internationally accepted
methods and techniques described in the literature and inspired by a previously published
economic evaluation. Methods and results are presented following the CHEERS standards.
The target population for LC screening is made up of adults without LC (confirmed
or suspected) at high risk of developing LC (ICD-10: C34), that is, smokers and ex-smokers.
Due to the characteristics of the existing clinical studies on the effectiveness of LC
screening programs, the economic evaluation does not extend to the population at risk of
LC due to other factors. The starting age of the cohort is 50 years and its evolution with or
without screening throughout the live of the subjects was analysed. The cohort is made up
of 100 000 subjects with a proportion of men and women corresponding to the prevalence
of smokers of more than 20 cigarettes/day among people aged 45 or over by sex in Spain.
A complete economic evaluation was carried out in which the costs and effects of the
LC screening program using LDCT and the comparator consisting of not screening were
compared. Two possible strategies are evaluated depending on the frequency of screening,
that is, annual screening program and biennial screening program. Sub-strategies according
to several age subgroups are also presented. Evaluating screening using biomarkers is
ruled out as there is currently insufficient scientific evidence on its effectiveness and safety.
The analysis is carried out for the whole life of the patients and, therefore, costs and
future effects are discounted at 3%.
The main outcome measure is quality-adjusted life years (QALYs). Other outcomes
of interest such as life years gained (LYG), number of cancers diagnosed, deaths from LC,
and incidental findings were also reported.
In the base case, the perspective of the analysis was that of the NHS, so only direct
healthcare costs were included: costs directly related to the screening program (identification of the population that meets the inclusion criteria, invitation to perform the test, LDCT and its interpretation), diagnosis (tests and consultation with professionals), treatment and monitoring of patients. Costs were expressed in 2021 Euros. The social perspective was included in the sensitivity analysis for those patients younger than 65 years at the time of diagnosis.
The alternatives in comparison were ordered from lowest to highest cost. The costs
and effects in terms of QALYs of the alternatives were compared two by two. The most
expensive and least effective alternatives were discarded from the analysis by strict
dominance. Subsequently, other alternatives were discarded due to extended dominance
by having a higher ICER for a lower cost. The ratios of the non-dominated alternatives
are presented. The ICER is interpreted considering the most recently estimated costeffectiveness threshold for Spain, €25 000/QALY.
The model and the analyzes were implemented with the Microsoft Excel 2013
computer program and the Visual Basic programming language
The Markov model represents the course of the disease with quarterly cycles and
covers the entire life of the patient (up to 100 years of age). The model consists of two parts:
the first part represents the natural history of the disease, where the cohort starts from the
“without LC” state. People can develop or evolve through different states before receiving
a diagnosis, which can be made by screening or for other reasons. After diagnosis, the
patient will be treated depending on the stage of the disease and will enter post-treatment
states.
The identification of the values of the model’s parameters was performed throughout
the review of the literature developed in this report, quasi-systematic reviews of the
bibliography and consults to statistical sources and experts. Certain assumptions discussed
and approved by the authors of the report, and validated by the panel of experts/external
reviewers, were also assumed in cases of insufficient scientific evidence.
The main parameters related to the performance of LDCT are sensitivity and
specificity. The values used in the base case are estimated from the 2-year follow-up of
the NELSON trial and the study by Hofer et al. Regarding costs, the costs estimated by
Ibarrondo et al. from a cohort of patients in the Basque Country was used in the base case.
The utilities are based on assumptions and on the values used by Hofer et al., in order to
be comparable with this study and with the study by Gómez-Carballo et al.
One-way and multi-way sensitivity analyses and a probabilistic sensitivity analysis
were performed. Sensitivity analyses are carried out on the most efficient and least
expensive alternative among the efficient strategies according to the base case analysis.
Budget impact analysis
In addition to the economic evaluation, a budget impact analysis was performed to estimate
the cost of implementing a LC screening program with LDCT.
The NHS perspective and a time horizon of three years were used, so only the
direct healthcare costs (in 2021 Euros) of the first three years of implementation of the
screening were considered. Discount rates were not applied to costs, as recommended by
the methodological guidelines.
The gross budget impact is presented for the current scenario in Spain (without
screening) and for several hypothetical scenarios (with screening, for two frequencies
(annual and biennial) and different age subgroups). The net budget impact for the Spanish
population (and by autonomous communities) aged 50-80 years was estimated from the
difference between the hypothetical scenario and the current one.
The costs included in the current scenario (without screening) were diagnostic and
treatment and follow-up costs. The different hypothetical scenarios (with screening)
included, in addition to these, the costs of screening, the investment in CT equipment, and
the cost related to the interpretation of LDCTs and preparation of reports by radiologists.
The population that generates such costs was estimated. In the case of no screening
program (current scenario), the incidence of new cases of LC in Spain in 2021 was used
and the cases of LC diagnosed by stages (I, II, IIIA, IIIB and IV) and the population that
receives each treatment were calculated, as well as the number of patients attending to a
primary care consultation or to the emergency room due to the presence of symptoms.
For the hypothetical scenarios (with screening), the population likely to be screened
that is invited and agrees to participate in the screening was first estimated. The participation
rates of the national colorectal cancer screening program were used as a reference, as LC
screening has not yet been implemented in Spain and no information on this is available.
Subsequently, the number of LDCT performed annually, the positive results (true positives
or false positives) for which other confirmatory tests should be performed, and the cases
of LC diagnosed by stage (whether through the screening or by other reasons) were
calculated, as well as the distribution among treatments.
In addition, the number of necessary CT equipment to be acquired and the time
invested in the LDCT interpretation reports by radiologists were estimated.
A sensitivity analysis was carried out for the most efficient strategy from the point of
view of cost-effectiveness, in which different key parameters were varied.
Results
Systematic review of the literatura on LC screening with LDCT
264 references were identified in all the databases (without duplicates) and 42 articles
were preselected. After the full text analysis, 15 studies (17 articles) were finally selected.
In addition, 19 studies (21 articles) from the SR by Snowsill et al. were considered, and 4
studies were identified through the alert system, as well as 4 studies manually. Finally, 46
articles corresponding to 42 studies are included in the review.
Most of the studies conclude that screening with LDCT is more expensive and more
effective than no screening and that, under certain conditions, it could be cost-effective.
Due to differences in smoking exposure, cost structure and cancer treatment, among other
elements, efficient scenarios may vary depending on the country. Consequently, few studies
point to the generalization of the results to other contexts and groups of patients. Two
studies carried out for Spain were published in 2021.
Diaz et al. developed a Markov-based microsimulation model to describe the natural
history of LC and evaluate various prevention strategies in men aged 35 years and older
(smoking cessation interventions), screening with LDCT from 50 years of age, or a
combination of both. in addition to not intervening. Compared with not intervening, the
only cost-effective “screening alone” strategy (<€25 000/QALY) would consist of a single
LDCT performed at age 55 (€17 352/QALY). Strategies consisting of two or three LDCT
from the age of 55 have ratios slightly above €25 000/QALY. However, there are several
combinations of smoking cessation and screening alternatives that could be cost-effective,
with ICERs below the threshold of €25 000 or even €21 000/QALY, compared to other
combinations, with not screening or with implementing non-combined strategies
Gomez-Carballo et al. designed a Markov model that compared screening with LDCT
versus not screening for a 19-year time horizon in a population aged 50 to 75, who were
smokers or ex-smokers, from the NHS perspective. The ICERs obtained both in the base
case (€2345/QALY) and in the sensitivity analyses are below any threshold reported in
the literature. According to the probabilistic sensitivity analysis, the screening would be
cost-effective with a probability of 79% for a threshold of €21 000/QALY.
Systematic review of the literatura on LC screening using biomarkers
32 references were identified in all the databases (without duplicates) and a single article
was selected. Sullivan et al. published a randomized controlled trial with 12 208 participants
at risk of developing LC, aged 50-75 years, in the United Kingdom. Participants underwent
either an EarlyCDT-Lung biomarker test, followed by a computed tomography (if the
first is positive), or no screening and treatment only of patients who present symptoms.
The incremental cost per early case (stage I/II) detected after 2 years was £116 000, with
screening using the EarlyCDT-Lung test being more expensive and more effective than
no screening. The major limitation of this economic evaluation is the short-term time
horizon. In addition, the authors themselves acknowledge the need to carry out a model
(not published) given the imbalance between the arms.
Cost-effectiveness analysis (de novo model)
Compared with the current ‘no screening’ alternative, all screening strategies are more
costly and more effective in terms of LYG, QALY, diagnosed lung cancers, and avoided
deaths from LC. The differences in LYG and QALY between alternatives are small.
Screening results in incidental findings of other health problems. The biennial screening is
less expensive and less effective than the annual screening by age ranges.
The greatest cost from the NHS perspective is the cost of screening. This cost is lower
in the biennial screening than in the annual one. From a social perspective, the greatest
cost corresponds to the loss of productivity of people with LC. Several alternatives are
ruled out either by strict domination or by extended domination. The estimated ICERs
are above the cost-effectiveness threshold of €25 000 per QALY. The smallest ratios are
the result of comparing ‘Biennial screening, from 50 to 70 years’ with ‘Do not screen’ (€34
235/QALY) and ‘Biennial screening, from 50 to 80 years’ versus ‘Biennial screening, from
50 to 70 years’ (€30 004/QALY).
Some variables affect the results according to the deterministic sensitivity analysis. The
discount rate stands out, as if it were lower than the currently recommended rate, the ICER
would be close to the cost-effectiveness threshold of €25 000/QALY. The cost-effectiveness
also depends on the protocol and the established positivity threshold to determine that
the LDCT shows a potential LC. The protocol followed in the NELSON trial provides the
lowest ICERs when comparing screening with not screening in subjects between 50 and 70
years. The results of the probabilistic sensitivity analysis for biennial screening in subjects
aged 50 to 70 years (more efficient strategy), compared to no screening, show that less
than 30% of the simulations are below the cost-effectiveness threshold, showing a great
uncertainty about the cost-effectiveness of this screening strategy.
Budget impact analysis
The gross budget impact (over 3 years) for the current scenario (without screening) was
estimated at €788.53 million. Annual screening with an age range of 50-80 years generates
the greatest budget impact (€4584.31 million) in gross terms, while the most efficient
screening strategy (biennial screening between 50-70 years), from the point of view of costeffectiveness analysis, generates a gross budget impact of €3030.80 million, which means a net budget impact of €2242.27 million with respect to the current scenario (assuming the acquisition of 100% of the CT equipment necessary for the potential demand of the program). By autonomous community, the net budget impact would be greater in those regions where the population likely to be screened is greater (Andalusia, Catalonia and Madrid), given that we assume equal prices for the entire national territory.
The sensitivity analysis shows that, for the biennial screening strategy between 50
and 70 years, the most relevant variations are observed when the response rate changes to
20% (net budget impact of €770.46 million) and when the participation rate increases to
80% (net impact of €3741.83 million). In addition, the net budget impact can be reduced
to €2136.68 million and €2030.44 million if the percentage of CT equipment purchased is
reduced to 50% or 0%, respectively.
Organisational aspects
Results
Information from the 8 RCTs included in the efficacy and safety domain was used to
perform subgroup analyses of the different screening modalities. As additional evidence, 6
pilot studies were included.
Characteristics of the studies included
The characteristics of the included RCTs, as well as the assessment of risk of bias and
quality of evidence are detailed in the efficacy and safety section. The GRAE evidence
profiles can be consulted in appendix 3.1.
The pilot studies focused on evaluating the implementation, feasibility and roll-out of an
LDCT-based LC programme on an annual basis. Most were conducted in the Americas (Canada
and Ontario) and Asia (Korea and China) and only one was conducted in Europe (France). Most
of them were regional/local experiences and only one was at national level (Korea).
Three studies recruited through primary care physicians, pulmonologists or electronic
registration (Canada, Oregon and France), with one adding radio, print, TV, poster and leaflet
information campaigns (France). One study (Henan, China) recruited through phone calls and
face-to-face interviews, combined with advertisements on social media and other community
media; one through advertisements in newspapers, public transport and public office notice
boards plus a questionnaire (Korea); and the last one by invitation (Yunan, China).
For candidate selection, three studies used the NLST age and smoking criteria (France,
Korea and Oregon) and two used risk prediction models: the PLCOm2012 (6-year LC
risk ≥2%) (Ontario) and the Harvard risk index (risk >1.5) (China). The age criterion
was set at 55-74 years, except in the Asian studies which extended the age limit to include
participants aged 40-74.
Smoking cessation interventions ranged from referral of smokers to an in-hospital
smoking cessation programme or alternatively a telephone cessation programme (Ontario),
smoking cessation counselling (Korea) or simple smoking cessation information (Oregon,
France).
Three pilot studies used the Lung-RADS system as an algorithm for the classification
and management of nodules detected on LDCT (Korea, Ontario, Oregon), two of them
using a computerised diagnostic support system for reading the images (Korea and
Ontario). One study was based on nodule volume (France) and the other on size and
density criteria (China).
The information provided on the resources employed was limited. They point to
minimal needs for specialised staff, multidisciplinary chest disease teams (pulmonology,
thoracic surgery, medical oncology, radiation oncology, thoracic pathology and thoracic
radiology), the use of multi-slice CT and supporting software/browsers for candidate
selection and/or image reading, and education and training needs, among others.
Clinical outcomes of included studies
In the efficacy and safety section, subgroup analyses were performed on participant and
organisational characteristics. Subgroup analyses on different screening strategies, such
as number of screening rounds or organisational differences, could not be performed as
studies could not be assigned to the appropriate categories or there were no significant
differences between studies with respect to the categories.
Clinical results of the pilot studies
In terms of eligibility, in 4 studies the percentage of the population assessed as high
risk and eligible for screening (eligible population) was greater than 65% of the enrolled
population (range: 66.5-98.4%) (France, China, Ontario, Oregon). In one it was close to
20% (China) and in another 8% (Korea). The participation rate (proportion of the eligible
population agreeing to participate and being screened) ranged from 40% to 100%.
The 3 pilot studies using the Lung-RADS algorithm (Korea, Ontario, Oregon)
reported that 84% of the screened population had a negative initial LDCT result, 6.2%
a positive result and 9.7% indeterminate. This means that in 16% of the population a
finding requiring diagnostic evaluation and/or follow-up was found. Among patients with
a positive initial LDCT result, a diagnosis of LC was confirmed in 17.3%. The overall
detection rate was 1.5% and depending on stage, 66-71% of detected LCs were in early
stage I-II and between 17-33% in stage III-IV. The rate of PF ranged from 14-17% and one
study reported 2 interval LCs. Between 0.1-0.6% of the screened population suffered an
adverse event related to the diagnostic procedures, with surgical mortality being 2%.
The 2 pilot studies using a nodule volume-based screening protocol (France, China)
indicated that 86% of the screened population had a negative initial LDCT result and
14% identified findings requiring diagnostic evaluation and/or follow-up (8.4% positive
and 5.5% indeterminate). The overall rate of LC detection was 1.1%, with 75-77% being
early stage I-II. The percentage of PF ranged from 3-9%.
Risk-benefit balance of different screening strategies
In relation to organisational variations in screening, RCTs comparing LDCT versus no
screening for LC were either largely comparable (inclusion criteria with respect to smoking),
or there was no statistically significant interaction (screening interval, size of screening
centre, age of CT team, gender or age of participants), or they could not be assigned to
appropriate categories (such as screening invitation processes or research strategy).
In conclusion, the available evidence is not sufficient to conclude whether any one
specific strategy in LC screening is favourable compared to other screening strategies (very
low-quality evidence; GRADE profiles appendix 3.1).
Social aspects
Results
19 studies were identified as relevant to the research question on the effectiveness of strategies
to inform people likely to be invited to a LC screening program. In the initial search of
the EUnetHTA report, 15 studies for this research were identified: 5 randomized trials, 3
controlled observational studies and 7 uncontrolled pre-post intervention (PPI) studies.
The last search in the EUnetHTA report was performed on July 24, 2020. Four additional
studies were identified in the search update performed for the adaptation of this report: one
randomized trial and three uncontrolled PPI studies.
Characteristics of the included studies
Two types of studies related to the research question were found: studies on informational
interventions or invitation to LC screening and, studies on strategies or tools for shared
decision-making on LC screening. Two RCTs and one non-randomized comparative study
compared different strategies/materials to inform or invite LC screening. The remaining 16
studies (3 RCTs, 4 observational studies, and 9 uncontrolled PPI studies) evaluated the effect
of different strategies or shared decision-making tools for people eligible for LC screening.
Summary of results relevant to the evaluation
Thirteen studies provided data on change in participants’ knowledge about early detection
of LC. Participant empowerment was assessed in nine studies, all focusing on the decision
conflict experienced by participants, while seven studies also investigated whether
participants were prepared to make an informed decision about early detection of LC.
Participant satisfaction with the information was assessed in three studies. Participation
rate was included as an outcome in eight studies.
Assessment of risk of bias and quality of evidence
The risk of bias at the study level was rated as low for two RCTs, and as high for four
RCTs. In these trials it was unclear whether the randomization sequence was adequately
generated and/or whether concealment was adequate. The risk of bias on improved
participant knowledge, empowerment and satisfaction, informed decision-making, and
screening participation rate was rated as low in RCTs.
At the outcome level, the risk of bias for observational studies and for single-arm
PPI (before-after) studies was rated as low for six studies on all reported outcomes
(improved knowledge, informed decision-making, empowerment of participants and rate
of participation in the projection). For the remaining seven studies the risk of bias, at the
outcome level, was high due to missing data, the presence of selective outcome reporting,
or the presence of bias in the selection of study participants. The GRADE evidence profiles
can be consulted in appendix 4.5.
Results on clinical efficacy and safety
Knowledge improvement
We included 14 studies (5 RCTs, 1 RCTs, 8 one arm). One RCT and four observational
studies showed significant differences in the improvement of knowledge with the use of
decisions aids as a information tool on screening.
This conclusion is based on moderate-quality evidence, which was downgraded by one
level due to the imprecision of the results of two studies with low numbers of participants.
Informed decision making
Seven studies (2 RCTs, 2 RCTs, and 3 pre-post) investigated the effect of different
informational materials for people eligible for LC screening. The use of decision aids to
promote participation in LC screening programs probably strengthens informed decision
making among participants compared to people who receive standard informational
materials.
This conclusion is based on moderate-quality evidence, as this should be downgraded
due to the imprecision of the results of two studies with low numbers of participants.
Participant empowerment
Results on participant empowerment come from four RCTs, one non-randomized
comparative study, and four one-arm “before-after” studies. One RCT showed a significant
decrease in decisional conflict with the use of decisional aids to inform LC screening
compared to using standard informational materials.
The conclusion is based on moderate-quality evidence, which was downgraded by one
level due to the imprecision of the results of a small RCT.
Participant satisfaction
The results on participant satisfaction come from three studies: one RCT, that investigated
the effect of different invitation strategies for LC screening, and two observational studies,
that investigated different types of informational materials for people eligible for screening.
Targeted invitation strategies for LC screening may result in little or no difference to
participants’ satisfaction with their decision to participate in screening compared to using
standard informational materials.
This conclusion is based on low-quality evidence, as the level of evidence from the
only available study was downgraded by 2 levels due to imprecision, as it was an RCT with
a very small sample size.
Screening participation rate
Information on the results of the screening participation rate comes from an observational
study, that investigated the effect of information leaflets on LC screening, two RCTs, that
investigated different invitation strategies for screening, and three studies, that evaluated
the effect of different informational materials for people eligible for screening. A targeted
invitation strategy to promote participation in PC screening programs may have little or
no influence on the participation rate, compared to the use of standard informational
materials.
This conclusion is based on low-quality evidence, which was downgraded two levels
due to the imprecision of data from a single, low-sample RCT.
Conclusion
Compared with the use of standard informational materials, the use of decision aids
increased the participants’ knowledge about LC screening and the decision certainty
regarding participation. The type of decision aid or the mode of implementation (face-toface or by telephone) does not seem to have a significant effect on the results.
The different screening invitation strategies do not affect the knowledge of LC
screening or the participant’s satisfaction. The number of RCTs and observational studies
evaluating the effectiveness of different information or invitation strategies on participation
in LC screening programs is limited.
Ethical aspects
Results
Of the 17 criteria analysed, 8 were considered to be of high or decisive importance: 2 related to the principle of autonomy, 1 to beneficence/non-maleficence, 2 to justice and equity and the remaining 3 to general aspects related to technology, effectiveness and accuracy.
The search for additional information retrieved 464 references from which 21 studies
were selected.
Among the ethical issues related to LC screening, most studies noted: 1) inequalities in
both access and eligibility due to gender, in racial and ethnic minorities (mostly black and
African-American population) and in certain vulnerable groups (low socio-economic and
educational status, patients with mental illness, etc.); and 2) issues related to the imbalance
in the presentation of information on the benefits and risks of screening for informed
decision making.
Ethical considerations, such as balancing the benefits and risks of the entire programme,
equity of access, respect for autonomy, and the right to privacy and confidentiality are of
crucial importance.
Environmental aspects
Results
No study referring to the environmental impact of the evaluated technology was identified
in the literature search.
One of the two companies consulted (Siemens) provided information through the
environmental declaration of its product. This statement shows that the design is respectful of
the environment in all phases of the product life cycle (including supply material, production
and delivery, use, maintenance, and end of life), meeting the requirements of international
standards for the design of medical equipment. The company develops the Siemens
Healthineers Environmental Management System, with the aim of ensuring environmental
protection, health management and safety, wich is implemented worldwide and is certified
under ISO 14001 and OHSAS 18001 standards
Authors' methods:
Efficacy and safety of LDCT screening
Methodology
For research questions 1 and 2, the results of the European collaborative network on health
technology assessment (EUnetHTA) report “Lung cancer screening in high risk groups”
were used as a starting point. Taking into account the time limit, specific literature searches
in the general and specific databases were updated (last update December 2021).
The selection of studies was peer-reviewed, independent and blinded, according to the
inclusion and exclusion criteria established in the protocol. Data extraction was performed
systematically in standardised tables. The validity of the studies and level of evidence was
assessed using specific scales or tools according to the study design. The overall quality of
the evidence (at the level of clinical outcome) was graded using the GRADE (Grading of
Recommendations Assessment, Development and Evaluation) methodology.
In addition to comparing the results of individual studies, we performed meta-analysis
and sensitivity analysis and examined effect modifiers using subgroup analyses, where
methodologically possible
Economic analysis
Methodology
Systematic review of cost-effectiveness
A review of the scientific literature on the cost-effectiveness of LC screening programs in
high-risk populations was carried out. A preliminary search identified 1 SR of economic
evaluations of LC screening using LDCT that was chosen for adaptation and update. No
SR of economic evaluations of biomarkers was identified, so a new review was carried out
without time restriction.
Original works evaluating the cost-effectiveness of LC screening were selected
according to the established inclusion and exclusion criteria. Completed economic
evaluation, where LC screening programs using LDCT or biomarkers (in addition to
LDCT) are analised, in adults (≥18 years) without LC (diagnosed or suspected) (CIE-10:
C34) with high risk of LC, weres selected.
The electronic databases consulted were MEDLINE, Embase, and Web of Science
(WOS). Two independent searches were performed, one for LDCT screening and one for
biomarker screening. The strategy carried out in the EUnetHTA report was replicated. This
is similar to the search applied by Snowsill et al., but the SIGN (Scottish Intercollegiate
Guidelines Network) economic filter was added. The search for economic evaluations of
biomarkers did not have any time restriction. No language of publication or age restrictions
were applied.
Two reviewers independently selected the studies by reading the titles and abstracts
first, and then the full texts. When there were doubts or discrepancies between reviewers,
these were resolved by discussion or with the help of a third reviewer.
The assessment of the methodological quality of the economic evaluations was carried
out with the CHEC instrument. Quality was assessed by one reviewer and checked by a
second reviewer. The economic evaluations carried out for Spain were also evaluated using
the guideline for the verification of economic evaluation requirements in Spain.
Data were extracted by one reviewer and checked by a second reviewer. The extracted
data were those related to the identification of the article (authors, date of publication,
country where the study was carried out, funding, etc.), to the design and methodology
(objective, design and duration of the study, characteristics of the patients, description of
the interventions, outcome measures, type of analysis, perspective, etc.), and to the results
(effectiveness results, costs, incremental cost-effectiveness ratio (ICER), etc.).
A narrative synthesis of results was carried out. The SR results of screening using
LDCT or biomarkers are presented separately. Foreign studies are also presented
separately from those carried out in Spain.
Organisational aspects
Methodology
Updating and adaptation of the EUnetHTA collaborative report. In addition, partial
quasi-systematic literature searches were carried out to locate informative material and
other relevant documentation for contextualisation.
As additional evidence, information was included from pilot experiences of
prospective LC screening programmes conducted in the public sector, with the aim of
locating information of interest in terms of screening protocols used, inclusion of smoking
cessation interventions, need for human and material resources, and the need for a
comprehensive and comprehensive approach to smoking cessation, among others, as well
as the performance of such strategies in a real screening programme setting and not only
in controlled and ideal settings as in the case of RCTs
Social aspects
Methodology
Independent systematic literature searches performed on the EUnetHTA collaborative
report “Lung cancer screening in high risk groups”, in the MEDLINE, Embase, Cochrane
Central Register of Controlled Trials, and Cochrane Database for SR, were updated. The
last search was performed on September, 2021
Ethical aspects
Methodology
The list of specific ethical aspects of screening technologies in the methodological manual
of the European Collaboration on Health Technology Assessment was taken as a starting
point (EUnetHTA Core Model®), which includes 17 criteria organised into 8 categories.
The relevance of each criterion was assessed by a panel of 5 bioethicists who independently and blindly weighted its importance on a numerical scale (1: least important and 9: most important or decisive). All criteria with an overall median score ≥7 (high or decisive importance) were selected.
A Pubmed search was conducted using search filters and terms specific to the field
of ethics to identify relevant information that had not been retrieved in the searches
conducted in the other domains of the report. The studies considered to be of interest were
those that could be applicable to our health system, discarding those that were carried out
in populations with important cultural, social, etc. differences with respect to our context.
A principle-based descriptive analysis was carried out, taking into account
the evidence considered in the previous sections of this report and the complementary
information identified in the targeted search.
Environmental aspects
Methodology
In order to answer the research question referring to the possible environmental
repercussions of the implementation of the LC screening program, a review of the
literature was carried out to identify and evaluate the articles that could provide relevant
information on the analysis of carbon, hydrological and ecological footprints originated
by this technology during its complete life cycle. A literature search was carried out in the
MEDLINE database applying a specific search strategy. In addition, information on the
environmental impact of the products was requested to manufacturers or distributors of
the evaluated technologies.
Details
Project Status:
Completed
Year Published:
2023
URL for published report:
https://sescs.es/wp-content/uploads/2015/09/66_SESCS_-AvaliaT_Cribado_Cancer_Pulmon_NIPO_cuerpo.pdf?x68814
Requestor:
Ministry of Health
English language abstract:
An English language summary is available
Publication Type:
Full HTA
Country:
Spain
MeSH Terms
- Lung Neoplasms
- Mass Screening
- Early Detection of Cancer
- Program Evaluation
Keywords
- .Cribado
- Neoplasias Pulmonares
- Planes y Programas de Salud
Contact
Organisation Name:
Canary Health Service
Contact Address:
Dirección del Servicio. Servicio Canario de la Salud, Camino Candelaria 44, 1ª planta, 38109 El Rosario, Santa Cruz de Tenerife
Contact Name:
sescs@sescs.es
Contact Email:
sescs@sescs.es
This is a bibliographic record of a published health technology assessment from a member of INAHTA or other HTA producer. No evaluation of the quality of this assessment has been made for the HTA database.